Article pubs.acs.org/jpr
Phosphoproteome Analysis of Formalin-Fixed and ParaffinEmbedded Tissue Sections Mounted on Microscope Slides Masaki Wakabayashi,†,# Hiroki Yoshihara,§,# Takeshi Masuda,§ Mai Tsukahara,† Naoyuki Sugiyama,†,§ and Yasushi Ishihama*,†,§ †
Graduate School of Pharmaceutical Sciences, Kyoto University, Sakyo-ku, Kyoto 606-8501, Japan Institute for Advanced Biosciences, Keio University, Daihoji, Tsuruoka, Yamagata 997-0017, Japan
§
S Supporting Information *
ABSTRACT: Formalin-fixed and paraffin-embedded (FFPE) sections mounted on microscope slides are one of the largest available resources for retrospective research on various diseases, but quantitative phosphoproteome analysis of FFPE sections has never been achieved because of the extreme difficulty of procuring sufficient phosphopeptides from the limited amounts of proteins on the slides. Here, we present the first protocol for quantitative phosphoproteome analysis of FFPE sections by utilizing phase-transfer surfactant-aided extraction/tryptic digestion of FFPE proteins followed by high-recovery phosphopeptide enrichment via lactic acid-modified titania chromatography. We established that FFPE sections retain a similar phosphoproteome to fresh tissue specimens during storage for at least 9 months, confirming the utility of our method for evaluating phosphorylation profiles in various diseases. We also verified that chemical labeling based on reductive dimethylation of amino groups was feasible for quantitative phosphoproteome analysis of FFPE samples on slides. Furthermore, we improved the LC− MS sensitivity by miniaturizing nanoLC columns to 25 μm inner diameter. With this system, we could identify 1090 phosphopeptides from a single FFPE section obtained from a microscope slide, containing 25.2 ± 5.4 μg of proteins. This protocol should be useful for large-scale phosphoproteome analysis of archival FFPE slides, especially scarce samples from patients with rare diseases. KEYWORDS: FFPE slide, formalin-fixed and paraffin-embedded, phase-transfer surfactant, protein extraction, phosphoproteome analysis, shotgun analysis, post-translational modifications
■
INTRODUCTION Protein phosphorylation on serine, threonine, and tyrosine residues is one of the most frequent post-translational modifications, and many basic cell functions, such as proliferation, differentiation, cell cycle control, and metabolism, depend upon phosphorylation-mediated signal transduction networks.1 Because of their critical importance in functional regulation, reversible phosphorylation−dephosphorylation reactions are now widely accepted as a major determinant of cell fate. Furthermore, alterations in phosphorylation profiles are closely related to the onset and progression of various diseases. For example, aberrant phosphorylation is one of the most common features of cancer cells, being associated with abnormal growth.2−4 Thus, an overall view of cellular protein phosphorylation networks is considered to be indispensable for a deep understanding of disease mechanisms, as well as for drug development and disease-related biomarker discovery. However, comprehensive and quantitative analysis of phosphorylated proteins in clinical samples is still challenging due to the limited availability of such samples, as well as the difficulty of rapidly freezing tissues (within seconds to minutes) during © 2013 American Chemical Society
operation in order to avoid hypoxia and stress signaling, although some quantitative studies on the phosphoproteome of patients have been reported.5,6 Phosphoproteomics technologies for large-scale data acquisition have been intensively developed in order to achieve comprehensive identification and quantitation of phosphorylated proteins. For the enrichment of digested phosphopeptides, immobilized metal ion affinity chromatography (IMAC)7−11 and metal oxide chromatography (MOC) using titania or zirconia12−18 have been well-developed. MOC-based phosphopeptide enrichment using titania was first applied by Ikeguchi and Nakamura12 and subsequently employed by other groups.15,19,20 We further developed an improved approach named aliphatic hydroxy acid-modified metal oxide chromatography (HAMMOC),17 in which hydrophilic hydroxy acids, such as lactic acid and glycolic acid, are used to provide higher selectivity for phosphopeptides. Use of the HAMMOC method allows the sample size necessary for phosphoproteome analysis Received: September 22, 2013 Published: December 12, 2013 915
dx.doi.org/10.1021/pr400960r | J. Proteome Res. 2014, 13, 915−924
Journal of Proteome Research
Article
LC−MS system with direct sample loading onto the LC column.
