Chemical Recognition of Oxidation-Specific Epitopes in Low-Density

Article pubs.acs.org/ac

Chemical Recognition of Oxidation-Specific Epitopes in Low-Density Lipoproteins by a Nanoparticle Based Concept for Trapping, Enrichment, and Liquid Chromatography−Tandem Mass Spectrometry Analysis of Oxidative Stress Biomarkers Elisabeth Haller,† Gerald Stübiger,‡ Daniel Lafitte,§ Wolfgang Lindner,† and Michael Lam ̈ merhofer*,∥ †

Department of Analytical Chemistry, University of Vienna, Währingerstrasse 38, 1090 Vienna, Austria Center of Physiology and Pharmacology, Medical University of Vienna, Schwarzspanierstraße 17, 1090 Vienna, Austria § Faculté de Pharmacie de Marseille, Université de Marseille, 27 Boulevard Jean Moulin, CS 30064−13385 Marseille, Cedex 5 France ∥ Institute of Pharmaceutical Sciences, Pharmaceutical (Bio)Analysis, University of Tübingen, Auf der Morgenstelle 8, 72076 Tübingen, Germany ‡

S Supporting Information *

ABSTRACT: Monitoring bioactive oxidized phospholipids (OxPLs), such as 1-palmitoyl-2-(5′-oxovaleroyl)-sn-glycero-3-phosphocholine (POVPC) and 1-palmitoyl-2-(9′-oxononanoyl)-sn-glycero-3-phosphocholine (PONPC), is of major interest as they play a crucial role in a variety of age related diseases, e.g., in the development and progression of atherosclerosis. Since they are in low abundance in samples like oxidized low-density lipoproteins (OxLDL) and human plasma, respectively, their analysis as risk biomarkers requires the combination of an efficient selective sample preparation with highly sensitive detection methods, such as liquid chromatography−electrospray ionization-tandem mass spectrometry (LC−ESI-MS/MS). In this study, a nanoparticle-based strategy for successful trapping and enrichment of aldehyde-containing oxidized phospholipids is presented. The concept involves a derivatization step with a bifunctional reagent containing both a hydrazide group for hydrazone formation with carbonyl-containing PLs and a thiol moiety for subsequent trapping on GNPs. After washing, the trapped analytes are quantitatively released from the nanoparticles’ surface by transimination with hydroxylamine. The released oxime-derivatives of the carbonylated-OxPLs are subsequently analyzed by LC−ESI-MS/MS in the selected reaction monitoring scan mode. Several parameters of this workflow were optimized. With the optimized nanoparticle-based extraction and enrichment step, very clean extracts of these biomarkers can be obtained and the detection limits can be significantly decreased from 2.76 and 2.65 nM for PONPC and POVPC, respectively, to 0.17 and 0.44 nM. The applicability of this nanoparticle-based sample preparation concept was demonstrated by successful extraction of oxidized phospholipids from biological samples, such as human plasma, MDAmodified LDL and Cu2+-oxidized LDL.

O

Human LDL mainly consists of phospholipids (PLs), cholesterol esters, triacylglycerols, and apolipoprotein B. Almost 80% of the mass of LDL is constituted by lipids whereby more than half of the lipid molecules contain polyunsaturated fatty acids (PUFAs) such as arachidonic and linoleic acid.19−21 They possess a high susceptibility to lipid peroxidation and in fact the detrimental biological activity of OxLDL primarily results from oxidation of 1-palmitoyl-2arachidonyl-sn-glycero-3-phosphocholine (PAPC) and 1-palmitoyl-2-linoleoyl-sn-glycero-3-phosphocholine (PLPC).22 Thereby, a series of structurally defined oxidation products are

xidative stress is involved in a variety of pathological conditions such as inflammation, atherosclerosis, and cancer.1 In the genesis of atherosclerosis, reactive oxygen species (ROS) lead to oxidative modifications of low-density lipoproteins (LDLs). Resultant chemical modifications comprise carbonylated apolipoproteins and oxidized phospholipids (OxPLs). Oxidized LDL (OxLDL) plays a significant role in the initiation and progression of atherosclerosis.2−13 According to the EPIC Norfolk study, an increased risk of development of atherosclerosis was associated with elevated levels of circulating oxidized LDL in human plasma.9,14−18 This study clearly emphasizes the potential utility of OxLDLs and OxPLs in plasma as diagnostic risk biomarkers for cardiovascular diseases (CVDs).9 © 2014 American Chemical Society

