University of Southern Denmark

Phosphotyrosine biased enrichment of tryptic peptides from cancer cells by combining pY-MIP and TiO2 affinity resins

Bllaci, Loreta; Torsetnes, Silje Bøen; Wierzbicka, Celina Katarzyna; Shinde, Sudhirkumar;Sellergren, Börje; Rogowska-Wrzesinska, Adelina; Jensen, Ole Nørregaard

Published in:Analytical Chemistry

DOI:10.1021/acs.analchem.7b02091

Publication date:2017

Document version:Accepted manuscript

Citation for pulished version (APA):Bllaci, L., Torsetnes, S. B., Wierzbicka, C. K., Shinde, S., Sellergren, B., Rogowska-Wrzesinska, A., & Jensen,O. N. (2017). Phosphotyrosine biased enrichment of tryptic peptides from cancer cells by combining pY-MIP andTiO2 affinity resins. Analytical Chemistry, 89(21), 11332–11340. https://doi.org/10.1021/acs.analchem.7b02091

Go to publication entry in University of Southern Denmark's Research Portal

Terms of useThis work is brought to you by the University of Southern Denmark.Unless otherwise specified it has been shared according to the terms for self-archiving.If no other license is stated, these terms apply:

• You may download this work for personal use only. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying this open access versionIf you believe that this document breaches copyright please contact us providing details and we will investigate your claim.Please direct all enquiries to puresupport@bib.sdu.dk

Download date: 12. Feb. 2022

Subscriber access provided by University Library of Southern Denmark

Analytical Chemistry is published by the American Chemical Society. 1155 SixteenthStreet N.W., Washington, DC 20036Published by American Chemical Society. Copyright © American Chemical Society.However, no copyright claim is made to original U.S. Government works, or worksproduced by employees of any Commonwealth realm Crown government in the courseof their duties.

Article

Phosphotyrosine biased enrichment of tryptic peptides fromcancer cells by combining pY-MIP and TiO2 affinity resinsLoreta Bllaci, Silje Bøen Torsetnes, Celina Katarzyna Wierzbicka, Sudhirkumar

Shinde, Börje Sellergren, Adelina Rogowska-Wrzesinska, and Ole Nørregaard JensenAnal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.7b02091 • Publication Date (Web): 03 Oct 2017

Downloaded from http://pubs.acs.org on October 23, 2017

Just Accepted

“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are postedonline prior to technical editing, formatting for publication and author proofing. The American ChemicalSociety provides “Just Accepted” as a free service to the research community to expedite thedissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscriptsappear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have beenfully peer reviewed, but should not be considered the official version of record. They are accessible to allreaders and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offeredto authors. Therefore, the “Just Accepted” Web site may not include all articles that will be publishedin the journal. After a manuscript is technically edited and formatted, it will be removed from the “JustAccepted” Web site and published as an ASAP article. Note that technical editing may introduce minorchanges to the manuscript text and/or graphics which could affect content, and all legal disclaimersand ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errorsor consequences arising from the use of information contained in these “Just Accepted” manuscripts.

1

Phosphotyrosine biased enrichment of tryptic peptides from cancer

cells by combining pY-MIP and TiO2 affinity resins

Loreta Bllaci,§# Silje B. Torsetnes,‡§# Celina Wierzbicka,¶ Sudhirkumar Shinde,¶ Börje Seller-gren,¶ Adelina Rogowska-Wrzesinska, § and Ole N. Jensen∗ §

§Department of Biochemistry and Molecular Biology and VILLUM Center for Bioanalytical Sciences, University of

Southern Denmark, DK-5230 Odense M, Denmark ¶Department of Biomedical Sciences, Malmö University, S-20506 Malmö, Sweden

ABSTRACT: Protein phosphorylation at distinct tyrosine residues (pY) is essential for fast, specific and accurate signal transduction in cells. Enrichment of pY-containing peptides derived from phosphoproteins is commonly facilitated by use of immobilized anti-pY antibodies prior to phosphoproteomics analysis by mass spectrometry. We here report on an al-ternative approach for pY-peptide enrichment using inexpensive pY imprinted polymer (pY-MIP). We assessed by mass spectrometry the performance of pY-MIP for enrichment and sequencing of phosphopeptides obtained by tryptic diges-tion of protein extracts from HeLa cells. The combination of pY-MIP and TiO2 based phosphopeptide enrichment provid-ed more than 90% selectivity for phosphopeptides. Mass spectrometry signal intensities were enhanced for most pY-phosphopeptides (approximately 70%) when using the pY-MIP-TiO2 combination as compared to TiO2 alone. pY consti-tuted up to 8% of the pY-MIP-TiO2 enriched phosphopeptide fractions. The pY-MIP-TiO2 and the TiO2 protocols yielded comparable numbers of distinct phosphopeptides, 1693 and 1842 respectively, from microgram levels of peptide samples. Detailed analysis of physicochemical properties of pY-MIP-TiO2 enriched phosphopeptides demonstrated that this proto-col retrieved phosphopeptides that tend to be smaller (<24 residues), less acidic and almost exclusively monophosphory-lated, as compared to TiO2 alone. These unique properties render the pY-MIP based phosphopeptide enrichment tech-nique an attractive alternative for applications in phosphoproteomics research.

