Gold Nanoparticle-Conjugated Anti-Oxidized Low-Density Lipoprotein

Jul 29, 2013 - Qun Liang, Han Liu, Tianyu Zhang, Yan Jiang, Ai-Hua Zhang. Untargeted lipidomics study of coronary artery disease by FUPLC-Q-TOF-MS...
0 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/ac

Gold Nanoparticle-Conjugated Anti-Oxidized Low-Density Lipoprotein Antibodies for Targeted Lipidomics of Oxidative Stress Biomarkers Helmut Hinterwirth,† Gerald Stübiger,‡ 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 § Institute of Pharmaceutical Sciences, University of Tübingen, Auf der Morgenstelle 8, 72076 Tübingen, Germany ‡

S Supporting Information *

ABSTRACT: Oxidized low-density lipoproteins (OxLDLs), in particular, oxidized phosphatidylcholines (OxPCs), are known to be involved in pathophysiological processes such as cardiovascular diseases and are described as potential biomarkers, for example, for atherosclerosis. In our study, we used the specific affinity of anti-OxLDL antibodies (Abs) conjugated to gold nanoparticles (GNPs) for extraction and enrichment of OxPCs via selective trapping of OxLDLs from plasma combined with the sensitive detection by liquid chromatography/tandem-mass spectrometry (LC−MS/MS). Successful bioconjugation chemistry of Abs via a bifunctional polyethylene glycol (PEG) spacer and protein G linkage, respectively, was controlled by measuring the surface plasmon resonance (SPR) spectra, size, and zeta potentials. Furthermore, the amount of Ab immobilized onto GNP via the PEG linker was determined. With the optimized immobilization chemistry, the ability and potential of the GNP-based extraction procedure was used for the determination of the dissociation constant, Kd, of the OxLDL binding to the GNP−Ab conjugate. Moreover, apparent Kd’s were determined for individual PCs and their oxidation products using the compound-specific selected reaction monitoring mode, which allows the characterization of the Ab affinity and, thus, assessment of the potential antigenicity of (Ox)PCs bound to OxLDLs. In summary, the application of GNP-based bioanalysis for selective targeting of OxLDLs and the fast and sensitive detection by LC−MS/MS offers new possibilities for targeted lipidomics in lipoproteins as well as for oxidative stress lipid biomarker screening.

T

genic properties.1,2,8−12 Specific immune receptors (e.g., CD36)13 and antibodies (Abs)14 have been described as selective agents targeting OxLDL, whereby Cu2+-oxidized LDL (CuLDL) is widely used as model to mimic OxLDL in vitro. Phosphatidylcholines (PCs) as a major group of membrane phospholipids (PLs) change their structure if oxidized and expose their head groups, whereby the change of the structural motif of oxidized PCs (OxPCs) is regarded as essential for Ab recognition.2,13,15 Nevertheless, there is still a necessity for the identification and screening of (new) biomarkers that are specific for CVD. In this context, the specificity of Abs against OxLDL can be taken as advantageous to improve the diagnosis of OxPLs and to increase our knowledge about their contribution to the antigenicity of oxidized lipoproteins in vivo. In addition, a significant challenge in targeted detection and quantitation of OxPLs represents their low levels in vivo (submicromolar levels), their structural complexity, and the diversity of (fragmented and nonfragmented) oxidation

oday, cardiovascular disease (CVD) represents a major cause of mortality worldwide. An estimated 17.3 million people died from CVD in 2008, representing 30% of all global deaths (World Health Organization, 09/2011). Next to ́ risk factors, such as hypercholesterolemia, hypertraditional ́ tension, smoking, gender, and age, both adaptive and innate immune mechanisms play a role in atherogenesis.1,2 Furthermore, studies demonstrate that even asymptomatic teenagers and young adults show a high prevalence of coronary atherosclerosis.3 Thus, the detection of specific biomarkers will imply a better possibility for early preclinical diagnosis of atherosclerosis and CVD prevention in the future. Great interest in this regard concerns biomarkers of oxidative stress, such as carbonylated and nitrosylated proteins, 8-hydroxyguanosine, and oxidized lipids.4−7 Minimally oxidized low-density lipoprotein (minimally modified LDL, mmLDL) and late forms of oxidized LDL (OxLDL) are known to have severe physiological effects in inflammatory diseases such as atherosclerosis, whereas lipids containing oxidized forms of polyunsaturated fatty acids (oxidized PUFAs, oxPUFAs) are described as potential biomarkers with pro-inflammatory and ultimately pro-athero© 2013 American Chemical Society

Received: June 14, 2013 Accepted: July 29, 2013 Published: July 29, 2013 8376

dx.doi.org/10.1021/ac401778f | Anal. Chem. 2013, 85, 8376−8384

Analytical Chemistry

Article

Scheme 1. Abs Immobilized onto Carboxylic Acid Functionalized GNPs via EDC/NHS Couplinga

a By changing the buffer from PBS to the amine-containing Tris/HCl buffer, the residual carboxylic groups were finally blocked with Tris. Thus, nonspecific binding could be reduced by suppression of ionic interactions.

