Post-translational Modification: Phosphoproteins

Table of Contents

Letter from the Editor
Index of Experts
Q1: What sample pre-processing steps do you recommend and why?
Q2: What instrumentation or setup is optimal for the detection and study of phosphoproteins?
Q3: What approaches do you use to selectively isolate phosphopeptides?
Q4: How do you identify phosphoproteins and other phosphorylated molecules?
Q5: How do you determine the extent of phosphorylation of a purified protein?
Publications

Download the PDF version here

Letter from the Editor

For this latest edition in Genome Technology's technical reference guide series, we've asked a handful of experts to share their expertise on protein phosphorylation.

Phosphorylation is one of the most common, well-studied, and important — at least for organisms in our domain — of post-translational modifications. In textbook parlance, phosphorylation refers to the addition of a phosphate to an amino acid chain. Back in the mid-1950s, George Burnett and Eugene Kennedy offered an account of "the first direct evidence of the enzymatic phosphorylation of a protein substrate." One year later, Edmond Fischer and Edwin Krebs published their groundbreaking paper, which demonstrated the general regulatory importance of reversible phosphorylation. A brave new world of biochemistry was ushered in by the finding, for which the duo won the Nobel Prize in 1992.

Today, it's a given that most regulatory pathways are governed by the reversible phosporylation of proteins. The phenomenon forms the basis of signaling networks, and can influence a host of cellular activities — sub-cellular localization, protein-protein interactions, and many others. Abnormal phosphorylation is recognized as being at the root of many human diseases, so proteins and their phosphates are of immense interest to basic and translational scientists alike.

In this guide, you'll find lab-tested tips on phosphoprotein sample processing and optimal instrumentation, along with the best techniques for isolating phosphopeptides and identifying sites of phosphorylation.

Finally, many thanks are due to Pawel Ciborowski, director of the proteomics program at the University of Nebraska Medical Center, for his key role in formulating the theme of this guide.

— Jennifer Crebs

Index of Experts

Catherine Fenselau
University of Maryland

Sergio de la Fuente van Bentem
University of Vienna, Austria

Benedikt Kessler
University of Oxford

Shao-En Ong
The Broad Institute of MIT and Harvard

Akhilesh Pandey
Johns Hopkins University

Scott C. Peck
University of Missouri at Columbia

Waltraud Schulze
Max-Planck Institute of Molecular Plant Physiology

Albert Sickmann
Rudolf-Virchow-Center University of Würzburg

Q1: What sample preprocessing steps do you recommend and why?

To reduce the complexity of the sample that you analyze, it is recommended to pre-fractionate your protein extract. For instance, you can focus on a particular organelle and purify it before looking at the phosphoprotein content. The lower complexity of your sample increases the chance of identifying the lower abundant phosphoproteins, like most signaling components. If you want to focus on a particular protein or a subgroup of the phosphoproteome, you can use protein-specific or phosphorylation site-specific antibodies to immunoprecipitate protein(s) from your sample.

— Sergio de la Fuente van Bentem

Some of the most popular methods [include] one- and two-dimensional gel electrophoresis; immunoprecipitation using anti-phospho-serine, -threoine or -tyrosine antibodies; and specific enrichment for phosphopeptides.

Each approach has its advantages and disadvantages. The gel-based separation methods can provide an excellent degree of protein separation, but have limitations in detection by mass spectrometry due to inefficient extraction of protein/peptide material from gels. Gel-free protein enrichment steps offer increased sensitivity for detection, but are often accompanied by a large amount of non-specific material. Proteolytic digestion prior to enrichment of phosphorylated material allows the use of peptide-based separation and isolation methods. However, the complexity of the mixture is far greater than compared to the initial protein material, which might cause difficulties in identification of phosphorylated peptide material.

— Benedikt Kessler

A suitable lysis buffer should contain adequate amounts of kinase and phosphatase inhibitors; using chilled buffers and processing the sample on ice should keep in vitro phosphorylation and dephosphorylation to a minimum. Sample fractionation, like subcellular fractionation or affinity purification of sub-proteomes, is a useful means of targeting specific proteins of interest.

