- Email: [email protected]
Development of a nucleotide sugar purification method using a mixed mode column & mass spectrometry detection
Journal of Pharmaceutical and Biomedical Analysis 115 (2015) 402–409
Contents lists available at ScienceDirect
Journal of Pharmaceutical and Biomedical Analysis journal homepage: www.elsevier.com/locate/jpba
Development of a nucleotide sugar purification method using a mixed mode column & mass spectrometry detection Heather Eastwood a,∗ , Fang Xia b , Mei-Chu Lo a , Jing Zhou a , John B. Jordan a , John McCarter a , Wesley W. Barnhart a , Kyung-Hyun Gahm a a b
Department of Molecular Structure & Characterization, Amgen, Inc., Thousand Oaks, CA 91320, United States ASK-Gene Pharma, Inc., Camarillo, CA 93012, United States
a r t i c l e
i n f o
Article history: Received 30 April 2015 Received in revised form 31 July 2015 Accepted 1 August 2015 Available online 4 August 2015 Keywords: Nucleotides Nucleotide-sugar Sugar phosphate Mixed-mode stationary phase High performance liquid chromatography Mass spectrometry
a b s t r a c t Analysis of nucleotide sugars, nucleoside di- and triphosphates and sugar-phosphates is an essential step in the process of understanding enzymatic pathways. A facile and rapid separation method was developed to analyze these compounds present in an enzymatic reaction mixture utilized to produce nucleotide sugars. The Primesep SB column explored in this study utilizes hydrophobic interactions as well as electrostatic interactions with the phosphoric portion of the nucleotide sugars. Ammonium formate buffer was selected due to its compatibility with mass spectrometry. Negative ion mode mass spectrometry was adopted for detection of the sugar phosphate (fucose-1-phophate), as the compound is not amenable to UV detection. Various mobile phase conditions such as pH, buffer concentration and organic modifier were explored. The semi-preparative separation method was developed to prepare 30 mg of the nucleotide sugar. 19 F NMR was utilized to determine purity of the purified fluorinated nucleotide sugar. The collected nucleotide sugar was found to be 99% pure. Published by Elsevier B.V.
1. Introduction Nucleotide sugars contain sugar or sugar derivatives connected through the glycosidic hydroxyl group to the 2nd phosphate of a nucleoside 5 -pyrophosphate. The biosynthesis of protein and lipidlinked oligosaccharides utilizes nucleotide sugars for glycosylation [1,2]. Therefore, analysis of nucleotide sugars is an important step in understanding the glycosylation mechanism in cells. Inhibition of fucosylation, the process of adding fucose sugar units to a molecule, is of interest in the field of antibody-dependent cellular cytotoxicity (ADCC). The antibody binds to the target antigen and engages CD16 on the effector cell. CD16 is a low affinity IgG receptor Fc III (Fc␥RIII) found on the surface of natural killer cells, such as monocytes and macrophages. Binding of the antibody causes the release of pore-forming proteins and proteases, which lyse the target cell. Absence of fucose on the antibody improves binding to the effector cell which, in turn, enhances cell lysis [3]. Fucosylation may be reduced through the use of a small molecule inhibitor of the fucosylation pathway.
∗ Corresponding author at: Amgen, Inc., One Amgen Center Drive, 29-2-C, Thousand Oaks, CA 91320, United States. Fax: +1 805 480 3016. E-mail address: [email protected] (H. Eastwood). http://dx.doi.org/10.1016/j.jpba.2015.08.001 0731-7085/Published by Elsevier B.V.
Various chromatographic methods have been reported for the separation of nucleotides, nucleotide-sugars and sugarphosphates. These include reversed-phase ion-pairing (RP-IP), porous graphite carbon (PGC), hydrophilic interaction liquid chromatography (HILIC) and mixed mode chromatography [1,2,4–10]. Guanosine diphosphate (GDP) sugars and their metabolites were separated by ion-pairing chromatography utilizing ODS-columns (C-18) and ion-pairing agents such as tributylamine with acetic acid [4]. Tetrabutylammonium hydrogen sulfate has been shown to be successful at separating these analytes as well. However, the ionpairing agent was not compatible with the mass spectrometer [5]. In addition, to these studies, separation of various nucleotide sugars, such as GDP-mannose, GDP-fucose and GDP-rhamnose were successfully performed using a C-18 reversed-phase column and triethylammonium acetate buffer as an ion-pairing reagent for increasing retention [1]. Unfortunately, this buffer has low volatility and would be difficult to remove from collected purified samples. Also, nucleoside triphosphates were not explored in the study. Columns with porous graphite carbon (PGC) were shown to separate nucleotides according to the base and the sugar moiety rather than the ionic phosphate groups, unlike ion-pairing reversed-phase chromatography. However, PGC columns required complex cleaning methods, which included boiling in TFA and rinsing with sulfite,
H. Eastwood et al. / Journal of Pharmaceutical and Biomedical Analysis 115 (2015) 402–409
403
Fig. 1. (a) Chemoenzymatic synthesis of GDP-TFMF from TFMF-fucose and the intermediate TFMF-1-phosphate. (b) Structures of ATP, GDP, ATP, ADP, GDP-fucose, fucose-1-P and fucose. (c) The schematic depiction of the Primesep SB stationary phase with strong embedded basic ion-pairing groups bound to silica.
as well as an acid/base treatment to achieve separation [2]. Changes in the column could lead to long equilibration time and reproducibility issues. One possible explanation for this was a slow change on the PGC surface from oxidation [6]. HILIC has been useful at separating hydrophilic analytes and has been successful for the separation of nucleotides, sugar phosphates and nucleotide sugars [7–9]. Recently, separation of nucleotides was achieved with a cyclofructan based column using ammonium acetate buffer in both gradient and isocratic methods [7]. Unlike
other HILIC stationary phases, cyclofructan offers dual retention mechanisms. Traditional hydrogen bonding/dipolar interactions can be supplemented by dynamic ion interaction effects for anionic analytes. This is due to cyclofructan binding certain buffer cations to form a positively charged cavity in the stationary phase which will bind the negatively charged phosphates of the nucleotides. While this mode is useful for nucleotide separations, it did not address separating nucleotide sugars or sugar phosphates that could potentially be present.
404
H. Eastwood et al. / Journal of Pharmaceutical and Biomedical Analysis 115 (2015) 402–409
Mixed mode columns are phases that provide at least two modes of interaction in one stationary phase, such as ion exchange with C18 properties [10]. This is beneficial due to the multiple controllable interactions on the column, which allow for better control of retention for various analytes [11]. Mixed mode has been mainly used to separate biologic molecules, nucleic acids, peptides and proteins [12–14]. Primesep has several commercially available mixed mode which offer selectivity in the separation of a broad array of chemical compounds and in a multitude of applications [12,15–19]. Among the several Primesep columns available, the Primesep SB column contains strong basic ion-pairing groups, improves retention of acidic compounds by anion-exchange mechanism, separates bases by ion-exclusion mechanism together with hydrophobicity and retains neutral compounds via a reversed-phase mechanism [20]. It has been shown to resolve analytes that are very polar and have highly ionic characteristics, such as etidronate disodium [21]. The column was used for stability-indicating method development and validation for analysis of etidronate disodium [21]. Another study showed successful separation of nucleotides and sugar phosphates [10]. Therefore, the Primesep SB column was chosen to separate nucleotides, sugar phosphates, and the nucleotide sugar in this study. In this manuscript, it is reported that the mixed-mode column without the ion-pairing in the mobile phase was successful at separating a sugar phosphate, nucleoside mono, di- and tri-phosphates, as well as a nucleotide sugar. Ammonium formate buffer was selected due to the compatibility with MS for detection and confirmation of the analytes. Fluorine NMR was utilized to determine purity of the purified fluorinated nucleotide sugar. Challenges and observations in the semi-preparative scale are also discussed.