products present in OxLDL. Moreover, the limited availability of suitable standards further adds to the difficulty of their reliable analysis during lipidomics studies.8,16,17 One simple method for extraction of lyso-PLs and PLs from plasma utilizes MeOH as the solvent, followed by a centrifugation step.18 Softionization techniques in mass-spectrometry (MS), such as electrospray ionization (ESI-MS) with direct infusion or coupled to liquid chromatography (LC−ESI-MS) are the number one techniques of choice in lipidomics to date.17,19−22 In addition to this, matrix-assisted laser desorption/ionization mass-spectrometry (MALDI-MS) and imaging mass spectrometry are getting more and more popular in lipidomics analysis of biological samples.23−25 Recently, we used a combined approach of MALDI-TOF MS and LC−ESI-MS/MS for the targeted screening of human clinical samples for OxPLs.26 In our current study, we combined the advantage of antibody specificity for extraction and enrichment of OxLDL with the sensitive detection and quantitation of OxPLs using LC−ESIMS/MS. Therefore, we used the benefits of gold nanoparticles (GNPs) for Ab immobilization (e.g., monoclonal anti-CuLDL Ab) for nanoimmunoprecipitation of OxLDL particles, followed by subsequent analysis of individual OxPLs of the lipoproteins. GNPs are a popular carrier for nanotrapping concepts because of their high surface-to-volume ratio as well as easy and low-cost controlled synthesis with narrow size distribution. The straightforward chemical functionalization of GNPs with thiol-containing molecules utilizes the strong dative bond, forming a self-assembling monolayer (SAM) with a high surface coverage on the gold surface.27 Furthermore, functionalized GNPs offer a broad range of bioconjugation chemistries for Ab immobilization via specific linkers28−33 or direct attachment via protein A/G.34 Stable colloidal immunoaffinity-modified nanoparticle suspensions can be pipetted like homogeneous solutions, unlike microparticulate suspensions. This facilitates miniaturization of sample preparation in microliter scales. Compared with SELDI with immobilized antibodies on the chip surface for capture, the advantage of the current analysis strategy is its feasibility of incorporation of a volume preconcentration factor,35 whereas Western blotting for analysis of protein biomarkers is time-consuming and lacks structural confirmation by MS of the current methodology.36

In our case, the biocompatible coating of GNPs with bifunctional polyethylene glycol (HS-PEG7−COOH) spacers was used for further immobilization of anti-OxLDL antibodies (GNP−PEG7−Ab) and compared with the direct attachment of Ab onto GNPs via protein G (GNP-ProtG−Ab). The nonspecific binding could be reduced for GNP−PEG7−Ab by simply changing the immobilization buffer and blocking the residual carboxylic groups with tris(hydroxymethyl)aminomethane (Tris), whereby ionic interactions can be suppressed (Scheme 1). Thereby, OxLDL was extracted (from standard solutions as well as from plasma) by the specific recognition of GNPconjugated Abs and enriched by simple centrifugation of the GNP−PEG7−Ab/antigen complex. After removal of the supernatant and washing of the functionalized GNPs, the precipitate was treated with MeOH and briefly subjected to ultrasonication. Thereby, the antibody−antigen (Ab/Ag) complex dissociates, and the GNP-immobilized Abs as well as proteins (e.g., ApoB from OxLDL) and buffer salts can be readily removed from solution by centrifugation as a pellet. On the other hand, (Ox)PLs remain dissolved in the supernatant directly available for LC−MS/MS analysis without the need of further time-consuming sample handling and enrichment steps (e.g. vacuum concentration). Furthermore, an enrichment step by volume reduction can be incorporated during the sample preparation process. Precursor ion scanning of the specific product ion at m/z 184.1 allows fast screening for the PC headgroup containing PLs (e.g., PCs or sphingomyelin (SM)) and their oxidation products. On the other hand, compoundspecific detection by selected reaction monitoring (SRM) mode allows the targeted detection and quantification of individual (Ox)PCs bound to the trapped lipoproteins. Finally, the application for analysis of OxLDL directly from human plasma demonstrates the appropriateness of this new (GNP-based) bioanalysis platform.

■

EXPERIMENTAL SECTION

Materials and Chemicals. Gold(III) chloride trihydrate (HAuCl4 ·3H2O), trisodium citrate, tris(hydroxymethyl)aminomethane (Tris), N-(3-dimethylaminopropyl)-N′-ethylcarbodiimide (EDC), N-hydroxysuccinimide (NHS), and O8377

dx.doi.org/10.1021/ac401778f | Anal. Chem. 2013, 85, 8376−8384

Analytical Chemistry

Article

Table 1. Optimized MS/MS Conditions for Compound-Specific Detection of PCs and Their Oxidation Productsa MS/MS conditions analyte

C:DB

precursor ion m/z

product ion m/z

DP(V)

CE(V)

CXP (V)

184.1 184.1

116 101

35 39

12 15

184.1 184.1 184.1 184.1

61 87 96 87

39 43 39 43

15 12 12 12

184.1 184.1 184.1 184.1 184.1

87 105 87 101 87

43 45 43 45 43

12 10 12 14 12

184.1 184.1 184.1 184.1 184.1 184.1 184.1 184.1 184.1 184.1 184.1 184.1 184.1 184.1 184.1

91 101 81 87 65 87 87 101 87 66 51 90 96 90 90

37 55 43 43 40 43 43 53 43 45 45 40 45 40 40

10 14 8 12 15 12 12 14 12 12 12 12 12 12 12

Lyso-PCs lyso-PC16 lyso-PC18

16:0 26:2

496.3 524.4

DMPC PMPC DPPC DAPC

28:0 30:0 32:0 40:0

678.6 706.5 734.7 846.7

OPPC PLPC POPC PAPC SAPC

33:1 34:2 34:1 36:4 38:4

746.6 758.6 760.6 782.7 810.6

POVPC PONPC PAZPC OPPC−OH PLPC−OH POPC−OH OPPC−OOH PLPC−OOH POPC−OOH PAPC−OH PAPC−OOH PLPC−(OOH)2 PEIPC SAPC−OOH SAPC−(OOH)2

21:0(Aldo) 25:0 (Aldo) 25:0 (Aldo) 33:1 (OH) 34:2 (OH) 34:1 (OH) 33:1 (OOH) 34:2 (OOH) 34:1 (OOH) 36:4 (OH) 36:4 (OOH) 34:2 (OOH)2