Received: July 30, 2014 Accepted: September 14, 2014 Published: September 15, 2014 9954

dx.doi.org/10.1021/ac502855n | Anal. Chem. 2014, 86, 9954−9961

Analytical Chemistry

Article

reversed-phase liquid chromatography for various (Ox)PL species, presence of several isobaric lipids exhibiting identical fragmentation pattern by releasing the PC headgroup (m/z 184.1), and significant ion suppression of target analytes in the ESI process in MS detection due to coeluted phospholipids. To overcome some of those problems, we herein propose a novel highly selective nanoparticle-based extraction of carbonylated OxPLs. It makes use of a selective derivatization reaction employing a bifunctional cross-linker having a hydrazide group for efficient reaction with carbonyl functionalities of analytes and a terminal thiol group (see Figure 2) for binding on gold

generated, such as 1-palmitol-2-(5′-oxovaleroyl)-sn-glycero-3phosphocholine (POVPC) and 1-palmitoyl-2-(9′-oxononanoyl)-sn-glycero-3-phosphocholine (PONPC) (see Figure 1) as

Figure 1. Concept of the nanoparticle-based extraction approach for carbonylated phospholipids, whereby n = 5 corresponds to POVPC and n = 9 to PONPC.

well as 1-palmitoyl-(2-glutaroyl)-sn-glycero-3-phosphocholine (PGPC) and 1-palmitoyl-2-(5,6-epoxy-isoprostane)-sn-glycero3-phosphocholine (PEIPC), which all represent major oxidation-specific epitopes of OxLDL.3,6,9,12,23,24 They exhibit a variety of pro-inflammatory and pro-atherogenic properties8 and are recognized by the innate immune system in humans.25 For example, the levels of POVPC have been found to be significantly increased in atherosclerotic lesions of rabbits.26 Recently, Ravandi et al. reported that the majority of identified phospholipids in OxLDL consists of nonoxidized PLs and only about 5−8% of the total phospholipid content is represented by oxidized phosphocholines (OxPCs).8 Accurate analysis of low-abundance oxidized phospholipids from complex biological samples is a challenging task. Commonly employed sample preparation techniques to enrich OxPLs of interest such as solid-phase extraction (SPE) face the problems of nonspecific trapping of lipids, sample losses, and risk of artificial oxidation. For example, ZrO2-coated silica particles (20 μm, 120 Å) have been used for the enrichment of non-oxidized PCs from biological matrixes based on Lewis acid−base interactions. However, this approach turned out to be inappropriate for oxidized phospholipids due to the harsh conditions employed.27 Thus, these procedures are not sufficiently sensitive and reliable for the analysis of very low abundance oxidized PL species in biological samples. The low concentrations of oxidized phospholipids make a specific enrichment of the target analytes for a successful quantification necessary. For carbonylated target analytes, carbonyl tagging has been proposed recently. One viable strategy for analysis of oxidative stress markers with a carbonyl moiety described for proteins is derivatization with 2,4-dinitrophenyl hydrazine (DNPH) and subsequent detection with antibodies directed against the DNP moiety.28,29 Also nanoparticle-based carbonyl traps have been recently reported for glycosphingolipids30 and by G. Stübiger et al. for carbonylated PLs.31 The former employed aminoxy-modified nanoparticles and the latter a derivatization of carbonylated PLs with ion-exchange tags and subsequent trapping on ion-exchange nanoparticles. Aside of the low abundance, further challenges of accurate analysis of carbonylated OxPLs in complex biological matrixes by electrospray ionization−tandem mass spectrometry hyphenated with liquid chromatography (LC−ESI-MS/MS) are the large excess of native over-oxidized species, poor selectivity of

Figure 2. Molecular structures of derivatizing agents.

nanoparticles (GNPs) for trapping of the target analytes POVPC and PONPC (see Figure 1). GNPs are ideal substrates for trapping due to their straightforward preparation and unique physicochemical properties and because of the high affinity of gold toward thiols (∼47 kcal/mol for the Au−S bond32) which can be exploited for thiol capture by selfassembling monolayer (SAM) formation. Practical utility and advantages of nanoparticles for sample preparation have already been demonstrated,30,33,34 also in combination with matrix-assisted laser desorption/ionization mass spectrometry (MALDI-MS),35−40 and LC−ESI-MS/ MS.41 Thereby, noncovalently bonded analytes are eluted from the nanoparticles after repetitive washing steps which may be associated with analyte losses. In the current concept, analytes are immobilized on the functionalized GNPs via dative bond and hydrazone formation (see Figure 1). It was supposed to give rise to clean extract and quantitative recoveries. However, dedicated release strategies had to be examined targeting either the strong dative Au−S bond or the covalent but reversible hydrazone functionality (see Figure 2). The potential and applicability of the concept to serve as an efficient sample preparation strategy for the LC−ESI-MS/MS analysis of oxidized phospholipids as biomarkers for oxidative stress was tested with human plasma samples as well as with MDAmodified and Cu2+-oxidized LDL.

