618
DOI 10.1002/pmic.201400102
Proteomics 2015, 15, 618–623
TECHNICAL BRIEF
Integrated approach using multistep enzyme digestion, TiO2 enrichment, and database search for in-depth phosphoproteomic profiling Dohyun Han1,2,3 , Jonghwa Jin1,2 , Jiyoung Yu1,2 , Kyunggon Kim1,2 and Youngsoo Kim1,2,3 1
Departments of Biomedical Sciences, Medical Research Center, Seoul National University College of Medicine, Seoul, Korea 2 Biomedical Engineering, Medical Research Center, Seoul National University College of Medicine, Seoul, Korea 3 Institute of Medical & Biological Engineering, Medical Research Center, Seoul National University College of Medicine, Seoul, Korea Protein phosphorylation is a major PTM that regulates important cell signaling mechanisms. In-depth phosphoproteomic analysis provides a method of examining this complex interplay, yielding a mechanistic understanding of the cellular processes and pathogenesis of various diseases. However, the analysis of protein phosphorylation is challenging, due to the low concentration of phosphoproteins in highly complex mixtures and the high variability of phosphorylation sites. Thus, typical phosphoproteome studies that are based on MS require large amounts of starting material and extensive fractionation steps to reduce the sample complexity. To this end, we present a simple strategy (integrated multistep enzyme digestion, enrichment, database search–iMEED) to improve coverage of the phosphoproteome from lower sample amounts which is faster than other commonly used approaches. It is inexpensive and adaptable to low sample amounts and saves time and effort with regard to sample preparation and mass spectrometric analysis, allowing samples to be prepared without prefractionation or specific instruments, such as HPLC. All MS data have been deposited in the ProteomeXchange with identifier PXD001033 (http://proteomecentral.proteomexchange.org/dataset/PXD001033).
Received: March 19, 2014 Revised: July 9, 2014 Accepted: August 22, 2014
Keywords: LC-MS/MS / Phosphoproteome / Post-translational modification / Proteome profiling / Systems biology
Additional supporting information may be found in the online version of this article at the publisher’s web-site
Phosphorylation is a reversible PTM that regulates proteins that mediate many essential processes, such as the cell cycle, proliferation, apoptosis, metabolism, signal transduction, and development [1]. Dysregulation of protein phosphorylation is a hallmark of various diseases [2, 3], necessitating comprehensive characterization of phosphorylation events. Large-scale phosphoproteomic studies that are based on MS Correspondence: Professor Youngsoo Kim, Department of Biomedical Engineering, Seoul National University College of Medicine, 28 Yongon-Dong, Chongno-Ku, Seoul 110-799, Korea E-mail: [email protected] Fax: +82-2-741-0253 Abbreviations: FASP, filter-aided sample preparation; GRAVY, grand average of hydropathy; HCD, higher-energy collisional dissociation; iMEED, integrated multistep enzyme digestion, enrichment, database search; TiO2 , titanium dioxide C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
provide an overall image of the phosphorylation-based signaling network in cells and tissues, which can be used to screen biomarkers and drug targets for disease diagnosis and treatment [2]. However, in-depth phosphoproteomic profiling is technically challenging due to the low abundance of phosphoproteins and high variability of phosphorylation sites [4]. To address these technical limitations, many strategies, including phosphopeptide enrichment [5, 6] and extensive peptide fractionation [7, 8], have been developed. Nevertheless, these approaches have common drawbacks, requiring large amounts of starting and MS injection material, extensive fractionation steps, and hundreds of hours of mass spectrometric measurement times to ensure that a sufficient number of phosphopeptides are identified. Thus, a complementary
Colour Online: See the article online to view Figs. 1–3 in colour. www.proteomics-journal.com
Proteomics 2015, 15, 618–623
workflow that effects simple, rapid, and comprehensive determination of the phosphoproteome must be developed. In this study, we describe an enhanced phosphoproteomic strategy (integrated multistep enzyme digestion, enrichment, database search, or iMEED) that improves coverage of the phosphoproteome from lower sample amounts for MS injection and with less experimental MS time than other approaches. As shown in Fig. 1, we combined multistep enzyme digestion and filter-aided sample preparation (FASP) with a multistep phosphopeptide enrichment method, followed by higher-energy collisional dissociation (HCD) on a Q-Exactive and multiple-database searches including SequestHT and MS Amanda [9]. Even compared with our previous studies using conventional phosphoproteomic methods, the iMEED method identified more phosphorylation sites. Approximately 1 × 107 BV-2 mouse microglial cells were cultured in 100 mm culture dish with DMEM, supplemented with 5% heat-inactivated FBS, 100 U/mL penicillin, and 100 mg/mL streptomycin, at 37⬚C and 5% CO2 and lysed in SDT buffer (100 mM Tris, pH 7.4, 4% SDS, and 0.1 M DTT). Protein concentration was measured using a BCA reducing agent compatibility assay kit (Thermo Scientific, Rockford, IL, USA). The procedure of sample preparation was performed in triplicate for biological replicates. Proteins were digested by multistep FASP as described [10, 11]. Proteins (500 g) were loaded onto a 30-k spin fil-
619 ter (EMD Millipore, Billerica, MA, USA), and buffer was exchanged with UA solution (8 M UREA in 0.1 M Tris, pH 8.5) by centrifugation. Reduced cysteines were alkylated with IAA solution (0.05 M iodoacetamide in UA solution) for 30 min at room temperature in the dark. Additional buffer was exchanged with 40 mM ammonium bicarbonate. The first protein digestion step was conducted at 37⬚C overnight with trypsin (enzyme-to-substrate ratio [w/w] of 1:100). Next, the peptides were collected by centrifugation. For the iMEED approach, after the filter units were washed sequentially with water, UA buffer, and 40 mM ammonium bicarbonate, the second digestion was performed with trypsin (enzyme-tosubstrate ratio [w/w] of 1:100). All resulting peptides were acidified with 1% TFA and desalted using homemade C18StageTips as described [10]. Desalted samples were dried in a vacuum centrifuge and stored at −80⬚C until titanium dioxide (TiO2 ) enrichment. For the iMEED approach, phosphopeptides were enriched using TiO2 beads as described with some modifications [10]. Peptides from 500 g of protein were dissolved with 200 L loading buffer (1 M glycolic acid, 80% acetonitrile, and 2% TFA) and mixed with 1 mg of TiO2 beads (GL Sciences, Japan) in a tube and rotated gently for 15 min. After centrifugation, the supernatants were transferred to new Eppendorf tubes and incubated with TiO2 beads again as described above. The beads were washed sequentially with 200 L 5 mM KH2 PO4 ,
Figure 1. Workflow of the iMEED approach.