594.5 650.6 666.6 762.6 774.7 776.6 778.6 790.7 792.6 798.7 814.8 822.6 828.8 842.6 874.6

Saturated PCs

Unsaturated PCs

Oxidized PCs

a

38:4 (OOH) 38:4 (OOH)2

DP = declustering potential, CE = collision energy, CXP = cell exit potential.

obtain citrate-stabilized GNPs with average diameters of 26.2 ± 4.4 nm. Carboxypegylation was carried out with 1 μL of HSPEG7−COOH solution per milliliter of GNP suspension to form a self-assembled monolayer (SAM). Excess of unbound ligands was removed by centrifugation and resuspension. Antibodies (1 μL of 1 mg mL−1 solution) were immobilized onto 1 mL of GNP−PEG7−COOH intermediate by EDC/ NHS linkage (final concentration of EDC was 1.2 mM; of NHS, 6 mM) to obtain GNP−PEG7−Ab. Unreacted carboxylic acid groups were finally capped with 50 mM TrisHCl, pH 7.5, followed by several washing steps. For the blank control, the carboxypegylated functionalized GNPs were reacted as well with EDC and NHS for 2 h and finally blocked with 50 mM TrisHCl, pH 7.5, followed by several washing steps to give GNP−PEG7−Tris. Protein G was immobilized by adsorption onto citratestabilized GNPs. Therefore, 100 μL of protein G solution (1 mg mL−1 in 50 mM PBS, pH 7.5) was added to 1 mL of GNP suspension. Unbound protein G was removed by 2× centrifugation and resuspension in buffer. Antibodies (1 μL of 1 mg mL−1 added to 1 mL of GNP-ProtG solution) were bound via their Fc region utilizing their high affinity to protein G. After reacting overnight at 4 °C, the unbound antibodies were removed by several centrifugation steps and washed with 50 mM of Tris/HCl buffer, pH 7.5. Extraction and Enrichment of Oxidized LDL. Ten microliters of GNP-conjugated Abs was incubated overnight at

(2-carboxyethyl)-O′-(2-mercaptoethyl)heptaethylene glycol (HS-PEG7−COOH, PEG7), Tween20, and fish gelatin were all obtained from Sigma-Aldrich (Vienna, Austria). Methanol (HPLC grade, MeOH) was received from VWR (Vienna, Austria). Plasma was received from Technoclone (Vienna, Austria). Protein G was a gift from Rainer Hahn, University of Natural Resources and Life Sciences, Vienna, Austria. Anti-Cu2+ oxidized LDL-antibody (anti-CuLDL−Ab; AB3230, rabbit polyclonal antibody) was obtained from Millipore (Temecula, CA, USA), anti-Nε-(carboxymethyl)lysine antibody (antiCML−Ab; KH024, mouse monoclonal antibody, IgG2a) was from TransGenic Inc. (Kobe, Japan), anti-malondialdehyde oxidized LDL antibody (anti-MDA−Ab; ab17354, mouse monoclonal antibody, IgG2b) was from Abcam plc (Cambridge, UK), and EO6 (mouse monoclonal antibody, IgM) was from Avanti (Alabaster, AL). 3,3′5,5′-Tetramethylbenzidine (TMB) was received from Sciotec (Tulln, Austria). Cu2+ oxidized human LDL (CuLDL) and native human LDL (natLDL) were obtained from Cell Biolabs, Inc. (San Diego, CA, USA). Preparation of GNPs and Immobilization of Antibodies. GNPs were prepared as described in previous work optimized for the immobilization chemistry of proteins.31 Briefly, 50 mL of gold(III) chloride trihydrate solution in bidistilled water (final concentration was 1.14 mM) was heated under stirring until boiling and subsequently reduced with 5 mL of trisodium citrate (final concentration was 2.28 mM) to 8378

dx.doi.org/10.1021/ac401778f | Anal. Chem. 2013, 85, 8376−8384

Analytical Chemistry

Article

Figure 1. Characterization of functionalized GNPs at distinct stages of surface modification of two different batches (1 and 2) by various methods. (a) SPR spectra of citrate-stabilized GNPs (GNP (citrate)), PEG-functionalized GNPs (GNP−PEG7−COOH and GNP−PEG7−Tris, respectively), and Ab-conjugated GNP (GNP−PEG7−Ab). (b) SPR spectra show a considerable wavelength shift of the SPR absorption band λ (Amax) and a decrease in its absorbance, A, with each surface modification step. (c) With each functionalization step, the hydrodynamic diameter of the resultant modified GNPs increases as measured with DLS. (d) All derivatized GNPs showed excellent colloidal stability as a result of zeta potentials (ZP) of at least −25 mV.

room temperature with 10 μL of CuLDL suspension or plasma spiked with CuLDL. Unbound LDL was removed by washing four times, centrifugation, and resuspension in 50 mM Tris/ HCl, pH 7.5. The bound CuLDL was extracted by resuspending the pellet in 20 μL of MeOH and short ultrasonication (∼5 min) to isolate the lipids. Thereby, GNPconjugated Abs and proteins (e.g., plasma proteins and apolipoprotein from LDL) as well as buffer salts were removed by centrifugation as a pellet. Finally, the supernatant containing the (Ox)PLs was directly analyzed by LC−ESI-MS/MS without further purification. As reference samples, pure CuLDL and CuLDL spiked to plasma were directly measured after extraction with methanol before incubation with the GNPconjugated Abs. For matrix spiking experiments, plasma was diluted 1:500 either with 50 mM Tris/HCl, pH 7.5 buffer for GNP−Ab extraction or with MeOH when directly measured to remove plasma proteins from the sample. LC−MS/MS Method. LC−ESI-MS/MS experiments were performed on an Agilent 1200 series hyphenated with a 4000 QTrap LC−MS/MS system (Applied Biosystems, Foster City, CA) via electrospray ionization (ESI) interface using a Phenomenex C8-column (Luna (2), 3 μm C8, 150 × 2 mm) and a HPLC in-line filter (0.5 μm depth filter × 0.004 i.d.). The following general settings were optimized in previous work by the group:37,38 ESI voltage was set to 4300 V, the temperature of the ion source was 500 °C, and the entrance potential was 10 V. Nitrogen was used as the nebulizer, heater, and curtain gas with the pressure set to 60, 50, and 10 psi, respectively. The HPLC system was equipped with a binary gradient pump, autosampler, vacuum degasser, and a temperature-controlled