■

MATERIALS AND METHODS Chemicals. Gold nanoparticles (final concentration was 1.64 × 10−6 mM with a size of 28.2 ± 3.1 nm as measured by DLS) were synthesized according to the Turkevich-Frens method as described in the Supporting Information. 4Mercaptobenzoic hydrazide (PACa) was a gift from D. Lafitte, Université d’Aix-Marseille, Marseille, France. 4-Aminobenzoic hydrazide (H2N-PACa, Amino-PACa) was purchased from Sigma-Aldrich (Vienna, Austria), 3-(2-Pyridyldithio) propionyl hydrazide (PDPH) from Thermo Fisher Scientific Inc. (Rockford, IL), and N-acetyl-Cys-Ala-Leu-Gln-Asp-hydrazide 9955

dx.doi.org/10.1021/ac502855n | Anal. Chem. 2014, 86, 9954−9961

Analytical Chemistry

Article

centrifuged in order to spin down the GNPs. The resulting clear supernatant (20 μL) was analyzed by LC−ESI-MS/MS interfaced with a switching valve after chromatographic separation, and the excess of releasing agent which eluted with the void volume was diverted to waste to avoid contamination of the mass spectrometer. Analysis of PONPC and POVPC in Human Plasma, MDA-LDL, and CuLDL Standards. Human plasma was diluted 1:500 in methanol, centrifuged, and filtered (0.22 μm) and was either directly derivatized with 10 μL of PACa (5.94 mM in MeOH) or used for dilutions in plasma-spiking experiments with MDA-LDL and CuLDL. A volume of 0.25−1 μL of MDA-LDL and CuLDL (each 1 mg/mL), respectively, were derivatized with 10 μL of PACa (5.94 mM in MeOH) and allowed to react overnight at RT. Methanol or diluted plasma, respectively, was added to achieve a total volume of 100 μL. A volume of 100 μL of the derivatized extracts were added to 1 mL of centrifuged GNPs. The immobilization reaction was allowed to proceed overnight to ensure complete binding. Afterward, the modified GNPs were washed with methanol by three consecutive centrifugation and resuspension steps. Finally, 20 μL of 6 mM NH2OH in MeOH/H2O (80/20; v/v) was added to the pellet of the centrifuged particles in order to release the trapped phospholipid. The particles were again centrifuged and the supernatant analyzed by LC−ESI-MS/MS. As reference samples, extracted phospholipids of MDA-LDL, CuLDL, and spiked samples were used without derivatization and analyzed by LC−ESI-MS/MS. LC−MS/MS Method. All measurements were carried out on an Applied Biosystems (AB Sciex, Foster City, CA) 4000 QTrap mass spectrometer hyphenated with an Agilent 1200 HPLC system (Waldbronn, Germany) equipped with a vacuum degasser, binary pump, autosampler, and a thermostated column compartment. The ESI source consisted of a Turbo V ionspray with an integrated heating device to increase ionization efficiency and was operated in SRM mode. The electrospray voltage was set to +4500 V, the ion source temperature to 500 °C, and the entrance potential to 10 V. Nitrogen was used as a nebulizer, heater, and curtain gas, with the pressure set to 60, 50, and 10 psi, respectively. The dwell time for each SRM transition was 10 ms, and the pause between two consecutive SRM transitions was set to 5 ms. Chromatographic separation was carried out on a Luna 5u C8 column from Phenomenex (Aschaffenburg, Germany) (150 mm × 3.0 mm, 5.0 μm, 100 Å) applying a flow rate of 0.3 mL/ min and an injection volume of 5 μL. Mobile phases (A) Millipore water and (B) methanol each with 10 mM ammonium acetate were employed. Gradient elution was performed from 30% to 100% B in 30 min and kept at 100% methanol for a further 10 min to ensure elution of lipids from the stationary phase.27,42 Afterward, the stationary phase was reequilibrated for 10 min with 30% mobile phase B. Moreover, in-between the HPLC column and ion source of the MS instrument, a switching valve was interfaced. The switching valve was programmed to discard the first 15 min of the chromatographic separation. Thereafter, the effluent from the column was directly transferred to the MS system. Analyte detection was performed in positive SRM mode by monitoring the target pseudomolecular ion in Q1 and the phosphocholine fragment at m/z 184.1 in Q3 which is specific for phosphatidylcholines. Compound-specific MS parameters for the SRM transitions such as declustering potential (DP),