2.3. HPLC instrumentation The analytical experiments were carried out on an Agilent 1200 LC consisting of a degasser, a binary pump, an autosampler, a diode array detector and a column thermostat. This was coupled with an Agilent 6140 quadrupole mass spectrometer. The semi-preparative purification experiments were carried out on an Agilent 1100 analytical/preparative hybrid liquid chromatography unit consisting of a degasser, an analytical quaternary pump, a preparative binary pump, an autosampler, a diode array detector and a fraction collection module. All data were processed using the Agilent Chemstation software (Agilent, Santa Clara, CA, USA). The HPLC columns were the Sielc Primesep SB column 5 m, 4.6 × 150 mM for the analytical method and 5 m, 21.2 × 150 mm for the semi-preparative method (Sielc, Wheeling, IL, USA). 2.4. HPLC method conditions
2. Experimental
For the analytical method, the mobile phase consisted of mixture of 50 mM ammonium formate buffer at pH 2.7 or 3.0 (a) and acetonitrile (b) (a:b, 85:15, v/v). The pH was adjusted using formic acid prior to addition of acetonitrile. Flow rate was 1.0 mL/min, and column temperature was ambient. A 10 L aliquot of 1.0 mg/mL sample was injected. For the semi-preparative method, mobile phase A was prepared by mixing 50 mM of ammonium formate buffer at pH 2.7 (a) and acetonitrile (b) (a:b, 90:10, v/v). Mobile phase B was prepared by mixing 100 mM of ammonium formate buffer at pH 2.7 (a) and acetonitrile (a:b, 85:15, v/v). Mobile phase A was used for the first 24 min. Then, mobile phase B was utilized until the end of the run at 35 min. Flow rate was 20.0 mL/min, and the column was at ambient temperature. 900 L of crude sample was injected and fractions collected.
2.1. Materials
2.5. NMR instrumentation
Standards utilized in this study included: adenosine triphosphate (Fisher Science Education, Nazareth, PA, USA), adenosine 5 -diphosphate, guanosine 5 -triphosphate sodium salt hydrate, guanosine 5 -diphosphate sodium salt, guanosine 5 -diphospho-l-fucose sodium salt, ˇ-l-fucose 1-phosphate bis (cyclohexylammonium) salt and d- (+)-fucose (Sigma–Aldrich, St. Louis, MO, USA). Inorganic pyrophosphatase was obtained from Fisher Scientific, Pittsburgh, PA. Ammonium formate, trizma base, HPLC grade acetonitrile and HPLC grade water were obtained from Sigma–Aldrich, St. Louis, MO, USA. Formic acid was purchased from EMD Biosciences, Billerica, MA, USA.
All NMR experiments were performed at 25 ◦ C on a 500 MHz Bruker Avance II+ NMR spectrometer equipped with a SEF cryoprobe (for 19 F direct detection and 1 H decoupling). All data were processed using the Bruker Topspin software (Bruker Biospin, Billerica, MA, USA).
2.2. Chemoenzymatic synthesis of guanosine diphosphate-trifluoromethyl fucose (GDP-TFMF) The primary compound of interest is a trifluoromethyl analogue of fucose attached to GDP, as shown in Fig. 1. The reaction for the chemoenzymatic synthesis of this GDP-trifluoromethyl fucose (GDP-TFMF) contains 5 mM adenosine triphosphate (ATP), 5 mM guanosine triphosphate (GTP), 5 mM of trifluoromethyl fucose (TFMF), 0.1 g/L in-house produced L-fucokinase/GDP-fucose pyrophosphorylase (FKP), 0.01 U/L inorganic pyrophosphatase, in a buffer of 100 mM Tris, pH 7.5 and 5 mM MgCl2 , resulting in a total volume of 20 mL. The reaction mixture was incubated at 37 ◦ C for 6 h with shaking at 150 rpm. At the end of the reaction, an equal volume of acetonitrile was added. The solution was vortexed and then incubated at room temperature for 10 min. Insoluble material was removed by centrifugation at 5000 rpm for 10 min [22]. Supernatant was transferred to a new tube for purification.
2.6. NMR method conditions Recovered samples were dissolved in water and lyophilized overnight. Samples were prepared as 50 mM stocks in DI water and were then diluted to 500 M concentration in 20 mM sodium phosphate, pH 7.4, 10% D2 O (for NMR field lock). An internal 19 F NMR standard was spiked at 500 M and 1D 19 F NMR spectra (with proton decoupling) were acquired. Peaks from the NMR standard were integrated and set as reference integrals. Remaining peaks were integrated and used to calculate actual experimental concentration of compound. 3. Results & discussion The Primesep SB column had been shown to separate sugar phosphates by a gradient method using aqueous ammonium formate and acetonitrile as the mobile phase [23]; therefore, this column was investigated to separate TFMF-1-P, GDP-TFMF, ADP, GDP, ATP and GTP using the aforementioned buffer. Initial method development utilized commercially available ADP, GDP, ATP, GTP, GDP-fucose (which is similar to GDP-TFMF) and fucose-1-P (which is similar to TFMF-1-P). The Primesep SB column provided adequate preliminary resolution between these analytes. Since the column provided desired results in separating GDP-fucose and fucose-1-P
H. Eastwood et al. / Journal of Pharmaceutical and Biomedical Analysis 115 (2015) 402–409
405
Retenon Factor (k)
20
15
10
5
0
25
50
100
Buffer Concentraon (mM) Fig. 2. Retention factors (k) of GTP (), GDP (×), ATP (䊏), ADP (䊉), GDP-fucose () and fucose-1-P (*) at different buffer concentrations. Column: Primesep SB (4.6 × 150 mM, 5 m). Mobile phase: NH4 HCO2 (pH 3.0)/CH3 CN (85:15). Flow-rate: 1.0 mL/min. UV detection at 254 nm and MS ESI negative mode for fucose-1-phosphate.
from the other analytes, further exploration was focused on Primesep SB. Adjustments in mobile phase buffer concentration and pH resulted in significant changes in retention time of the analytes, as detailed in the following sections. 3.1. Buffer study Altering buffer type and/or concentration can offer different selectivities for separation methods. Separation of sugarphosphates have been reported using ammonium formate with the Primesep SB [10,23] and ammonium acetate with a reversed phase column [24]. It was also reported that sodium phosphate buffer with a strong anion-exchange column was used for nucleotide separations [10]. It was also reported that sodium phosphate buffer [10] was used for nucleotide separations. For this study, ammonium formate was selected, since it was successful for separating similar analytes such as ADP, ATP, GDP, GTP [7] and GDP-fucose [12]. Retention factors (k) of ATP, GDP, ATP, ADP, GDP-fucose and fucose-1-P at different buffer concentrations are shown in Fig. 2. It was found that increasing ammonium formate concentration resulted in significant difference between the di- and tri-phosphate analytes. At 50 mM ammonium formate, the k of ATP and GTP were 43 and 82 respectively, whereas ADP and GDP were 7 and 13 respectively. At 100 mM, ADP and GDP were not as well sepa-
rated as at 50 mM. In addition, the analytes of interest (GDP-fucose and fucose-1-P) were resolved better from ADP, GDP, ATP and GTP at 50 mM compared to 100 mM. The 25 mM buffer provided substantial separation among analytes, but would not be practical for the preparative scale under isocratic conditions due to the long retention times. Generally, increasing concentrations of ammonium formate in the mobile phase was found to decrease retention time of all the analytes studied. This behavior could be attributed mainly to the ion exchange behavior [25]. The increased ammonium ions in the mobile phase could compete for the negatively charged phosphate groups of the analytes, resulting in shorter retention times. In addition, the increased formate ions may compete with the analyte for the positively charged moiety of the stationary phase. 3.2. pH study Effect of mobile phase pH on retention time was explored. While keeping the buffer concentration (50 mM) and the acetonitrile percentage (15%) constant, pH was adjusted using formic acid. Results demonstrated that pH affected retention time. The Primesep SB column is always positively charged throughout the pH range of the mobile phases explored in this study [21]. While the buffer range for ammonium formate is 2.8–4.8, the pH values tested in this study were in the range of 2.0–5.0. Within these values, it can
Fig. 3. Retention factors (k) of GTP (), GDP (×), ATP (䊏), ADP (䊉), GDP-fucose () and fucose-1-P (*) at different pH values of NH4 HCO2 in the mobile phase. pH was adjusted using formic acid. Same chromatographic conditions as in Fig. 2 using 50 mM NH4 HCO2 .