C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.proteomics-journal.com
620
D. Han et al.
30% ACN, 350 mM KCl and 200 L 40% acetonitrile, 2% TFA and resuspended in 100 L 80% ACN, 2% TFA. The beads were loaded onto homemade StageTip-C8 in 200 L yellow tips that were preconditioned with 80% acetonitrile, 2% TFA, washed once with the same buffer, and eluted with 100 L elution buffer 1 (5% ammonia) and 100 L elution buffer 2 (10% ammonia and 40% acetonitrile). Ammonia and acetonitrile were evaporated using a vacuum centrifuge, and the peptides were acidified in 1% TFA and desalted using homemade StageTip-C18. Desalted phosphopeptide samples were dried in a vacuum centrifuge and stored at −80⬚C until LC-MS/MS analysis. The phosphopeptide samples from the iMEED approach were analyzed on an LC-MS/MS instrument, consisting of a Nanoflow Easy-nLC 1000 (Proxeon Biosystems, Odense, Denmark) that was connected to a Q Exactive mass spectrometer (Thermo Scientific, Bremen, Germany) through a nanoelectrospray ion source (denoted as HR-HCD), as described with some modifications [12]. Peptides were separated on a two˚ 5 m column system, comprising a trap column (100 A, ˚ particle, 75 m × 3 cm) and an analytical column (200 A, 3 m particle, 75 m × 10 cm) with a 200 min gradient from 2 to 40% acetonitrile at a flow rate of 300 nL/min, and analyzed on a mass spectrometer. MS spectra were acquired on an Orbitrap analyzer with a mass range of 300–1800 m/z and 70 000 resolutions at m/z 200 (Q Exactive). HCD scans were acquired at a resolution of 15 000. HCD peptide fragments were acquired at 27% NCE. Each fraction was analyzed in quadruplicate. All raw files were processed in Proteome Discoverer, version 1.4 using the SequestHT or MS Amanda algorithm [9] against the IPI mouse database (version 3.87, 59 534 entries), containing forward and reverse protein sequences. MS/MS searches were performed with the following parameters: full enzyme digest using trypsin (After KR/-) with up to two missed cleavages; a precursor ion mass tolerance of 10 ppm (monoisotopic mass) for iMEED; HCD fragment ion mass tolerance of 0.02 Da; carbamidomethylation as a fixed modification; and oxidation of methione and phosphorylation of serine, thereonine, and tyrosine as variable modifications. The Percolator semisupervised machine learning algorithm [13] in Proteome Discoverer was used to merge the search results from multiple engines (SequestHT and MS Amanda) and subsequently filter peptide spectrum matches at a false discovery rate < 1% using q-values. In addition, peptide spectrum matches that received a search engine rank of 1 were considered for further analysis. Location of phosphorylation sites from HCD (Q-Exactive) was determined using the PhosphoRS 3.1 algorithm [14] in Proteome Discoverer. A summary of the iMEED datasets is shown in Supporting Information Fig. 1. Phosphopeptides with a PhosphoRS probability ࣙ 0.99 were considered to have confident phosphorylation site localization. The pI values of the phosphopeptides were calculated using pI Calculator (http://trac.nbic.nl/picalculator/). Detailed information on the sample preparation and analysis versus our pre C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Proteomics 2015, 15, 618–623
vious method is described in Supporting Information. All raw MS data files and Proteome Discoverer output files have been deposited into the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via the PRIDE partner repository with the following dataset identifier: PXD001033. To increase the phosphopeptide enrichment and coverage of the phosphoproteome using low amounts of starting material compared with our established workflow [10], we developed the iMEED approach in the BV-2 cell line (Fig. 1 and Supporting Information Table 1). In biological replicate 1, 500 g of BV-2 cell lysate was used as starting material for the enzymatic digestions and multistep TiO2 enrichment. The first round of TiO2 enrichment (E1) from the initial peptide digests was performed as described [10] with some modifications, including the use of batch-mode enrichment and sequential elution. One-twentieth of an aliquot (1/20; 5%) of the E1 sample (amount enriched from 25 g of starting material) per injection was analyzed by HR-HCD and a SequestHT database search demonstrating that 5573 phosphopeptides were identified from the quadruplicate technical replicates using 20% of sample. Among these phosphopeptides, 4114 (73.8%) were phosphopeptides with confident localization (PhosphoRS 3.1 probability ࣙ 0.99). To compare with our established approach [10], one-tenth of an aliquot (1/10) of the sample (denoted as pE1) that was enriched using our previous methods from 500 g of starting material was analyzed in quadruplicate technical replicates. In pE1 dataset, we identified 3552 unique phosphopeptides in total, 1719 of which contained confident phosphorylation sites (Supporting Information Table 2). The average PhosphoRS peptide score of all identified phosphopeptides was 263.8 in fraction E1 of iMEED versus 94.1 in fraction pE1 from previous methods (Supporting Information Fig. 2). Despite the small MS injection amount, 2.4 times as many phosphopeptides were identified with confidence and a 2.8-fold higher PhosphoRS score was observed in E1 of iMEED compared with pE1, which might be attributed to the optimization of TiO2 enrichment and use of HR-HCD. To increase the recovery of phosphopeptides, we performed a second round of TiO2 enrichment (E2) using the flowthrough of E1. In addition, a third TiO2 enrichment (E3) step was performed using peptide samples from the second digestion. We identified 3043 and 3136 unique phosphopeptides in the E2 and E3 samples, respectively, 2324 (76.4%) and 2541 (81%) of which were identified as confident phosphopeptides. Notably, inconsistent with another study [5], a similar number of phosphopeptides was identified from samples E2 and E3, which presumably resulted from the properties of sample E3, which contained newly derived phopshopeptides from the second digestion. In the multistep enrichment, three fractions (E1, E2, E3) yielded 8071 unique phosphopeptides, 5940 of which (73%) were matched to confident phosphopeptides. Consequently, we identified 3.5 times as many confident phosphopeptides compared with our established workflow [10] (Fig. 2A). www.proteomics-journal.com
Proteomics 2015, 15, 618–623
621
Figure 2. Evaluation of the iMEED approach. (A) Comparison of identified unique phosphosites in each enriched fraction of biological replicate 1. The numbers of phosphor-serine (blue), phosphor-threonine (red), and phosphor-tyrosine (green) sites with confident localization (PhosphoRS probability ࣙ 0.99) are also shown. The multistep enrichment comprised a first (E1), second (E2), and third (E3) round of TiO2 enrichment. The number of phosphorylation sites identified from sample pE1 using our previous method is shown as the control. Percentages of identified phosphopeptides with confident localization from all phosphopeptides are shown with a dashed line. (B) Comparison between SequestHT_only and the combination of SequestHT and MS Amanda. In three fractions of the three biological replicates, the numbers of unique phosphopeptides identified using SequestHT_only and the combination of SequestHT and MS Amanda are shown in blue and red, respectively. (C) Overlap of unique phosphopeptide with high-confidence localization identified using the iMEED approach is shown in the Venn diagram. (D) Percentages of overlap between phosphopeptides identified in each round of the multistep enrichment. Overlap percentages between technical replicates are shown as blue bars, whereas overlap percentages between biological replicates are shown as red bars. Error bars represent percentage distributions of overlapping peptides between technical replicates or biological replicates.
