Title:

One drop chemical derivatization - DESI-MS analysis for metabolite structure identification

Abstract:

Structural elucidation of metabolite is an important part during the discovery and development process of new pharmaceutical drugs. LC-MS/MS techniques are usually the technique of choice for structural identification but can’t always provide precise structural identification of the studied metabolite (e.g., site of hydroxylation, site of glucuronidation, etc). In order to identify those metabolites, different approaches are used combined with MS data including nuclear magnetic resonance (NMR), H/D exchange and chemical derivatization followed by LC-MS. Those techniques are often time consuming and/or require extra sample pre-treatment.

In this paper a fast and easy to set up tool using DESI-MS for metabolite identification is presented.In the developed method, analytes in solution are simply dried on a glass plate with printed Teflon spots and then a single drop of derivatization mixture is added, once the spot is dried, the derivatized compound is analyzed. Six classic chemical derivatizations were adjusted to work as a one drop reaction and applied on a list of compound with relevant functional groups. Then two successive reactions on a single spot of amoxicillin were tested andthe methodology described was successfully applied on an in-vitro incubated alprazolam metabolite. All reactions and analysis were performed within an hour and gave useful structural information by derivatizing functional groups, making the method a time saving and efficient tool for metabolite identification if used in addition or in some cases as an alternative to common methods.

Introduction:

Metabolite identification is an important part during the discovery and development process of new pharmaceutical drugs. Structural elucidation of metabolites can be valuable for the optimization of lead candidates in drug discovery and provide critical information to assess the toxicology or pharmacology profile of compounds in drug development.[1-3]

Mass spectrometry (MS)coupled with chromatographic separation techniques such asultra-performance liquid chromatography (UPLC) areroutinely used for metabolite identification.[4-7]More sophisticated mass spectrometers whichallow tandem mass spectrometry (MS/MS) and multiple-stage mass spectrometry (MSn) experiments as well as high resolution (HR) instruments are extensively used.[8-12]

Usually metabolites are identified in comparison with the parent drug, however,even advanced UPLC-HRMS/MS or UPLC-HRMSn instruments cannot always provide full structural identification of the studied metabolite(e.g. site of hydroxylation, O- or N-glucuronidation).[12, 13]Therefore, different approaches are used in combination with MS data,including nuclear magnetic resonance (NMR),ion mobility spectrometry (IMS), H/D exchange and chemical derivatizationif full structural information is required and cannot be provided by MS fragmentation alone.[6,13-18]

NMR is a routinely used powerful analytical technique capable of providing unambiguous chemical structures, but it requires an important amount of purified metaboliteand may require significant measuring times. As an alternative or in addition to NMR, chemical derivatization coupled with LC-MS/MS has been proven to be a practical strategy for metabolite identification.Derivatization provides information on specific functional groups newly formed or altered during the metabolism process and can occasionally help resolving structural changes by giving a different fragmentation pattern.[19-24]

Reactions are usually carried out at microscale on crude samplesprior to analysis and can be coupled with on-line H/D exchange to facilitate structural identification and interpretation of MS/MS data.[17]However, derivatization can betime consuming and the derivatized mixture often needs additional sample pre-treatment prior to separation by liquid chromatography.

In this paper,an easy to set upand time saving method coupling derivatization and DESI-MS analysis is presented as an additional toolformetabolite identification. DESI (desorption electrospray ionization) is an ambient ionization technique that wasfirst introduced by Cooks in 2004.[25] In this technique very similar to ESI,an electrospray is simply directed onto the surface of interest under ambient conditions and extractsanalytes from the surface by a “droplet pick-up” mechanism into the MS via an extended inlet. Further details aboutthe principle of DESIcan be foundin the literature.[25, 26]In the pharmaceutical industry DESI-MS is mainly used for drug and metabolite imaging in tissues and drug tablet screening with the advantage of working under atmospheric conditions with minimal sample preparation.[26, 27] Another common application, reactive DESIis used to perform online derivatization by the addition of a reagent in the spraying solvent.[26] Although reactive DESI can be very useful for imagingand screening, the reactions that can be performed in this way are limited and not always useful for structural identification. Moreover, the spraying solvent needs to be changed for every new reaction. Therefore, reactive DESI is not suitable for the experiments conducted in this paper.