column compartment (Agilent, Waldbronn, Germany). The mobile phase consisted of methanol/Millipore water (80:20; v/v) containing 10 mM ammonium acetate (A) and methanol containing 10 mM ammonium acetate (B). The following gradient was run: 0−10 min, 55−100% B; 10−20 min, 100% B; 20−21 min, 100−55% B; 21−30 min, 55% B. The flow rate was 0.2 mL min−1, and the column temperature was set to 25 °C. The injection of 5 μL of the sample was performed with a needle wash. The scan range in the positive mode was from 300 to 1000 m/z. Phosphatidylcholines (PCs) were detected by precursor ion scan with m/z 184.1 in the positive ion mode (DP = 101 V, CE = 35 V, CXP = 15 V). Compound-specific MS/MS parameters for selected reaction monitoring (SRM) are shown in Table 1. Data analysis was performed with Analyst 1.5 software (Applied Biosystems, Foster City, CA). An overview about the analytes, their compound names and abbreviations, and their structure is given in Supporting Information Table S1 and Figure S1. Immunoassay: Determination of Kd. The dissociation constant Kd for anti-CuLDL−Ab was additionally determined with sandwich enzyme-linked immunosorbent assay (ELISA). In brief, anti-CuLDL−Ab was dotted onto microtiter plates and incubated overnight at 4 °C with CuLDL. Unbound CuLDL was removed with washing solution (50 mM Tris/HCl, pH 7.5, 0.1% (v/v) Tween20), and the residual microtiter plate was blocked with blocking buffer (50 mM Tris/HCl, pH 7.5, 0.1% (v/v) Tween20, and 2% (w/v) fish gelatin). The bound CuLDL was finally detected with anti-CuLDL−Ab and a secondary horseradish peroxidase (HRP)-linked anti-IgG−Ab. 8379

dx.doi.org/10.1021/ac401778f | Anal. Chem. 2013, 85, 8376−8384

Analytical Chemistry

Article

Figure 2. Specificity of distinct Ab−GNP conjugates for CuLDL vs natLDL and its dependence on the immobilization chemistry for different antibodies via PEG spacer (GNP−PEG7−Ab) vs direct attachment via protein G binding (GNP-ProtG−Ab). Shown are the extraction efficiencies for natLDL and CuLDL as the sum of the extracted (Ox)PCs determined with a precursor ion scan. GNP−PEG7 and GNP-ProtG (without antibody immobilized) served as negative controls, respectively. LDL represents the direct measurements of standards without nanobead extraction.

(ζ-potential, ZP) by electrophoretic mobility measurements (Figure 1). Monitoring wavelength shifts and absorbance variances of the SPR band of GNPs (wavelength maxima around 530 nm) is useful as fast quality control method for the GNP antibody immobilization process and its reproducibility (Figure 1a). With each incremental step of surface chemistry, a significant shift of the absorption band to higher wavelength and, concomitantly, a decrease in the absorbance maximum are observed, in accordance with the Mie theory (Figure 1b). Citrate-capped GNPs reveal the SPR band at 526 nm, which is shifted to 532 nm after self-assembly of the bifunctional PEG linker and to 536 nm with the subsequent step of Ab immobilization and Tris-end-capping of carboxylates, respectively, which indicates a slight increase in NP size. Furthermore, the DLS measurements confirm the increase in the size with each successive modification step (Figure 1c, Supporting Information Figure S2a). The hydrodynamic diameter as determined by DLS shows an increase from 34.3 ± 0.1 nm for citrate-stabilized GNPs to 68 ± 3 nm for carboxy-pegylated GNPs and 106 ± 7 nm for the corresponding GNP−Ab conjugates, respectively. The accurate size of supporting citratecapped GNPs was determined by transmission electron microscopy to be 26.2 ± 4.4 nm, and the concentration was calculated as cGNP = 2.04 × 10−9 M.27 All NP-suspensions showed zeta potentials lower than −25 mV, indicating satisfactory colloidal stability (Figure 1d, Supporting Information Figure S2b). It is also worthwhile mentioning that the linkage of the PEG spacers to the GNP surface is extremely stable from a chemical viewpoint. Careful washing after synthesis is sufficient to avoid PEG contamination in the LC−MS system. For extraction purposes, the binding capacity of the GNP− Ab conjugate is of prime relevance to obtain sufficient sensitivity for subsequent MS analysis. Thus, the amount of Ab immobilized via PEG spacer onto GNPs was quantified in triplicate with enzyme-linked anti-IgG−Ab. The quantification was based on a calibration curve obtained by plotting of different anti-CuLDL−Ab concentrations on a microtiter plate (see the Experimental Section). With this approach, the amount of bound Ab per GNP could be determined as 1.7 ± 0.4 Ab/GNP.