406
H. Eastwood et al. / Journal of Pharmaceutical and Biomedical Analysis 115 (2015) 402–409
Fig. 4. Retention factors (k) of GTP (), GDP (×), ATP (䊏), ADP (䊉), GDP-fucose () and fucose-1-P (*) using different percentages of CH3 CN in the mobile phase at pH 3.0. All other chromatographic conditions are the same as in Fig. 3.
be assumed that changes in retention are mainly due to the effect of the mobile phase on the analyte rather than on the charge state of the stationary phase. As shown in Fig. 3, the analytes with more ionizable phosphate groups retained longer as the mobile phase pH increased. The analytes in this study that contain the most phosphate groups, the nucleoside triphosphates (ATP and GTP), were retained longer than all other analytes studied. Nucleoside diphosphates (ADP and GDP) and the sugar-phosphate (fucose-1-P) also followed this trend and continued to show increased retention as the pH increased. However, GDP-fucose did not follow the trend; though its greatest retention was at pH 4.0, it did decrease at pH 5.0. While pH 5.0 offered the greatest differences in retention factors, pH 3.0 provided the adequate resolution and a more practical run time for the preparative run. Nucleotide tri-phosphates eluted at over 40 min at pHs greater than 3.0. The pH value of 2.0 also
offered sufficient separation, but it was at the lower boundary of the recommended operating pH range (2.0–4.5) of the Primesep SB column and outside of the buffering capacity of ammonium formate buffer.
3.3. Organic solvent percentage study Percentage of the organic additive in the mobile phase could be used to manipulate retention time, while keeping the buffer concentration constant. A previous study has demonstrated separation of sugar-phosphates (including frucose-6-phosphate, ribose-6phosphate, etc.) utilizing ammonium formate buffer, acetonitrile, and the Primesep SB column [23]. It was observed that changing acetonitrile concentration had a marginal impact in comparison to changes in buffer concentration.
Fig. 5. Analytical separation of the reaction mixture containing TFMF-1-P, GDP-TFMF, ADP, GDP, ATP and GTP*. (a) UV at 254 nm. (b) MS ESI negative mode. (c) Extracted ion chromatogram (EIC) of TFMF-1-P.(d) EIC of GDP-TFMF. Column: Primesep SB (4.6 × 150 mm, 5 m). Mobile phase: 50 mM NH4 HCO2 (pH 2.7)/CH3 CN (85:15). Flow-rate: 1.0 mL/min. 10 L injection. *GTP not included in figure due to excessive retention time.
H. Eastwood et al. / Journal of Pharmaceutical and Biomedical Analysis 115 (2015) 402–409
407
Fig. 6. Semi-preparative purification of the reaction mixture with modified step gradient. Column: Primesep SB (21.2 × 150 mm, 5 m). Mobile Phase A: 50 mM NH4 HCO2 (pH 2.7)/ CH3 CN (90:10). Mobile Phase B: 100 mM NH4 HCO2 (pH 2.7)/CH3 CN (85:15). Step gradient at 100% A from 0 to 24 min, then switched to 100% B and held for 16 min. Total run time: 40 min. Flow-rate: 20 mL/min. 900 L injection. UV at 254 nm.
For our study, changing the acetonitrile percentage had less of an impact on the retention of fucose-1-P as noted in the aforementioned study. However, GDP, GTP, and GDP-fucose had a more pronounced reduction in retention time, possibly due to their more hydrophobic nature, as shown in Fig. 4. Acetonitrile percentage adjustment was useful in fine tuning the method for the preferred
retention times. A range of 10–15% acetonitrile was explored in the mobile phase to facilitate a shortened total run time and maintain adequate separation between GDP-fucose and fucose-1-P. Since acetonitrile provided desired results in separating the GDP-TFMF and TFMF-1-P from other analytes, which will be explained in the following sections, additional organic solvents were not explored.
Fig. 7. 19 F NMR for purity assessment of GDP-TFMF. (a) 500 M internal 19 F NMR standard, N-trifluoroacetyl-d-glucosamine. (b) 500 M internal 19 F NMR standard spiked in and peaks integrated and set as a reference integral. The sum of the GDP-TFMF peak area integrals in both pyranose and furanose forms was compared to the standard integrals.
408
H. Eastwood et al. / Journal of Pharmaceutical and Biomedical Analysis 115 (2015) 402–409
3.4. Final conditions for analytical HPLC-MS and semi-preparative methods After studying buffer concentration, pH and acetonitrile percentage, the final conditions were determined. The mobile phase, consisting of 50 mM ammonium formate at pH 2.7 with 15% acetonitrile, was successful at separating GTP, ATP, GDP, ADP, GDPfucose and fucose-1-P standards using the Primesep SB column. Four of the standards (GDP, ADP, GDP-fucose and fucose-1-P) were separated within 15 min. Negative mode ES-API MS detection was necessary to detect fucose-1-P due to the lack of a chromophore. The same HPLC–MS method was then used to analyze the reaction mixture containing TFMF-1-P and GDP-TFMF, as shown in Fig. 5. This method successfully separated these two compounds from the other components (unreacted TFMF, ADP, GDP and ATP) in the reaction mixture. Using the same mobile phase composition as the analytical method, a semi-preparative separation method was initially developed. However, GDP and GDP-TFMF closely eluted, which would make it difficult to collect GDP-TFMF (data not shown). To further separate these close eluting analytes, the acetonitrile percentage was adjusted from 15% to 10%. Since ATP and GTP eluted much later using 10% acetonitrile, a step gradient was applied to decrease the retention time. After 24 min, mobile phase was changed to 100 mM ammonium formate with 15% acetonitrile. The resultant chromatogram is shown in Fig. 6, and method details are explained in the legend. Using the step gradient semi-preparative method, the overall run time was decreased from ∼120 min to 40 min. This method was successfully applied to purify the TFMF-1-P and GDPTFMF from the reaction mixture.
3.5. Purity assessment using 19 F NMR After isolating the desired compound, GDP-TFMF, purity determination was necessary. Since there was no pure authentic compound available, 19 F NMR was applied using an internal standard, N-trifluoroacetyl-d-glucosamine, to assess the purity of the final product. First, a 19 F spectrum of the standard compound was acquired (Fig. 7A). Next, a solution containing 500 M of the internal standard and ∼500 M of the purified GDP-TFMF was prepared. Assignments of the pyranose and furanose forms of the sugar were made based on previous experimental data (primarily 1 H NMR, data not shown). 19 F NMR spectra were obtained, and the signals for both the standard and compound were integrated based on peak area (Fig. 7B). The concentration of GDP-TFMF was determined by comparing integral values for signals from both compounds. The purity of GDP-TFMF was determined to be approximately 99%, with a yield of 30 mg.