to be reduced, and thousands of unique phosphosites can be routinely identified from 100 μg of proteins in a single LC− MS/MS run.17,21 Recent advances in LC−MS/MS instruments have also expanded phosphoproteome coverage, leading to identification and quantitation of 14000−20000 phosphorylation sites from more than 10 mg of HeLa cell lysates, using IMAC or titania chromatography.22,23 Large-scale phosphoproteomics has also been used to identify around 30000 phosphopeptides from mammalian tissues, although the large sample size required (>100 mg)24−26 means that this approach is unsuitable for evaluation of small clinical tissue specimens. Application of the HAMMOC method should enable largescale identification of phosphopeptides in clinical samples, and this would surely provide valuable insights into the relationship between altered phosphorylation profiles and various diseases. Clinical specimens from human patients are usually stored as formalin-fixed and paraffin-embedded (FFPE) tissue blocks in order to maintain their histomorphological structures. FFPE tissues are often sectioned and mounted on microscope slides, commonly for immunohistochemical analysis, and such slides are cumulatively preserved in almost all medical institutions. Thus, they represent a huge repository of clinical information. Also, considering that the availability of fresh specimens in general is extremely poor, archives of FFPE tissue slides are one of the largest and most attractive resources for retrospective research, especially on rare diseases, including various cancers. Many clinical researchers have great expectations for the application of proteomic approaches to FFPE tissues, and, after the first report of shotgun-based FFPE proteome analysis,27 publications in this area have dramatically increased during the past several years. However, formalin-fixation and paraffinembedding are serious obstacles to efficient protein retrieval,28 although various methods have been suggested for more efficient protein extraction, such as Liquid Tissue,29 Qproteome,30 Rapigest SF,31 as well as the filter-aided sample preparation (FASP) method.32 In our previous study, we introduced phase transfer surfactants (PTS) as effective protein-solubilizing reagents that can provide unbiased extraction even for the membrane proteome.33,34 The PTSaided method proved more effective than Rapigest SF34 and FASP method35 for protein extraction. Therefore, we considered that our PTS method could have potential applicability for FFPE protein extraction. No phosphoproteomic study of FFPE tissue sections has yet been reported due to the lack of a sophisticated standard method offering sufficient protein recovery. Nevertheless, two phosphoproteomic studies of FFPE tissue blocks were recently reported. Gamez-Pozo et al. identified less than 100 phosphopeptides from 200 μg of FFPE proteins.36 Ostasiewicz et al. successfully identified 7718 phosphopeptides with the FASP method, using a large amount of extracted proteins (5000 μg) and a long analytical time (24 h) for LC−MS/MS analysis.32 In the case of FFPE sections on microscope slides, the tissues are tightly attached to the slides, greatly increasing the difficulty of harvesting enough proteins for subsequent phosphopeptide enrichment, so a phosphoproteomic approach has not previously been applied. In this study, we demonstrate for the first time that quantitative phosphoproteome analysis of FFPE tissue sections mounted on microscope slides is feasible. Furthermore, we present a method for highly sensitive phosphoproteome analysis of individual FFPE slides by using a miniaturized
■
MATERIALS AND METHODS
Materials