The detection was carried out with 3,3′,5,5′-tetramethylbenzidine (TMB) as substrate. After the enzyme reaction was stopped with sulfuric acid, the absorbance was recorded at a wavelength of 450 nm using an ELISA plate reader (PerkinElmer 1420 Multilabel Counter). Immunoassay: Determination of Ab Bound per GNP. For the determination of the amount of Ab bound per GNP, GNP−PEG7−Ab suspension was incubated with a HRP-labeled anti-IgG−Ab. After several washing steps, centrifugation, and resuspension of the GNP−PEG7−Abs, the bound anti-IgG− Ab-linked HRP was detected with TMB. The amount of Ab per GNP was calculated by comparison with the calibration curve obtained by plotting of different anti-CuLDL−Ab concentrations onto a microtiter plate and detection with the same secondary antibody. Size and Zeta Potential Measurements. GNPs and the conjugates were characterized at each immobilization step by measuring the size and size distribution by dynamic light scattering (DLS) and their zeta potentials (ZP) by electrophoretic mobility measurements using a Zetasizer instrument (Malvern Zetasizer Instruments Nano series, Prager Instruments). Calculation of logP Values. Log P values were calculated using ACD/logP DB (ACD/Laboratories, 7.00 Release, Product version 7.07).

■

RESULTS AND DISCUSSION Characterization of Functionalized GNPs. GNP-antibody conjugates developed herein are dedicated for selective extraction of OxLDL from plasma samples. Thus, important properties for their suitability are Ab specificity for OxLDL as the target protein over natLDL, reasonable binding capacity and sufficient colloidal stability of the functionalized GNP suspensions. To ensure these properties, various methods of nanoparticle (NP) analysis were untilized for quality control and to follow the progress of the immobilization chemistry of Abs onto GNPs. Thus, a thorough characterization of the NPs at each step of functionalization was carried out. These include measurements of surface plasmon resonance (SPR) spectra, determination of size and size distributions by dynamic light scattering (DLS) as well as characterization of zeta potentials 8380

dx.doi.org/10.1021/ac401778f | Anal. Chem. 2013, 85, 8376−8384

Analytical Chemistry

Article

Specificity of Distinct GNP-Antibody Conjugates. The effect of immobilization chemistry via PEG spacer and direct attachment via protein G binding was studied for different commercially available anti-OxLDL antibodies for the extraction of CuLDL and natLDL, respectively (Figure 2). For this purpose, the sum of the intensities for the precursor ion scan of m/z 184.1 corresponding to the PC-headgroup containing PLs in methanolic extracts was measured for both CuLDL and natLDL trapped with the distinct GNP−Ab conjugates. The direct analysis of natLDL and OxLDL for their content of PCs without enrichment by NP-based trapping reveals nearly the same PC content for both of them. A significant enrichment of natLDL and OxLDL was furnished when the two lipoprotein solutions were extracted with GNP−PEG7−COOH as a negative control (without Ab immobilized) prior to extraction of PLs with MeOH. These similar signal intensities for both kinds of LDL indicate, however, a high nonspecific binding. On the contrary, all the GNP−Ab conjugates tested clearly showed a specific binding and enrichment of the OxLDL without taking advantage of possible enrichments by volume, whereas natLDL showed solely weak adsorption. Although no significant differences in extraction efficiencies and specificities for (Ox)PCs were observed from the results shown in Figure 2, the covalent immobilization strategy of anti-CuLDL−Ab via PEG7 ligand was found to be superior over protein G binding. Therefore, this method was used as the preferred strategy for GNP−AB conjugate preparation during the further steps of our study. Reduction of Nonspecific Binding. PEG ligands are often utilized for creation of biocompatible coatings because of their protein-repellent properties from surfaces.39,40 Furthermore, the hydrophilic HS-PEG7−COOH spacer (log P = −2.07 ± 0.72) used in our experiments contributes to the exceptional colloidal stability of the resultant surface-functionalized GNPs in aqueous solution (ZP (GNP−PEG7−COOH) = −31 ± 2 mV) and was found to minimize nonspecific binding of lipophilic PLs (e.g., log P (DMPC) = 8.75 ± 0.63) to the GNP surface. However, the carboxylic acid groups of the PEG spacer on the outer GNP surface are capable of nonspecifically binding proteins via ionic interactions. Accordingly, residual carboxylic acid groups were capped with Tris after immobilization of Abs onto GNPs to reduce nonspecific binding (Scheme 1, Figure 3). The nonspecific binding was tested for HS-PEG7−COOHmodified GNPs without immobilized Abs as a negative control in comparison with the corresponding Tris-blocked GNPs. Figure 3 shows the difference in the signal intensities of precursor ion scans for PCs summed up in each case before and after blocking with Tris. A significant reduction of nonspecific binding was obtained when the residual carboxylic groups were capped. It was considerably lower than the specific binding with GNP−PEG7−Ab. It is evident that the GNP−Ab conjugate has higher binding capacity, especially in the lower concentration range, as compared with the negative control (GNP−PEG7− Tris). Generally, the total nonspecific binding on GNP− PEG7−Ab is thought to be much lower than in the case of GNP−PEG7−Tris because of additional steric hindrance for protein interaction owing to immobilized Abs on the GNP surface. Determination of GNP−PEG7−Ab Affinity and Binding Capacity. To characterize the specific binding properties of the optimized GNP−Ab conjugate, a binding assay with CuLDL was performed. Thus, a dilution series of CuLDL was, after

Figure 3. Reduction of nonspecific binding by blocking of residual carboxylic groups of GNP−PEG7−COOH with Tris (GNP−PEG7− Tris) as measured by a decrease in the summed signal intensities of the precursor ion scan for extracted phosphocholines. In contrast to GNP−PEG7−Ab, no significant CuLDL binding was observed for GNP−PEG7−Tris in the low concentration range.

