Anal Bioanal Chem (2012) 402:1041–1056 DOI 10.1007/s00216-011-5547-5

ORIGINAL PAPER

Zirconium arsenate-modified silica nanoparticles for specific capture of phosphopeptides and direct analysis by matrix-assisted laser desorption/ionization mass spectrometry Pei-Xuan Zhao & Xiao-Feng Guo & Hong Wang & Chu-Bo Qi & He-Shun Xia & Hua-Shan Zhang

Received: 29 June 2011 / Revised: 14 October 2011 / Accepted: 31 October 2011 / Published online: 22 November 2011 # Springer-Verlag 2011

Abstract In this paper, we report, as far as we are aware, the first use of zirconium arsenate-modified silica nanoparticles (ZrAs-SNPs) for specific capture of phosphopeptides, followed by matrix-assisted laser desorption/ ionization mass spectrometric (MALDI MS) analysis. Under the optimized enrichment conditions, the efficiency and specificity of ZrAs-SNPs were evaluated with tryptic digests of four standard proteins (α-casein, β-casein, ovalbumin, and bovine serum albumin) and compared with those of titanium arsenate-modified silica nanoparticles (TiAs-SNPs). The results showed that more selective enrichment of multiply phosphorylated peptides was observed with ZrAs-SNPs than with TiAs-SNPs whereas TiAs-SNPs resulted in slightly better recovery of singly phosphorylated peptides. ZrAs-SNPs were chosen for direct capture of phosphopeptides from diluted human serum of healthy and adenocarcinoma individuals. Our experimental profiling of serum phosphopeptides revealed that the level of phosphorylated fibrinogen peptide A was up-regulated in the serum of adenocarcinoma patients in comparison with healthy adults. This suggests the possibility of using ZrAsSNPs for discovery of biomarkers of the pathogenesis process of tumors.

P.-X. Zhao : X.-F. Guo : H. Wang (*) : H.-S. Zhang Department of Chemistry, Key Laboratory of Analytical Chemistry for Biology and Medicine (Ministry of Education), Wuhan University, Wuhan 430072, China e-mail: [email protected] C.-B. Qi : H.-S. Xia Department of Pathology, Hubei Cancer Hospital, Wuhan 430079, China

Keywords Zirconium arsenate-modified silica nanoparticles . Phosphopeptide enrichment . Singly/multiply phosphorylated peptides . Human serum . MALDI MS

Introduction Protein phosphorylation is one of the most important posttranslational modifications controlling a variety of cellular processes, for example cell proliferation, differentiation, signal transduction, and metabolism [1–3]. Identification of phosphorylated proteins and characterization of the precise sites of phosphorylation within these proteins at the molecular level are important for fundamental understanding of these biological processes. Generally, proteolytic digests of phosphoproteins can be characterized by matrixassisted laser desorption/ionization mass spectrometry and electrospray ionization mass spectrometry (MALDI MS and ESI MS) [4–6]. Unfortunately, satisfactory results cannot be obtained by direct MS analysis of protein digests, because of the low abundance of phosphoproteins in tissues or cells and ion suppression by non-phosphopeptides in the digestion products. Therefore, selective enrichment of phosphopeptides from the enzymatic digests of proteins becomes a required step before MS analysis. Immobilized metal ion affinity chromatography (IMAC) is the method most frequently applied for enrichment of phosphopeptides. The method was originally based on the affinity of the phosphate groups of phosphopeptides for multivalent metal ions (traditionally Fe3+ ) immobilized on chromatographic supports via functional chelators, for example iminodiacetic acid (IDA) and nitrilotriacetic acid (NTA) [7–10]. However, the selectivity of conventionally used IMAC adsorbents is

1042

limited, because non-phosphopeptides with carboxyl groups may have affinity for immobilized metal ions similar to that of phosphopeptides, which results in reduced efficiency and sensitivity of phosphopeptide enrichment. Methyl esterification has been introduced to block non-specific binding of acidic residues and enhance the selectivity of phosphopeptide enrichment, but the low yield of derivatization and side reactions compromise the sensitivity of this approach and increase the complexity of samples [11, 12]. A simple and feasible solution for this problem is to develop novel affinity materials (metal ions, ligands, supports, etc.) with high efficiency and specificity for enrichment of phosphopeptides. Recently, use of Zr4+ and Ti4+ to replace Fe3+ as the immobilized metal ion has been proved to have higher selectivity for phosphopeptides over acidic residues compared with traditional Fe3+-IMAC beads [13–15], because of the anion-exchange properties of zirconium(IV) and titanium(IV) in acidic solution [16, 17]. Zr4+-IMAC and Ti4+IMAC for phosphopeptide enrichment have been compared, and the results indicated that isolation of singly phosphorylated peptides with Ti4+-IMAC is more selective than with Zr4+-IMAC whereas Zr4+-IMAC preferentially enriched multiply phosphorylated peptides [14, 18]. The choice of support is also important. Various silica materials have been used in the construction of IMAC materials for phosphopeptide enrichment, because of the ease of derivatization of silica [8, 15, 19, 20]. Thus, silica nanoparticles, with high surface-to-volume ratio, were chosen as the support in this work. From the literature [21, 22], p-hydroxybenzenearsonic acid (HBA) as a specific precipitant has been widely used for determination of zirconium and titanium, because arsenate strongly retains Zr4+ and Ti4+. In addition, in recent years phosphate has been widely used as novel ligand for immobilization of Zr4+ and Ti4+ in the construction of affinity materials [13– 15]. Because arsenic has similar properties to phosphorus, and arsenate has a similar structure to phosphate, arsenate may also be suitable for use as a promising ligand in the preparation of new IMAC materials. Adsorbents containing arsenate as ligand have not yet been investigated for immobilization of metal ion for potential enrichment of phosphopeptides in proteomics. Herein, zirconium and titanium arsenate-modified silica nanoparticles (ZrAsSNPs and TiAs-SNPs) were prepared in this study. The enrichment conditions were initially optimized by using ZrAs-SNPs and α-casein digest; the efficiency and specificity of ZrAs-SNPs for enrichment of phosphopeptides were then evaluated with complex peptide mixtures and compared with those of TiAs-SNPs. Finally, ZrAsSNPs were successfully used for profiling of serum phosphopeptides in human serum from healthy individuals and patients with adenocarcinoma.