Sodium deoxycholate (SDS), sodium lauroyl sarcosinate (SLS), dithiothreitol (DTT), iodoacetamide (IAA), triethylammonium bicarbonate (TEAB), DL-lactic acid, lysyl endopeptidase (LysC), ethyl acetate, acetonitrile, acetic acid, methanol, ethanol, xylene, trifluoroacetic acid (TFA), and piperidine were purchased from Wako (Osaka, Japan). Heavy and light (13CD2O or 12CH2O) formaldehyde were obtained from Taiyo Nippon Sanso Corporation (Tokyo, Japan) and SigmaAldrich (St. Louis, MO), respectively. Titania was from GL Sciences (Tokyo, Japan). Modified trypsin was from Promega (Madison, MA). C8 and C18 Empore products were purchased from 3M (St. Paul, MN). Rapigest SF37 was from Waters (Milford, MA). Liquid Tissue MS Protein Prep kit was purchased from Expression Pathology (Gaithersburg, MD). Qproteome FFPE Tissue kit was obtained from QIAGEN (Hilden, Germany). FFPE mouse liver tissue slides were from Genostaff (Tokyo, Japan). Water was obtained from a Millipore Milli-Q system (Bedford, MA). Sample Preparation
Fresh liver tissues were dissected from C57BL/6 mice and stored at −80 °C until use. FFPE mouse liver tissue sections with surface areas of less than 40 mm2 and a thickness of 5 μm were stored at 4 °C. For protein extraction, freshly frozen tissues were soaked in PTS buffer (12 mM SDC, 12 mM SLS in 100 mM Tris-HCl, pH 9.0) and minced into small pieces. FFPE tissue sections were deparaffinized and rehydrated by successive washes in xylene (3×), 100% (2×), 96% (2×), 70% ethanol (2×) and water. After air-drying, the tissue sections were percolated with PTS buffer, Liquid Tissue buffer,29 extraction buffer of the Qproteome FFPE tissue kit, or Rapigest SF solution for 90 min, and harvested with a scalpel blade. In the cases of Liquid Tissue MS Protein Prep kit, Qproteome FFPE tissue kit, and Rapigest SF reagent, the following treatments were performed according to the manufacturer’s instructions. In the case of the PTS-based method, the collected tissues were incubated on a heating block at 99 °C for 60 min and then sonicated for 10 min. Recovered proteins were quantified with a Pierce BCA Protein Assay kit and subjected to reduction and alkylation with 10 mM DTT and 55 mM IAA, respectively. The sample solutions were diluted 5-fold with 50 mM ammonium bicarbonate and digested with LysC for 3 h, followed by overnight trypsin digestion. After digestion, an equal amount of ethyl acetate was added. The mixture was acidified with 0.5% TFA (final concentration) and then agitated for 2 min and centrifuged at 15800g for 2 min to completely separate the aqueous and organic phases. The aqueous phase was collected and desalted with C18-StageTips.16,38 Phosphopeptide Enrichment
Phosphopeptide enrichment was performed with titania HAMMOC as previously described, with some modifications.17 Briefly, C8-StageTips packed with titania beads were equilibrated with 20 μL of 80% acetonitrile containing 0.1% TFA and 300 mg/mL of lactic acid as a selectivity enhancer (solution A). The digested samples were diluted with an equal volume of solution A and loaded onto the HAMMOC tip. After successive washes with 20 μL of solution A and 20 μL of 916
dx.doi.org/10.1021/pr400960r | J. Proteome Res. 2014, 13, 915−924
Journal of Proteome Research
Article
Figure 1. PTS-based protein extraction procedure. (A) FFPE sections were deparaffinized by the conventional method and percolated with PTS buffer for easy harvesting with a scalpel blade. (B) Recovered proteins were heated at 99 °C to break protein cross-links. (C) Residual paraffin was eliminated with ethyl acetate during the removal of PTS.
solution B (80% acetonitrile with 0.1% TFA), phosphopeptides were eluted with 20 μL of 0.5% piperidine. The eluted fractions were acidified with 3 μL of 10% TFA, desalted with C18StageTip, and concentrated in a vacuum evaporator. The resulting phosphopeptides were dissolved in 4% acetonitrile with 0.5% TFA and injected into the LC−MS/MS system. For high-sensitivity analysis, the dissolved phosphopeptides were directly injected onto a cartridge column (10 cm length × 50 μm ID) packed with Reprosil-Pur C18-AQ materials (3 μm, Dr. Maisch, Ammerbuch, Germany) using a column loader cell (Nikkyo Technos, Tokyo, Japan) under a nitrogen pressure of 7 MPa. The cartridge column was directly connected to the analytical column.