Protein phosphorylation is a widespread and universal regulatory mechanism in prokaryote and eukaryote cell signaling and metabolism.1–5 Highly sensitive mass spec-trometry (MS) techniques now makes phosphopeptide sequencing and quantitation feasible at a routine basis, giving rise to the rapidly growing field of phosphoprote-omics.6–8

The development of chemical, biochemical and immu-nological methods for phosphoprotein and phosphopep-tide isolation, enrichment and separation has received great attention in the context of phosphoproteomics. The most successful methods for phosphopeptide enrichment rely on the affinity of inorganic moieties towards the phosphate groups located on serine (pS), threonine (pT) and tyrosine (pY) residues in peptides and proteins. Fe3+- and Ti4+-immobilized metal ion affinity chromatography (IMAC) and titanium dioxide (TiO2) are widely used methods for global phosphopeptide enrichment prior to MS.9–15 Despite their successful application in global phosphoproteome studies, affinity based methods such as IMAC and TiO2 methods lack the capability to discrimi-nate between the three common types of phosphorylated residues in eukaryotes, i.e. pS, pT and pY. In particular, phosphorylation at tyrosine residues is of immense inter-

est due to its role in cell signaling and regulation of me-tabolism, cellular growth, development and differentia-tion.16,17 The event of pY is of low abundance (<1%) com-pared to pS (90%) and pT (10%) in most eukaryote cells.18 Many tyrosine kinases are membrane bound receptors of growth factors and hormones which initiate signal trans-duction.19–21 Aberrant phosphorylation of tyrosine residues on such receptors may lead to cellular transformation and cancer.16 The most efficient and specific bioanalytical method for enrichment of pY-peptides and proteins relies on anti-pY antibodies yielding thousands of pY-peptide identifications.22–24 Some drawbacks related to the use of anti-pY antibodies include the requirement of large amounts of starting material (up to hundreds of mg), var-iability of antibody quality and specificity/selectivity, and cost.22,25

Advents in polymer chemistry and molecular mimicry26 gave rise to the concept of using molecularly imprinted polymers (MIPs) for molecular recognition and separation of peptides and proteins.27–31 Recently, we introduced a new approach for phosphopeptide enrichment featuring neutral, urea-based imprinted phosphate receptors.32–34 The resulting MIPs could in principle address some of the aforementioned deficiencies. In this context, phosphoty-

Page 1 of 11

ACS Paragon Plus Environment

Analytical Chemistry

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

2

rosine imprinted polymers (pY-MIPs) were used to selec-tively enrich pY-peptides spiked at low levels into proteo-lytic digests. However, this approach is yet to be demon-strated on native biological samples. To investigate this, we produced a pY-MIP based on a modified version of the procedure reported by Emgenbroich et al.32 and tested it on simple and complex tryptic peptide mixtures, the lat-ter derived from human cervical cancer cell (HeLa) pro-tein extract. The method development involved assess-ment of the properties of the pY-MIP alone and in combi-nation with our established TiO2 protocol for phospho-peptide enrichment at the low microgram level. We find that the sequential use of pY-MIP and TiO2 enrichment provide some distinct advantages for enhanced detection of pY-peptides in phosphoproteomics.

EXPERIMENTAL

Synthesis of pY-MIP. Molecularly imprinted polymer with pY affinity was prepared as described previously with some modifications.32–35 (Supporting Information).

Preparation of Semi-complex Peptide Mixture. The semi-complex peptide mixture, prepared as described elsewhere,14 was brought to a final concentration of 1 pmol/µL per digested protein, and spiked with a mixture containing four pY- and four pS-peptides (Supporting Information).

Cancer Cell Protein Extract. Three 10 cm dishes of human cervical cancer epithelial cells, HeLa, were cul-tured and grown to 95% confluence. Stimulation with sodium pervanadate (PVD)-supplemented media was carried for a duration of 30 min. Cells were then washed and scraped/transferred into 15 mL falcon tubes, and cen-trifuged. The resulting protein lysate (approximately 1 mg) was reduced and alkylated, then digested with endo-peptidase Lys-C for 3 h prior to overnight trypsin diges-tion. See Supplementary Information for details. A vol-ume of 2 µL tryptic peptide mixture was used for accurate quantification of peptide concentration by amino acid composition analysis (AAA).37

Sample Desalting. Peptide samples were desalted after TiO2 enrichment and prior to pY-MIP enrichment (see Supplementary Information). Samples were then dried in a vacuum concentrator and stored until further use.

Enrichment Strategies. The experimental work con-sisted of four protocols: the reference enrichment TiO2, pY-MIP and two combined protocols: pY-MIP-TiO2 and TiO2-pY-MIP. The workflow and experimental details for analysis of the semi-complex peptide sample are provided in Supporting Information, Figure S1.

Direct pY-MIP. The dry sample resulting from desalting of 200 µg peptides (determined by AAA) was first dis-solved in 2 µL of 10% TFA, then added 4µL of Milli-Q wa-ter and mixed thoroughly, before 194 µL MeCN was added slowly drop-wise to give a final composition of 97% MeCN in 0.1% TFA as the pY-MIP loading solution. pY-MIP columns were prepared and operated similarly to the

desalting columns. They were manually packed in dispos-able 200 µL pipette tips which were plugged with a layer of C8 EmporeTM extraction disk to retain pY-MIP parti-cles. A suspension of pY-MIP in MeCN was added to the plugged pipette tip until the length of the column reached 1.5 cm. Solvents and fractions were collected in 1.5 mL low-binding vials with perforated lids where a 200 µL tip could fit and be stable upon centrifugation. Passing of liquid through the column was operated by centrifugal forces spun at 1000 rpmi for 10−15 min. Columns were first activated with 200 µL of MeCN, and conditioned with 200 µL of 97% MeCN in 0.1% TFA. The reconstituted sample described above was then added, and the column was washed with 50 µL of 97% MeCN in 0.1% TFA. The collec-tion vial was replaced by a new one and elution was per-formed in two steps using two solutions. First with 200 µL 95% MeOH in 0.1% TFA, followed by 200 µL of 50% MeOH in 1% TFA. Eluents were then pooled, dried using a vacuum concentrator and stored at -20◦C until further use.

TiO2 enrichment. TiO2 enrichment of 200 µg peptides was performed as described by Palmisano et al.38 The de-tailed experimental procedure is in the Supporting Infor-mation.