(CALQD) from piCHEM (Graz, Austria). Malondialdehydemodified low density lipoprotein (MDA-LDL) and copper (Cu2+) oxidized human LDL (CuLDL) were purchased from Cell Biolabs, Inc. (San Diego, CA). 1-Palmitoyl-2-(5oxovaleroyl)-sn-glycero-3-phosphocholine (POVPC) and 1palmitoyl-2-(9-oxononanoyl)-sn-glycero-3-phosphocholine (PONPC) were obtained from Avanti Polar Lipids, Inc. (Alabaster, AL). Derivatization of Aldehyde Containing Phospholipids. The derivatization protocol for PONPC and POVPC was optimized with regard to reagent excess required for various hydrazides such as PACa, H2N-PACa, PDPH, and CALQD (Figure 2). Thus, 0.5 μL of PONPC and POVPC (each 0.1 mM in MeOH), respectively, were derivatized by adding 49.5 μL of a 1 μM to 2 mM solution of the hydrazide corresponding to excess ratios of 1 to 2000 mol equivalents with respect to the aldehyde obtaining a final volume of 50 μL. The derivatization was carried out in methanol for 48 h under constant shaking and room temperature (note, derivatization should be performed under inert atmosphere, e.g., under argon to avoid artificial oxidation during sample preparation). Subsequently, the solutions were analyzed by LC−ESI-MS/MS monitoring the previously optimized SRM transitions (see Table S-2 in the Supporting Information). Binding of PACa and Amino-PACa on GNPs. A binding study of PACa and H2N-PACa, respectively, to gold nanoparticles was performed. Thereby, different volumes of GNPs (0.01−2 mL nanoparticle suspension corresponding to 1−3 × 10−3 nmol GNPs) were centrifuged for 15 min at 13 400 rpm and the supernatant discarded in order to remove the excess of citrate in solution. Subsequently, 100 μL of PACa and H2NPACa (each 0.21 mM in MeOH) was added to the centrifuged GNPs and immobilized overnight under constant shaking. Afterward, the nanoparticle suspension was centrifuged for 5 min at 13 400 rpm and the absorbance of the supernatant (100 μL) analyzed by UV−Vis spectroscopy at 269 nm. Furthermore, the binding study was repeated by using 0.5 μL of 0.1 mM PONPC (in MeOH) derivatized with 100 μL of PACa (0.21 mM in MeOH) as ligand and adding the derivative to different volumes of centrifuged GNPs (0.01−2 mL). Supernatants were analyzed by LC−ESI-MS/MS, which allows differentiation between product and underivatized PACa reagent. Investigation of Various Releasing Agents. A volume of 0.5 μL of 0.1 mM PONPC in MeOH was derivatized with 100 μL of 0.21 mM PACa and H2N-PACa, respectively, according to the previously optimized conditions, and quantitative conversion was verified by LC−ESI-MS/MS. Subsequently, 1 mL of gold nanoparticle suspension was centrifuged, the supernatant discarded, and the derivatized phospholipid reaction solution added. The immobilization of the derivatized phospholipid on GNPs was allowed to proceed overnight under constant shaking. Afterward, the GNP-analyte conjugates were washed by three consecutive steps of centrifugation and resuspension in 1 mL of methanol in order to remove unspecifically bound molecules. LC−ESI-MS/MS analysis of the supernatant of the last washing step was performed to ensure the complete depletion of unbound molecules and thus avoid false positive results. Subsequently, various releasing agents were studied by adding 20 μL of the reagents (6 mM in MeOH/H2O, 80/20; v/v) to the centrifuged particles. The nanoparticle suspension was subjected to ultrasonication for 1 min and was shaken overnight. Subsequently, the particles were 9956

