ProtocolPhosphorylation Events

Genome to Kinome: Species-Specific Peptide Arrays for Kinome Analysis

See allHide authors and affiliations

Science Signaling  20 Jan 2009:
Vol. 2, Issue 54, pp. pl1
DOI: 10.1126/scisignal.254pl1

Abstract

Tools for conducting high-throughput kinome analysis do not exist for many species. For example, two commonly used techniques for monitoring phosphorylation events are phosphorylation-specific antibodies and peptide arrays. The majority of phosphorylation-specific antibodies are for human or mouse targets, and the construction of peptide arrays relies on information from phosphorylation databases, which are similarly biased toward human and mouse data. This is a substantial obstacle because many species other than mouse represent important biological models. On the basis of the observation that phosphorylation events are often conserved across species with respect to their relative positioning within proteins and their biological function, we demonstrate that it is possible to predict the sequence contexts of phosphorylation events in other species for the production of peptide arrays for kinome analysis. Through this approach, genomic information can be rapidly used to create inexpensive, customizable, species-specific peptide arrays for high-throughput kinome analysis. We anticipate that these arrays will be valuable for investigating the conservation of biological responses across species, validating animal models of disease, and translating research to clinical applications.

Introduction

Phosphorylation represents the pivotal mechanism for regulation of cellular processes, and kinases are one of the most biologically important classes of enzymes (1). The regulatory roles of kinases in cellular pathways and disease, as well as their conserved catalytic cleft, make them attractive targets for drug therapy (2). The therapeutic and biological importance of kinases has prompted the development of novel strategies for quantification of their activity. For example, through array technology, it is possible to address global kinase activity of a given species—that is, its kinome. A challenge to this approach is to present appropriate kinase substrates on the array. Although the physiological substrates of most kinases are proteins, proteins are problematic for the construction of stable arrays. However, because the specificity of many kinases is dictated by residues surrounding the site of phosphorylation, a logical alternative is to use peptides representing these sequences as substrates (3, 4). Peptides modeled on the site of phosphorylation can be excellent kinase substrates, with initial rates of the catalyzed reaction (Vmax) and Michaelis-Menten constants (Km) close to those of the natural substrate (5, 6). Peptides are easily produced, relatively inexpensive, chemically stable, and highly amenable to array technology (7, 8). The use of peptide arrays to quantify global kinase activity has been demonstrated (912). To date, peptide arrays created for kinome analysis have been based on phosphorylation events characterized for a particular species and used for investigation of that same species.

Publicly available online databases such as Phosphosite (www.phosphosite.org) contain information on manually curated and literature-based serine, threonine, and tyrosine phosphorylation events. Search results for a specific protein return short peptide sequences corresponding to characterized phosphorylation sites. Where available, information on the corresponding kinase and descriptions of the biological function of the modification are also included. Because many proteins undergo phosphorylation at distinct sites to regulate discrete aspects of their function, information pertaining to each of these phosphorylation events is present within the database. On the array, each phosphorylation event can be presented as a separate peptide, offering a unique opportunity to comprehensively describe dynamic patterns of phosphorylation of a particular protein in response to a stimulus. A limitation of most of these databases is that the information is almost exclusively for human and mouse proteins and kinases.

Creation of a peptide array for kinome analysis depends on information emerging from previous phosphoproteome investigations. This represents a limiting obstacle for species whose phosphoproteomes have yet to be defined. The inability to conduct kinome analysis in these species is important for several reasons. Although mice offer the opportunity for generation of mutants to study kinase function and the consequences of kinase ablation, and although the recent characterization of the mouse kinome has been heralded for its ability to “enhance the exploration of the roles of all kinases in mouse models of human diseases” (13), the value of the mouse as a model for human disease has many limitations (14). Mice are the species of choice for biomedical investigations because of the availability, cost, and ease of generating genetically defined strains, but these selection criteria do not ensure effective disease models. An increasing number of studies demonstrate divergent mechanisms, outcomes, and pathologies between mice and humans (15) and emphasize the challenges of translating mouse research into clinical applications (16). This has prompted many researchers to invest greater resources into the development of other animal models that more accurately reproduce relevant aspects of human physiology (17). From such investigations, it is clear that specific large animals, such as cows, pigs, and sheep, often serve as more effective models of particular human diseases (1417). For example, the pig is a valuable model for whooping cough (18).