TiO2-pY-MIP. Following TiO2 enrichment (Supplemen-tary Information) samples were dried in a vacuum con-centrator, re-dissolved and applied for pY-MIP (vide su-pra). Eluents were dried by a vacuum concentrator and stored at -20 ◦C.

pY-MIP-TiO2. Samples were first processed with pY-MIP (vide supra) above. The resulting eluents were dried, re-dissolved in TiO2-loading solution and incubated with TiO2 beads (TiO2 enrichment described in Supplementary Information). The amount of TiO2 beads was determined based on the amount of peptides in the pY-MIP eluate, thus one additional replicate was prepared for AAA analy-sis. The peptide amount was found to be 2 µg and 0.3 mg TiO2 beads were used for enrichment. Washing and elu-tion were performed as described previously, however, volumes were reduced to 100 µL and 50 µL for washing and elution, respectively. Eluates were desalted as de-scribed below and, finally, dried by a vacuum concentra-tor and stored at -20◦C.

MALDI MS. MALDI MS was performed using an Ultra-flextreme mass spectrometer and processed with FlexA-nalysis software (Bruker Daltonics, Bremen, Germany).

Reversed-phase nanoLC-ESI-MS/MS. Samples were chromatographically separated by an EASY-nLC system (Thermo Scientific, Odense, Denmark) coupled online to a LTQ-Orbitrap Velos mass spectrometer (Thermo Scien-tific, San Jose CA). Peptides were separated on a 20 cm long fused silica capillary column (75 µm ID), packed in-house with ReproSil-Pur C18 AQ, 3 µm reversed-phase material (Dr. Maisch, Ammerbuch-Entringen, Germany). Peptides were eluted following a 130 min gradient of mo-bile phase A (0.1% FA) to 34% mobile phase B (95% MeCN in 0.1% FA), flow rate of 250 nL/min. The LTQ-Orbitrap Velos was operated in positive ion mode with data-

Page 2 of 11

ACS Paragon Plus Environment

Analytical Chemistry

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

3

dependent acquisition. The full scan was acquired in the Orbitrap with an automatic gain control (AGC) target value of 1 × 106 ions and a maximum fill time of 500 ms. Full-MS scans were acquired with resolution of 60,000 FWHM followed by 10 MS/MS scans of the most intense ions, also acquired with a mass resolution at 15000 (HCD normalized collision energy = 35; activation time 10 ms). Raw data were viewed in Xcalibur v2.0.7 (Thermo Fisher Scientific, USA).

Data Analysis. Data analysis of the raw data was per-formed with Proteome Discoverer v1.4 (Thermo Scien-tific) and searched against the Swiss-Prot human v3.53 database (20,199 entries) using an in-house Mascot server (v2.3, Matrix Science Ltd, London, UK). Database search-es were performed as follows: precursor mass tolerance of 10 ppm, product ion mass tolerance of 0.05 Da, Cys car-bamidomethylation as fixed modification, oxidation of Met and phosphorylation on S/T/Y as variable modifica-tions and two missed cleavages of trypsin were allowed. Only peptides with q−value of 0.01 (Percolator), search engine 1, Mascot rank 1 and cut−off value of Mascot score ≥ 22 were considered for further analysis. Phosphopep-tides were filtered for pRS values ≥ 95.

Relative quantification of phosphopeptides was achieved by using the Precursor Ion Area (PIA) feature of Proteome Discoverer. Normalization of data for pairwise comparison of two sets of triplicate data, e.g. pY-MIP-TiO2 vs TiO2, was performed in GraphPad Prism v7 using a scaling factor. The 0% level was set to the lowest value among the minima in the six intergroups and the 100% level was set to the highest value among the maxima in the six intergroups. Comparative analysis of phosphopep-tide intensities between protocols was performed using the average across the three replicates of each protocol. Paired t-tests were performed in GraphPad Prism v7.

Phosphorylation Motif Analysis. Sequence logos were visualized with IceLogo 1.2.39 Aligned sequences for Ice-Logo analysis were extracted using the motif−X algo-rithm.40,41 Subtractive IceLogo analysis was performed by using the TiO2 dataset as background and comparing it to pY-MIP-TiO2 TiO2-pY-MIP and pY-MIP datasets using a p-value setting of 0.05.

Figure 1. MALDI MS spectra of a simple peptide mixture analyzed before and after phosphopeptide enrichment. (A) Control, untreated peptide mixture, (B) Eluate obtained by pY-MIP processing of the simple mixture, (C) superim-posed spectra obtained by TiO2, pY-MIP-TiO2 and TiO2-pY-MIP enrichment of the simple peptide mixture shown in blue, orange and green, respectively. Five spectra panels in (D) show four pY-peptides. DRVpYIHPF (m/z: 1126.51), GADDSYpYTAAR (m/z: 1198.44), GADDpYpYTAR (m/z: 1278.41), TRDIpYETDpYpYRK (m/z: 1862.68) and one detect-ed pS-peptide AVPSPPPApSPR (m/z: 1154.55). Phosphopep-tides in the spectra are marked with their m/z value, pY-peptides are labelled by red asterisks.

Page 3 of 11

ACS Paragon Plus Environment

Analytical Chemistry

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

4

RESULTS

Phosphopeptide Enrichment and MS analysis. We initially tested the pY-MIP performance for phosphopeptide enrichment by using a simple tryptic peptide mixture containing regular (non-phosphorylated) peptides and a series of phosphorylated peptides, four of which were spiked-in synthetic pY-peptides (see the experimental workflow and synthetic peptides in Figure S1 and Table S1, Supporting Information). Untreated control sample and the pY-MIP processed samples were analyzed by MALDI MS (Figure 1, A-B). The pY-MIP sample processing protocol reduced the complexity of the MALDI mass spectrum, i.e. fewer ion signals were observed as compared to the control sample. Several pY-peptide signals were observed as indicated by asterisks in Figure 1B. Several other peptide ion signals were also observed. Based on these initial observations and additional pY-MIP assessment by LC-MS/MS experiments (see below) we concluded that the pY-MIP alone did not have sufficient capacity and specificity to allow for enrichment of phosphopeptides from very complex peptide mixtures, such as those derived from human cell lysate.

Next, we hypothesized that the combination of pY-MIP and TiO2 enrichment could provide the capacity, specific-ity and selectivity for enrichment of pY-peptides in phos-phoproteomics experiments.