agitated for 1 min and then incubated for 60 min at room temperature to achieve a labeling efficiency of more than 99%. The reaction was stopped by adding 16 μL of 1% ammonium hydroxide on ice and agitating the mixture for 1 min. Differentially labeled peptides were mixed and subjected to PTS removal and desalting steps as described above. Conventional NanoLC−MS System
NanoLC−MS/MS analyses were conducted by using a QSTAR system (QSTAR-XL mass spectrometer (ABSCIEX, Foster City, CA) equipped with an Agilent 1100 nanoflow pump (Waldbron, Germany) and a HTC-PAL autosampler (CTC Analytics, Zwingen, Switzerland), an Orbitrap system (LTQOrbitrap mass spectrometer, ThermoFisher Scientific, Bremen, Germany) equipped with a Dionex UltiMate 3000 pump with a FLM-3000 flow manager (Germering, Germany) and a HTCPAL autosampler, or a TripleTOF 5600 system (TripleTOF 5600 mass spectrometer equipped with a Dionex UltiMate 3000 RSLCnano pump and a HTC-PAL autosampler). Reprosil-Pur C18-AQ materials were packed into a self-pulled needle (150-mm length × 100 μm ID, 6-μm opening) with the
Stable Isotope Dimethyl Labeling
For stable isotope dimethyl labeling,39,40 Tris-HCl and ammonium bicarbonate were replaced with TEAB throughout the process. A solution of the digested peptides in 100 μL of PTS solution was mixed with 4 μL of 4% 13CD2O or 12CH2O, and then 4 μL of freshly prepared 0.6 M sodium cyanoborohydride was immediately added. The mixture was 917
dx.doi.org/10.1021/pr400960r | J. Proteome Res. 2014, 13, 915−924
Journal of Proteome Research
Article
Phosphorylated sites were unambiguously determined based on the presence of site-determining ions.48 Peptide quantitation was performed by Mass Navigator based on the integrated peak areas and the heavy- and light-labeled peptide ratio (H/L ratio) was calculated in individual runs.
nitrogen-pressurized column loader cell to prepare an analytical column needle with “stone-arch” frit.41 The injection volume was 5 μL, and the flow rate was 500 nL/min. The mobile phases consisted of (A) 0.5% acetic acid and (B) 0.5% acetic acid in 80% acetonitrile. A three-step linear gradient of 5−10% B in 5 min, 10−40% B in 60 min, and 40−100% B in 5 min, followed by 100% B for 10 min was employed. Spray voltages of 2400 V (in the QSTAR and Orbitrap systems) and 2300 V (in the TripleTOF 5600 system) were applied. The mass scan ranges were m/z 350−1400 (in the QSTAR system) or 300− 1500 (in the Orbitrap and TripleTOF 5600 systems), and the top three (QSTAR) or top ten (Orbitrap and TripleTOF 5600) precursor ions were selected in each MS scan for subsequent MS/MS scans. A lock mass function was used for the Orbitrap system to obtain constant mass accuracy during the gradient.42
■
RESULTS AND DISCUSSION
A Novel Protein Extraction Method from FFPE Tissue Sections on Microscope Slides
In this study, we developed a novel protein extraction method from FFPE sections mounted on microscope slides. This method was based on our PTS method and included three important steps. First, after conventional deparaffinization and rehydration, FFPE sections were percolated with PTS, which made it easy to strip the tightly attached tissues from the glass slides (Figure 1A). Second, mimicking antigen retrieval techniques used in immunohistochemistry,49,50 the harvested samples were heated at 99 °C to break protein cross-links formed by formalin fixation51 and to increase the accessibility of the proteins to digestive enzymes (Figure 1B). Finally, residual paraffin was further eliminated by ethyl acetate used for the removal of PTS (Figure 1C), resulting in higher recovery of digested peptides at the desalting step and more efficient enrichment of phosphopeptides in the HAMMOC method. In order to verify the performance of the method, the amount of proteins extracted from one FFPE section and the number of peptides identified by LC−MS/MS-based proteomic analysis were compared among the PTS-based method and previously reported reagents: Liquid Tissue, Qproteome, and RapiGest SF (Table 1). The PTS-based protocol provided comparable