— Shao-En Ong

As with all mass spectrometric analysis, sample preparation is key. Except that it is even more relevant in the case of phosphorylation given the low stoichiometry of phosphorylation, the low abundance of signaling molecules, and the analytical issues that make it difficult to visualize phosphopeptides. First of all, phosphopeptides are protected from phosphatases using specific inhibitors. Our emphasis is on enriching the protein(s) or peptides of interest — this is done using phosphospecific antibodies (we generally purify tyrosine phosphorylated proteins using a mixture of antiphosphotyrosine antibodies), phosphopeptide enrichment (these days, we are using TiO2 quite successfully) or protein-specific antibodies. If this is not feasible and the analysis is a global one, we like to either analyze only a subproteome at a time (e.g. nuclear fraction) or fractionate the cell lysates using chromatography (we use Agilent's recently introduced macroporous reversed phase C18 column) followed by enrichment of phosphopeptides. One can also first fractionate peptides by strong cation exchange followed by phosphopeptide enrichment.

— Akhilesh Pandey

Regardless of whether we are investigating individual proteins by immunoprecipitation or performing large-scale analyses by LC-MS/MS, sub-cellular fractionation is an essential step for both sensitivity and depth of coverage. If total samples are extremely complex, we often use bulk-liquid denaturing isoelectric focusing (e.g. using Bio-Rad's Rotofor) to prefractionate 20 mg to 100 mg of protein by pI prior to phosphopeptide enrichment procedures or running 2D gels. Finally, we pay a lot of attention to sample preparation. We use either methanolchloroform extraction or phenol back-extraction followed by precipitation in ammonium acetate and methanol to remove contaminants from our protein preparations. Cleaner samples separate more reliably during subsequent chromatographic steps.

— Scott Peck

Cell lysis combined with phosphatase inhibitors.

— Albert Sickmann

We are interested in membrane proteins, therefore our first pre-processing step is an enrichment of membrane proteins over a two-phase system. Secondly, we like to reduce the complexity of the protein mixture by either separating it over a 1D-SDS gel and then process individual gel slices or by doing an off-line SCX fractionation prior to phosphopeptide enrichment.

— Waltraud Schulze

Q2: What instrumentation or setup is optimal for the detection and study of phosphoproteins?

In our experience, the Qq-TOF (QSTAR by Applied Biosystems) provides molecular ions and sequence ions that retain phosphor groups when using electrospray or nanospray. MALDI on our Shimadzu Axima is also quite satisfactory.

— Catherine Fenselau

If you want to see the phosphoproteins, you can separate them by two-dimensional gel electrophoresis. An advantage of this is that you can match the molecular weight and other features with the protein of which you later identify the peptides by mass spectrometry or by using antibody detection. This is important since a peptide sequence of the protein does not give you information about the molecular weight of the protein from which it is derived.

— Sergio de la Fuente van Bentem

Generally, any mass spectrometer with the capability of performing MS/MS can be used for the determination of phosphorylation sites in peptides. A typical setup consists of a nano-LC-MS/MS system, including a liquid chromatography step, usually using a reversed-phase column. This is then directly coupled to a tandem mass spectrometer, such as an ion trap, an orthogonal quadrupole time-of-flight instrument (QTOF, QSTAR), or a triple quadrupole instrument. MALDI instruments capable of performing MS/MS (MALDI-TOF/TOF) have also been successfully used for this task, especially in combination with liquid chromatography (LC-MALDI).

— Benedikt Kessler

I think all detection methodologies are partially overlapping and complementary. For the comprehensive analysis of protein phosphorylation, multiple phosphopeptide enrichment tools and versatile mass spectrometry analytical capabilities such as CID and ETD fragmentation methods are required. Using one technology or instrument, it is difficult to obtain comprehensive phosphorylation data — but this applies to proteomic analysis in general. When we use CID as the method of dissociating peptides in an ion trap mass spectrometer, we regularly use MS3 mode because of the predominant neutral loss observed with serine/threonine phosphorylated peptides. Using a newer method for fragmenting peptides called electron transfer dissociation, we have been able to obtain data on serine/threonine phosphorylation sites that we did not get by using CID in an ion trap. We also use a QTOF mass spectrometer for our phosphorylation analysis. Also, replicate analysis allows one to go deeper — we have found only a 70 percent overlap in the phosphopeptides identified in two runs on the same instrument using the same parameters.

— Akhilesh Pandey

Our current setup for phosphopeptide analysis is an LTQ ion trap instrument in which we do MS3 fragmentation of those ions that display a neutral loss. Recently, we upgraded to an LTQ-Orbitrap, which has a higher sensitivity and better mass accuracy at the same scan speed and thus will give more confidence in the data analysis. In addition, ion trap instruments with the option for ETD fragmentation are very promising approaches to unambiguously identify the sites of phosphorylation by an additional different fragmentation method.