P. Zhao et al.

Experimental Materials and reagents α-Casein, β-casein, ovalbumin, bovine serum albumin (BSA), 2,5-dihydroxybenzoic acid (DHB), trifluoroacetic acid (TFA), and zirconium oxychloride (ZrOCl2·8H2O) were obtained from Sigma (St Louis, MO, USA). Sequencinggrade trypsin was purchased from Promega (Madison, WI, USA). 3-Glycidoxypropyltrimethoxysilane (GLYMO), ammonium bicarbonate (NH4HCO3), urea, dithiothreitol, and iodoacetamide were purchased from BioRad (Hercules, CA, USA). HPLC-grade acetonitrile (ACN) was obtained from Merck (Darmstadt, Germany). Titanium sulfate (Ti(SO4)2), nitric acid (HNO3 65%, GR grade) and hydrofluoric acid (HF 40%, GR grade) were purchased from Sinopharm Chemical Reagent (Shanghai, China). Sodium p-hydroxyphenylarsenate was obtained from Shanghai Biochemistry Pharmaceutical Factory (Shanghai, China). All aqueous solutions were prepared in high-purity water (18.2 MΩ cm) obtained from a Milli-Q system (Millipore, Milford, MA, USA). All other chemicals and reagents were of the highest grade commercially available. Calibration solutions (0– 10 mg L−1) for As were prepared from 1000 mg L−1 As in 2% HNO3 (National Institute of Metrology, P.R. China) by dilution with deionized water. Spherical silica nanoparticles of particle size 30 nm were obtained from Wuhan Shuaier Photoelectron New Material (Wuhan, China). Preparation of ZrAs-SNPs and TiAs-SNPs The procedures for preparation of ZrAs-SNPs are illustrated in Fig. 1a–d. First, p-hydroxybenzenearsonic acid (HBA) was prepared by acidification of sodium p-hydroxyphenylarsenate (Fig. 1a), and the product was then dried under vacuum before its use to modify silica nanoparticles. Caution must be exercised when using arsenate in the synthesis, because it is toxic; wear protective gloves and mask and perform the synthesis in a fume hood. Second, silica nanoparticles were washed with 20% HNO3, 0.5 mol L−1 NaCl, H2O, acetone, and diethyl ether, and dried at 150 °C for 4 h under vacuum before chemical derivatization. The activated silica nanoparticles were then derivatized with 3-glycidoxypropyltrimethoxysilane (GLYMO) in accordance with a previously reported method (Fig. 1b) [23]. Briefly, silica nanoparticles (1.0 g) were added Fig. 1 Schematic diagram of the preparation of ZrAs-SNPs for b enrichment of phosphopeptides. Detailed description of each step: (a) preparation of HBA, (b) derivatization of silica nanoparticles with GLYMO, (c) modification of HBA on the surface of the epoxy-silica, (d) immobilization of Zr4+ on the arsenate-modified silica nanoparticles, and (e) capture of phosphopeptides by ZrAs-SNPs

Zirconium arsenate-modified silica nanoparticles

1043

1044

P. Zhao et al.

to 25 mL sodium-dried toluene, 3 mL GLYMO, and 0.1 mL calcium hydride-dried triethylamine under dry nitrogen gas protection. The mixture was stirred at room temperature for 5 h and then heated at 110 °C under reflux for 16 h. The material was then filtered, washed with toluene, acetone, and diethyl ether, and dried at 110 °C under vacuum. The epoxysilica was further reacted with HBA (Fig. 1c). Epoxy-silica (0.2 g) and 0.5 g HBA were dispersed in 10 mL absolute ethanol with addition of 0.5 mL dry triethylamine. The mixture was the heated under reflux for 4 h to prepare arsenate-modified silica nanoparticles. After washing with toluene, H2O, 10% HAc, H2O, and MeOH, the particles were dried at 60 °C under vacuum. Finally, 20 mg arsenate-modified silica nanoparticles were incubated with 20 mL 100 mmol L−1 ZrOCl2 solution at room temperature overnight with gentle stirring to prepare zirconium arsenate-modified silica nanoparticles (ZrAs-SNPs) (Fig. 1d). After centrifugation at 17,000g for 5 min, the deposit was rinsed several times with water and then dispersed in 30% (v/v) ACN containing 0.1% (v/v) TFA solution before use. The TiAs-SNPs were prepared by use of the same procedure as for ZrAs-SNPs, except for incubation of arsenate-modified silica nanoparticles in Ti(SO4)2 solution instead of ZrOCl2 solution.

analyzed by inductively coupled plasma optical emission spectrometry (ICP–OES). For digestion, 50 mg arsenatemodified silica nanoparticles and silica nanoparticles (blank) were accurately weighted into polytetrafluoroethylene (PTFE) vessels and then 1 mL concentrated HNO3 and 5 mL HF were added. The samples were digested on a hot plate under watch glass covers. After cooling, 2–3 drops of concentrated HNO3 and 2 mL deionized water were added, and the clear digests obtained were diluted to 25 mL with deionized water before analysis by ICP–OES (IRIS Intrepid II XSP, Thermo Electron, USA) for arsenic concentration. Sample preparation Tryptic digestion of standard proteins Bovine α-casein and β-casein were each dissolved in 50 mmol L−1 pH 8.2 ammonium bicarbonate (NH4HCO3) buffer and treated with trypsin at an enzyme-to-substrate ratio of 1:50 (w/w) at 37 °C for 16 h. Ovalbumin and bovine serum albumin (BSA) (4 mg) were separately dissolved in 1 mL denaturing buffer containing 8 mol L−1 urea and 50 mmol L−1 NH4HCO3 for 3 h. This protein solution was then mixed with 20 μL 50 mmol L−1 dithiothreitol and incubated at 37 °C for 2 h. Iodoacetamide (50 mmol L−1, 40 μL) was added and the incubation continued for an additional 30 min at room temperature in the dark. Finally, the mixture was diluted 10-fold with 50 mmol L−1 NH4HCO3 and incubated at 37 °C for 16 h with trypsin at an enzyme-to-substrate ratio of 1:40 (w/w).