We investigated the sequential use of pY-MIP and TiO2 (pY-MIP-TiO2 protocol) or TiO2 and pY-MIP (TiO2-pY-MIP protocol) and compared their performance to the TiO2 protocol used as benchmark. Initial tests using MALDI MS (Figure 1C) were encouraging as they demon-strated enhanced selectivity for pY-peptides and reduced sample complexity, although at reduced signal-to-background levels for the pY-peptides. However, the comparison of superimposed spectra (insets in Figure 1D) exhibit a higher relative signal from most phosphopep-tides for pY-MIP-TiO2 than the two other methods (TiO2 and TiO2-pY-MIP).

Assessment of pY-MIP by LC-MS/MS. Next, we ap-plied LC-MS/MS to assess the performance of pY-MIP based protocols for phosphopeptide enrichment from a complex tryptic peptide mixture derived from HeLa cell protein extract. We treated the HeLa cells with the phos-phatase inhibitor sodium pervanadate to enhance protein tyrosine phosphorylation levels.36 We tested pY-MIP, pY-MIP-TiO2, and TiO2-pY-MIP protocols and compared them to our standard TiO2 enrichment protocol (Figure 2).

The combined pY-MIP-TiO2 protocol identified 1693 phosphopeptides, whereof 136 were phosphorylated only on tyrosine (Table 1). The standard TiO2 protocol re-trieved 1842 phosphopeptides, 142 thereof phosphorylated exclusively on tyrosine. Thus, serial use of pY-MIP prior to TiO2 performed as well as TiO2 alone, both protocols identifying on the order of 1700-1800 phosphopeptides (pRS >=95%). The use of TiO2 prior to pY-MIP was slight-ly less efficient

Figure 2. Workflow for phosphopeptide enrichment and method assessment using a HeLa cell lysate. HeLa cells were stimulated with sodium pervanadate prior to harvesting to block the activity of tyrosine phosphatases. Cells were lysed and protein digested with trypsin. Of the twelve ali-quots (200 µg peptide) six were enriched with TiO2 (left side of the scheme) while the other six were desalted and treated with pY-MIP (right side of the scheme). Next, out of the six TiO2-enriched samples three were left without further pro-cessing (protocol 1), whilst the other three were processed with pY-MIP (protocol 2). Similarly, out of the six pY-MIP treated samples three were not processed further (protocol 4); whilst the three were subjected to TiO2 based enrichment (protocol 3). All samples were finally analyzed by nLC-MS/MS. Also depicted is the pY-MIP phosphate binding site featuring two urea groups engaging in quadruple hydrogen bonds with a phosphotyrosine peptide.

32

and identified fewer phosphopeptides (1328). (Table 1, Figure S2, Supporting Information).

Having determined that the serial use of pY-MIP prior to TiO2 exhibited rather good characteristics for phos-phopeptide recovery, we set out to further study the re-covery of pY-, pS- and pT-peptides by using a quantitative approach. We compared the phosphopeptide ion intensi-ties of the pY (n=100) pS (n=998) and pT (n=56) peptides that were detected and identified by both the pY-MIP-

Page 4 of 11

ACS Paragon Plus Environment

Analytical Chemistry

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

5

TiO2 protocol and the TiO2 protocol. For this analysis, only peptides with one type of phosphorylated residue, on one or more sites, were considered (Figure 3).

Table 1. Phosphopeptide identification by affinity enrichment using four protocols (three technical replicates each).

Note. Method selectivity values, in the lower panel, were calculated as follows: ���������� � 100 � �� ��� � ���⁄ ,

For each individual phosphopeptide, we determined the ratio of normalized ion intensities observed by using the two protocols (Figure 3). If there was no discrimina-tion or differences in binding affinity/avidity between the TiO2 protocol and pY-MIP-TiO2 protocol for these phos-phopeptides, then the phosphopeptide ratios should dis-tribute evenly in the logarithmic diagram, with 50% being between 0 and 1 and 50% being above 1. Figure 3 shows that this indeed is the case for pS-peptides. The distribu-tion of pY-peptides, on the other hand, is biased towards higher absolute recovery, i.e. higher phosphopeptide ion intensities for 72% of the pY-peptides when using the pY-MIP-TiO2 protocol as compared to the TiO2 protocol. This

implies that pY-MIP-TiO2 has a higher affinity for a ma-jority of the pY-peptides than does TiO2. Interestingly, the common set of pT-peptides, although a small number, also exhibited higher ion signal intensities when using pY-MIP-TiO2 rather than TiO2 alone. Differences in ion intensities between pY-MIP-TiO2 and TiO2 were signifi-cant (P<0.0001) for pY and pT peptides (for pY: M=3.7 and SD =6.5, for pT: M =7.9 and SD =10.8), but not for pS- (P>0.7, M =0.1, SD =10.1). The TiO2-pY-MIP protocol did not provide any advantages with regard to quantitative enrichment of phosphopeptides (Figure S3 and additional statistics in Supporting Information).

In summary, pY-MIP has interesting properties as a first-line enrichment method for complex peptide sam-ples from a human cancer cell line, when followed by TiO2 enrichment. The total number of phosphopeptides and the number of pY-peptides are comparable to those obtained by TiO2 alone, despite added sample handling steps. Importantly, ion signal intensities were enhanced for a majority of pY-peptides as compared to TiO2 en-richment alone. This was also observed for some pT-peptides, but not for pS-peptides indicating that pY-MIP resin selects bulkier phosphoamino acid residue side chains such as pY and pT.

Physicochemical Properties of Phosphopeptides En-riched by TiO2 and pY-MIP-methods.

Next, we studied the physicochemical properties of the phosphopeptides that were enriched by TiO2, pY-MIP, pY-MIP-TiO2 and TiO2-pY-MIP protocols. Previous stud-ies suggest a slight preference of TiO2 towards acidic phosphopeptides35,42 in contrast to IMAC which was re-ported to enrich predominantly multiphosphorylated peptides.43,44 For this analysis we considered only phos-phopeptides that were detected in at least two technical replicates, and were unique to TiO2, pY-MIP-TiO2 and TiO2-pY-MIP (Figure S4). Given the limited number of phosphopeptides detected by pY-MIP, we considered all the identified phosphopeptides (258 phosphopeptides, minimally detected in two replicates).