Next, we sought to maximize coverage of the phosphoproteome relative to the established workflow. We compared the number of confident phosphopeptides that were found using several database search engines, combined with SequestHT and MS Amanda, with those found in SequestHT only (Fig. 2B). The multiple-database search identified an average of 7% more unique confident phosphopeptides than the single database search in all fractions (Fig. 2B). Finally, to determine the advantages and reproducibility of the iMEED approach, we tested this method in three biological replicates with four technical replicates of the BV-2 cell line, analyzing them in 36 LC-MS/MS runs (Fig. 2C and Supporting Information Table 3). A total of 8807 unique phosphopeptides were identified with high confidence; 3365 phosphopeptides were common to all three biological replicates, wherein 6355, 5395, and 5717 unique phosphopeptides were C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
detected in replicates 1, 2, and 3, respectively (Fig. 2C). Approximately 75% of the unique phosphopeptides overlapped with those identified among the four technical replicates in each sample (Fig. 2D and Supporting Information Table 4). For each sample, there was on average an overlap of 68% of all identified unique phosphopeptides between all pairs of biological replicates. These results suggest that iMEED is a reasonable and reproducible method. Further, the technical and biological variability was expressed by calculating the Pearson correlation value (R) with MS1 intensities for the technical and biological replicates. First, the MS1 intensity of phosphoproteins was calculated by summing the normalized intensities of all phosphopeptides that were assigned to the protein group with high confidence. Supporting Information Fig. 3 summarizes the correlation of phosphoprotein intensities of the four technical replicates in www.proteomics-journal.com
622
D. Han et al.
the three biological replicates for each sample. The R value of the technical replicates ranged from 0.891 to 0.984 in the E1 samples, 0.788 to 0.957 in the E2 samples, and 0.813 to 0.987 in the E3 samples. In addition, the average R value of the biological replicates was 0.91, 0.65, and 0.89 in the E1, E2, and E3 samples, respectively. Then, we determined its quantitative reproducibility between biological replicates, including enrichments and batchto-batch variation, by examining the level of independent phosphopeptide intensity. As shown in Supporting Information Tables 5 and 6, the R value of the biological replicates ranged from 0.917 to 0.932 in the E1 samples, 0.8 to 0.88 in the E2 samples, and 0.873 to 0.912 in the E3 samples. Notably, summing the MS1 intensities of phosphopeptides from fractions E1, E2, and E3 demonstrated high reproducibility between biological replicates, with an average Pearson correlation of 0.945, reflecting the tremendous potential of iMEED with regard to label-free quantitation of phosphopeptides. Consequently, the technical and biological variations were significantly low, indicating that our novel strategy is robust and highly reproducible. Three rounds of TiO2 enrichment, combined with multistep FASP, were sufficient and effective in identifying phosphopeptides from whole-cell lysates in depth. Only 9% of phosphopeptides appeared in all three fractions (E1, E2, and E3), whereas 91% of phosphopeptides were detected in 1 or 2 fractions (Supporting Information Fig. 4). These results indicate that our approach improves the comprehensiveness of the phosphoproteome in mass-limited samples. Particularly, the iMEED method improved the identification of multiphosphorylated peptides, although many studies have reported that TiO2 tends to isolate monophosphorylated peptides, not multiply phosphorylated peptides [15]. Without using adducts, such as citric acid [15], to improve the detection of multiphosphorylated peptides, nearly 30% of
Proteomics 2015, 15, 618–623
unique phosphopeptides that were identified were multiphosphorylated by iMEED. As shown in Supporting Information Fig. 5, most multiply phosphorylated peptides were detected following the first (E1) round of TiO2 enrichment, whereas with the second round (E2), an average of 60 such peptides were identified in three biological replicates. However, the proportion of multiply phosphorylated peptides in the third round (E3) of TiO2 enrichment from the second digestion was higher, indicating that further enrichment of the second digestion improves the identification of multiphosphorylated peptides. Further, the phosphopeptides that were enriched in each of the multistep-enriched samples had dis-similar chemical properties (Supporting Information Table 7). In samples E1 and E3, most of the identified phosphopeptides were acidic phosphopeptides with low pI values. The highest proportion (30%) of phosphopeptides in samples E1 and E3 had pI values ranging from 3.5 to 4 (Fig. 3A). The percentage of phosphopeptides with pI values below 4 decreased from 60% in E1 and E3 to 14% in E2. Only 4.5% of all phosphopeptides that were identified in E1 and E3 were basic, with pI values higher than 7, whereas this percentage rose to 16% in the E2 sample (Fig. 3A). An average of 53 phosphopeptides with pI values greater than 8.5 were identified in E2, versus ten phosphopeptides in E1 and E3. This distribution of phosphopeptides by pI value indicates that our multistep enrichment strategy increases coverage of the phosphoproteome. Moreover, the grand average of hydropathy (GRAVY) values—the average hydropathy values for all amino acids in a protein—differed between fractions. In the E1 and E3 samples, 12 and 18% of all identified phosphopeptides had GRAVY values below −2, versus 6.6% in the E2 sample (Fig. 3B). However, in sample E2, 30% of total phosphopeptides had GRAVY values higher than 0, compared with 15% in E1 and E3. These differences suggest that hydrophilic and
Figure 3. Distribution of pI and GRAVY values in multistep enrichment. (A) Histogram of percentage of identified phosphopeptides by pI value in the E1 (blue), E2 (green), and E3 (red) fractions. Cumulative percentages of phosphopeptides from pI 1.5 to 10 are shown with dashed lines. (B) Histogram of percentage of identified phosphopeptides by GRAVY value in the E1 (blue), E2 (green), and E3 (red) fractions. Cumulative percentages of phosphopeptides with GRAVY values of −4 to 3.5 are shown with dashed lines.