incubation, extracted with GNP−PEG7−Ab, and the supernatants were discarded. After several washing steps, the bound lipids were extracted from the NP precipitate with MeOH of the same volume as the original sample (i.e., no volume enrichment factor was incorporated). All extracted samples as well as the standard solutions before extraction were measured by LC−ESI-MS/MS with precursor ion scan monitoring of PCs. The sum of the signal intensities was then plotted against the initial CuLDL concentration of the solution. Thereby, a saturation curve was obtained for the sum of the (Ox)PCs (Figure 4a, blue dashed curve). It can be seen that the slope of the saturation curve is nearly the same as for the initial CuLDL sample without extraction (Figure 4a, blue line) which indicates that OxLDL can be efficiently extracted with nearly quantitative yields in the lower concentration range. At higher concentrations, however, the GNP−Ab conjugate reaches saturation, and hence, the recovered concentration after extraction by the GNP−Ab conjugate gets lower (see Figure S3 and the Supporting Information). The same experiment was repeated with human plasma as the matrix (Figure 4a). CuLDL was spiked at the same concentration levels and analyzed by LC−ESI-MS/MS once without GNP-based extraction (just workup by addition of MeOH and analysis from the supernatant; red line) and once after GNP-based extraction and elution of (Ox)PLs with MeOH (red dashed line). It can be seen from Figure 4a that the signal intensities of the initial solutions without GNP-based extraction are significantly higher. This observation can be explained by the fact that plasma already contains a high amount of lipids (e.g. 1−2 mg mL−1 LDL), which are detected by the MS method, whereas after incubation with the GNP−Ab conjugate, only the portion of (Ox)PCs bound to the specifically extracted CuLDL is detected. Whatsoever, the isotherm of the NP-based plasma extraction shows a saturation behavior, as expected, which indicates that the GNP−Ab conjugate is functional and specific for OxLDL. The binding capacity of the GNP conjugate can be adjusted with this setup for quantitative yield by increasing the GNP−Ab conjugate concentration. If more sensitivity is required, a lower volume of MeOH for extraction of the lipids or a higher concentration of the GNP−Ab conjugates can be used. 8381

dx.doi.org/10.1021/ac401778f | Anal. Chem. 2013, 85, 8376−8384

Analytical Chemistry

Article

Figure 4. (a) Extraction efficiencies of GNP−Ab conjugate for CuLDL in standard solutions and in spiked plasma samples. Sum of precursor ion intensities before extraction (full lines) and after extraction with GNP−Ab conjugate (dashed lines) in standard solution (blue) and plasma (red). (b) Langmuir adsorption isotherm (standard solution) (B, bound (mg mL−1); F, free (mg mL−1)) and (c) Scatchard plot.

Figure 5. Apparent dissociation constants, Kd, for several PCs and their oxidation products.

MW, for CuLDL of ∼3500 kDa was used for calculation, the MW of LDL is in the range of 2.4−3.9 MDa4). It documents once more that the GNP−Ab conjugate is functional. The minor difference (less than a factor of 4) may be due to immobilization, different accessibilities of binding sites in the two distinct setups (GNP−Ab conjugate vs. sandwich ELISA), or slight differences in the experimental conditions. Furthermore, the nonlinear Scatchard plot indicates that there is still some nonspecific binding or negative cooperativity (i.e., binding of the antigen to the recognition site reduces the affinity of the antigen on the second binding site on the antibody) (Figure 4c). In general, however, it is hard to distinguish between nonspecific binding and negative cooperativity. The maximal binding capacity was determined to be RLmax = 2.45 ± 0.05 μg mL−1 (7.00 ± 0.14 × 10−10 M) of GNP−Ab solution. With a NP concentration of ∼cGNP = 2.04 × 10−9 M, the bound amount of antigen was calculated as 0.34 ± 0.01 molecules of CuLDL per GNP−PEG7−Ab conjugate. Up to now, in several studies the antigenicity of (Ox)PLs was determined by competitive immunoassays using anti-(OxLDL)

In quantitative terms, the above NP-based Ab/Ag reaction can be described by the Langmuir adsorption isotherm (eq 1): [RL] =

[RL]max [L] Kd(RL) + [L]

or

Γb =

Γmax[L] Kd(RL) + [L]

(1)

wherein [RL] is the Ab−Ag complex, [L] is the equilibrium concentration of free antigen Ag; [RL]max is the maximal binding capacity (saturation capacity); Γb and Γmax are specific properties (i.e., here, signal intensities of the precursor ion scan) for the bound species at a given ligand concentration and at saturation, respectively; and Kd(RL) is the dissociation constant. The dependence of the sum of precursor ion intensities on CuLDL concentration (Figure 4a) was used as the calibration function to determine the concentrations of the extracted (bound) and free CuLDL. Fitting eq 1 to the isotherm data (of Figure 4b) for binding of CuLDL to the GNP−Ab conjugate provides a dissociation constant, Kd, of 4.13 ± 0.18 × 10−10 M (1.45 ± 0.06 × 10−03 mg mL−1). This Kd value is on the same order of magnitude as determined by immunoassay (Kd = 1.55 ± 0.26 × 10−9 M; a molecular weight, 8382

dx.doi.org/10.1021/ac401778f | Anal. Chem. 2013, 85, 8376−8384

Analytical Chemistry

Article

(Kd = 1.55 ± 0.26 × 10−9 M), demonstrates the specificity of the GNP−Ab conjugate for OxLDL binding. The extraction efficiency based on a maximal binding capacity of 1.45 ± 0.06 μg mL−1 GNP−Ab solution (0.34 ± 0.18 CuLDL per GNP− PEG7−Ab) was found sufficient for direct LC−ESI-MS/MS analysis of the bound OxPLs. In addition, the determination of apparent Kd’s for individual PCs and OxPCs offers new possibilities in targeted lipidomics for the simultaneous screening of the antigenicity of multiple lipid biomarkers. In accordance with the literature, the degree of oxidation and also the fatty acid chain lengths of PCs were found to be essential for antibody recognition, corroborating the importance of distinct molecular motives within PLs in the antigenicity of OxLDL.