4. Conclusion Successful analytical and semi-preparative HPLC methods were developed to resolve TFMF-1-P and GDP-TFMF from an enzymatic reaction mixture containing GTP, ATP, GDP, ADP, GDP-fucose and fucose-1-P within a reasonable run time, after determining concentration and pH of the ammonium formate buffer and the percentage of acetonitrile. Ammonium formate buffer was necessary as it allowed for the use of MS ES (-) to detect the sugar-phosphate, TFMF-1-P. Purification was performed to collect the nucleotide sugar (GDP-TFMF). The purity was determined utilizing 19 F NMR. Based on the experimental results and the application of the mixed mode column with volatile organic buffer, this technique can be utilized to separate similar sugar-phosphates and nucleotide sugars. This method is not only analytically useful but also appli-
cable to the preparative chromatography scale, allowing for easy recovery of the purified material with minimal salt content. Acknowledgment The authors would like to acknowledge Randy Jensen for his assistance with some of the NMR experiments conducted in this research. References [1] J. Rabina, M. Maki, E. Savilahti, N. Jarinen, L. Penttila, R. Renknon, Analysis of nucleotide sugars from cell lysates by ion-pair solid-phase extraction and reversed-phase high-performance liquid chromatography, Glycoconj. J. 18 (2001) 799. [2] M. Pabst, J. Grass, R. Fischl, R. Leonard, C. Jin, G. Hinterkorner, N. Borth, F. Altmann, Nucleotide and nucleotide sugar analysis by liquid chromatography–electrospray ionization–mass spectrometry on surface–conditioned porous graphitic carbon, Anal. Chem. 82 (2010) 9782. [3] N. Okeley, S. Alley, M. Anderson, T. Boursalian, P. Burke, K. Emmerton, S. Jeffrey, K. Klussman, C. Law, D. Sussman, B. Toki, L. Westendorf, W. Zeng, X. Zhang, D. Benjamin, P. Senter, Development of orally active inhibitors of protein and cellular fucosylation, PNAS 110 (2013) 5404. [4] J. Martin Buescher, S. Moco, U. Sauer, N. Zamboni, Ultrahigh performance liquid chromatography–tandem mass spectrometry method for fast and robust quantification of anionic and aromatic metabolites, Anal. Chem. 82 (2010) 4403. [5] K. Nakajima, S. Kitazume, T. Angata, R. Fujinawa, K. Ohtsubo, E. Miyoshi, N. Taniguchi, Simultaneous determination of nucleotide sugars with ion-pair reversed-phase HPLC, Glycobiology 20 (2010) 865. [6] A. Toernkvist, K. Markides, L. Nyholm, Chromatographic behaviour of oxidised porous graphitic carbon columns, Analyst 128 (2003) 844. [7] N. Padivitage, M. Dissanayake, D. Armstrong, Separation of nucleotides by hydrophilic interaction chromatography using the FRULIC-N column, Anal. Bioanal. Chem. 405 (2013) 8837. [8] H. Hinterwirth, M. Lammerhofer, B. Preinerstorfer, A. Gargano, R. Reischl, W. Bicker, O. Trapp, L. Brecker, W. Lindner, Selectivity issues in targeted metabolomics: separation of phosphorylated carbohydrate isomers by mixed-mode hydrophilic interaction/weak anion exchange chromatography, J. Sep. Sci. 33 (2010) 3273. [9] J. Ito, T. Herter, E. Baidoo, J. Lao, M. Vega-Sanchez, A. Smith-Moritz, P. Adams, J. Keasling, B. Usadel, C. Petzold, J. Heazlewood, Analysis of plant nucleotide sugars by hydrophilic interaction liquid chromatography and tandem mass spectrometry, Anal. Biochem. 448 (2014) 14. [10] L. Nunes, E.S. Martinez-Barajas, R. Powers, P. Sagar, B. Davis, M. Paul, Inhibition of SnRK1 by metabolites: tissue-dependent effects and cooperative inhibition by glucose 1-phosphate in combination with trehalose 6-phosphate, Plant Physiol.Biochem. 63 (2013) 89. [11] V. Orlovsky, Y. Zelechonok, Evolution of Mixed-Mode Chromatography Chromatography Today, (Aug/Sept 2011), pp. 24. [12] J.S.O. McCullagh, D. Juchelka, R. Hedges, Analysis of amino acid 13C abundance from human and faunal bone collagen using liquid chromatography/isotope ratio mass spectrometry, Rapid Commun. Mass Spectrom. 20 (2006) 2761. [13] R. Bischoff, L. McLaughlin, Nucleic acid resolution by mixed-mode chromatography, J. Chromatogr. 296 (1984) 329. [14] W.S. Hancock, Use of Mixed-Mode, High-performance liquid chromatography for the separation of peptide and protein mixtures, J. Chromatogr. 206 (1981) 71. [15] M. Davies, A study of retention and overloading of basic compounds with mixed-mode reversed-phase/cation-exchange columns in high performance liquid chromatography, J. Chromatogr. A 1138 (2007) 65. [16] J. Li, S. Shao, M. Jaworsky, P. Kurtulik, Simultaneous determination of cations, zwitterions and neutral compounds using mixed-mode reversed-phase and cation-exchange high-performance liquid chromatography, J. Chromatogr. A 1185 (2008) 185. [17] H. Licea-Perez, S. Wang, C. Bowen, Development of a sensitive and selective LC-MS/MS method for the determination of &␣lph␣;-fluoro-&et␣;-alanine, 5-fluorouracil and capecitabine in human plasma, J. Chromatogr. B 877 (2009) 1040. [18] M. Rance, S. Wilson, &et␣;-Diketones: peak shape challenges and the use of mixed-mode high-performance liquid chromatography methodology to obtain fast, high resolution chromatography, J. Chromatogr. A 1217 (2010) 1912. [19] A. Miller, H. Brown, T. Degg, K. Allen, B. Keevil, Measurement of plasma 5-hydroxyindole acetic acid by liquid chromatography tandem mass spectrometry—comparison with HPLC methodology, J. Chromatogr. B 878 (2010) 695. [20] Sielc. http://www.sielc.com/Products Primesep SB.html [21] X. Liu, J. Fang, N. Cauchon, P. Zhou, Direct stability-indicating method development and validation for analysis of etidronate disodium using a mixed-mode column and charged aerosol detector, J. Pharm. Biomed. Anal. 46 (2008) 639.
H. Eastwood et al. / Journal of Pharmaceutical and Biomedical Analysis 115 (2015) 402–409 [22] W. Wang, T. Hu, P. Frantom, T. Zheng, B. Gerwe, D. Soriano del Amo, S. Garret, R. Seidel III, P. Wu, Chemoenzymatic synthesis of GDP-L-fucose and the Lewis X glycan derivatives, PNAS 106 (2009) 16096. [23] J. Cruz, C. Emery, M. Wuest, D. Kramer, B. Lange, Metabolite profiling of Calvin cycle intermediates by HPLC–MS using mixed-mode stationary phases, Plant J. 55 (2008) 1047.
409
[24] J. Ding, C. Sorensen, Q. Zhang, H. Jiang, N. Jaitly, E. Livesay, Y. Shen, R. Smith, T. Metz, Capillary LC coupled with high-mass measurement accuracy mass spectrometry for metabolic profiling, Anal. Chem. 79 (2007) 6081. [25] A. Stalcup, K. Gahm, A sulfated cyclodextrin chiral stationary phase for high-performance liquid chromatography, Anal. Chem. 68 (1996) 1369.