— Waltraud Schulze

Q3: What approaches do you use to selectively isolate phosphopeptides?

For a highly specific isolation procedure we use an adapted protocol of immobilized metal affinity chromatography, which has been described by Ficarro and others (Ficarro et al., 2002). Esterification of the peptides prior to IMAC can result to a 90 percent pure phosphopeptide fraction. A difficulty is to reach a high level of esterification, but you can focus on this 90 percent pure set of phosphopeptides that have been esterified completely. Alternatively, you can immunoprecipitate a subset of phosphopeptides from your complex sample by specific antibodies.

— Sergio de la Fuente van Bentem

A common technique to enrich for phosphopeptides is based on IMAC, [which] relies on the affinity of phosphate groups for metal ions, such as Fe3+ or Ga3+, bound to tethered chelating compounds bound to a solid support. A complication of this approach is that aspartate and glutamate side chains also exert affinity to IMAC, and therefore get enriched at the same time. To overcome this problem, more recent protocols propose to convert the carboxylic acid groups of peptides to methylesters.

Alternatively, a variety of chemical tagging and capturing methods were developed and successfully used to enrich and isolate phosphopeptides. For instance, a base-catalyzed beta-elimination of the phosphate group from pSer and pThr followed by a subsequent Michael addition of ethaneditiol allows [one] to add a biotin tag to phosphopepties, and hence their isolation via a streptavidine column. This approach, however, is not suitable to isolate pTyr-containing peptides, and other chemical methods using a solid support were developed.

— Benedikt Kessler

I favor phosphotyrosine antibodies in the enrichment of proteins and peptides phosphorylated on tyrosine. There are several commercial antibodies available and they appear to work quite well. For general phosphopeptide enrichment, I like using the titanium dioxide based supports for LC-MS analysis. There has been considerable interest in TiO2 columns in the past few years and they have been successfully applied in several labs.

— Shao-En Ong

We are using TiO2 for enrichment of phosphopeptides these days. In situations where we are trying to identify proteins that are inducibly tyrosine phosphorylated, we generally couple the SILAC method (stable isotope labeling with amino acids in cell culture) to affinity enrichment using antiphosphotyrosine antibodies. In this case, we are actually only enriching phosphoproteins and not phosphopeptides; the SILAC data helps ascertain which proteins are inducibly tyrosine phosphorylated, although we do not generally identify the actual phosphotyrosine containing peptides themselves. We can also selectively isolate/analyze the phosphopeptides in the mass spectrometer — we generally do this by neutral loss scanning (for phospho-serine/-threonine peptides) or scanning for the immonium ion of phosphotyrosine (for phosphotyrosine peptides). Another key point is to use several different proteases on the same sample to maximize identification of phosphopeptides.

— Akhilesh Pandey

We use strong cation exchange with a volatile ammonium formate buffer system as described in Gruhler et al. (2005). The advantage is that large numbers of fractions/samples can be rapidly dried, eliminating time-consuming desalting steps, which also may lead to sample loss. Afterwards, we use commercial IMAC columns to specifically enrich for phosphopeptides. We compared columns from a number of companies, and most seem to work equally well. One comment is that sample amount should be a consideration. All IMAC columns have a certain level of irreversible binding; so if the input sample is limiting, one should either use commercial products with smaller resin volumes or self-pack individual microcolumns. With our sample preparations, we obtain 80 percent to 90 percent pure phosphopeptides. Unlike some reports, we do not see non-specific contamination with peptides containing poly-acidic residues. As with many aspects of biochemistry, variation depending on the source tissue or sample handling may occur.

— Scott Peck

IMAC, TiO2, ZrO2, and strong cation exchange.

— Albert Sickmann

In our institute, we have developed procedures for phosphoprotein enrichment using aluminum hydroxide. For the subsequent enrichment of phosphopeptides, we are using IMAC resin or titanium dioxide.

— Waltraud Schulze

Q4: How do you identify phosphoproteins and other phosphorylated molecules?

One conservative approach is to recognize peptide pairs that differ by 80 Da from pairs of isoform spots in 2D gels.

— Catherine Fenselau

We identify phosphoproteins by isolating phosphopeptides from complex trypsin-digested peptide mixtures. The peptide sequences are determined by nano-liquid chromatography-MS/MS/MS on an LTQ linear ion trap mass spectrometer, which is equipped with a nanoelectrospray ion source. This gives you the nature of the phosphoprotein as well as the phosphorylated residues. Moreover, this technique is ideal to study the phosphoproteome and signaling networks on a large scale.