The peptides were first grouped as either acidic, basic, neutral or hydrophobic (Figure 4, upper panel).

Page 5 of 11

ACS Paragon Plus Environment

Analytical Chemistry

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

6

Figure 3. Effect of pY-MIP-TiO2 enrichment on precursor ion intensities. pY-, pT- and pS-peptides, which were com-monly detected by pY-MIP-TiO2 and TiO2 were considered. Intensity refers to the average precursor ion area (PIA). Note that one pS-peptides was not included.

Peptides were assessed as hydrophobic when constitut-ed by 25% or more hydrophobic residues. The relative prevalence of peptides within these groups is represented by relative frequencies (RF) obtained by normalizing to the total number of peptides for each method. The chem-ical differences between pY-MIP-methods and reference method (TiO2) were then statistically tested through their RFs by one-way Analysis of Variance (ANOVA, p <0.05) followed by Dunnett’s test for multiple comparisons. The relative enrichment of acidic peptides (Figure 4, upper panel) was lower than the reference (TiO2) for all the three pY-MIP based methods. More specifically, the dif-ferences between pY-MIP and TiO2 (30 ±3 vs. 49±1%) and pY-MIP-TiO2 vs. TiO2 (27±3 vs. 49±1%) were statistically

determined as extremely significantly different (P<0.001), whilst TiO2-pY-MIP vs. TiO2 (42±2 vs. 49±1%) were signif-icantly different (P<0.05). The relative enrichment of basic peptides (Figure 4, upper panel) was higher for two of the three pY-MIP-methods, i.e. stand-alone pY-MIP and pY-MIP-TiO2 (Table S2 and Supporting Information).

The relative enrichment of neutral peptides (Figure 4, upper panel) was characterized by significantly higher RFs for the two combined methods compared to the ref-erence (TiO2-pY-MIP vs. TiO2: 21±1 vs. 13±1%, P<0.001, pY-MIP-TiO2 vs. TiO2: 21±12 vs. 13±1%, P<0.001) whilst no significant difference was found for TiO2 and pY-MIP (13 ±1 vs. 12±1%, P >0.05). Similarly, when hydrophobicity was considered, the combined methods pY-MIP-TiO2 and TiO2-pY-MIP yield significantly more hydrophobic pep-tides compared to TiO2 and pY-MIP: (41±2 vs. 19±1%, P<0.001), and (25±2 vs. 19±1%, P <0.01) (Figure 4, upper panel). Also, the two stand-alone methods, pY-MIP and TiO2 differ from each other, but with a lower statistical significance (15±1 vs. 19± 1, P<0.05).

Page 6 of 11

ACS Paragon Plus Environment

Analytical Chemistry

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

7

The analysis of charge distribution in phosphopeptides originating from the four methods suggests that pY-MIP and pY-MIP-TiO2, unlike TiO2, are not biased towards acidic peptides. This agrees with our previous report de-scribing pS-MIP based phosphopeptide enrichments.35 The two combined methods pY-MIP-TiO2 and TiO2-pY-MIP recover more uncharged and hydrophobic residues than the two stand-alone methods (pY-MIP and TiO2).

Next, we investigated the phosphopeptides to deter-mine the degree of phosphorylation and length of pep-tides for each of the four enrichment protocols (Figure 4, lower panel).

Phosphorylated peptides were divided into mono- or multiphosphorylated groups and into peptides of length 6-24 residues and peptides of length ≥25 residues. The RFs for these groups were then statistically tested with Dunnett's method for multiple comparisons in ANOVA (one-way, p<0.05) similarly to previous section. We ob-served that pY-MIP-based methods have a distinct prefer-ence for monophosphorylated and relatively short phos-phopeptides (Figure 4, lower panel and Supporting In-formation, Table S2).

We conclude that pY-MIP may act as size-filter for tryp-tic phosphopeptides retrieved from a biological sample. As we have concluded in our recent report, this is related to the broad pore size distribution of the material with a significant number of pores in the low meso- to mi-croporous regime (1-20 nm).45 The result of the filtering effect in this case however, is that the three pY-MIP-based methods demonstrate similar size profiles, with a shorter average peptide length compared to those derived from TiO2. Also, small tryptic peptides have higher chances of bearing a singular phosphorylated site resulting in a high-er probability of pY-MIP-based methods recovering monophosphorylated peptides. Such complementarity of pY-MIP may be advantageous, allowing researchers to tailor enrichment strategies for particular subsets of phosphopeptides.

Motif Analysis. Protein kinases bind to their substrates by recognizing certain amino acid motifs (consensus se-quences) surrounding a phosphorylation site (specificity determinants) and are broadly classified as proline-directed, acidophilic and basophilic.46–48 To assess wheth-er the observed chemical differences of phosphopeptides derived by our four analytical methods would be reflected in the sequence motifs and classes of kinase substrates, we analyzed the amino acid sequences with IceLogo.39 We performed a subtractive sequence logo analysis using the TiO2 dataset as the reference and comparing to the pY-MIP, TiO2-pY-MIP, and pY-MIP-TiO2 datasets.

pS and pT motifs were enriched for proline (P) in the position next to the phosphorylated site (+1) and depleted for acidic amino acid residues (E,D) at that position (Fig-ure 5A-C). This agrees with the bias observed in our pre-vious report on pS-MIP based phosphopeptide enrich-ments 35 and suggests that a proline residue next to the phosphorylation site may promote binding to the pY-MIP

groove, possibly by introducing a kink in the peptide chain.

Figure 4. Effect of enrichment methods on the physico-chemical properties of phosphopeptides. Bar charts of the upper panel show the relative frequencies of phospho-peptides classified as acidic, basic, neutral and hydrophobic. In the lower panel phosphopeptides are grouped according the level of phosphorylation and length. Bar charts were plot-ted using the set of the unique phosphopeptides of TiO2, pY-MIP-TiO2, TiO2-pY-MIP and total phosphopeptides of pY-MIP. These peptides were detected in minimum two of three replicates per method, and one count per replicate. Statisti-cal analysis consisted of one-way ANOVA (p<0.05), and Dunnett's multiple comparison test. Statistical significance: * (significant, P<0.05), **(highly significant, P<0.01), ***(extremely significant, P<0.001).