Contents lists available at ScienceDirect
Journal of Pharmaceutical and Biomedical Analysis journal homepage: www.elsevier.com/locate/jpba
Development of a nucleotide sugar purification method using a mixed mode column & mass spectrometry detection Heather Eastwood a,∗ , Fang Xia b , Mei-Chu Lo a , Jing Zhou a , John B. Jordan a , John McCarter a , Wesley W. Barnhart a , Kyung-Hyun Gahm a a b
Department of Molecular Structure & Characterization, Amgen, Inc., Thousand Oaks, CA 91320, United States ASK-Gene Pharma, Inc., Camarillo, CA 93012, United States
a r t i c l e
i n f o
Article history: Received 30 April 2015 Received in revised form 31 July 2015 Accepted 1 August 2015 Available online 4 August 2015 Keywords: Nucleotides Nucleotide-sugar Sugar phosphate Mixed-mode stationary phase High performance liquid chromatography Mass spectrometry
a b s t r a c t Analysis of nucleotide sugars, nucleoside di- and triphosphates and sugar-phosphates is an essential step in the process of understanding enzymatic pathways. A facile and rapid separation method was developed to analyze these compounds present in an enzymatic reaction mixture utilized to produce nucleotide sugars. The Primesep SB column explored in this study utilizes hydrophobic interactions as well as electrostatic interactions with the phosphoric portion of the nucleotide sugars. Ammonium formate buffer was selected due to its compatibility with mass spectrometry. Negative ion mode mass spectrometry was adopted for detection of the sugar phosphate (fucose-1-phophate), as the compound is not amenable to UV detection. Various mobile phase conditions such as pH, buffer concentration and organic modifier were explored. The semi-preparative separation method was developed to prepare 30 mg of the nucleotide sugar. 19 F NMR was utilized to determine purity of the purified fluorinated nucleotide sugar. The collected nucleotide sugar was found to be 99% pure. Published by Elsevier B.V.
1. Introduction Nucleotide sugars contain sugar or sugar derivatives connected through the glycosidic hydroxyl group to the 2nd phosphate of a nucleoside 5 -pyrophosphate. The biosynthesis of protein and lipidlinked oligosaccharides utilizes nucleotide sugars for glycosylation [1,2]. Therefore, analysis of nucleotide sugars is an important step in understanding the glycosylation mechanism in cells. Inhibition of fucosylation, the process of adding fucose sugar units to a molecule, is of interest in the field of antibody-dependent cellular cytotoxicity (ADCC). The antibody binds to the target antigen and engages CD16 on the effector cell. CD16 is a low affinity IgG receptor Fc III (Fc␥RIII) found on the surface of natural killer cells, such as monocytes and macrophages. Binding of the antibody causes the release of pore-forming proteins and proteases, which lyse the target cell. Absence of fucose on the antibody improves binding to the effector cell which, in turn, enhances cell lysis [3]. Fucosylation may be reduced through the use of a small molecule inhibitor of the fucosylation pathway.
∗ Corresponding author at: Amgen, Inc., One Amgen Center Drive, 29-2-C, Thousand Oaks, CA 91320, United States. Fax: +1 805 480 3016. E-mail address: [email protected] (H. Eastwood). http://dx.doi.org/10.1016/j.jpba.2015.08.001 0731-7085/Published by Elsevier B.V.
Various chromatographic methods have been reported for the separation of nucleotides, nucleotide-sugars and sugarphosphates. These include reversed-phase ion-pairing (RP-IP), porous graphite carbon (PGC), hydrophilic interaction liquid chromatography (HILIC) and mixed mode chromatography [1,2,4–10]. Guanosine diphosphate (GDP) sugars and their metabolites were separated by ion-pairing chromatography utilizing ODS-columns (C-18) and ion-pairing agents such as tributylamine with acetic acid [4]. Tetrabutylammonium hydrogen sulfate has been shown to be successful at separating these analytes as well. However, the ionpairing agent was not compatible with the mass spectrometer [5]. In addition, to these studies, separation of various nucleotide sugars, such as GDP-mannose, GDP-fucose and GDP-rhamnose were successfully performed using a C-18 reversed-phase column and triethylammonium acetate buffer as an ion-pairing reagent for increasing retention [1]. Unfortunately, this buffer has low volatility and would be difficult to remove from collected purified samples. Also, nucleoside triphosphates were not explored in the study. Columns with porous graphite carbon (PGC) were shown to separate nucleotides according to the base and the sugar moiety rather than the ionic phosphate groups, unlike ion-pairing reversed-phase chromatography. However, PGC columns required complex cleaning methods, which included boiling in TFA and rinsing with sulfite,
H. Eastwood et al. / Journal of Pharmaceutical and Biomedical Analysis 115 (2015) 402–409
403
Fig. 1. (a) Chemoenzymatic synthesis of GDP-TFMF from TFMF-fucose and the intermediate TFMF-1-phosphate. (b) Structures of ATP, GDP, ATP, ADP, GDP-fucose, fucose-1-P and fucose. (c) The schematic depiction of the Primesep SB stationary phase with strong embedded basic ion-pairing groups bound to silica.
as well as an acid/base treatment to achieve separation [2]. Changes in the column could lead to long equilibration time and reproducibility issues. One possible explanation for this was a slow change on the PGC surface from oxidation [6]. HILIC has been useful at separating hydrophilic analytes and has been successful for the separation of nucleotides, sugar phosphates and nucleotide sugars [7–9]. Recently, separation of nucleotides was achieved with a cyclofructan based column using ammonium acetate buffer in both gradient and isocratic methods [7]. Unlike
other HILIC stationary phases, cyclofructan offers dual retention mechanisms. Traditional hydrogen bonding/dipolar interactions can be supplemented by dynamic ion interaction effects for anionic analytes. This is due to cyclofructan binding certain buffer cations to form a positively charged cavity in the stationary phase which will bind the negatively charged phosphates of the nucleotides. While this mode is useful for nucleotide separations, it did not address separating nucleotide sugars or sugar phosphates that could potentially be present.
404
H. Eastwood et al. / Journal of Pharmaceutical and Biomedical Analysis 115 (2015) 402–409
Mixed mode columns are phases that provide at least two modes of interaction in one stationary phase, such as ion exchange with C18 properties [10]. This is beneficial due to the multiple controllable interactions on the column, which allow for better control of retention for various analytes [11]. Mixed mode has been mainly used to separate biologic molecules, nucleic acids, peptides and proteins [12–14]. Primesep has several commercially available mixed mode which offer selectivity in the separation of a broad array of chemical compounds and in a multitude of applications [12,15–19]. Among the several Primesep columns available, the Primesep SB column contains strong basic ion-pairing groups, improves retention of acidic compounds by anion-exchange mechanism, separates bases by ion-exclusion mechanism together with hydrophobicity and retains neutral compounds via a reversed-phase mechanism [20]. It has been shown to resolve analytes that are very polar and have highly ionic characteristics, such as etidronate disodium [21]. The column was used for stability-indicating method development and validation for analysis of etidronate disodium [21]. Another study showed successful separation of nucleotides and sugar phosphates [10]. Therefore, the Primesep SB column was chosen to separate nucleotides, sugar phosphates, and the nucleotide sugar in this study. In this manuscript, it is reported that the mixed-mode column without the ion-pairing in the mobile phase was successful at separating a sugar phosphate, nucleoside mono, di- and tri-phosphates, as well as a nucleotide sugar. Ammonium formate buffer was selected due to the compatibility with MS for detection and confirmation of the analytes. Fluorine NMR was utilized to determine purity of the purified fluorinated nucleotide sugar. Challenges and observations in the semi-preparative scale are also discussed.