dx.doi.org/10.1021/ac502855n | Anal. Chem. 2014, 86, 9954−9961

Analytical Chemistry

Article

hydrazone derivatives, further experiments were focused on these two cross-linkers. Trapping on GNPs. A binding study of PACa and H2NPACa on gold nanoparticles was performed in order to gain information under which conditions a complete trapping of the derivatized analyte can be achieved. In this setup, a constant quantity of free carbonyl reagent (21 nmol) was mixed with variable amounts of GNPs in the range between 1 to 3 × 10−3 nmol, which corresponds to 0.01−2 mL nanoparticle suspension of the concentration 1.64 × 10−6 mM. Figure 4A)

collision energy (CE), and cell exit potential (CXP) were optimized by direct infusion (see Table S3 in Supporting Information). Data evaluation and instrument control were carried out by Analyst software, version 1.5.2 (Applied Biosystems).

■

RESULTS AND DISCUSSION Nanoparticle-Based Derivatization-Bind-Elute Sample Preparation Concept. Initial experiments have been carried out in which the cross-linkers shown in Figure 2 have been immobilized on gold nanoparticles by SAM formation. The resultant functional GNPs would be directly suitable for trapping of the target carbonylated PLs. However, it was found that the colloidal stability of GNPs modified with the tested hydrazide reagents (e.g., PACa) was insufficient. Such derivatized particles quickly aggregated so that the hydrazide group was not accessible anymore for reaction with carbonyls. In the following we therefore focus exclusively on the approach shown in Figure 1. Derivatization Yields. A set of different cross-linking reagents (PACa, H2N-PACa, PDPH, CALQD) (for structures see Figure 2) incorporating a gold affinity molecular group such as a thiol or amine moiety for subsequent selective trapping was examined for their derivatization efficiencies using PONPC and POVPC as test compounds. Product yields are strongly influenced by structure and reactivity of reactants as well as reaction conditions. Plain methanolic reaction media have been selected due to the lipophilicity of the phospholipids and acid as well as base catalysis was avoided to preclude side product formation. For instance, acidic additives catalyze hydrolysis and acetal formation, which decreases the reaction yield. Figure 3

Figure 4. Illustration of the binding and saturation curve, respectively, of PACa and H2N-PACa on gold nanoparticles (A) and comparison of adsorption behavior of free PACa and PONPC derivatized with PACa (B) (note, the detected PONPC-derivate has two PACa molecules connected via a disulfide bond attached).

shows the binding curve obtained by plotting the amount of bound molecules in nanomoles versus the employed amount of nanomoles of gold nanoparticles (note, molarity of GNPs was determined by mass balance calculations). From these data, the required amount of gold nanoparticles with a size of 28.2 ± 3.1 nm for efficient analyte trapping as well as passivation of the surface with excess of reagent to prevent undesirable binding events was evaluated. In other words, the results reveal the amount of GNPs needed for quantitative trapping of (reacted and unreacted) reagent. Fitting a Langmuir isotherm model to the obtained data provided saturation capacities (i.e., maximal binding capacities) of 4.12 nmol for the amine compound and 21 nmol for the thiol containing ligand (i.e., quantitative trapping). Considering the surface area of the nanoparticles (A ≈ 2500 nm2 for d = 28.2 ± 3.1 nm) as well as the amount of GNPs when complete ligand binding is achieved, these values correspond to 5.3 PACa and 3.6 H2N-PACa molecules per nm2 GNP surface, which is consistent with already published data using inductively coupled plasma mass spectrometry (ICPMS) measurements for determination of the ligand coverage.43 These results clearly demonstrate the higher binding capacities of GNPs for PACa as well as the formation of a more dense and well-ordered selfassembled monolayer compared to H2N-PACa. The binding of H2N-PACa on the gold surface is based on weak electrostatic interactions between the protonated amine and the surfacebound citrate and AuCl4−/AuCl2− ions, respectively, and more strongly bound complexes such as [AuCl(NH2R)] are formed.44,45 Consequently, the orientation of the aminofunctionalized PACa molecule is suggested to be random and, thus, significantly diminishes the final surface coverage.46 Moreover, the weakly bound H2N-PACa molecules can be easily removed from the gold surface during workup procedures which decreases the analyte recovery. The length of the ligand,47 its bulkiness,48 and the terminal group has a significant impact on the surface coverage. Consequently, the same binding experiment was repeated

Figure 3. Reaction yields for PONPC with various different derivatizing agents (PACa, H2N-PACa, PDPH, CALQD) (corresponding results for POVPC are confirmative and are reported in the Supporting Information).