Digestion of arsenate-modified silica nanoparticles for arsenic determination The arsenate-modified silica nanoparticles were mineralized by acid digestion, and then the digested samples were Table 1 Detailed information about the observed phosphopeptides derived from α-casein, β-casein, and ovalbumin

a

pS, bM, and dQ* represent phosphorylated serine, oxidation on methionine, and pyroglutamylation on N-terminal Gln, respectively c

Phosphopeptide ion peak is [M+Na]+

Protein

[M+H]+

Phosphorylation sites

Peptide sequence

α-Casein

1466.51 1482.60 1562.04 1660.79 1832.83 1847.69 1927.69 1943.70 1951.95 2703.75 2720.91 2736.90

1 1 2 1 1 1 2 2 1 5 5 5

TVDMEpSTEVFTKa TVDMEpSTEVFTKb EQLpSTpSEENSKKc VPQLEIVPNpSAEER YLGEYLIVPNpSAEER DIGSEpSTEDQAMEDIK DIGpSEpSTEDQAMEDIK DIGpSEpSTEDQAMEDIK YKVPQLEIVPNpSAEER Q*MEAEpSIpSpSpSEEIVPNpSVEAQKd QMEAEpSIpSpSpSEEIVPNPNpSVEQK QMEAEpSIpSpSpSEEIVPNpSVEAQK

3008.03 2061.83 3122.27 2088.91 2511.13 2901.36

4 1 4 1 1 1

NANEEEYSIGpSpSpSEEpSAEVATEEVK FQpSEEQQQTEDELQK RELEELNVPGEIVEpSLpSpSpSEESITR EVVGpSAEAGVDAASVSEEFR LPGFGDpSIEAQCGTSVNVHSSLR FDKLPGFGDpSIEAQCGTSVNVHSSLR

β-Casein Ovalbumin

Zirconium arsenate-modified silica nanoparticles Fig. 2 MALDI mass spectra obtained from the tryptic digest of α-casein (0.1 μmol L−1, 20 μL). (a) Direct analysis, and subsequent ZrAs-SNPs enrichment of phosphopeptides from α-casein digest by use of a loading buffer of (b) 0.01% TFA, (c) 0.1% TFA, (d) 1% TFA, (e) H2O, and (f) 50 mmol L−1 NH4HCO3. All buffers contained 50% ACN. The phosphopeptides derived from α-casein are marked with asterisks

1045

1046

P. Zhao et al.

Fig. 3 Effect of ACN concentration (a, 30%; b, 50%; c, 70%) on specific capture of phosphopeptides from α-casein digest (0.1 μmol L−1, 20 μL) using ZrAs-SNPs. All buffers contained 0.1% TFA. The phosphopeptides derived from α-casein are marked with asterisks

All of the proteolytic digests obtained were lyophilized by use of a vacuum concentrator and stored in the freezer at −40 °C until further use. Preparation of human serum samples Human serum samples were obtained from Hubei Cancer Hospital. They were collected from eight healthy adults and eight adenocarcinoma patients in accordance with their

standard clinical procedures. The cancer samples were donated by three rectal adenocarcinoma, two gastric adenocarcinoma, one colon adenocarcinoma, one lung Fig. 4 Effect of DHB concentration (a, 10; b, 20; c, 50; d, b 100 mg mL−1) on specific capture of phosphopeptides from α-casein digest (0.1 μmol L−1, 20 μL) using ZrAs-SNPs. All buffers contained 0.1% TFA and 50% ACN. The phosphopeptides derived from αcasein are marked with asterisks

Zirconium arsenate-modified silica nanoparticles

1047

1048

P. Zhao et al.

Zirconium arsenate-modified silica nanoparticles

R Fig. 5

MALDI mass spectra obtained from phosphopeptides captured by ZrAs-SNPs from the tryptic digests of four standard proteins (αcasein, β-casein, ovalbumin and BSA) in the molar ratios (a) 1:1:1:0, (b) 1:1:1:1, (c) 1:1:1:10, and (d) 1:1:1:100. Phosphopeptides derived from α-casein, β-casein, and ovalbumin are marked with asterisks, hash symbols, and inverted triangles, respectively