C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.proteomics-journal.com
Proteomics 2015, 15, 618–623
hydrophobic phosphopeptides are more likely to be identified by multistep enrichment and that multistep enrichment identifies disparate populations of phosphopeptides. We have found evidence to support combining multistep FASP and multistage enrichment to reduce sample manipulation and improve the robustness and reproducibility of the overall workflow without prefractionation or the need for specific instruments. Further, in addition multiple-database searches, analysis by high-resolution MS yields more highconfidence phosphopeptides than our previous method [10]. The iMEED method is a relatively easy and efficient approach that identifies phosphopeptides from less MS injection material compared with other recently reported approaches [5, 6, 15] (Supporting Information Table 8). Thus, iMEED has generated the most comprehensive phosphoproteome of murine microglia to date, comprising 8807 unique phosphopeptides in 36 LC-MS/MS runs from 300 g of MS injection material, representing an approximately twofold improvement (4000 unique phosphopeptides) over our previous method in the same MS analysis time. In addition, compared with the human, rat, and mouse phosphorylation sites in the PhosphoSitePlus database (www.phosphosite.org), 1262 phosphosites were unrecorded (Supporting Information Fig. 6). In summary, the iMEED approach has tremendous potential in identifying the phosphoproteome in a comprehensive, simple, and reproducible manner by combining multistep FASP, multistep enrichment, and a multiple-database search, based on high-resolution MS. Moreover, our strategy enables one to perform a rapid in-depth analysis of the phosphoproteome using relatively small sample amounts. Thus, the remaining 80% of a sample can be used in other proteomic approaches, such as multiple reaction monitoring. Due to the considerably small MS injection amounts that we used in this study, we are developing a sequential enrichment method of various PTMs per sample for future experiments to further minimize the sample amount that is needed and the instrument run time. The MS proteomics data in this paper have been deposited in the ProteomeXchange Consortium (http://proteomecentral. proteomexchange.org) via the PRIDE partner repository [16]: dataset identifier PXD001033. This work was supported by a National Research Foundation of Korea grant (No. 2011 –0030740) and the Proteogenomic Research Program, funded by the Korean government [MISP]. This work was also supported by the Industrial Strategic Technology Development Program (#10045352), funded by the Ministry of Knowledge Economy (MKE, Korea). The authors have declared no conflict of interest.
References [1] Hunter, T., Signaling–2000 and beyond. Cell 2000, 100, 113– 127.
C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
623 [2] Harsha, H. C., Pandey, A., Phosphoproteomics in cancer. Mol. Oncol. 2010, 4, 482–495. [3] Iliuk, A. B., Tao, W. A., Is phosphoproteomics ready for clinical research? Clin. Chim. Acta 2013, 420, 23–27. [4] Engholm-Keller, K., Larsen, M. R., Technologies and challenges in large-scale phosphoproteomics. Proteomics 2013, 13, 910–931. [5] Yue, X. S., Hummon, A. B., Combination of multistep IMAC enrichment with high-pH reverse phase separation for indepth phosphoproteomic profiling. J. Proteome Res. 2013, 12, 4176–4186. [6] Tsai, C. F., Hsu, C. C., Hung, J. N., Wang, Y. T. et al., Sequential phosphoproteomic enrichment through complementary metal-directed immobilized metal ion affinity chromatography. Anal. Chem. 2014, 86, 685–693. [7] Zarei, M., Sprenger, A., Metzger, F., Gretzmeier, C., Dengjel, J., Comparison of ERLIC-TiO2, HILIC-TiO2, and SCX-TiO2 for global phosphoproteomics approaches. J. Proteome Res. 2011, 10, 3474–3483. [8] Engholm-Keller, K., Hansen, T. A., Palmisano, G., Larsen, M. R., Multidimensional strategy for sensitive phosphoproteomics incorporating protein prefractionation combined with SIMAC, HILIC, and TiO(2) chromatography applied to proximal EGF signaling. J. Proteome Res. 2011, 10, 5383– 5397. [9] Dorfer, V., Pichler, P., Stranzl, T., Stadlmann, J. et al., MS Amanda, a Universal Identification Algorithm Optimized for High Accuracy Tandem Mass Spectra. J. Proteome Res. 2014, 13, 3679–3684. [10] Han, D., Moon, S., Kim, Y., Ho, W. K. et al., Comprehensive phosphoproteome analysis of INS-1 pancreatic betacells using various digestion strategies coupled with liquid chromatography-tandem mass spectrometry. J. Proteome Res. 2012, 11, 2206–2223. [11] Wisniewski, J. R., Zougman, A., Nagaraj, N., Mann, M., Universal sample preparation method for proteome analysis. Nat. Methods 2009, 6, 359–362. [12] Han, D., Moon, S., Kim, Y., Kim, J. et al., In-depth proteomic analysis of mouse microglia using a combination of FASP and StageTip-based, high pH, reversed-phase fractionation. Proteomics 2013, 13, 2984–2988. [13] Spivak, M., Weston, J., Bottou, L., Kall, L. et al., Improvements to the percolator algorithm for Peptide identification from shotgun proteomics data sets. J. Proteome Res. 2009, 8, 3737–3745. [14] Taus, T., Kocher, T., Pichler, P., Paschke, C. et al., Universal and confident phosphorylation site localization using phosphoRS. J. Proteome Res. 2011, 10, 5354–5362. [15] Zhao, X., Wang, Q., Wang, S., Zou, X. et al., Citric acidassisted two-step enrichment with TiO2 enhances the separation of multi- and monophosphorylated peptides and increases phosphoprotein profiling. J. Proteome Res. 2013, 12, 2467–2476. [16] Vizca´ıno, J. A, Cot ˆ e, ´ R. G., Csordas, A., Dianes, J. A. et al., The PRoteomics IDEntifications (PRIDE) database and associated tools: status in 2013. Nucleic Acids Res. 2012, doi: 10.1093/nar/gks1262.