Sites of protein phosphorylation and their subsequent biological consequences are often conserved; therefore, it should be possible to predict the sequence contexts of phosphorylation events in proteins of other species on the basis of genomic information. To demonstrate the feasibility of this approach, we selected a set of phosphorylation events that represent the major signal transduction pathways, with an emphasis on pathways and processes relating to innate immunity. We then created a bovine peptide array for analysis of the bovine kinome, and used the results of application of this array to analyze the dynamic patterns of phosphorylation associated with exposure of bovine monocytes to lipopolysaccharide (LPS). Below, we describe in detail the procedure for this type of analysis, which can be used to create peptide arrays for the kinome analysis of other species or to explore phosphorylation events associated with other physiological processes. A brief outline of the construction and application of the bovine peptide arrays (Fig. 1) is followed by step-by-step instructions.

Fig. 1

Recommended steps for the development and application of a custom, species-specific peptide array for kinome analysis.

To determine the degree of amino acid sequence conservation surrounding human and bovine phosphorylation sites, we searched nearly 1000 peptides representing human phosphorylation sites of interest against the NCBI-NR protein database (ftp://ftp.ncbi.nih.gov/blast/db) with the Blastp program to generate orthologous bovine peptides. Blastp results revealed that about half of the bovine sequences had 100% identity to the query sequences and that the majority of the remaining peptides had limited sequence divergence (Table 1). A comparison of the protein descriptions for the query and hit sequences confirmed that they referred to orthologous proteins (19). These results indicate that phosphorylation sites between human and bovine orthologous proteins are not absolutely conserved, which suggests that a human array would be of limited value for bovine kinome analysis. Peptides representing 298 proposed bovine phosphorylation sites, two negative control peptides, and seven positive control peptides were printed to a 307-peptide array for kinome analysis. Peptides 15 amino acids in length, with the phosphoacceptor residue flanked by seven residues on either side, were printed in blocks such that in the array each peptide was presented in triplicate to facilitate assessment of technical reproducibility. The proteins and phosphorylation events represented on the array, as well as the specific peptide sequences, are included in the Supporting Online Materials (tables SA1 and SA2).

Table 1 Analysis of bovine kinase target peptides selected from queries of human peptides.
View this table:

We analyzed the phosphorylation events associated with exposure of bovine monocytes to activation of Toll-like receptor 4 (TLR4) by bacterial lipopolysaccharide (LPS). The TLR signaling pathway is well characterized (20) and a similar investigation has been reported with human cells and a human peptide array (9), thus providing an established framework to validate the bovine peptide arrays. The dynamic patterns of phosphorylation were evaluated in LPS-treated cells relative to media-treated control cells (Fig. 2). Individual peptide signals were reproducible with one-sample t tests returning confidence levels of P < 0.05 for about 75% of the peptides. Because several peptides representing phosphorylation targets attributed to TLR signaling were represented on the array, the phosphorylation of these peptides was consistent with activation of TLR signaling reported in studies with cells from other species. For example, phosphorylation was detected at several key phosphorylation sites after LPS stimulation: Thr209 of interleukin-1 receptor-associated kinase 1 (IRAK1) (21), Ser337 of nuclear factor κB (NF-κB) p105 isoform (22), Ser870 of NF-κB p100 (23), Ser276 of NF-κB p65 (24), and Ser396 of interferon regulatory factor 3 (IRF-3) (25) had high levels of phosphorylation after LPS stimulation. Residue numbers refer to the positioning of the phosphoacceptor residue within the bovine protein. Selected data from the peptide arrays of LPS-stimulated bovine monocytes are presented in Table 2.

