Article pubs.acs.org/ac
Extended Coverage of Singly and Multiply Phosphorylated Peptides from a Single Titanium Dioxide Microcolumn Masaki Wakabayashi,† Yutaka Kyono,† Naoyuki Sugiyama, and Yasushi Ishihama* Graduate School of Pharmaceutical Sciences, Kyoto University, 46-29, Yoshida-Shimo-Adachi-Cho, Sakyo-ku, Kyoto, 606-8501, Japan S Supporting Information *
ABSTRACT: We developed a novel approach to enlarge phosphoproteome coverage by selective elution depending on the number of phosphoryl group of peptides from a single titanium dioxide (TiO2) microcolumn using hydrophilic interaction chromatography (HILIC). In this approach, acidic methylphosphonate buffer including organic solvent is used for selective elution of singly phosphorylated peptides from an aliphatic hydroxy acid-modified metal oxide chromatography (HAMMOC) microcolumn and basic elution conditions with phosphate, ammonium hydroxide, and pyrrolidine are then employed for eluting multiply phosphorylated peptides retained by the HAMMOC microcolumn. Finally, we successfully identified 11 300 nonredundant phosphopeptides from triplicate analyses of 100 μg of HeLa cell lysates using this approach. This simple strategy made it possible to accomplish comprehensive and efficient phosphoproteome analysis from limited sample amounts loaded onto a single HAMMOC microcolumn without additional fractionation or enrichment approaches.
R
further improvement is required to map the entire phosphoproteome of cells. To solve the problem, Thingholm et al. developed SIMAC (Sequential Elution from IMAC) for sequential separation of singly and multiply phosphorylated peptides,18 where IMAC was used as a first filter to capture multiply phosphorylated peptides and the flow-through fraction was loaded onto a TiO2 column for isolating singly phosphorylated peptides. However, since the use of two chromatography columns causes lower throughput and recovery, it constitutes a barrier to large-scale phosphoproteome analysis. Previously, we reported a novel phosphopeptide enrichment method named aliphatic hydroxy acid-modified metal oxide chromatography (HAMMOC), in which hydrophilic hydroxy acids, such as lactic acid and glycolic acid, are used to provide higher selectivity for phosphopeptides.19 This approach made it possible to dramatically reduce the sample size necessary for phosphoproteome analysis. We further succeeded in enlarging the recovery of phosphopeptides from a single HAMMOC microcolumn using successive and selective release (SSR) with 5% ammonium hydroxide, 5% piperidine, and 5% pyrrolidine solution,20 and thousands of unique phosphosites can be routinely identified from 100 μg of cell lysates or tissue proteins without any fractionation.20−22 In addition, much wider phosphoproteome coverage was obtained compared to other phosphopeptide enrichment methods such as DHB/phthalate-
eversible protein phosphorylation is one of the most important post-translational modifications and plays a key role in various cellular events such as cell growth, metabolism, signal transduction, and apoptosis.1,2 Therefore, characterization of cellular phosphorylation status is crucial to elucidate the complex cellular signaling network and understand disease and drug action mechanisms. However, the abundance of most phosphorylated proteins is not only low but also dynamically changes depending on physiological states and cellular activities. Therefore, one of the major challenges in phosphoproteome mapping is to develop novel methods for efficiently capturing phosphorylated proteins or peptides of low copy number.3 Recent advances in mass spectrometry (MS)-based proteomics coupled with a specific enrichment method of phosphopeptides such as phosphoramidate chemistry (PAC),4 immobilized metal ion affinity chromatography (IMAC),5,6 strong cation exchange (SCX) chromatography,7 and metal oxide chromatography (MOC)8−10 allow the identification of thousands of phosphopeptides from complex biological samples. With further improvement of these technologies, several large-scale phosphoproteome analyses have been reported.11−16 However, it takes several days to measure the phosphoproteome, because fractionation prior to LC/MS is essential to increase the coverage of the phosphoproteome and it is still not satisfactory even after exhaustive prefractionation. As a consequence, no single method of selective, efficient, and comprehensive phosphopeptide enrichment is cited. At present, it has been estimated that more than 100 000 phosphorylation sites may exist in the human proteome17 and complete phosphoproteome analysis has never been done. Therefore, © 2015 American Chemical Society
Received: March 31, 2015 Accepted: September 24, 2015 Published: September 24, 2015 10213
DOI: 10.1021/acs.analchem.5b01216 Anal. Chem. 2015, 87, 10213−10221
Article
Analytical Chemistry titania MOC,23 IMAC, and even the combination of MOC and IMAC. However, the identification efficiency for multiply phosphorylated peptides was still insufficient compared to that for singly phosphorylated ones. Recently, it was reported that the use of glycerol and citric acid with TiO2 also enhanced the identification of multiply phosphorylated peptides.24,25 Furthermore, monodisperse microsphere-based Ti4+-IMAC was developed to achieve minimum adsorption of nonphosphorylated peptides and further efficient binding of phosphopeptides. In combination with SCX prefractionation, 9117 unique phosphosites were identified from 400 μg of MCF-7 cell lysates.26,27 The method, however, could not recover many multiply phosphorylated peptides as well (0.75) sites were accepted as unambiguous sites. The total peak areas of the identified peptides were calculated using the nonlabel quantification function of Mass Navigator.