www.proteomics-journal.com
DOI 10.1002/pmic.201400102
Proteomics 2015, 15, 618–623
TECHNICAL BRIEF
Integrated approach using multistep enzyme digestion, TiO2 enrichment, and database search for in-depth phosphoproteomic profiling Dohyun Han1,2,3 , Jonghwa Jin1,2 , Jiyoung Yu1,2 , Kyunggon Kim1,2 and Youngsoo Kim1,2,3 1
Departments of Biomedical Sciences, Medical Research Center, Seoul National University College of Medicine, Seoul, Korea 2 Biomedical Engineering, Medical Research Center, Seoul National University College of Medicine, Seoul, Korea 3 Institute of Medical & Biological Engineering, Medical Research Center, Seoul National University College of Medicine, Seoul, Korea Protein phosphorylation is a major PTM that regulates important cell signaling mechanisms. In-depth phosphoproteomic analysis provides a method of examining this complex interplay, yielding a mechanistic understanding of the cellular processes and pathogenesis of various diseases. However, the analysis of protein phosphorylation is challenging, due to the low concentration of phosphoproteins in highly complex mixtures and the high variability of phosphorylation sites. Thus, typical phosphoproteome studies that are based on MS require large amounts of starting material and extensive fractionation steps to reduce the sample complexity. To this end, we present a simple strategy (integrated multistep enzyme digestion, enrichment, database search–iMEED) to improve coverage of the phosphoproteome from lower sample amounts which is faster than other commonly used approaches. It is inexpensive and adaptable to low sample amounts and saves time and effort with regard to sample preparation and mass spectrometric analysis, allowing samples to be prepared without prefractionation or specific instruments, such as HPLC. All MS data have been deposited in the ProteomeXchange with identifier PXD001033 (http://proteomecentral.proteomexchange.org/dataset/PXD001033).
Received: March 19, 2014 Revised: July 9, 2014 Accepted: August 22, 2014
Keywords: LC-MS/MS / Phosphoproteome / Post-translational modification / Proteome profiling / Systems biology
Additional supporting information may be found in the online version of this article at the publisher’s web-site
Phosphorylation is a reversible PTM that regulates proteins that mediate many essential processes, such as the cell cycle, proliferation, apoptosis, metabolism, signal transduction, and development [1]. Dysregulation of protein phosphorylation is a hallmark of various diseases [2, 3], necessitating comprehensive characterization of phosphorylation events. Large-scale phosphoproteomic studies that are based on MS Correspondence: Professor Youngsoo Kim, Department of Biomedical Engineering, Seoul National University College of Medicine, 28 Yongon-Dong, Chongno-Ku, Seoul 110-799, Korea E-mail: [email protected] Fax: +82-2-741-0253 Abbreviations: FASP, filter-aided sample preparation; GRAVY, grand average of hydropathy; HCD, higher-energy collisional dissociation; iMEED, integrated multistep enzyme digestion, enrichment, database search; TiO2 , titanium dioxide C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
provide an overall image of the phosphorylation-based signaling network in cells and tissues, which can be used to screen biomarkers and drug targets for disease diagnosis and treatment [2]. However, in-depth phosphoproteomic profiling is technically challenging due to the low abundance of phosphoproteins and high variability of phosphorylation sites [4]. To address these technical limitations, many strategies, including phosphopeptide enrichment [5, 6] and extensive peptide fractionation [7, 8], have been developed. Nevertheless, these approaches have common drawbacks, requiring large amounts of starting and MS injection material, extensive fractionation steps, and hundreds of hours of mass spectrometric measurement times to ensure that a sufficient number of phosphopeptides are identified. Thus, a complementary
Colour Online: See the article online to view Figs. 1–3 in colour. www.proteomics-journal.com
Proteomics 2015, 15, 618–623
workflow that effects simple, rapid, and comprehensive determination of the phosphoproteome must be developed. In this study, we describe an enhanced phosphoproteomic strategy (integrated multistep enzyme digestion, enrichment, database search, or iMEED) that improves coverage of the phosphoproteome from lower sample amounts for MS injection and with less experimental MS time than other approaches. As shown in Fig. 1, we combined multistep enzyme digestion and filter-aided sample preparation (FASP) with a multistep phosphopeptide enrichment method, followed by higher-energy collisional dissociation (HCD) on a Q-Exactive and multiple-database searches including SequestHT and MS Amanda [9]. Even compared with our previous studies using conventional phosphoproteomic methods, the iMEED method identified more phosphorylation sites. Approximately 1 × 107 BV-2 mouse microglial cells were cultured in 100 mm culture dish with DMEM, supplemented with 5% heat-inactivated FBS, 100 U/mL penicillin, and 100 mg/mL streptomycin, at 37⬚C and 5% CO2 and lysed in SDT buffer (100 mM Tris, pH 7.4, 4% SDS, and 0.1 M DTT). Protein concentration was measured using a BCA reducing agent compatibility assay kit (Thermo Scientific, Rockford, IL, USA). The procedure of sample preparation was performed in triplicate for biological replicates. Proteins were digested by multistep FASP as described [10, 11]. Proteins (500 g) were loaded onto a 30-k spin fil-
619 ter (EMD Millipore, Billerica, MA, USA), and buffer was exchanged with UA solution (8 M UREA in 0.1 M Tris, pH 8.5) by centrifugation. Reduced cysteines were alkylated with IAA solution (0.05 M iodoacetamide in UA solution) for 30 min at room temperature in the dark. Additional buffer was exchanged with 40 mM ammonium bicarbonate. The first protein digestion step was conducted at 37⬚C overnight with trypsin (enzyme-to-substrate ratio [w/w] of 1:100). Next, the peptides were collected by centrifugation. For the iMEED approach, after the filter units were washed sequentially with water, UA buffer, and 40 mM ammonium bicarbonate, the second digestion was performed with trypsin (enzyme-tosubstrate ratio [w/w] of 1:100). All resulting peptides were acidified with 1% TFA and desalted using homemade C18StageTips as described [10]. Desalted samples were dried in a vacuum centrifuge and stored at −80⬚C until titanium dioxide (TiO2 ) enrichment. For the iMEED approach, phosphopeptides were enriched using TiO2 beads as described with some modifications [10]. Peptides from 500 g of protein were dissolved with 200 L loading buffer (1 M glycolic acid, 80% acetonitrile, and 2% TFA) and mixed with 1 mg of TiO2 beads (GL Sciences, Japan) in a tube and rotated gently for 15 min. After centrifugation, the supernatants were transferred to new Eppendorf tubes and incubated with TiO2 beads again as described above. The beads were washed sequentially with 200 L 5 mM KH2 PO4 ,
Figure 1. Workflow of the iMEED approach.