2.3. HPLC instrumentation The analytical experiments were carried out on an Agilent 1200 LC consisting of a degasser, a binary pump, an autosampler, a diode array detector and a column thermostat. This was coupled with an Agilent 6140 quadrupole mass spectrometer. The semi-preparative purification experiments were carried out on an Agilent 1100 analytical/preparative hybrid liquid chromatography unit consisting of a degasser, an analytical quaternary pump, a preparative binary pump, an autosampler, a diode array detector and a fraction collection module. All data were processed using the Agilent Chemstation software (Agilent, Santa Clara, CA, USA). The HPLC columns were the Sielc Primesep SB column 5 m, 4.6 × 150 mM for the analytical method and 5 m, 21.2 × 150 mm for the semi-preparative method (Sielc, Wheeling, IL, USA). 2.4. HPLC method conditions
2. Experimental
For the analytical method, the mobile phase consisted of mixture of 50 mM ammonium formate buffer at pH 2.7 or 3.0 (a) and acetonitrile (b) (a:b, 85:15, v/v). The pH was adjusted using formic acid prior to addition of acetonitrile. Flow rate was 1.0 mL/min, and column temperature was ambient. A 10 L aliquot of 1.0 mg/mL sample was injected. For the semi-preparative method, mobile phase A was prepared by mixing 50 mM of ammonium formate buffer at pH 2.7 (a) and acetonitrile (b) (a:b, 90:10, v/v). Mobile phase B was prepared by mixing 100 mM of ammonium formate buffer at pH 2.7 (a) and acetonitrile (a:b, 85:15, v/v). Mobile phase A was used for the first 24 min. Then, mobile phase B was utilized until the end of the run at 35 min. Flow rate was 20.0 mL/min, and the column was at ambient temperature. 900 L of crude sample was injected and fractions collected.
2.1. Materials
2.5. NMR instrumentation
Standards utilized in this study included: adenosine triphosphate (Fisher Science Education, Nazareth, PA, USA), adenosine 5 -diphosphate, guanosine 5 -triphosphate sodium salt hydrate, guanosine 5 -diphosphate sodium salt, guanosine 5 -diphospho-l-fucose sodium salt, ˇ-l-fucose 1-phosphate bis (cyclohexylammonium) salt and d- (+)-fucose (Sigma–Aldrich, St. Louis, MO, USA). Inorganic pyrophosphatase was obtained from Fisher Scientific, Pittsburgh, PA. Ammonium formate, trizma base, HPLC grade acetonitrile and HPLC grade water were obtained from Sigma–Aldrich, St. Louis, MO, USA. Formic acid was purchased from EMD Biosciences, Billerica, MA, USA.
All NMR experiments were performed at 25 ◦ C on a 500 MHz Bruker Avance II+ NMR spectrometer equipped with a SEF cryoprobe (for 19 F direct detection and 1 H decoupling). All data were processed using the Bruker Topspin software (Bruker Biospin, Billerica, MA, USA).
2.2. Chemoenzymatic synthesis of guanosine diphosphate-trifluoromethyl fucose (GDP-TFMF) The primary compound of interest is a trifluoromethyl analogue of fucose attached to GDP, as shown in Fig. 1. The reaction for the chemoenzymatic synthesis of this GDP-trifluoromethyl fucose (GDP-TFMF) contains 5 mM adenosine triphosphate (ATP), 5 mM guanosine triphosphate (GTP), 5 mM of trifluoromethyl fucose (TFMF), 0.1 g/L in-house produced L-fucokinase/GDP-fucose pyrophosphorylase (FKP), 0.01 U/L inorganic pyrophosphatase, in a buffer of 100 mM Tris, pH 7.5 and 5 mM MgCl2 , resulting in a total volume of 20 mL. The reaction mixture was incubated at 37 ◦ C for 6 h with shaking at 150 rpm. At the end of the reaction, an equal volume of acetonitrile was added. The solution was vortexed and then incubated at room temperature for 10 min. Insoluble material was removed by centrifugation at 5000 rpm for 10 min [22]. Supernatant was transferred to a new tube for purification.
2.6. NMR method conditions Recovered samples were dissolved in water and lyophilized overnight. Samples were prepared as 50 mM stocks in DI water and were then diluted to 500 M concentration in 20 mM sodium phosphate, pH 7.4, 10% D2 O (for NMR field lock). An internal 19 F NMR standard was spiked at 500 M and 1D 19 F NMR spectra (with proton decoupling) were acquired. Peaks from the NMR standard were integrated and set as reference integrals. Remaining peaks were integrated and used to calculate actual experimental concentration of compound. 3. Results & discussion The Primesep SB column had been shown to separate sugar phosphates by a gradient method using aqueous ammonium formate and acetonitrile as the mobile phase [23]; therefore, this column was investigated to separate TFMF-1-P, GDP-TFMF, ADP, GDP, ATP and GTP using the aforementioned buffer. Initial method development utilized commercially available ADP, GDP, ATP, GTP, GDP-fucose (which is similar to GDP-TFMF) and fucose-1-P (which is similar to TFMF-1-P). The Primesep SB column provided adequate preliminary resolution between these analytes. Since the column provided desired results in separating GDP-fucose and fucose-1-P
H. Eastwood et al. / Journal of Pharmaceutical and Biomedical Analysis 115 (2015) 402–409
405
Retenon Factor (k)
20
15
10
5
0
25
50
100
Buffer Concentraon (mM) Fig. 2. Retention factors (k) of GTP (), GDP (×), ATP (䊏), ADP (䊉), GDP-fucose () and fucose-1-P (*) at different buffer concentrations. Column: Primesep SB (4.6 × 150 mM, 5 m). Mobile phase: NH4 HCO2 (pH 3.0)/CH3 CN (85:15). Flow-rate: 1.0 mL/min. UV detection at 254 nm and MS ESI negative mode for fucose-1-phosphate.
from the other analytes, further exploration was focused on Primesep SB. Adjustments in mobile phase buffer concentration and pH resulted in significant changes in retention time of the analytes, as detailed in the following sections. 3.1. Buffer study Altering buffer type and/or concentration can offer different selectivities for separation methods. Separation of sugarphosphates have been reported using ammonium formate with the Primesep SB [10,23] and ammonium acetate with a reversed phase column [24]. It was also reported that sodium phosphate buffer with a strong anion-exchange column was used for nucleotide separations [10]. It was also reported that sodium phosphate buffer [10] was used for nucleotide separations. For this study, ammonium formate was selected, since it was successful for separating similar analytes such as ADP, ATP, GDP, GTP [7] and GDP-fucose [12]. Retention factors (k) of ATP, GDP, ATP, ADP, GDP-fucose and fucose-1-P at different buffer concentrations are shown in Fig. 2. It was found that increasing ammonium formate concentration resulted in significant difference between the di- and tri-phosphate analytes. At 50 mM ammonium formate, the k of ATP and GTP were 43 and 82 respectively, whereas ADP and GDP were 7 and 13 respectively. At 100 mM, ADP and GDP were not as well sepa-
rated as at 50 mM. In addition, the analytes of interest (GDP-fucose and fucose-1-P) were resolved better from ADP, GDP, ATP and GTP at 50 mM compared to 100 mM. The 25 mM buffer provided substantial separation among analytes, but would not be practical for the preparative scale under isocratic conditions due to the long retention times. Generally, increasing concentrations of ammonium formate in the mobile phase was found to decrease retention time of all the analytes studied. This behavior could be attributed mainly to the ion exchange behavior [25]. The increased ammonium ions in the mobile phase could compete for the negatively charged phosphate groups of the analytes, resulting in shorter retention times. In addition, the increased formate ions may compete with the analyte for the positively charged moiety of the stationary phase. 3.2. pH study Effect of mobile phase pH on retention time was explored. While keeping the buffer concentration (50 mM) and the acetonitrile percentage (15%) constant, pH was adjusted using formic acid. Results demonstrated that pH affected retention time. The Primesep SB column is always positively charged throughout the pH range of the mobile phases explored in this study [21]. While the buffer range for ammonium formate is 2.8–4.8, the pH values tested in this study were in the range of 2.0–5.0. Within these values, it can
Fig. 3. Retention factors (k) of GTP (), GDP (×), ATP (䊏), ADP (䊉), GDP-fucose () and fucose-1-P (*) at different pH values of NH4 HCO2 in the mobile phase. pH was adjusted using formic acid. Same chromatographic conditions as in Fig. 2 using 50 mM NH4 HCO2 .