The subtractive sequence motif analysis confirms that the pY-MIP based methods recover pS/pT-phosphopeptides that are less acidic than those recovered by TiO2 alone (Figure 5A-C). Subtractive sequence motif analysis for pY-phosphopeptides provided a different re-sult (Figure 5 D-F). Small hydrophobic amino acid resi-

Page 7 of 11

ACS Paragon Plus Environment

Analytical Chemistry

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

8

dues (A,V) were enriched in the -1 and +2 positions. Inter-estingly, K was enriched in the -1 position for the pY-MIP-TiO2 method, whereas T was enriched in the +1 position (Figure 5E). Although this sequence motif analysis was based on a rather small number of pY-peptides it suggests that hydrophobic interactions play a role in substrate binding of the pY-MIP.

We conclude that phosphopeptides obtained by the four enrichment protocols derived primarily from the activity of proline-directed kinases. The incorporation of pY-MIP in the combinatorial methods is associated with a reduction on the representation of motifs recognized by acidic-kinases compared to the reference TiO2; a result consistent with the analysis of chemical properties reveal-ing depletion of acidic phosphopeptides when pY-MIP was incorporated in the workflows. However, IceLogo did not reveal any observable difference in regulation of basic residues in +1 position, which is usually a signature of basophilic kinase sites. Interestingly, the apparent en-richment of T-residues in proximity to pS/pT (position +2) and adjacent to pY (position +1) by pY-MIP-TiO2 could be of interest in targeting dual specificity kinases.47

Figure 5. Sequence logos of phosphorylation sites obtained by subtractive sequence motif analysis against the TiO2 dataset. IceLogos centered on pS and pT residues were determined using default values of probability (p=0.05). A nine and five amino acid residue window was used for pS/pT (A-C) and pY (D-F) sequences respectively.

DISCUSSION

We thoroughly analyzed the performance of pY-MIP-based phosphopeptide enrichment by using a complex cancer cell (HeLa) protein extract. The potential of using pY-MIP for the enrichment of pY-peptides was previously investigated for simple peptide mixtures and brain lysates spiked with synthetic peptides.33 Yet, to the best of our knowledge there is no report describing the use of pY-MIP for enrichment of endogenous,34 low abundant pY-peptides, directly from a biological sample. Our present work using HeLa cell lysates resulted in the development of three pY-MIP-based workflows: direct (stand-alone) pY-MIP, pY-MIP-TiO2 and TiO2-pY-MIP. Their perfor-

mance for enrichment of pY-, pS-, and pT-peptides was compared to TiO2 including assessment of enrichment efficiency, physicochemical properties of recovered pep-tides, and kinase substrate motifs.

The main finding of this work is that pY-MIP quantita-tively enrich pY-peptides when succeeded by TiO2, in the form of a combined method pY-MIP-TiO2. Nearly 70% of pY-peptides which were detected simultaneously by pY-MIP-TiO2 and the reference TiO2 method, had higher signal intensities than their TiO2 counterpart (Figure 3). The pY-MIP-mediated enrichment of pT-peptides is a novel finding of high interest, since pT is 9-10-fold less common than pS-sites.9 This cross reactivity can have different causes. The intrinsic side chain selectivity of the pY-MIP has been extensively proven at the amino acid level and is ascribed to a tight neutral binding site, bind-ing the phosphate group in a cleft-like manner. The inter-actions with peptides however also involve non-specific binding, notably hydrophobic effects.35 This results in a preference for hydrophobic sequences, , e.g. proline rich sequences, that is known to be more abundant in cancer-related pT motifs.49 Finally, pY-MIP-TiO2 did not show superiority over TiO2 for the quantitative enrichment of pS-peptides. Both methods showed comparable represen-tation of the common pS-peptides (Figure 3).

The alternative TiO2-pY-MIP method was inferior to TiO2 both by qualitative and quantitative performance. All resulting phosphopeptides exhibited lower ion inten-sities as compared to their counterpart enriched by TiO2 (Figure S3). This reduction of intensities can be related to peptide losses during successive steps of sample prepara-tion.

Our attempts to develop a one-step stand-alone pY-MIP enrichment method did not succeed at this time; pY-MIP bound excessively to non-phosphorylated peptides, possibly due to interactions between pY-MIP and hydro-phobic regions and oxyanion-groups in the peptides. The unspecific peptide binding of pY-MIP when presented to complex peptide mixture, without prior preparation or fractionation, is in agreement with the results of Chen et al.,30 who used a chemically similar MIP (imprinted for pS). From our results, we conclude that pY-MIP-TiO2 en-rich pY–peptides, based on pY-MIP-affinity, and it is comparable to the stand-alone TiO2 protocol, despite a few additional sample handling steps. Importantly, the combined method provides an access to distinct parts of phosphoproteome since the overlap of phosphopeptides enriched by these two approaches (pY-MIP-TiO2 and TiO2) was only about 50%. This new pY-MIP-TiO2 strate-gy is applicable to quantitative analysis of differentially phosphorylated proteins in biological studies, e.g. to study aberrant cell signaling in health and disease.

Chemical biases of phosphopeptide enrichment meth-ods are widely reported and constitute an ongoing topic of research in phosphoproteomics.50 Diverse chemical interactions between resins and amino acid residues and sequences lead to preferential binding of certain subsets of phosphopeptides. Since pY-MIP and TiO2 exploit dif-

Page 8 of 11

ACS Paragon Plus Environment

Analytical Chemistry

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

9

ferent chemical properties, we anticipated physicochemi-cal variability among enriched phosphopeptides.