Highly Sensitive nanoLC−MS System
Miniaturized analytical columns were constructed using fused silica capillary tubing (200 mm length × 25 μm ID) according to the previous reports.43−45 Briefly, a sol−gel solution of 5:1 silicate/formamide was prepared by mixing 50 μL of Kasil1 with 10 μL of formamide and introduced to the midpoint of a section of capillary tubing by capillary action. Polymerization was induced using a soldering iron heated at 300 °C, and excess sol−gel solution was washed out with 0.5% acetic acid by use of the nitrogen-pressurized column loader cell. The frit was washed with methanol for 60 min, and then Reprosil-Pur C18AQ materials were packed as described above. Emitter tips were formed using a model P-2000 laser-based micropipet puller (Sutter Instrument Co., Novato, CA). Direct sample injection onto the cartridge column was performed as described above. The Dionex UltiMate 3000 RSLCnano pump was used to deliver mobile phases at the flow rate of 32 nL/min. A spray voltage of 1800 V was applied in the TripleTOF 5600 system. Other LC−MS conditions were the same as those of the conventional systems described above.
Table 1. Amount of Extracted Proteins and Number of Identified Peptides from FFPE Tissue Sections on Microscope Slides Obtained with Various Protein Extraction Methodsa
Data Processing
Mass Navigator v1.2 (QSTAR and Orbitrap; Mitsui Knowledge Industry, Tokyo, Japan) and ABSCIEX MS Data Converter (TripleTOF 5600) were used to create peak lists on the basis of the recorded fragmentation spectra. Peptides and proteins were identified by Mascot v2.3 (Matrix Science, London, U.K.) against UniprotKB/Swiss-prot release 2011_10 (09-Oct-2011), 2012_01 (25-Jan-2012), or 2013_06 (29-May-2013) with a precursor mass tolerance of 0.25 Da (QSTAR), 3 ppm (Orbitrap), or 20 ppm (TripleTOF 5600), a fragment ion mass tolerance of 0.25 Da (QSTAR), 0.8 Da (Orbitrap), or 0.1 Da (TripleTOF 5600) and strict trypsin specificity,46 allowing for up to two missed cleavages. Cysteine carbamidomethylation was set as a fixed modification, and methionine oxidation was allowed as a variable modification. Formylation or dimethylation of N-termini and ε-amino groups of lysine and phosphorylation of serine, threonine, and tyrosine were set as variable modifications in accordance with the experimental situation. Peptides were considered to be identified if the Mascot score was over the 95% confidence limit based on the “identity” score of each peptide and at least three successive yor b-ions with two or more y-, b- and/or precursor-origin neutral loss ions were observed, based on the error-tolerant peptide sequence tag concept.47 A randomized decoy database created by a Mascot Perl program estimated less than 1% falsepositive rate for identified peptides within the criteria.
instrument Qstar XL
TripleTOF 5600
protein recovery from FFPE sections (μg/slide)
no. of identified peptides
no. of formylated peptides (% of total peptides)
PTS Liquid Tissue Qproteome PTS
23.6b 22.0b
1943 1654
23 (1.8%) 27 (2.5%)
27.2b 29.9b
6457
212 (3.3%)
Rapigest SF
38.5b
3194
105 (3.3%)
protein extraction method
a
Triplicate LC−MS/MS analyses were performed for each sample. b Sequential tissue sections dissected from an identical tissue block were used.