Fig. 2

Phosphorylation of substrate targets on peptide arrays by cellular lysates from bovine monocytes stimulated with (A) LPS or (B) media. Purified CD14+ bovine monocytes (15 × 106) were cultured overnight, then stimulated with LPS (100 ng/ml) or cultured in media for 4 hours at 37°C. Monocyte lysates were prepared and incubated with peptide arrays in the presence of [γ-32P]ATP. After original slide images were acquired by a scanner, normalized, and corrected for background noise, then signal strength for each spot was calculated. The average signal for each peptide was calculated from 18 replicates, and this value was used to generate pseudo-images of peptide arrays for each condition assayed.

Table 2 Differential phosphorylation of select peptides in LPS-stimulated monocytes.
View this table:

Materials

Note: All reagents are from Sigma Aldrich unless otherwise indicated.

Adenosine triphosphate (ATP)

Aim V medium (Gibco, Invitrogen)

Antibody to bovine CD14 (VMRD)

Aprotinin

Bovine serum albumin (BSA)

Brij-35

CD14 microbeads (Miltenyi Biotec)

Cover slides (Fisher Scientific)

Culture cluster 6-well plate (Corning, Corning, NY)

EDTA

EGTA

Fetal bovine serum (FBS) (Gibco)

NaCl

NaF

Na3VO4

[γ-32P]ATP (2 mCi/ml) (Perkin Elmer)

Glycerol

Lipopolysaccharide (LPS)

Leupeptin

MgCl2

Microcentrifuge tubes (Eppendorf)

Microtiter 96-well filtration plates (Millipore, Bedford, MA)

Peptide and arrays [JPT Peptide Technologies (http://www.jpt.com)]

Phenyl methyl sulfonyl fluoride (PMSF)

Screwcap 50-ml centrifuge tube (Corning, Corning, NY)

Sodium pyrophosphate

Tris-HCl (pH 7.5)

Triton X-100

Equipment

ArrayVision software (GE Healthcare)

Barnstead International NANOpure II 4 Module water filtration system

Flow cytometer (BD FACSCalibur)

GeneSpring software (Agilent Technologies, Santa Clara, CA)

Noncontact printer [Nanoplotter (GESIM, Großerkmannsdorf, Germany) equipped with eight piezoelectric NanoTips (GESIM)]

Phosphoimager screen (Kodak)

ImageQuant TL v2005 software (Amersham, GE Healthcare)

Typhoon scanner (GE Healthcare)

Recipes

Recipe 1: Monocyte Culture Medium

Add heat-inactivated FBS to a final volume of 10% to Aim V medium.

Recipe 2: LPS Solution

Prepare at 100 μg/ml in distilled H2O. Store at –20°C.

Recipe 3: Lysis Buffer

Aprotinin (1 μg/ml)

EDTA (1 mM)

EGTA (1 mM)

NaCl (150 mM)

NaF (1 mM)

Na3VO4 (1 mM)

Leupeptin (1 μg/ml)

PMSF (1 mM)

Tris-HCl (pH 7.5, 20 mM)

Triton (1%)

Sodium pyrophosphate (2.5 mM)

Prepare 70 μl per peptide array.

Recipe 4: Activation Mix

ATP (50 μM)

Brij-35 (0.05% v/v)

BSA (0.25 mg/ml)

[γ-32P]ATP (2 mCi/ml)

Glycerol (50%)

MgCl2 (60 mM)

Prepare 10 μl per peptide array.

Recipe 5: Tris-Buffered Saline (TBS)

NaCl (0.2 M)

Tris (0.1 M)

Prepare 45 ml per peptide array in distilled water.

Recipe 6: Primary Wash Solution

Add Triton X-100 to a final concentration of 1% in TBS (Recipe 5). Prepare 45 ml per peptide array.

Recipe 7: Secondary Wash Solution

NaCl (2 M)

Triton X-100 (1%)

Prepare 90 ml per peptide array in distilled deionized H2O (distilled water passed through a Barnstead International NANOpure II 4 Module filtration system).

Instructions

All peptides were synthesized and printed by JPT Peptide Technologies (http://www.jpt.com). Thus, the detailed instructions include information for identifying the peptides of interest (via bioinformatics resources) and then using the purchased peptide arrays. The peptides were spotted in a grid pattern on a block (Fig. 3). Each block contains 298 bovine test peptides, two negative control peptides, and seven positive control proteins [histones 1 through 4, bovine myelin basic protein (MBP), and α/β casein]. Each array contains three replicate blocks in the same configuration.