illustrates reaction yields achieved with PACa, H2N-PACa, CALQD, and PDPH as a function of excess of derivatizing agent. It can be seen that PACa and H2N-PACa gave much higher yields (>96.5%) than the other two reagents CALQD (up to 48%) and PDPH (up to 82%). The electron-donating group in the p-position of PACa and H2N-PACa and its mesomeric effect increase the nucleophilicity of the hydrazide group in this conjugated aromatic system and significantly enhances the reactivity of these two reagents. To achieve complete conversion, a more than 100-fold molar excess of the hydrazide reagents was necessary for PACa and H2N-PACa. Excess reagent molecules bind to the GNPs as well and help passivate the surface preventing nonspecific binding of thiols and other compounds from biological samples. Because only PACa and H2N-PACa have been capable of completely converting PONPC and POVPC into the corresponding 9957

dx.doi.org/10.1021/ac502855n | Anal. Chem. 2014, 86, 9954−9961

Analytical Chemistry

Article

Table 1. Enrichment Factors (EFs) and the Corresponding Recoveries Using PACa and H2N-PACa As Derivatization Agents and Various Different Releasing Agentsa PACa H2N-PACa a

EF recovery (%) EF recovery (%)

cysteamine

PACa

H2N-PACa

CH3ONH2

NH2OH

NaBH4

HCOOH

0.44 8.8 ± 0.7 0.26 5.2 ± 3.2

3.87 77.3 ± 10.4 1.41 28.2 ± 14.1

1.46 29.2 ± 1.8 1.07 21.3 ± 5.8

1.25 25.0 ± 3.6 0.26 5.2 ± 3.0

4.69 93.8 ± 22.8 1.65 33.1 ± 11.5

0.13 2.6 ± 3.6 0.06 1.3 ± 0.4

1.4 23.6 ± 3.6 0.19 3.85 ± 0.5

Theoretical enrichment factor of 5; experiments performed in triplicate.

reached 77% and 28% for PACa- and H2N-PACa-bonded PONPC. Except for a small amount of PONPC, the released reaction product was PONPC-(PACa)2 in the case of PACabonded analyte which could stem from both ligand exchange and transimination. Analyte release was structurally less uniform in the case of H2N-PACa-bonded analyte (see the Supporting Information, Figure S-2). Recoveries were lower with H2N-PACa as elution agent. Since it appeared that transimination was more effective than thiol−thiol ligand exchange, other agents capable for transimination reaction were examined, e.g., methoxyamine and hydroxylamine. Methoxyamine gave disappointing results, however, hydroxylamine turned out to be an efficient releasing agent. Recoveries of 94% and 33%, respectively, for PACa- and H2N-PACa-bonded PONPC could be achieved. Previously, Ansar et al. found out, that sodium borohydride can successfully remove thiolates from GNPs due to the higher binding affinity of the hydride (118 kcal/mol) than that of the thiolate (∼47 kcal/mol).32,56 Although they could demonstrate that NaBH4 can quantitatively remove thiolates, readsorption takes place. Due to readsorption processes, analysis of the released phospholipids was performed immediately after addition of NaBH4. Although this procedure enables elution of the bound ligand as PONPC alcohol, the low recovery due to readsorption and nonuniform reaction product formation significantly hampered the subsequent quantitative determination.3 Last but not least, acidic hydrolysis of the hydrazone bond and thus elution of the original analyte without tag from the GNPs was considered as strategy.57 For this purpose, formic acid was added to the nanoparticle suspension releasing the trapped phospholipid. Unfortunately, side reactions such as acetal and hemiacetal formation were observed impeding the subsequent quantitative analysis. Overall, due to the higher stability of oximes, which commonly show 102 to 103-fold higher hydrolytic stability than the analogous hydrazones, hydroxylamine turned out to be the most potent competitor and most efficient releasing agent studied. Furthermore, it turned out to be first choice because it yields a single uniform defined reaction product, namely exclusively the oxime derivative of PONPC (PONPC=NHOH), and no hydrolysis of the acyl chain at the sn-2 position was observed under the reaction conditions. Implementing a volume enrichment factor along with high recoveries ensures significant sensitivity enhancements of the proposed nanoparticle-based sample pretreatment and enrichment approach consisting of derivatization with PACa, trapping with GNPs, elution with hydroxylamine and LC-ESI-MS/MS analysis in SRM mode. Calibration, Assay Performance, and Application for Biological Samples. To verify that the developed assay can be used for quantitative analysis, calibration functions with standard solutions of PONPC and POVPC have been