protein recovery to the other three reagents based on BCA protein assay. As for the number of identified peptides, the PTS-based protocol gave the best result (1943 peptides in the QSTAR-XL and 6457 peptides in the TripleTOF 5600 system) among the four protocols. In each method, protein concentration was measured prior to removal of protein extraction reagents. During the elimination process of RapiGest SF reagent, some digested peptides might be coprecipitated with the acid-induced large aggregates of the reagents,37 leading to loss and overestimation of the peptides injected into LC− MS/MS. In the case of Qproteome, no identifiable peptide was observed, although a comparable amount of proteins was detected in BCA protein assay. We no longer use this kit 918
dx.doi.org/10.1021/pr400960r | J. Proteome Res. 2014, 13, 915−924
Journal of Proteome Research
Article
Figure 2. Phosphoproteome analysis of FFPE and fresh mouse liver tissues. FFPE and freshly frozen tissues were extracted according to the PTSaided procedure, and 100 μg aliquots of the proteins were used for enrichment of phosphopeptides by the HAMMOC method. Each sample was analyzed in duplicate by an Orbitrap system. (A) Area-proportional Venn diagram of the unique phosphopeptides identified from three samples each of FFPE and fresh specimens. (B) Hexagonal radar chart for six characters of identified phosphopeptides. GRAVY score indicates the hydrophobicity of peptides.60 (C, D) The number of proteins corresponding to the indicated keywords in the classifications of Cellular Component and Biological Process, respectively, in UniprotKB.
based protein extraction/digestion and HAMMOC, and LC− MS measurement was done in duplicate. From FFPE and freshly frozen tissues, 1413 and 1197 unique phosphopeptides in total were identified, respectively (phosphopeptide lists and annotated MSMS spectra are in Supplementary Tables S2 and S3, Supporting Information), and 654 phosphopeptides (33.4% of the total number) overlapped between the two specimens (Figure 2A). Identified phosphopeptides showed quite similar physicochemical properties among the two samples, including pI value, peptide length, GRAVY score, and amino acid components (Figure 2B). Peak areas of overlapped peptides were also compared between the two. The median value of the FFPE-to-fresh ratios is 0.84 and the log2 (FFPE/fresh ratio) values of more than 90% of phosphopeptides converged between the median plus and minus 2 (Supplementary Figure 1, Supporting Information), indicating that there was no significant difference in phosphoproteome profiles. For further characterization of the influence of the preservation method on the phosphoproteome profile, identified phosphopeptides were categorized by UniprotKB keywords. The classifications of Cellular Component and Biological Process showed no marked difference between FFPE and freshly frozen samples (Figure 2C,D). However, the content of phosphopeptides with Cterminal lysine was apparently lower in FFPE sections than in freshly frozen tissues (42.8 versus 49.7%), while the opposite was the case for peptides with C-terminal arginine (48.5 versus 43.8%). This result was surely due to various modifications of ε-amino groups of lysine residues during formalin fixation, as
because the buffer constituents are not disclosed. For the identification of FFPE-derived peptides, formylation on protein N-termini and ε-amino groups of lysine residues was set as a variable modification in the Mascot search criteria. However, since the content of formylated peptides was low and did not differ significantly among the tested methods (1.8−3.3%), we did not consider the modification in the following analysis. Taking into consideration the entire sequence of operations from protein extraction to peptide identification, we consider that the PTS-based method is the best for proteomic analysis of FFPE tissue sections. The simplicity and cost-effectiveness of the novel PTS-based protocol are also major advantages for routine application. We have also evaluated the FASP method with less than 100 μg of lysate proteins, for comparison with the PTS protocol. The number of peptides identified with the FASP method was much smaller (3.6-fold less), probably due to protein adsorption on the filtration membrane (Supplementary Table S1, Supporting Information), as reported by Leon et al.35 Phosphoproteome Analysis of FFPE Tissue Sections and Fresh Frozen Tissues
In the HAMMOC method, 100 μg of extracted proteins is sufficient for one enrichment process, and this corresponds to three to five FFPE sections with the PTS-based method. Therefore FFPE proteins were harvested, combined, and subjected to HAMMOC. In order to characterize the properties of phosphopeptides from FFPE proteins, freshly frozen tissues were used as a control. Triplicate analyses were done for PTS919
dx.doi.org/10.1021/pr400960r | J. Proteome Res. 2014, 13, 915−924
Journal of Proteome Research
Article
Figure 3. Phosphoproteome analysis of FFPE slides stored for various periods. FFPE slides of mouse liver were stored for 0, 1, 5, or 9 months (M), and analyzed with the PTS- and HAMMOC-based LC−MS/MS system. Each sample was analyzed in duplicate with an Orbitrap system. (A) Hexagonal radar chart for six characters of identified phosphopeptides. (B) Overlap percentage of identified phosphopeptides from 0, 1, 5, or 9 Mpreserved FFPE slides. (C, D) The numbers of proteins corresponding to indicated keywords in the classifications of Cellular Component and Biological Process, respectively, in UniprotKB.