In the developed method,a drop of analyte in solution is simply dried on a glass plate with printed Teflon spots, after which a single drop of derivatization mixture is added. Once the spot has dried, the derivatized compound is analyzed by DESI-MS and the observed mass shifts induced by the derivatizationallow the analyst to deduce the presence or absence of the investigated functional group(s).

To optimize the derivatization procedure,a list of compounds with relevant functional groups (Fig. 1) was first selected and used to adjust reactionparameters for sixclassicalchemical derivatizationstrategiesthat can be executed as “one-drop” reactions for the derivatization of carboxylic acid, primary and secondary amine, alcohol, ketone and aldehyde functional groups. Nextamoxicillin, a test compound chosen for its variety of functional groups (amide, primary amine, carboxylic acid and phenol), was selected to perform successive derivatizations on a single spot. Finally,the derivatization of a metabolite obtained bymicrosomalincubation of alprazolam was performed as an example for drug metabolism studies.

Material and methods:

Chemicals and reagents:

Methanol, ethanol, acetonitrile and formic acid, all analytical grade, were purchased from Merck (Darmstadt, Germany). 3-amino-1-phenyl-propan-1-ol, 4-androstene-3,17-dione, 5-hydroxy-L-tryptophan, 4-hydroxy-tolbutamide, 2-piperazinoethylamine, 2,4,6-triphenylpyrylium tetrafluoroborate, α-[2-(methylamino)ethyl]benzyl alcohol, acetic anhydride, benzoyl chloride, carbobenzyloxy-N-epsilon-tosyl-L-lysine, diclofenac sodium salt, hydroxylamine hydrochloride, N-phenyl-p-phenylenediamine, paracetamol, picolinic acid N-oxide, progesterone, pyridine, (S)-(+)-1-methyl-3-phenylpropylamine, triethylamine andtrimethylsilyldiazomethane 2M in diethylether were purchased from Sigma-Aldrich. Ultrapure water was produced with a MilliQ system (Millipore, Billerica, MA, USA).

Sample preparation:

Amoxicillin was dissolved in 1% of formic acid in water, the other compounds shown in Fig.1 weredissolved in methanol/water (95/5; v/v)at a concentration of 5mM.Prior to derivatization,2µLof compound was depositedon a glass plate spot (micro-24TM glass slide, Prosolia, inc., Indianapolis, IN, US)using a 10 µL syringe (Hamilton, Bonaduz AG, Bonaduz, Switzerland) and allowed to dry in a fume hood.All chemical derivatizations were carried out by adding 2µL(hereinafter referred as a drop) of reagent mixture on a spot using a syringe and then allowed to dry again prior to analysis, reducing sample preparation to two steps of a few minutes.A detailed flowchart summarizing the method is presented in Fig.2.

Derivatization reactions:

Acetylation was performed by adding a drop of pyridine/acetic anhydride (80/20, v/v) for the derivatization of amine and alcohol functional groups or a drop of water (pH=3 adjusted with formic acid)/acetic anhydride (80/20, v/v)) for the derivatization of amine functional groups only.

Methylationwas performed by adding a drop of trimethylsilyldiazomethane (2M in diethylether diluted 1/25 (v/v) in methanol).

Benzoylation was performed by adding a drop of acetonitrile/benzoyle chloride (80/20, v/v).

Oximationwas performedby adding a drop of aqueous hydroxylamine hydrochloride solution (0.1 M).