with PACa-derivatized PONPC. As it can be seen in Figure 4B), the additional phospholipid residue significantly influences the binding behavior on gold nanoparticles, most probably due to steric hindrance during the reorganization of the selfassembled monolayer. Generally, long alkyl chains have been reported to be beneficial for SAM formation due to additional hydrophobic interactions.49 However, our results indicate that binding affinity is in fact lower (lower steepness of initial isotherm) but quantitative trapping can be accomplished with about the same amount of GNPs. According to these results, 1.65 pmol of gold nanoparticles is needed to completely bind 21 nmol of PACa, especially when derivatized with PONPC. Reversibility of Trapping and Optimization of Elution. The practical utility of the combined derivatization/trapping concept for sensitive analysis of carbonyl-group carrying PLs greatly depends on the success of quantitative release of the trapped analytes from the GNPs. Ligand or place exchange is one option for release and chemical reversibility of the reaction of hydrazone formation another one. Thereby, an additional mass tag is introduced (as compared to the underivatized analyte) which could be favorable for MS ionization50 or just provide a mass shift of the analyte peak toward higher m/z values in the MS spectrum avoiding interferences with employed matrixes in MALDI-TOF-MS analysis. The effectiveness of ligand exchange of thiolate molecules bound to the GNP surface is primarily determined by parameters such as structure of bound and incoming ligand and its concentration.51 Incomplete ligand substitution, readsorption, irreversible aggregation, and insolubility of new capping molecules in the nanoparticle suspension are some of the reported problems.51,52 Moreover, herein compatibility with the subsequent analysis by LC−ESI-MS/MS has to be considered as well. On the other hand, full reversibility of analyte immobilization by hydrazone formation can be accomplished by transimination reactions.53−55 Thereby, an amine containing ligand displaces the initial cross-linker and consequently releases the analyte from the gold nanoparticles. Thus, various releasing agents and strategies for elution of trapped thiol- and amino-terminated oxidized phospholipidderivatives from the gold nanoparticle surface were examined. The results are summarized in Table 1. In any case, the releasing agent was added in large excess and a volume enrichment factor of 5 was implemented. Hence, the maximally achievable enrichment factor at 100% recovery is 5. The initial idea was to use a thiol ligand exchange reaction and cysteamine was tested for this purpose. However, it turned out that recovery was poor for both PACa and H2N-PACa (90%. Only the target carbonylated PLs have been detected while no other phosphatidylcholines have been found in extracts. Moreover, this sample preparation technique offers the advantage of enrichment of low abundant carbonylated biomolecules from complex matrix, so that subsequent quantification can be performed with significantly lower detection limits (factor of >10 has been shown herein). Moreover, the entire procedure is fully compatible with mass spectrometry. The sample preparation protocol involves simple pipetting and spinning steps and thus can be readily implemented in a clinical analysis laboratory. The applicability of this nanoparticle-based extraction procedure with subsequent LC−ESI-MS/MS for biological samples was demonstrated by successfully analyzing PONPC and POVPC in human plasma, MDA-LDL, and CuLDL.

■

ASSOCIATED CONTENT

* Supporting Information S

Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*Phone: +49 7071 29 78 793. Fax: +49 7071 29 4565. E-mail: [email protected]. Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS Financial support of the “NanoMALDI” project by the Austrian BMVIT via the “Austrian Nano-Initiative” and “MNTERA.NET” (Grant No. 828701) is gratefully acknowledged.