Abs and isolated (oxidized) lipid species as competitive antigens. Thereby, the change of the structural motif of the PC headgroup within OxPCs is described to be essential for Ab recognition, in contrast to their unoxidized precursor species (e.g., PAPC).13,15,41 Our current approach, however, allows the simultaneous determination of apparent Kd’s (if standards are available, real Kd’s, as well) as measurement of antibody affinity and, thus, of the antigenicity of individual OxPLs, which are specifically extracted from OxLDL using the GNP−Ab conjugate (note that Kd’s are termed apparent because the concentrations of the individual (Ox)PCs are not known; only that of the CuLDL concentration is known). To do so, Langmuir isotherms were constructed for individual PCs and OxPCs extracted from OxLDL by monitoring the MS intensities of compound-specific SRM transitions in positive ion mode and plotting the intensities measured for the respective bound PL species (Γb) versus the CuLDL concentration (Supporting Information Figure S4, S5). Figure 5 shows the estimates of Kd for several known PCs and their oxidation products as obtained by extraction with GNP− PEG7−Ab and compound-specific SRM detection. In general, with an increase in the hydroxylation and peroxidation rate, a decrease in the apparent Kd and, thus, an increase in the antibody affinity is observed for the oxidation products of OPPC (PC 33:1), PLPC (34:2), and POPC (34:1), respectively. This fact clearly signifies the importance of the degree of oxidation in long-chain OxPCs for recognition of OxLDL by the antibody. On the other hand, short-chain OxPCs containing ω-terminal aldehydic fatty acid residues (e.g. PONPC, POVPC) showed comparably low Kd’s. Interestingly, the latter OxPCs were also found as major OxPL species in plasma samples of hyperlipidemic patients during our previous study.26 In contrast, no increase in antibody affinity was found for OxPCs derived from PUFA-containing PLs with longer chain lengths, such as PAPC (36:4) and SAPC (38:4), or for saturated or unoxidized PCs, such as DMPC, PMPC, PPPC, and DAPC, respectively. This finding is in accordance with the postulation of Friedman et al., who described the influence of antibody recognition of OxLDL with the structure and fatty acid chain lengths of OxPLs.15

■

ASSOCIATED CONTENT

S Supporting Information *

Plots of size (DLS) and zeta-potential measurements, overview about (Ox)PCs structures, and further Langmuir plots. This material is available free of charge via the Internet at http:// pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*Phone: +49 7071 29 78793. 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 “Nano-MALDI” project by the Austrian BMVIT via the “Austrian Nano-Initiative” and “MNT-ERA.NET” (Grant no. 828701 to G.S. and M.L.) is gratefully acknowledged.

■

■

CONCLUSIONS In this paper, we describe the establishment and evaluation of a novel GNP-conjugated anti-OxLDL antibody platform combined with the sensitive detection of LC−ESI-MS/MS for the specific screening of OxLDL-related lipid biomarkers of oxidative stress. In contrast to conventional approaches, the implementation of optimized GNP−PEG7−Ab conjugates allows the selective extraction and enrichment of low-abundant oxidized phospholipids (OxPLs) bound to OxLDL and specifically trapped from plasma with the GNP−PEG7−Ab conjugate. After GNP−Ab conjugate-based trapping of OxLDL, simple methanolic extraction dissolves the lipids in the supernatant, which is directly compatible with LC−MS analysis without further workup, whereas GNPs, proteins, and buffer salts are depleted via centrifugation. The bioconjugation chemistry was monitored using SPR spectra, size, and zeta potential measurements as fast quality control methods. Furthermore, the amount of Abs bound per GNP could be determined by indirect approach with secondary anti-IgG−Ab with 1.7 ± 0.4 Ab/GNP. Determining the dissociation constant, Kd, of CuLDL with 4.13 ± 0.18 × 10−10 M, which was approved by immunoassay

REFERENCES

(1) Binder, C. J.; Chang, M.-K.; Shaw, P. X.; Miller, Y. I.; Hartvigsen, K.; Dewan, A.; Witztum, J. L. Nat. Med. 2002, 8, 1218−1226. (2) Weismann, D.; Binder, C. J. Biochim. Biophys. Acta, Biomembr. 2012, 1818, 2465−2475. (3) Tuzcu, E. M.; Kapadia, S. R.; Tutar, E.; Ziada, K. M.; Hobbs, R. E.; McCarthy, P. M.; Young, J. B.; Nissen, S. E. Circulation 2001, 103, 2705−2710. (4) Suzuki, Y. J.; Carini, M.; Butterfield, D. A. Antioxid. Redox Signaling 2010, 12, 323−325. (5) Kasai, H.; Kawai, K.; Li, Y.-s. Genes Environ. 2008, 30, 33−40. (6) Oskolkova, O. V.; Afonyushkin, T.; Bochkov, V. N. In Oxidized Phospholipids in Health and Pathology, Transworld Research Network: Trivandrum, India, 2011; pp 25−46. (7) Jacob, K. D.; Noren, H. N.; Trzeciak, A. R.; Evans, M. K. Mech. Ageing Dev. 2013, 134, 139−157. (8) Bochkov, V. N.; Oskolkova, O. V.; Birukov, K. G.; Levonen, A.L.; Binder, C. J.; Stoeckl, J. Antioxid. Redox Signaling 2010, 12, 1009− 1059. (9) Greig, F. H.; Kennedy, S.; Spickett, C. M. Free Radical Biol. Med. 2012, 52, 266−280. (10) Postle, A. D. Eur. J. Lipid Sci. Technol. 2009, 111, 2−13. (11) Itabe, H. J. Clin. Biochem. Nutr. 2012, 51, 1−8. (12) Berliner, J. A.; Leitinger, N.; Tsimikas, S. J. Lipid Res. 2009, 50 (Suppl), S207−212. 8383