Pyridination reagent was prepared as follows: 2.0 mg of 2,4,6-triphenylpyrylium tetrafluoroborate was dissolved in ethanol/triethylamine, 92.7/7.3 v/v. The derivatization was performed by adding a drop of the reagent mixture.

Successive derivatizations:

DESI-MS:

According to previously described procedure, a drop of amoxicillin in acidified water was depositedon a Teflon spot and allowed to dry for 15 min. Then subjected to methylation (5min drying time) and acetylation (10 min drying time) successively.

Classical derivatization approach:

A head to head comparison was made with a classicalderivatizationapproach in solution in order to compare time and reagent consumption.[24, 28] In the latter method100 µL of amoxicillin in acidified water was first dried under nitrogen gas and reconstituted in 100 µL of DMSO. 50 µL of methanol containing 1.4 equivalent of trimethylsilyldiazomethane was added. After standing at room temperature for 30 min the mixture was mixed with 100 µL of water and injected onto an UPLC system to check if the reaction was successful.The first 20 seconds of the short LC run were splitted to waste in order to avoid source contamination of the MS system with the reagent. 2µ L were injected on an Acquity UPLC system (Waters Corporation, Milford, MA, USA) equipped with an AcquityUPLC BEH C18 (1.7µm) 50 x 2.1 mm ID column (Waters Corporation, Wexford, Ireland). 0.1% formic acid in water (solvent A) and 0.1% formic acid in acetonitrile (solvent B) were used as eluent with a gradient from 5% B to 100% B in 5 min, held at 100% B for 1 min and then back to 5% B in 1 min.For the second reaction, water and methanol was removed from an aliquot of the first reaction mixture by a nitrogen gas flow and 50 µL of pyridine/acetic anhydride (1:1 v/v) was added. After 15 min reaction at room temperature the mixture was mixed with 100 µL of water and injected to detect the additional acetylation.

Alprazolam Incubation and fraction collection:

Incubations were performed in triplicate at 37°C in a shaking water bath for 20 min. The incubates of 1mL final volume contained 0.2M of potassium phosphate buffer (pH=7.4), 10pmol/mL of recombinant CYP3A4, 20µM of alprazolam and an NADPH generating system consisting of 0.5mg of glucose-6-phosphate, 0.358µL of glucose-6-phosphate dehydrogenase, 5µg of MgCl2.6H2O (100 mg.mL) and 0.125mg of NADP. After incubation, the reaction was stopped by adding 1mL of DMSO in the enzyme mixture. Incubates were pooledtogether and centrifuged at 10,000g for 10 min, then the supernatant was directly injected on the HPLC system.

4-hydroxyalprazolam was collected using an Xterra RP18 (5µm) 250x4.6 mm ID column(WatersCorporation, Milford, MA, USA) after a single injection of 250 µL on an Accela UHPLC system (Thermo Electron,Bremen, Germany). A 0.1% formic acid in water (solvent A) and 0.1% formic acid in acetonitrile/methanol (50/50, v/v, solvent B) gradient was used to isolate the 4-hydroxyalprazolam metabolite. The gradient was 20% B for 1 min, to 75% B in 9 min, to 100% B in 4 min, held at 100% B for 1 min, then back to 20% B in 1 min and held at 20% B for 2 min at a constant flow of 1 mL/min-1.The peak of 4-hydroxyalprazolam was collected at 11.88 min and used without further treatment. One drop of the 400 µL collected was deposited on a Teflon spot and then acetylated with acetic anhydride in pyridine. The spot of 4-hydroxyalprazolam was analyzed by DESI-MS before and after derivatization.

Mass spectrometry:

All experiments were carried out on a hybrid quadrupole-orbitrap mass spectrometer (Q-ExactiveTM, Thermo Fisher Scientific inc., Bremen, Germany). The Q-Exactive instrument was operated in positive and negative ion mode and the data obtained were processed by the instrument software (Xcalibur version 2.2). The ion injection time and the number of microscans were set to 200ms and 4, respectively. Full-MS scans were acquired with a mass resolution of 70 000 at m/z 200.

