Pooling Aqueous Humor Samples: Bias in 2D-LC-MS/MS Strategy? Patricia Escoffier,†,‡ Luc Paris,§ Bahram Bodaghi,| Martin Danis,†,‡,§ Dominique Mazier,†,‡,§ and Carine Marinach-Patrice*,†,‡ Universite´ Pierre et Marie Curie-Paris6, UMR S 945, Paris F-75013, France, INSERM, UMR S 945, Paris F-75013, France, AP-HP, Groupe Hospitalier Pitie´-Salpeˆtrie`re, Service Parasitologie-Mycologie, Paris F-75013, France, and AP-HP, Groupe Hospitalier Pitie´-Salpeˆtrie`re, Service Ophtalmologie, Paris F-75013, France Received July 23, 2009
The proteomic analysis of body fluids presents a major challenge in studies of human diseases. Traditional techniques for protein separation require large volumes and large amounts of protein, which may be difficult to obtain for certain fluids, such as the aqueous humor (AH). Two-dimensional liquid chromatography (2D-LC-MS/MS), adapted for peptides separation from complex protein mixtures, provides an alternative approach in proteomic analysis with a potential utility in biomarker research. We investigated several different 2D-LC-MS/MS methods for use with the AH of patients with cataract, traditionally used as a control group in studies of ocular diseases. We compared analyses of individual samples with analyses of pools of proteins or peptides, and found that the investigation strategy used strongly influenced protein identification. We identified 71 proteins related to extracellular proteins highly abundant in serum (e.g., albumin or transferrin) and involved in various functions, such as transport and metabolism, together with intracellular (myeloblastin) or organelle-specific proteins (cytochrome c). An evaluation of the advantages and disadvantages of each method suggested that individual analyses and the use of peptide mixtures should be favored as complementary techniques in the search for biomarkers in ocular diseases. Keywords: Human • aqueous humor • proteome • 2D-LC/MS/MS • pooling • biomarker discovery
Introduction The intraocular fluid filling the anterior and posterior chambers of the eye is called the aqueous humor (AH). It is produced by the ciliary process and secreted from anterior segment tissues into the posterior chamber through the active transport of ions and solutes.1,2 The AH provides various functions, including maintaining intraocular pressure, providing nutrients for the nonvascularized ocular tissues (lens, cornea, trabecular meshwork), removing the products of ocular tissue metabolism (lactate, pyruvate, carbon dioxide), transporting ascorbic acid into the anterior segment to protect against oxidation and a possible role in the local immune response.3 AH is produced continually, with a turnover of between 30 min and 2 h, and the balance between production and drainage determines intraocular pressure.2 AH is derived from plasma in the capillary network of ciliary processes by passive mechanisms (diffusion and ultrafiltration of lipids and water-soluble substances) and active transport (secretion), and * To whom correspondence should be addressed. Carine MarinachPatrice, Universite´ Pierre et Marie Curie, INSERM UMR S 945, 75013 Paris, France. E-mail,
[email protected]; tel, 33(0)140779736; fax, 33(0)145838858. † Universite´ Pierre et Marie Curie-Paris6. ‡ INSERM. § AP-HP, Groupe Hospitalier Pitie´-Salpeˆtrie`re, Service ParasitologieMycologie. | AP-HP, Groupe Hospitalier Pitie´-Salpeˆtrie`re, Service Ophtalmologie. 10.1021/pr9006602
2010 American Chemical Society
its composition is therefore essentially the same as that of plasma. AH thus contains many proteins with a high abundance in serum,4 including transferrin, antibodies and albumin. The AH proteome has been poorly analyzed to date, invariably by two-dimensional electrophoresis (2D-E) techniques.5,6 Previous studies have aimed to compare proteins from control subjects and patients with acute corneal rejection or myopia. The ability of several multiplex immunoassays to detect certain biomarkers in control AH samples has recently been assessed, using clinically available samples.7 In this study, 90 predetermined analytes were screened in a procedure involving the identification of known proteins extracted from a database, or specific proteins, such as potential biomarkers, by 2D-E. In the corresponding previous studies, as in most investigations in ocular diseases, ocular fluids from patients undergoing cataract surgery were used as a control. AH samples from normal eyes would be better controls, but samples of this type cannot be obtained, for ethical reasons. Although it remains possible that cataract affect AH composition, by increasing the abundance of certain proteins,8,9 the AH proteome of cataract patients was also used as a control in this study. We first investigated the global proteome of the aqueous humor as a first step toward the identification of biomarkers for ocular diseases. The gold standard for separating complex proteomic samples is currently 2D gel electrophoresis, despite the limitations of this technique. Membrane proteins and proteins with extreme pI (isoelectric point) values may go undetected, together with very large or Journal of Proteome Research 2010, 9, 789–797 789 Published on Web 11/25/2009
research articles
Escoffier et al. in a particular context, as this study was based on determination of a specific proteome through qualitative analysis (presence or absence of proteins). We assessed the advantages and disadvantages of sample pooling and individual analysis strategies for 2D-LC-MS/MS experiments.