C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.proteomics-journal.com
620
D. Han et al.
30% ACN, 350 mM KCl and 200 L 40% acetonitrile, 2% TFA and resuspended in 100 L 80% ACN, 2% TFA. The beads were loaded onto homemade StageTip-C8 in 200 L yellow tips that were preconditioned with 80% acetonitrile, 2% TFA, washed once with the same buffer, and eluted with 100 L elution buffer 1 (5% ammonia) and 100 L elution buffer 2 (10% ammonia and 40% acetonitrile). Ammonia and acetonitrile were evaporated using a vacuum centrifuge, and the peptides were acidified in 1% TFA and desalted using homemade StageTip-C18. Desalted phosphopeptide samples were dried in a vacuum centrifuge and stored at −80⬚C until LC-MS/MS analysis. The phosphopeptide samples from the iMEED approach were analyzed on an LC-MS/MS instrument, consisting of a Nanoflow Easy-nLC 1000 (Proxeon Biosystems, Odense, Denmark) that was connected to a Q Exactive mass spectrometer (Thermo Scientific, Bremen, Germany) through a nanoelectrospray ion source (denoted as HR-HCD), as described with some modifications [12]. Peptides were separated on a two˚ 5 m column system, comprising a trap column (100 A, ˚ particle, 75 m × 3 cm) and an analytical column (200 A, 3 m particle, 75 m × 10 cm) with a 200 min gradient from 2 to 40% acetonitrile at a flow rate of 300 nL/min, and analyzed on a mass spectrometer. MS spectra were acquired on an Orbitrap analyzer with a mass range of 300–1800 m/z and 70 000 resolutions at m/z 200 (Q Exactive). HCD scans were acquired at a resolution of 15 000. HCD peptide fragments were acquired at 27% NCE. Each fraction was analyzed in quadruplicate. All raw files were processed in Proteome Discoverer, version 1.4 using the SequestHT or MS Amanda algorithm [9] against the IPI mouse database (version 3.87, 59 534 entries), containing forward and reverse protein sequences. MS/MS searches were performed with the following parameters: full enzyme digest using trypsin (After KR/-) with up to two missed cleavages; a precursor ion mass tolerance of 10 ppm (monoisotopic mass) for iMEED; HCD fragment ion mass tolerance of 0.02 Da; carbamidomethylation as a fixed modification; and oxidation of methione and phosphorylation of serine, thereonine, and tyrosine as variable modifications. The Percolator semisupervised machine learning algorithm [13] in Proteome Discoverer was used to merge the search results from multiple engines (SequestHT and MS Amanda) and subsequently filter peptide spectrum matches at a false discovery rate < 1% using q-values. In addition, peptide spectrum matches that received a search engine rank of 1 were considered for further analysis. Location of phosphorylation sites from HCD (Q-Exactive) was determined using the PhosphoRS 3.1 algorithm [14] in Proteome Discoverer. A summary of the iMEED datasets is shown in Supporting Information Fig. 1. Phosphopeptides with a PhosphoRS probability ࣙ 0.99 were considered to have confident phosphorylation site localization. The pI values of the phosphopeptides were calculated using pI Calculator (http://trac.nbic.nl/picalculator/). Detailed information on the sample preparation and analysis versus our pre C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Proteomics 2015, 15, 618–623
vious method is described in Supporting Information. All raw MS data files and Proteome Discoverer output files have been deposited into the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via the PRIDE partner repository with the following dataset identifier: PXD001033. To increase the phosphopeptide enrichment and coverage of the phosphoproteome using low amounts of starting material compared with our established workflow [10], we developed the iMEED approach in the BV-2 cell line (Fig. 1 and Supporting Information Table 1). In biological replicate 1, 500 g of BV-2 cell lysate was used as starting material for the enzymatic digestions and multistep TiO2 enrichment. The first round of TiO2 enrichment (E1) from the initial peptide digests was performed as described [10] with some modifications, including the use of batch-mode enrichment and sequential elution. One-twentieth of an aliquot (1/20; 5%) of the E1 sample (amount enriched from 25 g of starting material) per injection was analyzed by HR-HCD and a SequestHT database search demonstrating that 5573 phosphopeptides were identified from the quadruplicate technical replicates using 20% of sample. Among these phosphopeptides, 4114 (73.8%) were phosphopeptides with confident localization (PhosphoRS 3.1 probability ࣙ 0.99). To compare with our established approach [10], one-tenth of an aliquot (1/10) of the sample (denoted as pE1) that was enriched using our previous methods from 500 g of starting material was analyzed in quadruplicate technical replicates. In pE1 dataset, we identified 3552 unique phosphopeptides in total, 1719 of which contained confident phosphorylation sites (Supporting Information Table 2). The average PhosphoRS peptide score of all identified phosphopeptides was 263.8 in fraction E1 of iMEED versus 94.1 in fraction pE1 from previous methods (Supporting Information Fig. 2). Despite the small MS injection amount, 2.4 times as many phosphopeptides were identified with confidence and a 2.8-fold higher PhosphoRS score was observed in E1 of iMEED compared with pE1, which might be attributed to the optimization of TiO2 enrichment and use of HR-HCD. To increase the recovery of phosphopeptides, we performed a second round of TiO2 enrichment (E2) using the flowthrough of E1. In addition, a third TiO2 enrichment (E3) step was performed using peptide samples from the second digestion. We identified 3043 and 3136 unique phosphopeptides in the E2 and E3 samples, respectively, 2324 (76.4%) and 2541 (81%) of which were identified as confident phosphopeptides. Notably, inconsistent with another study [5], a similar number of phosphopeptides was identified from samples E2 and E3, which presumably resulted from the properties of sample E3, which contained newly derived phopshopeptides from the second digestion. In the multistep enrichment, three fractions (E1, E2, E3) yielded 8071 unique phosphopeptides, 5940 of which (73%) were matched to confident phosphopeptides. Consequently, we identified 3.5 times as many confident phosphopeptides compared with our established workflow [10] (Fig. 2A). www.proteomics-journal.com
Proteomics 2015, 15, 618–623
621
Figure 2. Evaluation of the iMEED approach. (A) Comparison of identified unique phosphosites in each enriched fraction of biological replicate 1. The numbers of phosphor-serine (blue), phosphor-threonine (red), and phosphor-tyrosine (green) sites with confident localization (PhosphoRS probability ࣙ 0.99) are also shown. The multistep enrichment comprised a first (E1), second (E2), and third (E3) round of TiO2 enrichment. The number of phosphorylation sites identified from sample pE1 using our previous method is shown as the control. Percentages of identified phosphopeptides with confident localization from all phosphopeptides are shown with a dashed line. (B) Comparison between SequestHT_only and the combination of SequestHT and MS Amanda. In three fractions of the three biological replicates, the numbers of unique phosphopeptides identified using SequestHT_only and the combination of SequestHT and MS Amanda are shown in blue and red, respectively. (C) Overlap of unique phosphopeptide with high-confidence localization identified using the iMEED approach is shown in the Venn diagram. (D) Percentages of overlap between phosphopeptides identified in each round of the multistep enrichment. Overlap percentages between technical replicates are shown as blue bars, whereas overlap percentages between biological replicates are shown as red bars. Error bars represent percentage distributions of overlapping peptides between technical replicates or biological replicates.