Fig. 3

Mapping of the bovine peptide array. One block comprises 300 peptides (including the negative controls) and the seven positive controls (histones 1 to 4, bovine MBP, and α/β casein; black spots). An array holds three replicate blocks of the same configuration. The positive controls are printed around the borders of each block for visualization purposes.

Each positive control is a full-length protein and is known to be phosphorylated in cells in the presence of ATP. We generally do not include the signals from these controls in the peptide array analysis; instead, these proteins are mainly included to aid in visualization and grid assignment of the blocks. In addition, to determine intraexperimental variability in substrate phosphorylation, we print each block of 300 peptides in triplicate. The final physical dimensions of the arrays are 19.5 mm by 19.5 mm, with each peptide spot having a diameter of ~350 μm and separated by 750 μm.

Peptide Selection and Design

1. Identify phosphorylation events regulating cellular processes or proteins of interest from the literature. The number of phosphorylation events identified should match the number of peptides planned for each block of the array.

Note: Although the arrays are customizable for specific events or processes, we recommend devoting a subset of the peptides to phosphorylation events representing a broader spectrum of cellular events to permit greater opportunity for novel discovery.

2. From publicly available databases, such as Phosphosite (http://www.phosphosite.org), obtain the sequences surrounding the phosphorylation sites of interest for human proteins.

3. Compare the selected human peptides against the NCBI-NR database, using Blastp (version 2.2.13), to obtain specific consensus sequences from another species. In the present example, the 15–amino acid peptides were queried against the NCBI NR protein database (ftp://ftp.ncbi.nih.gov/blast/db/nr.00.tar.gz) using the default Blastp settings (BLOSUM562 matrix) with the organism field set to Bos taurus.

Note: For the specific application described here, Blastp was set to search bovine protein records and retrieve short exact matches for the list of human peptides. This approach is applicable to species for which annotated genomic material is available.

4. Compare the protein descriptions for the query and hit sequences to verify that they refer to orthologous proteins.

Using the Peptide Arrays for Kinome Analysis

We analyzed the phosphorylation events associated with exposure of magnetic-activated cell sorter (MACS)–purified bovine monocytes stimulated with either media or LPS (26) to validate the arrays. The isolation and assessment of the purity of the monocytes were performed according to standard procedures. Blood was collected by venipuncture from mature cattle, using 0.3% EDTA as an anticoagulant. Monocytes were isolated by the MACS purification method with CD14 microbeads (26). Monocytes were plated in Aim V medium supplemented with 1% heat-inactivated FBS. Purity of monocytes was assessed by flow cytometry after cell labeling with an antibody to bovine CD14.

These cells and these stimuli were chosen because the signaling pathways activated by TLR4 have been characterized by traditional methodologies and provide a template for the validation of initial array results. The kinome analysis described can be applied to the appropriate population of defined cells under the desired treatment conditions. General instructions are provided, along with specific details for using bovine monocytes.

1. Culture the cells for the desired time before stimulation.

Note: The purified bovine monocytes were cultured for 19 hours before stimulation.

2. Stimulate the cells with the desired conditions.

Note: Purified monocytes (15 × 106) were stimulated with Monocyte Culture Medium (Recipe 1) or Monocyte Culture Medium supplemented with LPS (100 ng/ml) (Recipe 2) for 4 hours at 37°C.

3. Pellet 15 × 106 cells in culture medium by centrifugation at 300g for 8 min.

Note: To ensure reproducibility across experiments, it is essential to use equal numbers of cells in each experiment.

4. Lyse the cell pellets (15 × 106 cells) with 100 μl of Lysis Buffer (Recipe 3), incubate on ice for 10 min, and then spin in a microcentrifuge for 10 min at 4°C.

Note: It is essential that the pellet is fully resuspended to ensure complete cell lysis, as this will influence the intensity of the signal.