adenocarcinoma, and one breast carcinoma patients. Briefly, blood samples (healthy and adenocarcinoma individuals) were collected in PET (polyethylene terephthalate) vacuum blood-collection tubes (ST750CG; Beijing Sekisui Trank Medical Technology, China) and centrifuged at 4 °C for 5 min at 2,000g. After collection, the serum samples were sub-divided and stored frozen at −80 °C until further use. Before use, 10 μL human serum was diluted with 20 μL 2,5-dihydroxybenzoic acid (DHB) solution at a concentration of 20 mg mL−1 in ACN–H2O–TFA 50:50:0.1 (v/v) and used directly for phosphopeptide enrichment. Capture of phosphopeptides by use of ZrAs-SNPs and TiAs-SNPs Peptide samples (tryptic digests of standard proteins or human serum) were first diluted with DHB solution. ZrAs-SNPs or TiAs-SNPs (20 mg mL−1, 5 μL) were then added. After vortex mixing for 30 min at room temperature, ZrAs-SNPs or TiAs-SNPs with trapped phosphopeptides were separated by centrifugation at 17,000g for 5 min. Subsequently, unbound impurities were removed by washing with 200 μL DHB solution then 30% ACN– 0.1% TFA. MALDI MS analysis For on-bead analysis of phosphopeptides after enrichment by MALDI MS, DHB (20 mg mL−1) in 50% ACN, 1% phosphoric acid was used as the matrix. The deposit of ZrAs-SNPs or TiAs-SNPs with trapped phosphopeptides was mixed with 2 μL DHB matrix, and 1 μL of the mixture was deposited on to the MALDI plate and left to air-dry at room temperature. Typically, spectra were obtained in the positive mode by use of a MALDI quadrupole timeof-flight (Q-TOF) mass spectrometer (SYNAPT G2 HDMS; Waters, USA) equipped with a 355-nm Nd: YAG laser and a sample target having the capacity to load 120 samples simultaneously. The laser power was adjusted to slightly above the threshold to obtain good resolution and signal-to-noise ratio (S/N). A peptide of known mass (Glu-Fibrinopeptide, EGVNDNEEGFFSAR, m/z 1570.68) was used as the internal mass standard. Mass spectrometric data analysis was performed using MassLynx V4.1 software (Micromass/Waters).

1049

Results and discussion Design and characterization of ZrAs-SNPs and TiAs-SNPs To develop a reliable purification procedure, novel affinity materials with excellent selectivity for phosphopeptide enrichment are essential. Among the factors affecting the design of IMAC materials, choice of metal ions and ligands are the primary considerations. The specific and strong affinity of Zr4+ and Ti4+ for phosphate makes them ideal immobilized metal ions on IMAC beads for phosphopeptide enrichment [13–15]. For immobilization of Zr4+ and Ti4+, arsenate is a promising ligand because of the high binding strength of HBA for zirconium and titanium [21, 22]. Arsenic is another member of Group VA in the periodic table of the elements with similar characteristics to those of phosphorus. However, arsenate as a potential ligand has not been used to chelate Zr4+ and Ti4+ in the preparation of new materials for proteome research. In this study, new adsorbents were prepared by chelation of Zr4+ and Ti4+ with arsenate-modified silica nanoparticles and used for specific capture of phosphopeptides from peptide samples. Furthermore, good biocompatibility for example the hydrophilicity of arsenate, the presence of GLYMO as spacer arm, and the high surface-to-volume ratio of silica nanoparticles, are all very suitable for use in phosphopeptide enrichment. Because the hybrid form (sp3) of the central atom of arsenic acid (H3AsO4) is the same as that of phosphoric acid (H3PO4) and metal ion coordinates to the oxygen atom of polyoxy anion, it is reasonable to speculate that the layer structure of metal(IV) arsenate is similar to that of metal (IV) phosphate, i.e. formed from MO6 octahedra [24, 25]. For ZrAs-SNPs (Fig. 1d), each Zr4+ coordinates to more than one arsenate group and the arsenate binds to more than one Zr4+. Therefore, the extremely strong chelation of Zr4+ stacks the original monolayer and provides a very stable, well-defined interface of zirconium arsenate sites, leaving free coordination sites for binding of phosphopeptides (Fig. 1e). Furthermore, the unique coordination properties of arsenate with Zr4+ can result in superior retentive strength, avoiding loss of bound metal ions during the enrichment procedure. The model of TiAs-SNPs with trapped phosphopeptides is the same as that of ZrAsSNPs, except that Zr4+ is replaced by Ti4+. ICP–OES experiments were performed to characterize the amount of derivatized arsenic on arsenate-modified silica nanoparticles, which reveals that the As is 1.70 mg g−1 (corresponding to 22.72 μmol g−1). The more arsenate groups derivatized on the surface of silica nanoparticles and the more Zr4+/Ti4+ loaded with strong retentive strength, the better the phosphopeptide enrichment

1050

obtained. Considering 100 μg ZrAs-SNPs/TiAs-SNPs and picomole levels of peptides in the sample used in the experiment, the loading capacity of Zr4+/Ti4+ is three orders of magnitude higher than the amount of phosphopeptides, which enables efficient and specific capture of phosphopeptides by the ZrAs-SNPs/TiAs-SNPs. Optimization of enrichment conditions using ZrAs-SNPs The binding properties of novel IMAC materials are unknown. Thus, as the first step, it is necessary to optimize the procedures for phosphopeptide enrichment to maximize phosphopeptide recovery while minimizing contamination from non-phosphopeptides. Initial optimization of the enrichment conditions (binding pH, organic solvent, additive, etc.) was performed by use of ZrAs-SNPs and α-casein digest. A list of phosphopeptides derived from α-casein, with their theoretical molecular masses, is shown in Table 1. The efficiency and specificity of phosphopeptide enrichment were evaluated from the number of phosphopeptides isolated from peptide mixture and from interference from non-phosphopeptides as a result of nonspecific competitive binding. Throughout this study, enrichment experiments were performed by the “batch method” by mixing peptide samples with affinity materials in microcentrifuge tubes. The beads with trapped phosphopeptides were transferred on to the target for ease of on-bead analysis by MALDI MS, avoiding the problem of incomplete elution and possible loss of the eluted phosphopeptides in the elution step. The pH is the most important factor affecting the efficiency and specificity of phosphopeptide enrichment. High binding selectivity of phosphopeptides over acidic peptides can be achieved with ZrAs-SNPs by adjustment of the pH of the loading buffers. Direct analysis of 2 pmol αcasein digest by MALDI MS resulted in detection of only four phosphopeptides, and the peaks of phosphopeptides were seriously suppressed by those of non-phosphopeptides in the spectrum (Fig. 2a). MALDI mass spectra obtained after ZrAs-SNPs enrichment of phosphopeptides from the same digest at different binding loading buffer pH are shown in Fig. 2b–f. All these enrichment experiments involved loading buffer (0.01%, 0.1%, or 1% aqueous TFA, water, or 50 mmol L−1 NH4HCO3 in 50% ACN) and washing buffer (200 mmol L−1 NaCl contained 50% ACN and 0.1% TFA). The results showed that use of 0.1% TFA as loading buffer resulted in recovery of the largest number of phosphopeptides (twelve) and the best specificity of phosphopeptide enrichment among the loading buffers of different pH (Fig. 2c). A low concentration of TFA was insufficient to protonate the carboxyl groups of acidic peptides (Fig. 2b) whereas a high concentration led to loss of the negative charge on the phosphate groups of the