Next, we sought to maximize coverage of the phosphoproteome relative to the established workflow. We compared the number of confident phosphopeptides that were found using several database search engines, combined with SequestHT and MS Amanda, with those found in SequestHT only (Fig. 2B). The multiple-database search identified an average of 7% more unique confident phosphopeptides than the single database search in all fractions (Fig. 2B). Finally, to determine the advantages and reproducibility of the iMEED approach, we tested this method in three biological replicates with four technical replicates of the BV-2 cell line, analyzing them in 36 LC-MS/MS runs (Fig. 2C and Supporting Information Table 3). A total of 8807 unique phosphopeptides were identified with high confidence; 3365 phosphopeptides were common to all three biological replicates, wherein 6355, 5395, and 5717 unique phosphopeptides were C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
detected in replicates 1, 2, and 3, respectively (Fig. 2C). Approximately 75% of the unique phosphopeptides overlapped with those identified among the four technical replicates in each sample (Fig. 2D and Supporting Information Table 4). For each sample, there was on average an overlap of 68% of all identified unique phosphopeptides between all pairs of biological replicates. These results suggest that iMEED is a reasonable and reproducible method. Further, the technical and biological variability was expressed by calculating the Pearson correlation value (R) with MS1 intensities for the technical and biological replicates. First, the MS1 intensity of phosphoproteins was calculated by summing the normalized intensities of all phosphopeptides that were assigned to the protein group with high confidence. Supporting Information Fig. 3 summarizes the correlation of phosphoprotein intensities of the four technical replicates in www.proteomics-journal.com
622
D. Han et al.
the three biological replicates for each sample. The R value of the technical replicates ranged from 0.891 to 0.984 in the E1 samples, 0.788 to 0.957 in the E2 samples, and 0.813 to 0.987 in the E3 samples. In addition, the average R value of the biological replicates was 0.91, 0.65, and 0.89 in the E1, E2, and E3 samples, respectively. Then, we determined its quantitative reproducibility between biological replicates, including enrichments and batchto-batch variation, by examining the level of independent phosphopeptide intensity. As shown in Supporting Information Tables 5 and 6, the R value of the biological replicates ranged from 0.917 to 0.932 in the E1 samples, 0.8 to 0.88 in the E2 samples, and 0.873 to 0.912 in the E3 samples. Notably, summing the MS1 intensities of phosphopeptides from fractions E1, E2, and E3 demonstrated high reproducibility between biological replicates, with an average Pearson correlation of 0.945, reflecting the tremendous potential of iMEED with regard to label-free quantitation of phosphopeptides. Consequently, the technical and biological variations were significantly low, indicating that our novel strategy is robust and highly reproducible. Three rounds of TiO2 enrichment, combined with multistep FASP, were sufficient and effective in identifying phosphopeptides from whole-cell lysates in depth. Only 9% of phosphopeptides appeared in all three fractions (E1, E2, and E3), whereas 91% of phosphopeptides were detected in 1 or 2 fractions (Supporting Information Fig. 4). These results indicate that our approach improves the comprehensiveness of the phosphoproteome in mass-limited samples. Particularly, the iMEED method improved the identification of multiphosphorylated peptides, although many studies have reported that TiO2 tends to isolate monophosphorylated peptides, not multiply phosphorylated peptides [15]. Without using adducts, such as citric acid [15], to improve the detection of multiphosphorylated peptides, nearly 30% of
Proteomics 2015, 15, 618–623
unique phosphopeptides that were identified were multiphosphorylated by iMEED. As shown in Supporting Information Fig. 5, most multiply phosphorylated peptides were detected following the first (E1) round of TiO2 enrichment, whereas with the second round (E2), an average of 60 such peptides were identified in three biological replicates. However, the proportion of multiply phosphorylated peptides in the third round (E3) of TiO2 enrichment from the second digestion was higher, indicating that further enrichment of the second digestion improves the identification of multiphosphorylated peptides. Further, the phosphopeptides that were enriched in each of the multistep-enriched samples had dis-similar chemical properties (Supporting Information Table 7). In samples E1 and E3, most of the identified phosphopeptides were acidic phosphopeptides with low pI values. The highest proportion (30%) of phosphopeptides in samples E1 and E3 had pI values ranging from 3.5 to 4 (Fig. 3A). The percentage of phosphopeptides with pI values below 4 decreased from 60% in E1 and E3 to 14% in E2. Only 4.5% of all phosphopeptides that were identified in E1 and E3 were basic, with pI values higher than 7, whereas this percentage rose to 16% in the E2 sample (Fig. 3A). An average of 53 phosphopeptides with pI values greater than 8.5 were identified in E2, versus ten phosphopeptides in E1 and E3. This distribution of phosphopeptides by pI value indicates that our multistep enrichment strategy increases coverage of the phosphoproteome. Moreover, the grand average of hydropathy (GRAVY) values—the average hydropathy values for all amino acids in a protein—differed between fractions. In the E1 and E3 samples, 12 and 18% of all identified phosphopeptides had GRAVY values below −2, versus 6.6% in the E2 sample (Fig. 3B). However, in sample E2, 30% of total phosphopeptides had GRAVY values higher than 0, compared with 15% in E1 and E3. These differences suggest that hydrophilic and
Figure 3. Distribution of pI and GRAVY values in multistep enrichment. (A) Histogram of percentage of identified phosphopeptides by pI value in the E1 (blue), E2 (green), and E3 (red) fractions. Cumulative percentages of phosphopeptides from pI 1.5 to 10 are shown with dashed lines. (B) Histogram of percentage of identified phosphopeptides by GRAVY value in the E1 (blue), E2 (green), and E3 (red) fractions. Cumulative percentages of phosphopeptides with GRAVY values of −4 to 3.5 are shown with dashed lines.