Our results suggest that pY-MIP and pY-MIP-TiO2, per-form similarly to each other and differently from TiO2. The most pronounced differences relate to the frequency of charged peptides, peptide length and level of phos-phorylation. The pY-MIP features charge neutral binding sites where phosphate recognition occurs through com-plementary multidentate hydrogen bonding.32 Thus, un-like the acidophilic nature of TiO2, the pY-MIP binds to target phosphopeptides independent of their charge state32−a property reflected in our results by roughly two-fold underrepresentation of acidic phosphopeptides by pY-MIP and pY-MIP-TiO2 as compared to TiO2 and TiO2-pY-MIP.

Monophosphorylated and short phosphopeptides were 20-30% more frequent among the three pY-MIP-methods than in TiO2 (Figure 4, lower panel). This is in agreement with previous results showing that the pY-MIP binding energy is sufficient for interaction with short peptides.32 In a pool of tryptic peptides with various lengths and levels of phosphorylation, the pY-MIP bias towards short and less phosphorylated peptides becomes prominent, poten-tially hampering the MS detection of multiphosphory-lated peptides in the pY-MIP-based workflows.

Our detailed assessment of phosphopeptide recovery led to the identification of motifs recognized predomi-nantly by proline-directed and acidophilic kinases–a commonly reported result.50 The identification of baso-philic kinase substrates is known to be more difficult. This issue has been overcome by e.g. SAX-mediated de-pletion of acidic phosphopeptides.51 Yet, in our study, the higher representation of basic phosphopeptides by pY-MIP and pY-MIP-TiO2 (compared to TiO2 and TiO2-pY-MIP) was not associated with increased identification of basophilic kinase motifs. Subtractive sequence motif analysis revealed a slight bias towards pY-phosphopeptides containing hydrophobic residues near the phosphorylation site.

CONCLUSIONS

We present the first thorough assessment of pY-MIP as an enrichment tool in phosphoproteomics using limited amounts of biological sample. We found that sequential phosphopeptide enrichment by pY-MIP and TiO2 recovers pY- and also pT- (but not pS-) peptides in a more quanti-tative manner than by TiO2 alone. This pY-MIP-TiO2 pro-tocol preferably recovers short, monophosphorylated pep-tides, with less acidic and more basic residues as com-pared to TiO2. These findings demonstrate that pY-MIP-TiO2 has analytical advantages and is particularly applica-ble to phosphoproteome analysis aimed at studies of low abundant pY- and pT-peptides.

ASSOCIATED CONTENT

Supporting Information

-Supplementary figures, tables and experimental details (pdf) -List of detected phosphopeptides and peptides (Excel). The Supporting Information is available free of charge on the ACS Publications website.

AUTHOR INFORMATION

Corresponding Author

Corresponding Author * E-mail: jenseno@bmb.sdu.dk

Present Addresses

‡ Current address: Akershus University Hospital, Oslo

Author Contributions

L.B. and S.B.T performed all experiments and initial data analysis. C.W. and S.S. generated MIPs. B.S., A.R.W. and O.N.J. designed and supervised this study. All authors con-tributed to data interpretation and the preparation of the manuscript. / All authors have given approval to the final version of the manuscript. / # L.B. and S.B.T contributed equally.

ACKNOWLEDGMENT

This work was supported by the EU-funded Marie Curie ITN project PEPMIP (PITN-GA-2010-264699) and the VILLUM Center for Bioanalytical Sciences at SDU. LB thanks Staffan Nilsson for valuable suggestions, and Pernille S. Lassen for assistance in cell culture preparations.

REFERENCES

(1) Fischer, E. H.; Krebs, E. G. J. Biol. Chem. 1955, 216 (1),

121–122. (2) Krebs, E. G.; Kent, A. B.; Fischer, E. H. J. Biol. Chem.

1958, 231 (1), 73–83. (3) Garnak, M.; Reeves, H. C. Science 1979, 203 (4385), 1111–

1112. (4) Picton, C.; Woodgett, J.; Hemmings, B.; Cohen, P. FEBS

Lett. 1982, 150 (1), 191–196. (5) Venerando, A.; Cesaro, L.; Pinna, L. A. FEBS J. 2017. (6) Jensen, O. N. Nat Rev Mol Cell Biol 2006, 7 (6), 391–

403. (7) Tsiatsiani, L.; Heck, A. J. R. FEBS J. 2015, 282 (14), 2612–

2626. (8) Aebersold, R.; Mann, M. Nature 2016, 537 (7620), 347–

355. (9) Scanff, P.; Yvon, M.; Pelissier, J. P. J. Chromatogr. 1991,

539 (2), 425–432. (10) Stensballe, A.; Andersen, S.; Jensen, O. N. Proteomics

2001, 1 (2), 207–222. (11) Zhou, H.; Ye, M.; Dong, J.; Han, G.; Jiang, X.; Wu, R.;

Zou, H. J. Proteome Res. 2008, 7 (9), 3957–3967.

Page 9 of 11

ACS Paragon Plus Environment

Analytical Chemistry

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

10

(12) Ye, J.; Zhang, X.; Young, C.; Zhao, X.; Hao, Q.; Cheng, L.; Jensen, O. N. J Proteome Res 2010, 9 (7), 3561–3573.

(13) Sano, A.; Nakamura, H. Anal Sci 2004, 20 (3), 565–566. (14) Larsen, M. R.; Thingholm, T. E.; Jensen, O. N.;

Roepstorff, P.; Jorgensen, T. J. Mol Cell Proteomics 2005, 4 (7), 873–886.

(15) Thingholm, T. E.; Larsen, M. R. In Methods in Molecular Biology; Humana Press Inc., 2016; Vol. 1355, pp 135–146.

(16) Hunter, T. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 1998, 353 (1368), 583–605.

(17) Hunter, T. Cold Spring Harb. Perspect. Biol. 2014, 6 (5), a020644.

(18) Mann, M.; Ong, S. E.; Grønborg, M.; Steen, H.; Jensen, O. N.; Pandey, A. Trends Biotechnol. 2002, 20 (6), 261–268.