P. Zhao et al.

phosphopeptides (Fig. 2d), so both are ineffective for phosphopeptide enrichment using ZrAs-SNPs. With neutral and alkaline binding buffers, phosphopeptides were not well retained by ZrAs-SNPs because of hydrolysis of immobilized Zr4+ (Fig. 2e, f). Finally, 0.1% TFA at the optimum pH was used as loading buffer for enrichment of phosphopeptides by ZrAs-SNPs. An appropriate proportion of ACN in the loading buffer is beneficial because it reduces hydrophobic interaction between non-phosphopeptides and the affinity materials, thereby improving the selectivity of ZrAs-SNPs for phosphopeptides. As shown in Fig. 3, twelve, twelve, and ten phosphopeptides were captured by ZrAs-SNPs when 2 pmol α-casein digest was loaded in buffer containing 30, 50, and 70% ACN, respectively, and 0.1% TFA. Increasing the proportion of ACN from 30% to 50% slightly enhanced the specificity of phosphopeptide enrichment (Fig. 3a, b). However, further increasing the ACN concentration to 70% resulted in recovery of fewer phosphopeptides from the tryptic digest of α-casein by ZrAs-SNPs compared with use of 50% ACN (Fig. 3b, c). Therefore, we chose 50% ACN with 0.1% TFA as the initial optimized loading buffer for specific capture of phosphopeptides. To further reduce the interaction between acidic nonphosphopeptides and ZrAs-SNPs, DHB was added to the loading and washing buffers during enrichment of phosphopeptides [26]. Tryptic digest of α-casein was loaded on to ZrAs-SNPs which were then washed with a buffer solution of 10, 20, 50, or 100 mg mL−1 DHB in ACN– H2O–TFA 50:50:0.1 (v/v). Figure 4 shows the results. Inclusion of DHB in the buffer system at a final concentration of 10 or 20 mg mL−1 efficiently reduced binding of non-phosphopeptides to ZrAs-SNPs while retaining the high binding affinity for phosphopeptides (Fig. 4a, b). At the same time, high concentrations of DHB (50 or 100 mg mL−1) in the loading and washing buffers resulted in isolation of fewer phosphopeptides from the αcasein digest by ZrAs-SNPs and low signal-to-noise ratio (S/N) of the phosphopeptide peaks (Fig. 4c, d). This is because a large excess of DHB molecules can effectively compete with both non-phosphopeptides and phosphopeptides for binding sites on ZrAs-SNPs. Taking both the number of phosphopeptides and the specificity of phosphopeptide enrichment into consideration, a solution of 20 mg mL−1 DHB in ACN–H2O–TFA 50:50:0.1 (v/v) was selected as the final optimized buffer for enrichment of Fig. 6 MALDI mass spectra obtained from phosphopeptides captured b by TiAs-SNPs from the tryptic digests of four standard proteins (αcasein, β-casein, ovalbumin, and BSA) in the molar ratios: (a) 1:1:1:0, (b) 1:1:1:1, (c) 1:1:1:10, and (d) 1:1:1:100. Phosphopeptides derived from α-casein, β-casein, and ovalbumin are marked with asterisks, hash symbols, and inverted triangles, respectively

Zirconium arsenate-modified silica nanoparticles

1051

1052

P. Zhao et al.

phosphopeptides by ZrAs-SNPs in all the experiments reported below. The results above revealed that phosphopeptides could be effectively identified by MALDI MS after ZrAs-SNPs enrichment under the optimized enrichment conditions. Specific capture of phosphopeptides from the tryptic digests of standard proteins To investigate the ability of ZrAs-SNPs to capture phosphopeptides from complex peptide mixtures, tryptic digests of phosphoproteins (α-casein, β-casein, and ovalbumin), and non-phosphoprotein (BSA) in different molar ratios (1:1:1:0, 1:1:1:1, 1:1:1:10, and 1:1:1:100) were used. Figure 5 shows the MALDI mass spectra obtained. In total, 17 phosphopeptides (12 phosphopeptides derived from αcasein, two from β-casein, and three from ovalbumin, 12α2β3O) from the tryptic digests of phosphoproteins (0.1 μmol L−1 for each phosphoprotein, 20 μL) were detected after ZrAs-SNPs enrichment (Fig. 5a). Detailed information about the phosphopeptides derived from αcasein, β-casein, and ovalbumin is shown in Table 1. In comparison with the results obtained from the tryptic digests of phosphoproteins (Fig. 5a), the ability of ZrAsSNPs to capture phosphopeptides specifically from the tryptic digests of four standard proteins at equal molar ratio (0.1 μmol L−1 for each protein, 20 μL) did not dramatically decline compared with the number of isolated phosphopeptides (12α2β3O) and S/N of the mass spectrum (Fig. 5b). Although peak intensities of phosphopeptides were reduced by increased nonspecific binding at low molar ratio (1:1:1:10) of phosphoproteins to non-phosphoprotein (0.1 μmol L−1 for each phosphoprotein and 1.0 μmol L−1 for BSA, 20 μL), 15 phosphopeptides (10α2β3O) definitely appeared in the mass spectrum (Fig. 5c). Furthermore, the peaks of multiply phosphorylated peptides (m/z 2703.75, 2720.91, 2736.90, 3008.03, and 3122.27) with a large number of phosphorylation sites (≥4) were clearly observable in the spectrum (Fig. 5a–c) without magnification, compared with previous studies [20, 26]. By further decreasing the molar ratio to 1:1:1:100 (0.01 μmol L−1 for