— Sergio de la Fuente van Bentem

The classical biochemical way to detect phosphoproteins is to use the radioactive isotope 32P or 33P, usually incorporated into an ATP analogue, such as g-[32P]-ATP. Alternatively, there are a number of antibodies specific for pSer, pThr, or pTyr that are currently available, allowing immunoprecipitation combined with western blot techniques to detect phosphorylated protein species. Unless site-specific antiphospho-antibodies are used, these methods normally do not allow to map or quantify phosphorylation sites present in proteins. Over the last couple of years, mass spectrometry has evolved to be the method of choice for detection, mapping, and quantitation of phosphorylation events in biomolecules. Phosphorylated protein material is digested with trypsin (or other proteolytic enzymes), followed by an enrichment step as described above, and analyzed by LC-MS/MS. Peptide species containing phosphate groups contain a mass tag (+79.97 Da). The analysis of MS/MS spectra allows to map the phosphorylation site to a single amino acid residue (see Steen et al., 2006).

— Benedikt Kessler

The standard proteomics approach is to use a phosphospecific enrichment step at the protein and/or peptide level to improve the chances of identifying phosphorylated species. This is followed by MS analysis specific to phosphorylation analysis. My approach is to use nanoscale LC-MS with a data-dependent acquisition method that first acquires a survey MS spectrum, then several tandem MS/MS spectra. If a characteristic neutral loss of a phosphate is detected in MS/MS, a further MS3 spectrum of that fragment is acquired and is typically more useful for peptide identification. Although proteomics focuses on large-scale analysis of proteins, one cannot dismiss the importance of complementary experimental approaches to identify phosphorylated proteins like western blotting and immunofluorescence with phospho-specific antibodies, or 32P labeling.

— Shao-En Ong

Our lab uses conventional molecular biology methods in addition to mass spectrometry to study signaling pathways. Therefore, we often confirm our findings from mass spectrometry by using phosphospecific antibodies — these antibodies can be generic (e.g. antiphosphotyrosine) or specific (e.g. anti-phosphoEGFR). Most of the time, though, we end up enriching for phosphoproteins, doing a digest, and identifying them by mass spectrometry. We also use 32P labeling of cells when necessary.

— Akhilesh Pandey

I think the answer depends on whether one wants to study an individual protein under different conditions or perform large-scale, unbiased screens. For individual proteins, we have had good success using monospecific peptide antibodies for immunoprecipitation followed by in-gel detection with the phosphospecific fluorescent stain, ProQ Diamond. For large-scale screens, we've been successful with both 2D gel electrophoresis of proteins and phosphopeptide enrichment by multidimensional chromatography followed by LCMS/MS. For the 2D gels, in vivo labeling of phosphoproteins with 33P-orthophosphate is very sensitive, but the low levels of signaling proteins coupled with generally low stoichiometry of phosphorylation means we rarely can align a radioactive spot with a stainable image for sequencing without using denaturing Rotofor to pre-enrich proteins by pI. This step greatly increases the amount of sample that can be loaded within the target region of the gel. For similar reasons, we only use large-format, single pI unit IPG strips when trying to align radioactive images for sequencing to maximize resolution. The problems with 2D gels, of course, are the limitations on the numbers and types of proteins that are resolved. In general, we are moving almost exclusively to using phosphopeptide pre-enrichment followed by LC-MS/MS in conjunction with iTRAQ labeling for relative quantification. The added benefit of this approach is immediate knowledge of which residues are undergoing differential phosphorylation during a response.

— Scott Peck

The first step in phosphopeptide identification is by an automated database search (we use Mascot), but we manually verify most of the phosphopeptides using in-house software. In the manual verification process, we check for neutral loss of 98 for serine and threonine phosphorylation and we check if the actual fragmentation patterns are consistent with the suggested sequence and/or phosphorylation site returned from the database search. Sometimes, false identifications can happen due to wrong charge state, assignment of the parent ions, or due to neutra loss-like behavior of oxidized methionine in the sequence.

— Waltraud Schulze

Q5: How do you determine the extent of phosphorylation of a purified protein?

An effective method to determine occupancy rates is an ongoing research effort in our lab.