reported previously.52 However, since the fixation process is complex and not well understood, the modifications were unpredictable and extremely difficult to confirm. These observations suggested that FFPE tissues maintain the native phosphorylation profile to a great extent, even though the number of observable peptides is slightly reduced. Thus, the PTS-based method is suitable for phosphoproteome analysis of FFPE tissue sections.
The classifications of Cellular Component and Biological Process in UniprotKB keywords (Figure 3C,D) were also similar for all storage periods. These data are in agreement with the findings in proteomic studies using FFPE tissue blocks32,52,53 and suggest that FFPE tissue sections can be retrospectively analyzed for at least nine months after preparation without any striking alteration of the phosphoproteome profile.
Phosphoproteome Analysis of FFPE Tissue Slides Stored for Different Periods
Quantitative Phosphoproteome Analysis of FFPE Tissue Sections
Most FFPE tissue sections used for histopathological diagnosis have been stored for a long period of time. To confirm the validity of retrospective analysis of FFPE tissues with our method, it is necessary to examine whether the phosphorylation profile changes during various storage periods in the FFPE state. Therefore, the phosphoproteome of mouse liver FFPE sections, preserved for 0, 1, 5, and 9 months, were analyzed using the PTS- and HAMMOC-based LC−MS/MS system. We found no significant difference in the physicochemical properties of the identified phosphopeptides (Figure 3A). Furthermore, the overlap percentage of identified phosphopeptides between any two arbitrarily selected periods remained almost the same (46−47%) for any combination (Figure 3B).
Quantitative phosphoproteomic analysis of clinical samples such as FFPE tissue sections was the next challenge, since this is indispensable to profile the global cellular signaling system. We therefore examined whether the combination of the stable isotope dimethyl labeling method with the PTS- and HAMMOC-based LC−MS/MS system was applicable to quantitative phosphoproteomics of FFPE tissue sections. One hundred microgram aliquots of extracted FFPE peptides were differentially modified with heavy and light formaldehyde in the presence of cyanoborohydride, resulting in differential isotopic dimethylation at N-termini and ε-amino groups of lysine residues. The mixture was subjected to HAMMOC-based LC− MS/MS analysis. From the duplicate analyses of five samples, 920
dx.doi.org/10.1021/pr400960r | J. Proteome Res. 2014, 13, 915−924
Journal of Proteome Research
Article
Figure 4. Quantitative phosphoproteome analysis of FFPE sections. Equal amounts of peptides were differentially dimethylated with heavy or light formaldehyde, mixed, and analyzed in duplicate on the TripleTOF 5600 system. (A) Average H/L ratios of phosphopeptide peak areas in each LC− MS/MS run. (B, C) Peak areas of heavily labeled phosphopeptides were plotted against the H/L ratios (B) with or (C) without normalization by the median value in each run.
Figure 5. Comparison between highly sensitive and conventional LC−MS systems for FFPE phosphopeptide identification. (A) Total ion current chromatograms of FFPE phosphopeptides from a single slide analyzed with the highly sensitive and conventional systems. (B) The numbers of phosphopeptides identified in the highly sensitive and conventional LC−MS systems. Data acquisition was performed using the TripleTOF 5600 system.