Materials and Methods
Figure 1. Procedures of AH sample pooling and individual analysis by 2D-LC-MS/MS method. For more details, see Materials and Methods.
very small proteins. The volume of human AH samples is typically between 50 and 150 µL for cataract surgery. However, in other ocular diseases, the sample volume may be much smaller, in some cases even less than 10 µL. It is therefore important to select the proteomic analysis method to be used for such approaches with care, based on a global proteomic analysis of a complex mixture with a limited volume and low protein concentration. Alternative methods must, therefore, be highly sensitive, with significant resolving power. Liquid chromatography (LC)-based separation techniques directly coupled to mass spectrometry (MS) were developed to overcome these disadvantages and to provide an alternative to 2D-E.10 Two-dimensional liquid chromatography (2D-LC/ MS) techniques adapted for peptide separation have been shown to identify a large number of proteins from samples with complex proteome.11 This strategy has been used to explore the vitreous humor proteome and has identified a candidate biomarker of proliferative vitreoretinopathy.12 This work resulted in the identification of 255 distinct proteins from the vitreous humor proteome, a larger number than previously identified by vitreous protein separation performed by 2D-E. 2D-LC-MS/MS provides more information about the proteins in samples with a low protein content than 2D-E, because reverse-phase separation can be carried out at nanoflow scale, and this approach can, therefore, be applied to AH analysis. Yu et al., like most researchers working in this field, applied a sample pooling strategy based on the analysis of proteins from the mixture of all patients samples in a particular group.13 Pooling has the potential advantage of decreasing overall variability due to differences between individuals. Sample pooling strategies for DIGE experiments have recently been investigated for quantitative proteomic purposes, and no evidence of bias triggered by sample pooling was found.14 However, no previous study has compared sample pooling and individual analysis by the 2D-LC-MS/MS method, estimating the gain and loss of information associated with each strategy. We determined the whole protein profile of AH from cataract patients, comparing the two methods with a view to their use in future studies aiming to identify biomarkers for ocular diseases. The first method involved analyzing each cataract sample separately, whereas the second involved pooling all the cataract samples (Figure 1). In the second method, two experimental procedures were confronted. The first procedure consisted in the digestion of all proteins from each sample separately before pooling the peptides (peptide pool) and the second procedure consisted in pooling proteins from each sample and digesting the protein pool (protein pool). All methods were compared 790
Journal of Proteome Research • Vol. 9, No. 2, 2010
Subjects and Samples. AH samples were collected at the Department of Ophthalmology, Pitie´-Salpeˆtrie`re Hospital, Paris, France. Sixty-three patients undergoing cataract surgery were enrolled in this study. AH samples were obtained by anterior chamber paracentesis with a 30-gauge needle. Samples of AH (20-200 µL) were immediately frozen at -80 °C in 1.5 mL Eppendorf microtubes and stored frozen until required. The total protein content of the AH samples was determined according to Bradford’s method (Bio-Rad Laboratories), following the kit manufacturer’s protocol. We subjected 16 AH samples to the following procedure (Figure 1). The 16 AH samples were separately subjected to insolution tryptic digestion before pooling the peptides (peptide mixture) or not (individual analysis). For protein mixture analysis, 1 µL from