406
H. Eastwood et al. / Journal of Pharmaceutical and Biomedical Analysis 115 (2015) 402–409
Fig. 4. Retention factors (k) of GTP (), GDP (×), ATP (䊏), ADP (䊉), GDP-fucose () and fucose-1-P (*) using different percentages of CH3 CN in the mobile phase at pH 3.0. All other chromatographic conditions are the same as in Fig. 3.
be assumed that changes in retention are mainly due to the effect of the mobile phase on the analyte rather than on the charge state of the stationary phase. As shown in Fig. 3, the analytes with more ionizable phosphate groups retained longer as the mobile phase pH increased. The analytes in this study that contain the most phosphate groups, the nucleoside triphosphates (ATP and GTP), were retained longer than all other analytes studied. Nucleoside diphosphates (ADP and GDP) and the sugar-phosphate (fucose-1-P) also followed this trend and continued to show increased retention as the pH increased. However, GDP-fucose did not follow the trend; though its greatest retention was at pH 4.0, it did decrease at pH 5.0. While pH 5.0 offered the greatest differences in retention factors, pH 3.0 provided the adequate resolution and a more practical run time for the preparative run. Nucleotide tri-phosphates eluted at over 40 min at pHs greater than 3.0. The pH value of 2.0 also
offered sufficient separation, but it was at the lower boundary of the recommended operating pH range (2.0–4.5) of the Primesep SB column and outside of the buffering capacity of ammonium formate buffer.
3.3. Organic solvent percentage study Percentage of the organic additive in the mobile phase could be used to manipulate retention time, while keeping the buffer concentration constant. A previous study has demonstrated separation of sugar-phosphates (including frucose-6-phosphate, ribose-6phosphate, etc.) utilizing ammonium formate buffer, acetonitrile, and the Primesep SB column [23]. It was observed that changing acetonitrile concentration had a marginal impact in comparison to changes in buffer concentration.
Fig. 5. Analytical separation of the reaction mixture containing TFMF-1-P, GDP-TFMF, ADP, GDP, ATP and GTP*. (a) UV at 254 nm. (b) MS ESI negative mode. (c) Extracted ion chromatogram (EIC) of TFMF-1-P.(d) EIC of GDP-TFMF. Column: Primesep SB (4.6 × 150 mm, 5 m). Mobile phase: 50 mM NH4 HCO2 (pH 2.7)/CH3 CN (85:15). Flow-rate: 1.0 mL/min. 10 L injection. *GTP not included in figure due to excessive retention time.
H. Eastwood et al. / Journal of Pharmaceutical and Biomedical Analysis 115 (2015) 402–409
407
Fig. 6. Semi-preparative purification of the reaction mixture with modified step gradient. Column: Primesep SB (21.2 × 150 mm, 5 m). Mobile Phase A: 50 mM NH4 HCO2 (pH 2.7)/ CH3 CN (90:10). Mobile Phase B: 100 mM NH4 HCO2 (pH 2.7)/CH3 CN (85:15). Step gradient at 100% A from 0 to 24 min, then switched to 100% B and held for 16 min. Total run time: 40 min. Flow-rate: 20 mL/min. 900 L injection. UV at 254 nm.
For our study, changing the acetonitrile percentage had less of an impact on the retention of fucose-1-P as noted in the aforementioned study. However, GDP, GTP, and GDP-fucose had a more pronounced reduction in retention time, possibly due to their more hydrophobic nature, as shown in Fig. 4. Acetonitrile percentage adjustment was useful in fine tuning the method for the preferred
retention times. A range of 10–15% acetonitrile was explored in the mobile phase to facilitate a shortened total run time and maintain adequate separation between GDP-fucose and fucose-1-P. Since acetonitrile provided desired results in separating the GDP-TFMF and TFMF-1-P from other analytes, which will be explained in the following sections, additional organic solvents were not explored.
Fig. 7. 19 F NMR for purity assessment of GDP-TFMF. (a) 500 M internal 19 F NMR standard, N-trifluoroacetyl-d-glucosamine. (b) 500 M internal 19 F NMR standard spiked in and peaks integrated and set as a reference integral. The sum of the GDP-TFMF peak area integrals in both pyranose and furanose forms was compared to the standard integrals.
408
H. Eastwood et al. / Journal of Pharmaceutical and Biomedical Analysis 115 (2015) 402–409
3.4. Final conditions for analytical HPLC-MS and semi-preparative methods After studying buffer concentration, pH and acetonitrile percentage, the final conditions were determined. The mobile phase, consisting of 50 mM ammonium formate at pH 2.7 with 15% acetonitrile, was successful at separating GTP, ATP, GDP, ADP, GDPfucose and fucose-1-P standards using the Primesep SB column. Four of the standards (GDP, ADP, GDP-fucose and fucose-1-P) were separated within 15 min. Negative mode ES-API MS detection was necessary to detect fucose-1-P due to the lack of a chromophore. The same HPLC–MS method was then used to analyze the reaction mixture containing TFMF-1-P and GDP-TFMF, as shown in Fig. 5. This method successfully separated these two compounds from the other components (unreacted TFMF, ADP, GDP and ATP) in the reaction mixture. Using the same mobile phase composition as the analytical method, a semi-preparative separation method was initially developed. However, GDP and GDP-TFMF closely eluted, which would make it difficult to collect GDP-TFMF (data not shown). To further separate these close eluting analytes, the acetonitrile percentage was adjusted from 15% to 10%. Since ATP and GTP eluted much later using 10% acetonitrile, a step gradient was applied to decrease the retention time. After 24 min, mobile phase was changed to 100 mM ammonium formate with 15% acetonitrile. The resultant chromatogram is shown in Fig. 6, and method details are explained in the legend. Using the step gradient semi-preparative method, the overall run time was decreased from ∼120 min to 40 min. This method was successfully applied to purify the TFMF-1-P and GDPTFMF from the reaction mixture.
3.5. Purity assessment using 19 F NMR After isolating the desired compound, GDP-TFMF, purity determination was necessary. Since there was no pure authentic compound available, 19 F NMR was applied using an internal standard, N-trifluoroacetyl-d-glucosamine, to assess the purity of the final product. First, a 19 F spectrum of the standard compound was acquired (Fig. 7A). Next, a solution containing 500 M of the internal standard and ∼500 M of the purified GDP-TFMF was prepared. Assignments of the pyranose and furanose forms of the sugar were made based on previous experimental data (primarily 1 H NMR, data not shown). 19 F NMR spectra were obtained, and the signals for both the standard and compound were integrated based on peak area (Fig. 7B). The concentration of GDP-TFMF was determined by comparing integral values for signals from both compounds. The purity of GDP-TFMF was determined to be approximately 99%, with a yield of 30 mg.