dx.doi.org/10.1021/ac401778f | Anal. Chem. 2013, 85, 8376−8384

Analytical Chemistry

Article

(13) Boullier, A.; Friedman, P.; Harkewicz, R.; Hartvigsen, K.; Green, S. R.; Almazan, F.; Dennis, E. A.; Steinberg, D.; Witztum, J. L.; Quehenberger, O. J. Lipid Res. 2005, 46, 969−976. (14) Hulthe, J. Clin. Chim. Acta 2004, 348, 1−8. (15) Friedman, P.; Horkko, S.; Steinberg, D.; Witztum, J. L.; Dennis, E. A. J. Biol. Chem. 2002, 277, 7010−7020. (16) Yang, K.; Han, X. Metabolites 2011, 1, 21−40. (17) Brouwers, J. F. Biochim. Biophys. Acta, Mol. Cell Biol. Lipids 2011, 1811, 763−775. (18) Zhao, Z.; Xu, Y. J. Lipid Res. 2010, 51, 652−659. (19) Li, M.; Zhou, Z.; Nie, H.; Bai, Y.; Liu, H. Anal. Bioanal. Chem. 2011, 399, 243−249. (20) Spickett, C. M.; Dever, G. BioFactors 2005, 24, 17−31. (21) Postle, A. D.; Wilton, D. C.; Hunt, A. N.; Attard, G. S. Prog. Lipid Res. 2007, 46, 200−224. (22) O’Donnell, V. B. Biochim. Biophys. Acta, Mol. Cell Biol. Lipids 2011, 1811, 818−826. (23) Sparvero, L. J.; Amoscato, A. A.; Dixon, C. E.; Long, J. B.; Kochanek, P. M.; Pitt, B. R.; Bayir, H.; Kagan, V. E. Chem. Phys. Lipids 2012, 165, 545−562. (24) Eberlin, L. S.; Ferreira, C. R.; Dill, A. L.; Ifa, D. R.; Cooks, R. G. Biochim. Biophys. Acta, Mol. Cell Biol. Lipids 2011, 1811, 946−960. (25) Fuchs, B.; Schiller, J. Eur. J. Lipid Sci. Technol. 2009, 111, 83−98. (26) Stübiger, G.; Aldover-Macasaet, E.; Bicker, W.; Sobal, G.; Willfort-Ehringer, A.; Pock, K.; Bochkov, V.; Widhalm, K.; Belgacem, O. Atherosclerosis (Amsterdam, Neth.) 2012, 224, 177−186. (27) Hinterwirth, H.; Kappel, S.; Waitz, T.; Prohaska, T.; Lindner, W.; Lammerhofer, M. ACS Nano 2013, 7, 1129−1136. (28) McKenzie, F.; Steven, V.; Ingram, A.; Graham, D. Chem. Commun. 2009, 2872−2874. (29) Wen, S.; Liu, D.-F.; Liu, Z.; Harris, S.; Yao, Y.-Y.; Ding, Q.; Nie, F.; Lu, T.; Chen, H.-J.; An, Y.-L.; Zang, F.-C.; Teng, G.-J. J. Lipid Res. 2012, 53, 829−838. (30) Oliveira, M. D. L.; Abdalla, D. S. P.; Guilherme, D. F.; Faulin, T. E. S.; Andrade, C. A. S. Sens. Actuators, B: Chem. 2011, B155, 775− 781. (31) Hinterwirth, H.; Lindner, W.; Laemmerhofer, M. Anal. Chim. Acta 2012, 733, 90−97. (32) Di Pasqua, A. J.; Mishler, R. E.; Ship, Y.-L.; Dabrowiak, J. C.; Asefa, T. Mater. Lett. 2009, 63, 1876−1879. (33) Masson, J.-F.; Battaglia, T. M.; Cramer, J.; Beaudoin, S.; Sierks, M.; Booksh, K. S. Anal. Bioanal. Chem. 2006, 386, 1951−1959. (34) James, A. E.; Driskell, J. D. Analyst 2013, 138, 1212−1218. (35) Dai, L.; Preston, R.; Bacica, M.; Kinhikar, A.; Bolanos, B.; Murphy, R. E. Bioconjugate Chem. 2013, 24, 53−62. (36) Jin, S.; Anderson, G. J.; Kennedy, R. T. Anal. Chem. 2013, 85, 6073−6079. (37) 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. (38) Gonzalvez, A.; Preinerstorfer, B.; Lindner, W. Anal. Bioanal. Chem. 2010, 396, 2965−2975. (39) Chen, H.; Yuan, L.; Song, W.; Wu, Z.; Li, D. Prog. Polym. Sci. 2008, 33, 1059−1087. (40) Knop, K.; Hoogenboom, R.; Fischer, D.; Schubert, U. S. Angew. Chem., Int. Ed. 2010, 49, 6288−6308. (41) Itabe, H.; Yamamoto, H.; Imanaka, T.; Shimamura, K.; Uchiyama, H.; Kimura, J.; Sanaka, T.; Hata, Y.; Takano, T. J. Lipid Res. 1996, 37, 45−53.

8384

dx.doi.org/10.1021/ac401778f | Anal. Chem. 2013, 85, 8376−8384