■
RESULTS AND DISCUSSION Fractionation of HAMMOC-Enriched Phosphopeptides by IMAC. First, to examine the properties of phosphopeptides enriched by HAMMOC and IMAC, HAMMOC-enriched phosphopeptides from 100 μg HeLa cell proteins were loaded onto IMAC columns. Then, the flowthrough and the eluate from IMAC columns were desalted by C18 StageTips and subsequently analyzed by nanoLC-MS/MS. Figure 1 and Table S1 show the number and total peak area of identified peptides in each fraction of the HAMMOC-IMAC method. The results obtained from a single HAMMOC and a single IMAC were set as controls. In total, 1293 unique phosphopeptides were identified in the flow-through fraction of the HAMMOC-IMAC method, and more than 97% of them (1257) were singly phosphorylated peptides. On the other hand, 1443 unique phosphopeptides were identified in the HAMMOC-IMAC eluate, of which more than half (749) were multiply phosphorylated peptides. Most of the multiply phosphorylated peptides were captured and eluted in the HAMMOC-IMAC, whereas singly phosphorylated peptides 10216
DOI: 10.1021/acs.analchem.5b01216 Anal. Chem. 2015, 87, 10213−10221
Article
Analytical Chemistry
Figure 3. Elution properties of peptides captured by titania-MOC tips. Tryptic digests from 100 μg HeLa cell lysates were loaded onto titania-MOC tips with 20 mM methylphosphonate (pH 2.0) including 70% ACN. The peptides bound to titania-MOC tips were released by 0.05−5.0 M TEAP buffer (pH 2.0) including 20−70% ACN, 5% ammonium hydroxide, and 5% pyrrolidine, in series. The wash and eluted fractions were analyzed by a LTQ-Orbitap XL system with self-pulled needle columns. Triplicate analyses were carried out for each fraction. Merged results and average values were shown for (A) the number and (B) total peak area of identified peptides. Mono-P, Multi-P, and N-P represent singly, multiply, and nonphosphorylated peptides, respectively. Aq: aqueous solution.
fractions was much lower than that in basic fractions, and many singly phosphorylated peptides still remained in the final basic fractions. The lower pH (1.5) changed this profile (shown in Figure S1) to elute more monophosphopeptides in the wash fraction, suggesting that phosphate groups were substantially uncharged at pH 1.5 and were not capable of retention as Lewis bases. Higher TEAP concentration up to 5.0 M did not contribute to enlarging the singly phosphorylated peptide coverage in acidic fractions (shown in Figure S2). Furthermore, the retention of non- and singly phosphorylated peptides on titania MOC tips was dramatically decreased upon increasing the ratio of water in the mobile phase (Figure 3 and Table S3). These results suggested that singly phosphorylated peptides were retained by titania resins by ligand exchange and/or hydrophilic interaction. As a result, singly phosphorylated peptides were predominantly observed in the acidic fractions, whereas multiply phosphorylated peptides as well as the rest of the singly phosphorylated peptides were eluted by the SSR basic solutions. We found that titania resin worked as a HILIC stationary phase under acidic conditions below pH 2.0 when appreciable organic solvent was present. In addition, methylphosphonate buffer (pH 2.0) with 20% ACN is useful for selective elution of singly phosphorylated peptides from titania-MOC tips. However, a lot of nonphosphorylated peptides were also detected in both the wash and the eluted fractions. As described above, we found that the use of lactic acid as a chelating enhancer is effective for improvement of phosphopeptide selectivity in titania-MOC (HAMMOC).19 Therefore, it was expected that highly selective and efficient elution of singly and multiply phosphorylated peptides from titania-MOC tips would