4. Conclusion Successful analytical and semi-preparative HPLC methods were developed to resolve TFMF-1-P and GDP-TFMF from an enzymatic reaction mixture containing GTP, ATP, GDP, ADP, GDP-fucose and fucose-1-P within a reasonable run time, after determining concentration and pH of the ammonium formate buffer and the percentage of acetonitrile. Ammonium formate buffer was necessary as it allowed for the use of MS ES (-) to detect the sugar-phosphate, TFMF-1-P. Purification was performed to collect the nucleotide sugar (GDP-TFMF). The purity was determined utilizing 19 F NMR. Based on the experimental results and the application of the mixed mode column with volatile organic buffer, this technique can be utilized to separate similar sugar-phosphates and nucleotide sugars. This method is not only analytically useful but also appli-
cable to the preparative chromatography scale, allowing for easy recovery of the purified material with minimal salt content. Acknowledgment The authors would like to acknowledge Randy Jensen for his assistance with some of the NMR experiments conducted in this research. References [1] J. Rabina, M. Maki, E. Savilahti, N. Jarinen, L. Penttila, R. Renknon, Analysis of nucleotide sugars from cell lysates by ion-pair solid-phase extraction and reversed-phase high-performance liquid chromatography, Glycoconj. J. 18 (2001) 799. [2] M. Pabst, J. Grass, R. Fischl, R. Leonard, C. Jin, G. Hinterkorner, N. Borth, F. Altmann, Nucleotide and nucleotide sugar analysis by liquid chromatography–electrospray ionization–mass spectrometry on surface–conditioned porous graphitic carbon, Anal. Chem. 82 (2010) 9782. [3] N. Okeley, S. Alley, M. Anderson, T. Boursalian, P. Burke, K. Emmerton, S. Jeffrey, K. Klussman, C. Law, D. Sussman, B. Toki, L. Westendorf, W. Zeng, X. Zhang, D. Benjamin, P. Senter, Development of orally active inhibitors of protein and cellular fucosylation, PNAS 110 (2013) 5404. [4] J. Martin Buescher, S. Moco, U. Sauer, N. Zamboni, Ultrahigh performance liquid chromatography–tandem mass spectrometry method for fast and robust quantification of anionic and aromatic metabolites, Anal. Chem. 82 (2010) 4403. [5] K. Nakajima, S. Kitazume, T. Angata, R. Fujinawa, K. Ohtsubo, E. Miyoshi, N. Taniguchi, Simultaneous determination of nucleotide sugars with ion-pair reversed-phase HPLC, Glycobiology 20 (2010) 865. [6] A. Toernkvist, K. Markides, L. Nyholm, Chromatographic behaviour of oxidised porous graphitic carbon columns, Analyst 128 (2003) 844. [7] N. Padivitage, M. Dissanayake, D. Armstrong, Separation of nucleotides by hydrophilic interaction chromatography using the FRULIC-N column, Anal. Bioanal. Chem. 405 (2013) 8837. [8] H. Hinterwirth, M. Lammerhofer, B. Preinerstorfer, A. Gargano, R. Reischl, W. Bicker, O. Trapp, L. Brecker, W. Lindner, Selectivity issues in targeted metabolomics: separation of phosphorylated carbohydrate isomers by mixed-mode hydrophilic interaction/weak anion exchange chromatography, J. Sep. Sci. 33 (2010) 3273. [9] J. Ito, T. Herter, E. Baidoo, J. Lao, M. Vega-Sanchez, A. Smith-Moritz, P. Adams, J. Keasling, B. Usadel, C. Petzold, J. Heazlewood, Analysis of plant nucleotide sugars by hydrophilic interaction liquid chromatography and tandem mass spectrometry, Anal. Biochem. 448 (2014) 14. [10] L. Nunes, E.S. Martinez-Barajas, R. Powers, P. Sagar, B. Davis, M. Paul, Inhibition of SnRK1 by metabolites: tissue-dependent effects and cooperative inhibition by glucose 1-phosphate in combination with trehalose 6-phosphate, Plant Physiol.Biochem. 63 (2013) 89. [11] V. Orlovsky, Y. Zelechonok, Evolution of Mixed-Mode Chromatography Chromatography Today, (Aug/Sept 2011), pp. 24. [12] J.S.O. McCullagh, D. Juchelka, R. Hedges, Analysis of amino acid 13C abundance from human and faunal bone collagen using liquid chromatography/isotope ratio mass spectrometry, Rapid Commun. Mass Spectrom. 20 (2006) 2761. [13] R. Bischoff, L. McLaughlin, Nucleic acid resolution by mixed-mode chromatography, J. Chromatogr. 296 (1984) 329. [14] W.S. Hancock, Use of Mixed-Mode, High-performance liquid chromatography for the separation of peptide and protein mixtures, J. Chromatogr. 206 (1981) 71. [15] M. Davies, A study of retention and overloading of basic compounds with mixed-mode reversed-phase/cation-exchange columns in high performance liquid chromatography, J. Chromatogr. A 1138 (2007) 65. [16] J. Li, S. Shao, M. Jaworsky, P. Kurtulik, Simultaneous determination of cations, zwitterions and neutral compounds using mixed-mode reversed-phase and cation-exchange high-performance liquid chromatography, J. Chromatogr. A 1185 (2008) 185. [17] H. Licea-Perez, S. Wang, C. Bowen, Development of a sensitive and selective LC-MS/MS method for the determination of &␣lph␣;-fluoro-&et␣;-alanine, 5-fluorouracil and capecitabine in human plasma, J. Chromatogr. B 877 (2009) 1040. [18] M. Rance, S. Wilson, &et␣;-Diketones: peak shape challenges and the use of mixed-mode high-performance liquid chromatography methodology to obtain fast, high resolution chromatography, J. Chromatogr. A 1217 (2010) 1912. [19] A. Miller, H. Brown, T. Degg, K. Allen, B. Keevil, Measurement of plasma 5-hydroxyindole acetic acid by liquid chromatography tandem mass spectrometry—comparison with HPLC methodology, J. Chromatogr. B 878 (2010) 695. [20] Sielc. http://www.sielc.com/Products Primesep SB.html [21] X. Liu, J. Fang, N. Cauchon, P. Zhou, Direct stability-indicating method development and validation for analysis of etidronate disodium using a mixed-mode column and charged aerosol detector, J. Pharm. Biomed. Anal. 46 (2008) 639.
H. Eastwood et al. / Journal of Pharmaceutical and Biomedical Analysis 115 (2015) 402–409 [22] W. Wang, T. Hu, P. Frantom, T. Zheng, B. Gerwe, D. Soriano del Amo, S. Garret, R. Seidel III, P. Wu, Chemoenzymatic synthesis of GDP-L-fucose and the Lewis X glycan derivatives, PNAS 106 (2009) 16096. [23] J. Cruz, C. Emery, M. Wuest, D. Kramer, B. Lange, Metabolite profiling of Calvin cycle intermediates by HPLC–MS using mixed-mode stationary phases, Plant J. 55 (2008) 1047.
409
[24] J. Ding, C. Sorensen, Q. Zhang, H. Jiang, N. Jaitly, E. Livesay, Y. Shen, R. Smith, T. Metz, Capillary LC coupled with high-mass measurement accuracy mass spectrometry for metabolic profiling, Anal. Chem. 79 (2007) 6081. [25] A. Stalcup, K. Gahm, A sulfated cyclodextrin chiral stationary phase for high-performance liquid chromatography, Anal. Chem. 68 (1996) 1369.