— Catherine Fenselau

For some proteins you can separate the phosphorylated form from the non-phosphorylated one by running them on a gel. By calculating the ratio between the two forms you can determine the fraction that is phosphorylated. A more sophisticated approach is to focus on the phosphopeptides of the protein and to compare the amount of the phosphorylated form with the same, but non-phosphorylated, peptide by mass spectrometry. This gives you the opportunity to measure changes in the extent of phosphorylation on a particular phosphosite during changing conditions, for instance stress application. A few labeling techniques have recently been developed for differential labeling of peptides, such as SILAC (stable isotope labeling by amino acids in cell culture) that has been described by Ong and others (Mol Cell Proteomics, 2002). This technique allows you to determine the phosphorylation status of each site of a particular protein and the changes during a stress treatment.

— Sergio de la Fuente van Bentem

This is a difficult one. I think one possible approach is to separate the purified protein using two-dimensional gel electrophoresis and analyze the resolving phosphoforms. The staining pattern will reveal the most abundant phosphoform. However, bear in mind that there could still be multiple phosphoforms within the same resolved gel spot as the phosphoforms could have the same number of phosphorylated residues but in different positions. Another useful approach is to analyze the protein mixture by top-down, intact protein mass spectrometry with a high-performance instrument. Several charge envelopes should be resolvable corresponding to the distinct phosphoproteins, separated in mass by the number of phosphates. The relative abundance of the charge envelopes could give an estimate of the extent of phosphorylation.

— Shao-En Ong

Most of our work thus far has been qualitative or has involved relative quantification. For relative quantification, we have used the SILAC method successfully. We are also using 18O-labeling, ICAT, and iTRAQ for relative quantification. Determining the extent of phosphorylation for a single protein is better done by biochemical methods mainly because you are analyzing the entire proteins and can also get absolute quantification-type of data. One can determine the extent of phosphorylation of a given site in a protein and this can be done in a number of ways including SILAC or AQUA (absolute quantification of abundance) methods, which use an internal standard phosphopeptide.

— Akhilesh Pandey

To date, we have only used 2D immunoblot analysis to address this type of question, mainly because most of the proteins we are studying do not have commercially available antibodies for ELISAs. The problem, of course, is that any particular phosphorylated state associated with a response is possibly a mixture of phosphorylated residues on the protein, and we can only measure the level of phosphorylation on the sites we know about. We will be able to design experiments properly only when we achieve sufficiently deep sequencing of the phosphoproteome under many different treatment conditions.

— Scott Peck

This is a challenging and very important question. Actually, we do not work with purified proteins, but try to estimate changes in phosphorylation states of proteins in complex mixtures. For that, we use protein correlation profiling as a label-free quantitative tool and alternatively we work with stable isotope labeling. In addition, in our institute we have developed methods for quantitative analysis of phosphorylation levels based on synthetic peptides that are subjected to ‘physiological' phosphorylation by kinases in native protein extracts. The degree of phosphorylation is then quantitatively monitored by MRM-experiments using a TSQ mass spectrometer. However, at this point it remains an in vitro assay.

— Waltraud Schulze

Publications

Amanchy R, et al. (2005) Phosphoproteome analysis of HeLa cells using stable isotope labeling with amino acids in cell culture (SILAC). J Proteome Research. 4, 1661-1671.

Amanchy R, et al. (2005) Stable isotope labeling with amino acids in cell culture (SILAC) for studying dynamics of protein abundance and post-translational modifications. Science's STKE. l2.

Gruhler A, et al. (2005) Quantitative phosphoproteomics applied to the yeast pheromone signaling pathway. Mol Cell Proteomics. 4(3): 310-27.

Chang EJ, et al. (2004) Analysis of protein phosphorylation by hypothesis-driven multiple-stage mass spectrometry. Anal Chem. 76(15): 4472-83.

Ficarro SB, et al. (2002) Phosphoproteome analysis by mass spectrometry and its application to Saccharomyces cerevisiae. Nat Biotechnol. 20(3): 301-5.

Goshe MB. (2006) Characterizing phosphoproteins and phosphoproteomes using mass spectrometry. Briefings in Functional Genomics and Proteomics. 4(4):363-376.

Mann M, Jensen ON. (2003) Proteomic analysis of post-translational modifications. Nat Biotechnol. 21(3):255-61.

Ong SE, et al. (2002) Stable isotope labeling by amino acids in cell culture, SILAC, as a simple and accurate approach to expression proteomics. Mol Cell Proteomics. 1(5):376-86.

Steen H, et al. (2006) Phosphorylation analysis by mass spectrometry. Mol Cell Proteomics. 5:172-81.

Steen H, Jebanathirajah JA, Springer M, Kirschner MW. (2005) Stable isotope-free relative and absolute quantitation of protein phosphorylation stoichiometry by MS. PNAS. 102(11):3948-53.