C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.proteomics-journal.com
Proteomics 2015, 15, 618–623
hydrophobic phosphopeptides are more likely to be identified by multistep enrichment and that multistep enrichment identifies disparate populations of phosphopeptides. We have found evidence to support combining multistep FASP and multistage enrichment to reduce sample manipulation and improve the robustness and reproducibility of the overall workflow without prefractionation or the need for specific instruments. Further, in addition multiple-database searches, analysis by high-resolution MS yields more highconfidence phosphopeptides than our previous method [10]. The iMEED method is a relatively easy and efficient approach that identifies phosphopeptides from less MS injection material compared with other recently reported approaches [5, 6, 15] (Supporting Information Table 8). Thus, iMEED has generated the most comprehensive phosphoproteome of murine microglia to date, comprising 8807 unique phosphopeptides in 36 LC-MS/MS runs from 300 g of MS injection material, representing an approximately twofold improvement (4000 unique phosphopeptides) over our previous method in the same MS analysis time. In addition, compared with the human, rat, and mouse phosphorylation sites in the PhosphoSitePlus database (www.phosphosite.org), 1262 phosphosites were unrecorded (Supporting Information Fig. 6). In summary, the iMEED approach has tremendous potential in identifying the phosphoproteome in a comprehensive, simple, and reproducible manner by combining multistep FASP, multistep enrichment, and a multiple-database search, based on high-resolution MS. Moreover, our strategy enables one to perform a rapid in-depth analysis of the phosphoproteome using relatively small sample amounts. Thus, the remaining 80% of a sample can be used in other proteomic approaches, such as multiple reaction monitoring. Due to the considerably small MS injection amounts that we used in this study, we are developing a sequential enrichment method of various PTMs per sample for future experiments to further minimize the sample amount that is needed and the instrument run time. The MS proteomics data in this paper have been deposited in the ProteomeXchange Consortium (http://proteomecentral. proteomexchange.org) via the PRIDE partner repository [16]: dataset identifier PXD001033. This work was supported by a National Research Foundation of Korea grant (No. 2011 –0030740) and the Proteogenomic Research Program, funded by the Korean government [MISP]. This work was also supported by the Industrial Strategic Technology Development Program (#10045352), funded by the Ministry of Knowledge Economy (MKE, Korea). The authors have declared no conflict of interest.
References [1] Hunter, T., Signaling–2000 and beyond. Cell 2000, 100, 113– 127.
C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
623 [2] Harsha, H. C., Pandey, A., Phosphoproteomics in cancer. Mol. Oncol. 2010, 4, 482–495. [3] Iliuk, A. B., Tao, W. A., Is phosphoproteomics ready for clinical research? Clin. Chim. Acta 2013, 420, 23–27. [4] Engholm-Keller, K., Larsen, M. R., Technologies and challenges in large-scale phosphoproteomics. Proteomics 2013, 13, 910–931. [5] Yue, X. S., Hummon, A. B., Combination of multistep IMAC enrichment with high-pH reverse phase separation for indepth phosphoproteomic profiling. J. Proteome Res. 2013, 12, 4176–4186. [6] Tsai, C. F., Hsu, C. C., Hung, J. N., Wang, Y. T. et al., Sequential phosphoproteomic enrichment through complementary metal-directed immobilized metal ion affinity chromatography. Anal. Chem. 2014, 86, 685–693. [7] Zarei, M., Sprenger, A., Metzger, F., Gretzmeier, C., Dengjel, J., Comparison of ERLIC-TiO2, HILIC-TiO2, and SCX-TiO2 for global phosphoproteomics approaches. J. Proteome Res. 2011, 10, 3474–3483. [8] Engholm-Keller, K., Hansen, T. A., Palmisano, G., Larsen, M. R., Multidimensional strategy for sensitive phosphoproteomics incorporating protein prefractionation combined with SIMAC, HILIC, and TiO(2) chromatography applied to proximal EGF signaling. J. Proteome Res. 2011, 10, 5383– 5397. [9] Dorfer, V., Pichler, P., Stranzl, T., Stadlmann, J. et al., MS Amanda, a Universal Identification Algorithm Optimized for High Accuracy Tandem Mass Spectra. J. Proteome Res. 2014, 13, 3679–3684. [10] Han, D., Moon, S., Kim, Y., Ho, W. K. et al., Comprehensive phosphoproteome analysis of INS-1 pancreatic betacells using various digestion strategies coupled with liquid chromatography-tandem mass spectrometry. J. Proteome Res. 2012, 11, 2206–2223. [11] Wisniewski, J. R., Zougman, A., Nagaraj, N., Mann, M., Universal sample preparation method for proteome analysis. Nat. Methods 2009, 6, 359–362. [12] Han, D., Moon, S., Kim, Y., Kim, J. et al., In-depth proteomic analysis of mouse microglia using a combination of FASP and StageTip-based, high pH, reversed-phase fractionation. Proteomics 2013, 13, 2984–2988. [13] Spivak, M., Weston, J., Bottou, L., Kall, L. et al., Improvements to the percolator algorithm for Peptide identification from shotgun proteomics data sets. J. Proteome Res. 2009, 8, 3737–3745. [14] Taus, T., Kocher, T., Pichler, P., Paschke, C. et al., Universal and confident phosphorylation site localization using phosphoRS. J. Proteome Res. 2011, 10, 5354–5362. [15] Zhao, X., Wang, Q., Wang, S., Zou, X. et al., Citric acidassisted two-step enrichment with TiO2 enhances the separation of multi- and monophosphorylated peptides and increases phosphoprotein profiling. J. Proteome Res. 2013, 12, 2467–2476. [16] Vizca´ıno, J. A, Cot ˆ e, ´ R. G., Csordas, A., Dianes, J. A. et al., The PRoteomics IDEntifications (PRIDE) database and associated tools: status in 2013. Nucleic Acids Res. 2012, doi: 10.1093/nar/gks1262.
www.proteomics-journal.com