each of cataract sample was mixed together and the mixture was subjected to in-solution tryptic digestion. The procedure was approved by the local Institutional Review Board (IRB). In-Solution Tryptic Digestion. For in-solution reduction, we added 5 µL of 45 mM dithiothreitol (DTT) in 50 mM ammonium bicarbonate to 10 µL of AH and incubated the mixture for 15 min at 95 °C. We added 5 µL of 100 mM iodoacetamide in 50 mM ammonium bicarbonate for protein alkylation and incubated the mixture in the dark, at room temperature, for 15 min. Proteins in solution were digested by overnight incubation with 20 ng/µL trypsin (Promega) at 37 °C, with a 1:50 (w/w) trypsin-to-protein ratio. Peptides were then dried under vacuum with a Speed-Vac (Thermo Savant) and stored at -20 °C until 2D-LC-MS/MS analysis. Mass Spectrometry. Peptides were solubilized in 20 µL of 0.1% formic acid in water. Then, a volume of 7 µL of the 16 HA samples was subjected separately to 2D-LC-MS/MS for individual analysis. For analyses of peptide mixtures, we mixed 3 µL of solubilized peptides from each sample together. We analyzed 7 µL of the digested protein mix. Peptides from protein and peptide mixtures were analyzed three times on an automated 2D nano-LC-MS/MS system (Ultimate 3000, LC Packing directly interfaced to HCT Ultra, Bruker Daltonics, with a nanoelectrospray ionization source). The samples were loaded onto a strong cation exchange (SCX) trap column (LC Packing, Dionex). A multistep ammonium acetate gradient (5 mM, 10 mM, 15 mM, 20 mM, 35 mM, 50 mM, 75 mM, 100 mM, 250 mM, 500 mM, 1 M) was used to transfer peptides from the SCX column onto an RP trap column (C18, LC Packing) for desalting at a flow rate of 20 µL/min. This trap column was sequentially connected in line with an analytical C18 column, from which each SCX LC fraction was eluted with a linear step gradient at a flow rate of 200 nL/min (with 0.1% formic acid in DDI water as the mobile phase). The Ultimate 3000 (LC Packing) was used to deliver mobile phase A (0.1% formic acid in water) and B (0.1% formic acid, 5% water, 95% acetonitrile). We used a linear gradient of 0-15% B over 5 min, followed by 15-60% B over 60 min, increasing to 100% B over 10 min. The HPLC pump and mass spectrometer were controlled by Esquire Control Software (Bruker Daltonics, Inc., Billerica, MA). The capillary voltage was kept at 2000 V, and the desolvation temperature
Pooling Aqueous Humor Samples
research articles
Figure 2. Protein concentration and number of proteins identified in aqueous humor samples from patients with cataract. No correlation was observed between the amount of protein and the number of proteins identified. The histogram indicates protein concentration and the curve indicates the number of proteins identified for each sample.
was maintained at 250 °C. The maximum ion injection time was set at 10 ms, with an ion charge control value of 200 000. MS/MS fragmentation of the six most intense precursor ions in the spectrum was performed automatically, with an exclusion window of 0.5 min. A blank was inserted between analyses and replicates. Data Processing and Database Search. The 12-step 2D LCMS/MS experiments were processed with Data Analysis software (Bruker Daltonics) to create a single MASCOT generic file (MGF). MGF files were used to search against human protein database NCBI (downloaded on December 2007 from ftp:// ftp/ncbi.nlm.nih.gov/ using MASCOT (Mascot Daemon, version 2.2, Matrix Science, London, U.K.)). The search parameters were as follows: 1+, 2+ and 3+ charged peptides, one missed cleavage, mass tolerance of (0.5 Da for peptides and MS/MS, carbamidomethyl and methionine oxidation were permitted. Proteins with one or more unique peptide identification (p e 0.5) were considered to be present in aqueous humor. Peptide assignments with a MASCOT score