Table 2 Overview of the results for specific capture of phosphopeptides from the tryptic digests of four standard proteins (α-casein, β-casein, ovalbumin, and BSA) in different molar ratios by use of different adsorbents

Average from three independent experiments

Molar ratio

each phosphoprotein and 1.0 μmol L−1 for BSA, 20 μL), four phosphopeptides (2α1β1O) could still be detected with high selectivity of phosphopeptide enrichment (Fig. 5d). Considering the level of phosphoproteins is two orders of magnitude lower than that of non-phosphoprotein in the peptide mixture, the performance of ZrAs-SNPs is good enough for specific capture of phosphopeptides. These facts further demonstrated the high efficiency and specificity of ZrAs-SNPs for enrichment of phosphopeptides from complex digests of proteins. Comparison of the performance of ZrAs-SNPs with TiAs-SNPs for enrichment of phosphopeptides Because the optimum enrichment conditions (TFA, ACN, DHB, etc.) mainly reduce non-specific binding of peptides to IMAC beads while enhancing the specific affinity for phosphopeptides, and the binding properties of Ti4+ are similar to those of Zr4+, it can be deduced naturally that the optimized buffer system for ZrAs-SNPs is also suitable for TiAs-SNPs. The MALDI mass spectra obtained from phosphopeptides captured by TiAs-SNPs from the tryptic digests of four standard proteins (α-casein, β-casein, ovalbumin, and BSA) in different molar ratios (1:1:1:0, 1:1:1:1, 1:1:1:10, and 1:1:1:100) are shown in Fig. 6, and the performance of ZrAs-SNPs in phosphopeptide enrichment compared with that of TiAs-SNPs is summarized in Table 2. Obviously, more selective isolation of multiply phosphorylated peptides was achieved with ZrAs-SNPs whereas TiAs-SNPs enabled slightly preferential enrichment of singly phosphorylated peptides (Table 2). The multiply phosphorylated peptides at m/z 2703.75, 2720.91, 2736.90, and 3008.03 with a large number of phosphorylation sites (≥4) appearing in Fig. 5a–c could not be detected when the tryptic digests of four standard proteins in the molar ratios 1:1:1:0, 1:1:1:1, and 1:1:1:10 are loaded on to TiAs-SNPs, as shown in Fig. 6a–c. However, the monophosphopeptide at m/z 1847.69 was more easily captured by TiAs-SNPs from the tryptic digests of four standard proteins in the molar ratios 1:1:1:0 and 1:1:1:1, respectively (Fig. 6a, b). When the molar ratio of each of the phosphoproteins to BSA was

No. of phosphorylated peptides captured by: ZrAs-SNPs

1:1:1:0 1:1:1:1 1:1:1:10 1:1:1:100

TiAs-SNPs

Singly

Multiply

Total

Singly

Multiply

Total

9 9 9 4

8 8 6 0

17 17 15 4

10 10 9 4

4 4 3 0

14 14 12 4

Zirconium arsenate-modified silica nanoparticles

1053

reduced to 1:1:1:100, the performance of phosphopeptide enrichment by ZrAs-SNPs and TiAs-SNPs was nearly equal (Fig. 5d and Fig. 6d). In contrast with the conclusions reached by Kweon and Hakansson using ZrO2 and TiO2 for enrichment of singly and multiply phosphorylated peptides [16], our results are in accordance with previous studies that also found Ti4+-IMAC to be more efficient than Zr4+-IMAC at recovering singly phosphorylated peptides whereas Zr4+IMAC was better than Ti4+-IMAC for recovery of multiply phosphorylated peptides [14, 18].

We suppose that the different selectivity of singly versus multiply phosphorylated peptides with Zr4+ and Ti4+immobilized affinity materials may be because of the different retention strength between immobilized metal ions and the phosphate groups of phosphopeptides. It is well known that Zr4+ and Ti4+ are metal ions with an inert gas configuration, that is to say, the electronic configuration outside the nucleus of Zr4+ and Ti4+ is similar to that of inert gas atoms. The metal ions of this configuration coordinate to ligands with Coulombic attraction. For the

Fig. 7 MALDI mass spectra obtained from a tryptic digest of phosphoproteins (α-casein, β-casein, and ovalbumin, in the molar ratio 1:1:1). (a) Direct analysis, (b) analysis of the residual solution after capture of phosphopeptides by use of TiAs-SNPs, (c) on-bead analysis of phospho-

peptides trapped by ZrAs-SNPs from the residual solution. Phosphopeptides derived from α-casein, β-casein, and ovalbumin are marked with asterisks, hash symbols, and inverted triangles, respectively. The peak of the calibration peptide at m/z 1570.68 is indicated with a plus symbol