phosphopeptides are slightly retained on a WAX column. Furthermore, in ERLIC mode, column retention is further enhanced depending on the phosphate number per peptide with additional organic solvent which promotes hydrophilic interaction of the phosphate group with the column. Since metal oxides such as titania and zirconia show anion exchange properties under acidic conditions,31 the ERLIC-like elution, according to the phosphate number per phosphopeptide, might be expected. We implemented ERLIC-mode conditions using titaniaMOC tips without lactic acid as a chelation enhancer. Tryptic digests from 100 μg HeLa cell proteins were loaded onto titania-MOC tips with 20 mM methylphosphonate buffer (adjusted to pH 2.0 with NaOH) including 70% ACN. After loading, the titania-MOC tips were washed by the same solution, and the wash fraction was collected. Peptides captured in MOC tips were eluted by a stepwise gradient of TEAP concentration and subsequently by the basic elution conditions which we have already developed previously as the successive and selective release (SSR) approach to maximize the phosphopeptide coverage including multiply phosphorylated peptides.20 The wash and the eluted fractions were desalted by C18 StageTips, followed by LC-MS/MS analysis. Figure 3 and Table S3 show the number and total peak area of identified peptides from 100 μg HeLa cell proteins by using titania-MOC tips. As a result, 858 nonredundant peptides were identified in the wash fraction, of which 752 were nonphosphorylated peptides. With increasing TEAP concentration in the elution buffer, the recovery of phosphopeptides was increased. This observation suggests that titania resin was working as an anionexchanger. However, the recovery of phosphopeptide in acidic 10217
DOI: 10.1021/acs.analchem.5b01216 Anal. Chem. 2015, 87, 10213−10221
Article
Analytical Chemistry
Figure 4. Separation of singly and multiply phosphorylated peptides from HeLa cell extracts by HILIC-mode Ti-HAMMOC. Tryptic peptides from 100 μg of HeLa cell lysates were purified using lactic acid-modified titania-MOC (HAMMOC) tips. Phosphopeptides captured by the titania microcolumn were released by (A) a stepwise gradient of 20−100 mM methylphosphonate buffer (pH 2.0) containing 20% ACN or (B) isocratic elution of 20 mM methylphosphonate buffer (pH 2.0) containing 20% ACN. Successive and selective release using 500 mM disodium hydrogen phosphate, 5% ammonium hydroxide, and 5% pyrrolidine was applied to phosphopeptides bound to metal oxide resin strongly. Each elution volume was 100 μL. Triplicate analyses were carried out by a LTQ-Orbitap XL system with self-pulled needle columns. Mono-p, Multi-P, and N-P represent singly, multiply, and nonphosphorylated peptides, respectively.
Comparison of Optimized HAMMOC with SIMAC (Sequential Elution from IMAC). In order to validate the performance, the optimized HAMMOC method with HILICmode and SSR elutions was compared to the SIMAC approach using tryptic digests from 100 μg of HeLa cell lysates. The results are shown in Figure 5 and Table S6. In SIMAC, 2072 nonredundant phosphopeptides were identified in total and 636 of them were multiply phosphorylated peptides. On the other hand, 2943 nonredundant phosphopeptides were identified by the optimized HAMMOC approach, 1041 of which were multiply phosphorylated peptides. The optimized HAMMOC method dramatically exceeded the SIMAC approach, in both numbers and total peak area of identified singly and multiply phosphorylated peptides. In addition, the phosphopeptide enrichment efficiency in SIMAC was 76% based on the peak areas, and more than half of the identified peptides were nonphosphorylated. It was considered that these results were caused by low recovery and selectivity for phosphopeptides by IMAC as described above. On the other hand, optimized HAMMOC exhibited a phosphopeptide enrichment efficiency of more than 92%. It was concluded that the optimized HAMMOC method with the additional HILIC-mode elution dramatically enhances the coverage of the phosphoproteome without combining other phosphopeptide enrichment approaches such as SIMAC. Improved Phosphoproteome Coverage Using Optimized HAMMOC Combined with a Monolithic Silica-C18