5310 redundant phosphopeptides were identified in total, and quantitative analysis was performed for 2793 phosphopeptides with quantifiable peak shapes. Figure 4A shows the average values of the heavy-to-light (H/L) ratios of peptide peak areas in each LC−MS/MS run. There was almost no deviation between duplicate runs of the same samples, and only small differences were observed among the five samples. In Figure
4B,C, the log10 (peak area) values of heavily labeled peptides were plotted against the log2 (H/L ratio) values with and without normalization by the median value in each run, respectively. The normalized log2 (H/L ratio) values of more than 99% of peptides converged between plus and minus 1, and the same was true for the raw values of more than 98% of peptides. Furthermore, such narrow distributions were 921
dx.doi.org/10.1021/pr400960r | J. Proteome Res. 2014, 13, 915−924
Journal of Proteome Research
Article
method can probably improve the identification depth of FFPE phosphoproteome. In case of using isobaric tags including tandem mass tags (TMT) and isobaric peptide tags for relative and absolute quantitation (iTRAQ) for peptide quantitation instead of dimethyl labeling, electrospray ionization in ammonium vapor phosphopeptides is conductive to comprehensive identification and accurate quantitation.57 For more detailed analysis, specific regions of FFPE sections have often been extracted by the laser capture microdissection (LCM) method58,59 to discriminate pathological cells from normal ones, since both probably coexist in a section. Therefore, it would be desirable to improve the sensitivity of FFPE phosphoproteome analysis even more, so that it can be applied to LCM samples. However, it was also reported that macrodissection, which involves only the rough removal of surrounding healthy tissues, was preferable for analyzing gene expression profiles of cancer tissues when the population of pathological cells is large enough.58 Macrodissection has the major advantage of affording a higher protein yield than LCM and could be readily combined with our current protocol, whereas further fractionation by LCM might decrease the protein recovery to less than the detection limits in the LC− MS system. Nevertheless, we believe that further improvement of the sensitivity and throughput performance of the present system should enable large-scale, comprehensive, quantitative phosphoproteome analysis of minute clinical specimens, including laser-microdissected FFPE tissues, in the near future.
maintained for the entire range of the peak area values. These data clearly demonstrated that the use of the PTS-based procedure made it possible to minimize sample loss in the protein extraction process from FFPE tissue sections and to reproducibly perform quantitative phosphoproteome analysis over a broad dynamic range of phosphopeptide amount. Highly Sensitive Phosphoproteome Analysis of a Single FFPE Section
In order to reduce the required sample size for FFPE phosphoproteome analysis, a highly sensitive LC−MS/MS system was employed. In this system, a miniaturized spray tip column of 25 μm ID was used as the analytical column, resulting in enhancement of the MS sensitivity.43,45 Furthermore, to avoid sample loss by adsorption onto the autosampler needle and switching valve, analytical samples were directly loaded onto the cartridge column. Then, the columns were connected directly to the analytical column. Figure 5A shows the total ion current chromatograms of phosphopeptides from individual FFPE sections obtained from microscope slides. This highly sensitive system provided a clear enhancement of the MS signals, resulting in identification of 1090 phosphopeptides from a single FFPE section, whereas only 111 phosphopeptides were identified by the conventional method (Figure 5B). Furthermore, the peak areas of 92 overlapped phosphopeptides were increased 9.1-fold on average in this highly sensitive system, and the total peak areas of identified phosphopeptides was enhanced 24.7-fold (Supplementary Figure 2, Supporting Information). These data indicated that this system dramatically increased the identification efficiency of phosphopeptides and is available for phosphoproteome analysis of tiny FFPE samples. At present, operation is manual, but it should be possible to automate the procedure for high-throughput analysis. In this study, we have achieved the first comprehensive and quantitative analysis of the phosphoproteome of individual FFPE tissue sections mounted on microscope slides, using the novel PTS-aided protocol. Our PTS method recovered proteins efficiently in an unbiased manner and almost completely removed residual paraffin, so that highly efficient phosphopeptide enrichment could be achieved. Furthermore, the highly sensitive nanoLC−MS system reduced the required sample size to that present in a single tissue section. Thus, the system should be applicable for retrospective research on long-termpreserved FFPE slides. Indeed, in terms of the identification rate, our system (>40 phosphopeptides/μg) showed outstanding results, in spite of using FFPE tissues, compared to previous tissue phosphoproteomics systems (