1054

P. Zhao et al.

Fig. 8 Three-dimensional view of MALDI MS profiling of phosphopeptides captured from the serum of eight healthy human adults and eight adenocarcinoma patients

same ligand (phosphate) and equal charges (+4) on the central ions the Coulombic attraction of Ti4+ to phosphate is stronger than that of Zr4+, because the ionic radius of Ti4+ is smaller than that of Zr4+ in the same group but different periods in the periodic table of the elements. As a result, TiAs-SNPs capture more singly phosphorylated peptides than ZrAs-SNPs, because the immobilized Ti4+ has higher affinity for monophosphopeptides compared with Zr4+, avoiding the loss of these peptides during the purification procedures. The much stronger retention of multiply phosphorylated peptides with immobilized Ti4+ makes them more difficult to elute with the DHB matrix (H3PO4–DHB) for on-bead analysis by MALDI MS [27, 28]. Together with their poorer ionization efficiency in the positive-ion mode, the multiply phosphorylated peptides at m/z 2703.75, 2720.91, 2736.90, and 3008.03 with a large number of phosphorylation sites (≥4) cannot be observed after treatment with TiAs-SNPs. To test our assumption described above, we designed a procedure as follows. TiAs-SNPs were used to Table 3 Phosphorylated fragments degraded from the protein fibrinogen in human serum

a

pS denotes phosphorylated serine

capture phosphopeptides from the tryptic digests of phosphoproteins (α-casein, β-casein, and ovalbumin with a molar ratio of 1:1:1), and the residual solution was then treated with ZrAs-SNPs. Figure 7a, b shows the MALDI mass spectra obtained from the tryptic digests of phosphoproteins without enrichment and the residual solution after incubation with TiAs-SNPs. It can be seen from Fig. 7b that most of the phosphopeptide peaks were missing in comparison with that in Fig. 7a, especially the three multiply phosphorylated peptides (m/z 2703.75, 2720.91, and 3122.27) with a large number of phosphorylation sites (≥4). Although the monophosphopeptide peak at m/z 1466.51 was still present, its ion signal intensity was very low. These facts indicated that phosphopeptides, including singly and multiply phosphorylated peptides, were efficiently captured by TiAsSNPs. Our previous experiment (Fig. 5a) proved that ZrAs-SNPs can effectively enrich both singly and multiply phosphorylated peptides from tryptic digests of phosphoproteins and these phosphopeptides can be eluted

Protein

[M+H]+

Phosphorylation sites

Peptide sequencea

Fibrinogen

1389.31 1460.39 1545.50 1616.57

1 1 1 1

DpSGEGDFLAEGGGV ADpSGEGDFLAEGGGV DpSGEGDFLAEGGGVR ADpSGEGDFLAEGGGVR

Zirconium arsenate-modified silica nanoparticles

by the DHB matrix (H3PO4–DHB) for on bead analysis by MALDI MS. However, only singly phosphorylated peptides (seven) with low S/N appeared in the spectrum when ZrAs-SNPs were used to capture phosphopeptides from the residual solution already treated with TiAs-SNPs (Fig. 7c). This phenomenon further demonstrated that the multiply phosphorylated peptides at m/z 2703.75, 2720.91, 2736.90, and 3008.03 with a large number of phosphorylation sites (≥4) were retained on TiAs-SNPs and could not be eluted by the DHB matrix (H3PO4–DHB) for on-bead analysis by MALDI MS. As a compromise between singly and multiply phosphorylated peptides, ZrAs-SNPs were used during enrichment of phosphopeptides from biological samples. Profiling of serum phosphopeptides by ZrAs-SNPs Human serum is the sample most commonly used for clinical diagnosis, searching for biomarkers, and recording the current physiological state of the body, because of its easy collection. However, profiling of serum phosphopeptides is still a big challenge because phosphorylated peptides are present at low abundance in human serum which contains thousands of peptides either degraded from larger proteins or secreted from cells and tissues [29]. Hence, the feasibility of using ZrAs-SNPs for specific capture of phosphopeptides directly from diluted human serum from eight healthy adults and eight adenocarcinoma patients was investigated. Figure 8 shows the MALDI mass spectra obtained from human serum from healthy and adenocarcinoma individuals pretreated with ZrAs-SNPs. The phosphopeptides at m/z 1389.31, 1460.39, 1545.50, and 1616.57 are derived from fibrinopeptide A (gi|229185, ADSGEGDFLAEGGGVR). Table 3 lists the detailed information about these phosphorylated fragments degraded from fibrinogen protein, whose sequences has been identified by tandem mass spectrometry combined with protein database searching [30, 31]. It was clearly seen that the phosphopeptides from human serum were expressed differently between healthy and adenocarcinoma individuals, especially the phosphopeptide peak at m/z 1616.57. Our phosphopeptide profiling experiments indicated that an upregulated level of phosphorylated fibrinogen peptide A was found in the serum of adenocarcinoma patients in comparison with that of healthy adults. Similar results have also been observed in urothelial carcinoma, ovarian cancer, and other diseases [32–35]. Proteases such as thrombin, neutrophil elastase, tryptase, matrix metalloproteinase, and cathepsin D/G may degrade fibrinogen into fragments [35, 36]. Therefore, investigation of the expression of phosphorylated fibrinogen fragments and activation of enzymes in cancer serum may help in the understanding of the pathogenesis process of tumors.

1055

Conclusions A novel approach utilizing ZrAs-SNPs to selectively enrich phosphopeptides is reported in this paper. Under the optimized enrichment conditions, the high efficiency and specificity of ZrAs-SNPs in phosphopeptide enrichment were demonstrated with tryptic digests from four standard proteins (α-casein, β-casein, ovalbumin, and BSA) in different molar ratios (1:1:1:0, 1:1:1:1, 1:1:1:10 and 1:1:1:100) and compared with those of TiAs-SNPs. Recovery of multiply phosphorylated peptides with a large number of phosphorylation sites (≥4) by ZrAs-SNPs was significantly than by TiAs-SNPs whereas TiAs-SNPs resulted in slightly better selectivity than ZrAs-SNPs for singly phosphorylated peptides. The different selectivity of singly versus multiply phosphorylated peptides between ZrAs-SNPs and TiAs-SNPs may be because of the different Coulomb attraction of immobilized Zr4+ and Ti4+ to the phosphate groups of phosphopeptides. Furthermore, ZrAsSNPs were used directly to capture phosphopeptides from diluted human serum from eight healthy adults and eight adenocarcinoma patients. The results showed that the phosphopeptides in human serum were expressed differently between healthy and adenocarcinoma groups. In view of their excellent performance in phosphopeptide enrichment, ZrAs-SNPs seem to be promising for specific capture of phosphopeptides in phosphoproteome analysis and may have a wide application in biomarker discovery and clinical research. Acknowledgments This work was supported by the National Natural Science Foundation of China (20775058 and 20835004) and the Special Research Fund for the Doctoral Program of Higher Education of China (20070486031).