A commercial DESI 2D source (OmniSpray 2D Source, Prosolia, inc., Indianapolis, IN, US) was used for all experiments. The spray solvent (0.1% of formic acid in methanol/water, 80/20, v/v) was directed onto the surface at an angle of 55° with respect to the glass plate. DESI source conditions in positive and negative ion mode were as follows: spray voltage: 2.0 kV, capillary temperature: 250°C, flow rate: 3µL.min-1.

Results and discussion:

Derivatization reactions:

Results from the derivatization reactions of the drugs presented in Fig. 1 are summarized in Table 1. Acetylation performed in pyridine derivatizesprimary and secondary aminesand phenol and aliphatic alcohol functional groups, whereas acetylation in water only derivatizesprimary and secondary aminesand phenol functional groups. Therefore 3-amino-1-phenyl-propan-1-ol, α-[2-(methylamino)ethyl]benzylalcohol and 4-hydroxy-tolbutamide only show a mass shift of +42 u in water and a mass shift +84 u in pyridine corresponding with one and two acetylations, respectively. In the case of 4-hydroxy-tolbutamide,the sulfonamide nitrogen of the sulfonylurea is reactive and subject to acetylation, but also to methylationand benzoylation.[29] This is rather exceptional since amide bonds are normallyresistant to derivatization. Other compounds with only amine and phenol functional groups (5-hydroxy-L-tryptophan, (S)-(+)-1-methyl-3-phenylpropylamine, N-phenyl-p-phenylenediamine, paracetamol and 2-piperazinoethylamine) give similar results in pyridine and acidified water. Primary and secondary amines with a delocalized lone electron pair,such as the aromatic and heterocyclic aromatic amines in diclofenac, 4-hydroxy-L-tryptophan and N-phenyl-p-phenylenediamineare, however, not derivatized which allows the reaction to discriminate them.

Benzoylation with benzoyl chloride is usually performed in a biphasic reaction medium with an aqueous phase containing a base such as sodium hydroxide.[30]Biphasic reactions can, however, not be performed in a drop size medium. Recent work reportsderivatization of phenol and aliphatic amines without alkali by mixing an equimolar quantity of benzoyl chloride and amine or phenol.[31]Therefore, the single drop reaction was done in a mix of acetonitrile and benzoyl chloride only. Benzoylation gives the same results as acetylation performed in water but is more subjected to steric effects.

The oximationderivatization successfully convertedthe carbonyl groups of 4-androstene-3,17-dione and progesterone into oximes. As an example,the DESI-MS spectrum of progesteronebefore and after oximationis presented in Fig. 4.

Classic methylation using trimethylsilyldiazomethane in methanol mainly derivatizes carboxylic acid and phenol functional groups such as thosepresent in carbobenzyloxy-N-epsilon-tosyl-L-lysine, paracetamol, diclofenac, picolinic acid N-oxide and 5-hydroxy-L-tryptophan but also generatesmethylated aliphatic alcohols, aminesand amidesas artifacts.[32] Usually thoseartifacts representedless than 2% in relative abundance of the pseudo-molecular ion or derivatized proton adduct.However, in some cases where the alcohol amine or amide is reactive, the methylated compound can be a major reaction product such as the sulfonamide nitrogen in the sulfonylurea of 4-hydroxy-tolbutamide.

Pyridination using 2,4,6-triphenylpyrylium tetrafluoroborate works selectively on primary aminesby forming a pyridiniumcationand thus enhances the sensitivity of N-phenyl-p-phenylenediamine, (S)-(+)-1-methyl-3-phenylpropylamine, 3-amino-1-phenyl-propan-1-ol, 5-hydroxy-L-tryptophan and 2-piperazinoethylamine.