be accomplished by preliminarily removing nonphosphorylated peptides using HAMMOC and subsequently eluting with acidic and basic eluents. We further compared two elution conditions in HILIC-mode HAMMOC. One was a stepwise gradient of 20−100 mM methylphosphonate concentration. The other was an isocratic elution of 20 mM methylphosphonate buffer (pH 2.0) including 20% ACN, and the maximum elution volume was set as 500 μL. In both conditions, the SSR approach using three kinds of basic eluents, 500 mM disodium hydrogen phosphate, 5% ammonium hydroxide, and 5% pyrrolidine, was applied to recover phosphopeptides bound to titania resin strongly. The results are shown in Figure 4 and Table S4. In both acidic elution conditions, singly phosphorylated peptides were selectively recovered, and multiply phosphorylated peptides were enriched in the basic fractions. Moreover, the addition of lactic acid to loading buffer dramatically reduced nonphosphorylated peptides in each fraction. The overlap of phosphopeptides between adjacent fractions exhibited a similar trend with elution via either a gradient of methylphosphonate concentration or isocratic elution with large volumes of methylphosphonate buffer (Table S5). However, regarding the ratio of multiply phosphorylated peptides, the isocratic condition (66.4% and 49.3%) was better than the gradient condition (48.4% and 35.2%) in the phosphate- and the amines-eluted fractions, respectively (Table S4). Therefore, we decided to employ the isocratic elution condition. 10218
DOI: 10.1021/acs.analchem.5b01216 Anal. Chem. 2015, 87, 10213−10221
Article
Analytical Chemistry
Figure 5. Comparison of phosphopeptide enrichment of HeLa cell extracts between HILIC-mode Ti-HAMMOC and SIMAC. Tryptic peptides from 100 μg of HeLa cell lysates were examined by lactic acid-modified Ti-HAMMOC and SIMAC. Triplicate analyses were carried out for each fraction by a LTQ-Orbitap XL system with self-pulled needle columns. Merged results and average values were shown for (A) the number of identified peptides and (B) the peak area values, respectively. Mono-P, Multi-P, and N-P represent singly, multiply, and nonphosphorylated peptides, respectively. (C) Area proportional Venn diagram of phosphopeptides identified with HILIC-mode Ti-HAMMOC and SIMAC.
Capillary Column. Finally, coping with the elevated sample complexity due to the improved recovery of phosphopeptides, the optimized HAMMOC method was combined with a highly efficient LC separation by utilizing a meter-long monolithic silica capillary column.47−50 Using a 4 h gradient, 6329 and 8169 unique phosphopeptides were identified from HILIC and SSR fractions which included 80.3% and 38.6% singly phosphorylated peptides, respectively (Figure 6A,B and Table S7). Taken together, we successfully identified 11 300 unique phosphopeptides including 5924 singly phosphorylated and 5376 multiply phosphorylated peptides from triplicate analysis of 100 μg each of HeLa cell lysates, and 7575 nonredundant phosphosites on 2115 proteins could be identified. The detailed information on identified phosphopeptides was summarized in Tables S8 and S9. Figure 6C shows the averaged physicochemical properties of identified phosphopeptides based on the amino acid sequences. Phosphopeptides eluted in the HILIC fraction included less acidic residues than those in SSR ones. This result indicated that both phosphoryl and carboxy groups are Lewis bases and peptides and that a lot of these groups interact strongly with titania, a Lewis acid. Furthermore, since charge balances of phosphopeptides would be important to some extent for initial binding to the titania surface, phosphopeptides eluted in SSR fractions tended to
include not only multiple phosphate groups but also more acidic residues. The longer peptide length and more hydrophilic property of phosphopeptides in SSR fractions were mainly derived from the higher contents of aspartate and glutamate residues, conferring multiple negative charges, as well as multiple phosphate groups. Using our optimized HAMMOC method, singly and multiply phosphorylated peptides could be separately eluted from a single HAMMOC microcolumn. As a result, multiply phosphorylated peptides could avoid ion suppression by coeluting singly phosphorylated peptides and have more chances to be identified in MS analysis. Consequently, we successfully identified more than 10 000 phosphopeptides, including 99 out of 504 protein kinases and 390 out of the 2287 proteins containing “transcription” in their Swissprot keywords (UniProtKB/Swiss-Prot Release of 24-Jul-2013) which are generally low-abundance proteins. These results would be easily improved by utilizing the combination of longer columns and gradients providing higher separation in nanoLC and thereby high sensitivity in MS analysis.48 Furthermore, our method requires only a tiny amount of proteins (