References 1. Graves JD, Krebs EG (1999) Pharmacol Ther 82:111–121 2. Hunter T (2000) Cell 100:113–127 3. Mann M, Ong SE, Gronborg M, Steen H, Jensen ON, Pandey A (2002) Trends Biotechnol 20:261–268 4. McLachlin DT, Chait BT (2001) Curr Opin Chem Biol 5:591–602 5. Schweppe RE, Haydon CE, Lewis TA, Resing KA, Ahn NG (2003) Acc Chem Res 36:453–461 6. Collins MO, Yu L, Choudhary JS (2007) Proteomics 7:2751–2768 7. Xu XQ, Deng CH, Gao MX, Yu WJ, Yang PY, Zhang XM (2006) Adv Mater 18:3289–3293 8. Pan CS, Ye ML, Liu YG, Feng S, Jiang XG, Han GH, Zhu JJ, Zou HF (2006) J Proteome Res 5:3114–3124 9. Dunn JD, Watson JT, Bruening ML (2006) Anal Chem 78:1574– 1580 10. Tsai CF, Wang YT, Chen YR, Lai CY, Lin PY, Pan KT, Chen JY, Khoo KH, Chen YJ (2008) J Proteome Res 7:4058–4069 11. Ficarro SB, McCleland ML, Stukenberg PT, Burke DJ, Burke DJ, Ross MM, Shabanowitz J, Hunt DF, White FM (2002) Nat Biotechnol 20:301–306

1056 12. Brill LM, Salomon AR, Ficarro SB, Mukherji M, Stettler-Gill M, Peters EC (2004) Anal Chem 76:2763–2772 13. Feng S, Ye ML, Zhou HJ, Jiang XG, Jiang XN, Zou HF, Gong BL (2007) Mol Cell Proteomics 6:1656–1665 14. Zhou HJ, Ye ML, Dong J, Han GH, Jiang XN, Wu RN, Zou HF (2008) J Proteome Res 7:3957–3967 15. Wu J-H, Zhao Y, Li T, Xu C, Xiao K, Feng Y-Q, Guo L (2010) J Sep Sci 33:1806–1815 16. Kweon HK, Hakansson K (2006) Anal Chem 78:1743–1749 17. Li Y, Xu XQ, Qi DW, Deng CH, Yang PY, Zhang XM (2008) J Proteome Res 7:2526–2538 18. Yu ZY, Han GH, Sun ST, Jiang XN, Chen R, Wang FJ, Wu RA, Ye ML, Zou HF (2009) Anal Chim Acta 636:34–41 19. Zhou HJ, Xu SY, Ye ML, Feng S, Pan CS, Jiang XG, Li X, Han GH, Fu Y, Zou HF (2006) J Proteome Res 5:2431–2437 20. Zhao P-X, Zhao Y, Guo X-F, Wang H, Zhang H-S (2011) J Chromatogr A 1218:2528–2539 21. Claassen A (1942) Chem Zentr 1:3124 22. Richter F (1941) Z Anal Chem 121:1–16 23. Larsson PO, Glad M, Hansson L, Mansson MO, Ohlson S, Mosbach K (1983) Adv Chromatogr 21:41–85 24. Stanghellini PL, Boccaleri E, Diana E, Alberti G, Vivani R (2004) Inorg Chem 43:5698–5703 25. Nonglaton G, Benitez IO, Guisle I, Pipelier M, Leger J, Dubreuil D, Tellier C, Talham DR, Bujoli B (2004) J Am Chem Soc 126:1497–1502

P. Zhao et al. 26. Larsen MR, Thingholm TE, Jensen ON, Roepstorff P, Jorgensen TJD (2005) Mol Cell Proteomics 4:873–886 27. Stensballe A, Jensen ON (2004) Rapid Commun Mass Spectrom 18:1721–1730 28. Hart SR, Waterfield MD, Burlingame AL, Cramer R (2002) J Am Soc Mass Spectrom 13:1042–1051 29. Petricoin EF, Belluco C, Araujo RP, Liotta LA (2006) Nat Rev Cancer 6:961–967 30. Hu LH, Zhou HJ, Li YH, Sun ST, Guo LH, Ye ML, Tian XF, Gu JR, Yang SL, Zou HF (2009) Anal Chem 81:94–104 31. Li Y, Qi DW, Deng CH, Yang PY, Zhang XM (2008) J Proteome Res 7:1767–1777 32. Theodorescu D, Wittke S, Ross MM, Walden M, Conaway M, Just I, Mischak H, Frierson HF (2006) Lancet Oncol 7:230– 240 33. Ogata Y, Hepplmann CJ, Charlesworth MC, Madden BJ, Miller MN, Kalli KR, Cilby WA, Bergen HR, Saggese DA, Muddiman DC (2006) J Proteome Res 5:3318–3325 34. Orvisky E, Drake SK, Martin BM, Abdel-Hamid M, Ressom HW, Varghese RS, An Y, Saha D, Hortin GL, Loffredo CA, Goldman R (2006) Proteomics 6:2895–2902 35. Ebert MPA, Niemeyer D, Deininger SO, Wex T, Knippig C, Hoffmann J, Sauer J, Albrecht W, Malfertheiner P, Rocken C (2006) J Proteome Res 5:2152–2158 36. Seydewitz HH, Matthias FR, Schondorf TH, Witt I (1987) Thromb Res 46:437–445