It was noticed that the one-drop derivatization yields were generally lower compared to those obtained for reactions performed in solution.Complete derivatization is, however, not necessary for structural identification. In case of sensitivity issues the yield could be improved by adding a second drop of reagent. For example one drop of water/acetic anhydride derivatized the amine function of 3-amino-1-phenyl-propan-1-ol with a yield of 53.7% (yield based on signal intensity) whereas a second drop increased the yield to 79.4%.

Other reactions such as Eschweiler-Clarke methylation, alkylation using ethyl chloroformate and esterification with pentafluorobenzyl bromide were also evaluated. It was noticed that a drop of reagent usually dried within a few minutes stopping the reaction before enough material had reacted. The reduction of N-oxide by titanium (III) chloride and the oxidation of alcohol with pyridiniumchlorochromate werealso tested but such reactions performed in an uncontrolled reaction medium systematically led to erratic results.

H/D exchange, like derivatization, can be a useful tool for structural identification. Therefore, as an additional experiment, an attempt was made to obtain H/D exchange spectra by spraying deuterated solvents using the DESI-MS setup on compounds listed in Figure 1. However, mainly the proton adduct ([M+H]+) was replaced by a deuterium ([M+D]+), which do not give any structural information. The exchange of the other exchangeable hydrogens was rather incomplete or absent which might be explained by time constraints for the reaction or back-exchange with hydrogens from the atmosphere.

Successive derivatizations:

The derivatization approach described also allows performing successive derivatizations of a compound on the same spot and in a short time frame. This was illustrated by performing successive methylation and acetylation on a single amoxicillin spot. The same spot was analyzed by DESI-MS before derivatization and after each reaction.

The first reaction showed methylation of the carboxylic acid (see proton adduct at m/z380.1279 Fig. 5b),while also someof the original drug remained(see proton adduct atm/z 366.1125Fig. 5b). The second reaction resulted in the amine and phenol acetylation of the methylated carboxylic acid as main product (see proton adduct at m/z464.1496 Fig. 5c). Acetylation of the phenol and amine of the remaining amoxicillin wasvisible at m/z 450.1338.

After each analysis by DESI-MS, enough material isleft on the spot to allow another reaction to be performed, reducing the amount of compound needed if different functional groups have to be investigated. Various combinations of reactions and their order of execution can also provide extra structural information even if spectratend to be more difficult to interpret with a growing number of derivatizedfunctional groups.

The same conclusion was obtained following the classical derivatization approach described in materials and methods: the majority of the compound was methylated (m/z 380; RT=1.4 min) after reaction with trimethylsilyldiazomethane with a small fraction underivatized (m/z 366; RT=0.5 min) (Fig. 6a); the remaining free amine and phenol of the methylated (m/z 464; RT=2.3) and underivatized amoxicillin (m/z 450; RT=1.9) from the first reaction are acetylated by acetic anhydride (Fig. 6b). When comparing both approaches, the DESI-MS method has the main advantage to be very easy to perform via the application of just a few drops, but also consumes less analyte (only 2 μL) and far less reagent (4μLtotal in this experiment) and was 4 times faster to perform than the fastest classical approach that could be applied for this example. DESI-MS is, however, intrinsically less sensitive than LC-ESI-MS since only a small fraction of the compound deposited is extracted from the surface and goes into the MS system, while compounds are concentrated and purified by the LC system before entering the ESI source. [33] However, advanced structure elucidation is usually only done on more important (major) metabolites. Therefore, the sensitivity will usually be more than sufficient to perform the experiment. Since only ± 2 µL is needed for the successive reactions one can also thoroughly concentrate the compound prior to the experiment to boost the sensitivity. This was not done in the current example because it was not needed and intended to show an equal comparison between both techniques.

In-vitro example (microsomal incubation):

Finally, the methodology described was applied on a metabolite collected from an in vitro metabolism sample. A hydroxylatedmetabolite of alprazolam was obtained by liquid chromatography separation and fraction collection, followed by one drop chemical derivatization – DESI-MS analysis.