■

REFERENCES

(1) Itabe, H. J. Clin. Biochem. Nutr. 2012, 51, 1−8. 9960

dx.doi.org/10.1021/ac502855n | Anal. Chem. 2014, 86, 9954−9961

Analytical Chemistry

Article

(28) Sellami, L.; Villard, C.; Brunel, J. M.; Lafitte, D. Patent Application WO2012152943A1, May 11, 2012. (29) Levine, R. L.; Garland, D.; Oliver, C. N.; Amici, A.; Climent, I.; Lenz, A.-G.; Ahn, B.-W.; Shaltiel, S.; Stadtman, E. R. Methods Enzymol. 1990, 186, 464. (30) Nagahori, N.; Abe, M.; Nishimura, S.-I. Biochemistry 2009, 48, 583−594. (31) Stübiger, G.; Wuczkowski, M.; Bicker, W.; Belgacem, O. Anal. Chem. 2014, 86, 6401−6409. (32) Di Felice, R.; Selloni, A. J. Chem. Phys. 2004, 120, 4906−4914. (33) El-Sayed, I. H.; Huang, X.; El-Sayed, M. A. Nano Lett. 2005, 5, 829−834. (34) Chang, C.-W.; Tseng, W.-L. Anal. Chem. 2010, 82, 2696−2702. (35) Wu, H.-P.; Yu, C.-J.; Lin, C.-Y.; Lin, Y.-H.; Tseng, W.-L. J. Am. Soc. Mass Spectrom. 2009, 20, 875−882. (36) Teng, C.-H.; Ho, K.-C.; Lin, Y.-S.; Chen, Y.-C. Anal. Chem. 2004, 76, 4337−4342. (37) Wang, M.-T.; Liu, M.-H.; Wang, C. R. C.; Chang, S. Y. J. Am. Soc. Mass Spectrom. 2009, 20, 1925−1932. (38) Su, C.-L.; Tseng, W.-L. Anal. Chem. 2007, 79, 1626−1633. (39) Son, J.; Lee, G.; Cha, S. J. Am. Soc. Mass Spectrom. 2014, 25, 891−894. (40) Stolowitz, M. L. Proteomics 2012, 12, 3438−3450. (41) Harkness, K. M.; Cliffel, D. E.; McLean, J. A. Analyst 2010, 135, 868−874. (42) Oskolkova, O. V.; Afonyushkin, T.; Preinerstorfer, B.; Bicker, W.; von Schlieffen, E.; Hainzl, E.; Demyanets, S.; Schabbauer, G.; Lindner, W.; Tselepis, A. D.; Wojta, J.; Binder, B. R.; Bochkov, V. N. J. Immunol. 2010, 185, 7706−7712. (43) Hinterwirth, H.; Kappel, S.; Waitz, T.; Prohaska, T.; Lindner, W.; Lammerhofer, M. ACS Nano 2013, 7, 1129−1136. (44) Lu, Y.; Thien Van, T.; Jiaqi, Z.; Na, L. In 2012 IEEE 6th International Conference on Nano/Molecular Medicine and Engineering (NANOMED), Bangkok, Thailand, November 4−7, 2012; pp 21−26. (45) Kumar, A.; Mandal, S.; Pasricha, R.; Mandale, A. B.; Sastry, M. Langmuir 2003, 19, 6277−6282. (46) Woehrle, G. H.; Brown, L. O.; Hutchison, J. E. J. Am. Chem. Soc. 2005, 127, 2172−2183. (47) Hadley, A.; Aikens, C. M. J. Phys. Chem. C 2010, 114, 18134− 18138. (48) Hong, R.; Fernandez, J. M.; Nakade, H.; Arvizo, R.; Emrick, T.; Rotello, V. M. Chem. Commun. 2006, 2347−2349. (49) Zhou, J.; Ralston, J.; Sedev, R.; Beattie, D. A. J. Colloid Interface Sci. 2009, 331, 251−262. (50) Guo, J.; Prokai, L. J. Proteomics 2011, 74, 2360−2369. (51) Zhao, P.; Li, N.; Astruc, D. Coord. Chem. Rev. 2013, 257, 638− 665. (52) Baranov, D.; Kadnikova, E. N. J. Mater. Chem. 2011, 21, 6152− 6157. (53) Rowan, S. J.; Cantrill, S. J.; Cousins, G. R. L.; Sanders, J. K. M.; Stoddart, J. F. Angew. Chem., Int. Ed. 2002, 41, 898−952. (54) Corbett, P. T.; Leclaire, J.; Vial, L.; West, K. R.; Wietor, J.-L.; Sanders, J. K.; Otto, S. Chem. Rev. 2006, 106, 3652−3711. (55) Barboiu, M. In Constitutional Dynamic Chemistry; Barboiu, M., Ed.; Springer: Berlin, Heidelberg, Germany, 2012; pp 33−53. (56) Ansar, S. M.; Ameer, F. S.; Hu, W.; Zou, S.; Pittman, C. U.; Zhang, D. Nano Lett. 2013, 13, 1226−1229. (57) Kalia, J.; Raines, R. T. Angew. Chem., Int. Ed. 2008, 47, 7523− 7526.

9961

dx.doi.org/10.1021/ac502855n | Anal. Chem. 2014, 86, 9954−9961