(19) Ullrich, A.; Schlessinger, J. Cell 1990, 61 (2), 203–212. (20) Fantl, W. J.; Johnson, D. E.; Williams, L. T. Annu Rev

Biochem 1993, 62, 453–481. (21) Schlessinger, J. Cell 2000, 103 (2), 211–225. (22) Di Palma, S.; Zoumaro-Djayoon, A.; Peng, M.; Post, H.;

Preisinger, C.; Munoz, J.; Heck, A. J. J Proteomics 2013, 91, 331–337.

(23) Zoumaro-Djayoon, A. D.; Heck, A. J.; Munoz, J. Methods 2012, 56 (2), 268–274.

(24) Ding, V. M.; Boersema, P. J.; Foong, L. Y.; Preisinger, C.; Koh, G.; Natarajan, S.; Lee, D. Y.; Boekhorst, J.; Snel, B.; Lemeer, S.; Heck, A. J.; Choo, A. PLoS One 2011, 6 (3), e17538.

(25) van der Mijn, J. C.; Labots, M.; Piersma, S. R.; Pham, T. V; Knol, J. C.; Broxterman, H. J.; Verheul, H. M.; Jimenez, C. R. J Proteomics 2015, 127 (Pt B), 259–263.

(26) Polyakov, M. V. Zhurnal Fizieskoj Khimii/Akad. SSSR 1931, 2, 799–805.

(27) Ye, L.; Mosbach, K. Chemistry of Materials. 2008, pp 859–868.

(28) Hoshino, Y.; Kodama, T.; Okahata, Y.; Shea, K. J. J. Am. Chem. Soc. 2008, 130 (46), 15242–15243.

(29) Vlatakis, G.; Andersson, L. I.; Muller, R.; Mosbach, K. Nature 1993, 361 (1978), 645–647.

(30) Kempe, M.; Glad, M.; Mosbach, K. J. Mol. Recognit. 1995, 8 (1–2), 35–39.

(31) Takeuchi, K.; Ito, F. Biol Pharm Bull 2011, 34 (12), 1774–1780.

(32) Emgenbroich, M.; Borrelli, C.; Shinde, S.; Lazraq, I.; Vilela, F.; Hall, A. J.; Oxelbark, J.; De Lorenzi, E.; Courtois, J.; Simanova, A.; Verhage, J.; Irgum, K.; Karim, K.; Sellergren, B. Chemistry (Easton). 2008, 14 (31), 9516–9529.

(33) Helling, S.; Shinde, S.; Brosseron, F.; Schnabel, A.; Muller, T.; Meyer, H. E.; Marcus, K.; Sellergren, B. Anal Chem 2011, 83 (5), 1862–1865.

(34) Chen, J.; Shinde, S.; Subedi, P.; Wierzbicka, C.; Sellergren, B.; Helling, S.; Marcus, K. J. Chromatogr. A 2016, 1471, 45–50.

(35) Chen, J.; Shinde, S.; Koch, M.-H.; Eisenacher, M.; Galozzi, S.; Lerari, T.; Barkovits, K.; Subedi, P.; Krüger, R.; Kuhlmann, K.; Sellergren, B.; Helling, S.; Marcus, K. Sci. Rep. 2015, 5, 11438.

(36) Thingholm, T. E.; Larsen, M. R.; Ingrell, C. R.; Kassem, M.; Jensen, O. N. J. Proteome Res. 2008, 7 (8), 3304–3313.

(37) Højrup, P. In Peptide Antibodies: Methods and Protocols; Houen, G., Ed.; Springer New York: New York, NY, 2015; pp 65–76.

(38) Palmisano, G.; Parker, B. L.; Engholm-Keller, K.; Lendal, S. E.; Kulej, K.; Schulz, M.; Schwämmle, V.; Graham, M. E.; Saxtorph, H.; Cordwell, S. J.; Larsen, M. R. Mol. Cell. Proteomics 2012, 11 (11), 1191–1202.

(39) Colaert, N.; Helsens, K.; Martens, L.; Vandekerckhove, J.; Gevaert, K. Nature methods. United States November 2009, pp 786–787.

(40) Schwartz, D.; Gygi, S. P. Nat Biotechnol 2005, 23 (11), 1391–1398.

(41) Chou, M. F.; Schwartz, D. Curr Protoc Bioinforma. 2011, Chapter 13, Unit 13.15-24.

(42) Bodenmiller, B.; Mueller, L. N.; Mueller, M.; Domon, B.; Aebersold, R. Nat Meth 2007, 4 (3), 231–237.

(43) Thingholm, T. E.; Jensen, O. N.; Larsen, M. R. Proteomics 2009, 9 (6), 1451–1468.

(44) Rosenqvist, H.; Ye, J.; Jensen, O. N. Methods Mol. Biol. 2011, 753, 183–213.

(45) Wierzbicka, C.; Torsetnes, S. B.; Jensen, O. N.; Shinde, S.; Sellergren, B. RSC Adv. 2017, 7 (28), 17154–17163.

(46) Kennelly, P. J.; Krebs, E. G. J Biol Chem 1991, 266 (24), 15555–15558.

(47) Pinna, L. A.; Ruzzene, M. Biochim. Biophys. Acta - Mol. Cell Res. 1996, 1314 (3), 191–225.

(48) Kreegipuu, A.; Blom, N.; Brunak, S.; Jarv, J. FEBS Lett. 1998, 430 (1–2), 45–50.

(49) Huang, K.-Y.; Horng, J.-C. Biochemistry 2015, 54 (40), 6186–6194.

(50) Matheron, L.; van den Toorn, H.; Heck, A. J.; Mohammed, S. Anal Chem 2014, 86 (16), 8312–8320.

(51) Dong, M.; Ye, M.; Cheng, K.; Song, C.; Pan, Y.; Wang, C.; Bian, Y.; Zou, H. J Proteome Res 2012, 11 (9), 4673–4681.

Page 10 of 11

ACS Paragon Plus Environment

Analytical Chemistry

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

11

TOC graphic.

Page 11 of 11

ACS Paragon Plus Environment

Analytical Chemistry

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960