5. Determine the protein concentration of the cellular lysates through Bradford assay or similar techniques.

Note: To ensure accuracy, an equivalent amount of protein must be applied to each peptide microarray.

6. Mix a 70-μl aliquot of the supernatant with 10 μl of Activation Mix (Recipe 4), and apply all 80 μl of this mixture to the peptide microarray as a bubble-free drop.

7. Apply a glass cover slip to sandwich and disperse the drop, and incubate the peptide microarray slide for 2 hours at 37°C.

8. Allow the cover slip to self-disassociate while washing the peptide microarray slide in 45 ml of Primary Wash Solution (Recipe 6). Wash the peptide microarray slides twice with 45 ml of Secondary Wash Solution (Recipe 7), and twice with 45 ml of distilled H2O.

Note: Microarrays prepared as described fit in a 50-ml screw-capped tube that can be filled with the appropriate wash solution and mechanically agitated at low speed. Change the position of the tube to ensure that all portions of the array are bathed in the wash solutions.

9. Air-dry the arrays, then expose to a phosphoimager screen for 1 week.

Detecting and Analyzing Phosphorylation of the Array Peptides

1. Set a Typhoon scanner at the highest sensitivity settings with a pixel size of 25 μm, and obtain array images from the phosphoimager screen.

2. Process the captured image of the phosphoimager screen using ImageQuant TL v2005 software and, following the manufacturer’s instructions, crop the images to the visible outlines of the peptide arrays in order to obtain individual peptide array images.

3. Create a protocol file in ArrayVision containing the coordinates of each spot and the measurements of spacing between spots and blocks, as well as the dimension of spots and blocks, according to the manufacturer’s instructions.

Note: For the peptide arrays described here, the parameters are horizontal and vertical spacing of 0.125 mm between spots, a distance of 0.125 mm between two blocks, and dimension of 0.4 mm for each spot.

4. Number the three replicate blocks on the array as 1, 2, and 3, with one set as the block adjacent to the slide label.

Note: Proper orientation of the slide on the image screen is necessary for the program to correlate the correct spot identifier to its intensity.

5. When loading a peptide array, rotate the array so that the label side is positioned to the left, and the corner with a row as well as a column of positive controls is oriented to the bottom right.

6. Calculate the background intensity for each spot as the average of pixels from four regions in the immediate vicinity of each spot.

7. Remove data points for the positive controls and blank spots, which were used for orienting the grid.

Note: The high intensity values for positive control spots would skew lower signal data from bovine peptides and preclude proper normalization and statistical analysis.

8. Perform global normalization to the 50th percentile on the raw data, using GeneSpring software.

9.Evaluate reproducibility of the signal by comparing peptide signals to a baseline value of 1 for the normalized data from the replicates, using a sample t test.

Troubleshooting

Reproducibility of Signal Intensity Across Arrays

Although the reproducibility of individual peptides on a single array is excellent, different arrays of the same treatment condition, performed at separate times, may differ in overall signal intensity. While the relative intensity of any given peptide is consistent across arrays, differences in absolute intensity across arrays complicate data analysis. We find that differences in overall signal intensity across arrays usually result from incomplete cell lysis. The degree of labeling of a given peptide depends on both the activity and quantity of the corresponding enzyme; therefore, it is essential that an identical number of cells are used for each experiment and that the cells are completely lysed. One potential source of incomplete lysis is incomplete resuspension of the cell pellet in the Lysis Buffer. Analysis of protein concentration by Bradford assay or other similar techniques will confirm that equal concentrations of protein are compared. Another factor that affects signal intensity is the radiolabeled ATP. To ensure consistent overall intensity across arrays, the quantity of radiolabeled ATP, as well as the ratio of labeled to unlabeled ATP, must be consistent across experiments.

High Background

To minimize background, it is necessary to remove any radioactivity not covalently linked to the peptides. Failure to adequately wash the slides results in high background, both generally and in localized regions, that overwhelms the signal from the peptides. Arrays should be gently agitated during the washes, and care should be taken to ensure that all portions of the array are bathed in the wash solutions.

Physical Damage to the Arrays

The arrays are susceptible to mechanical disruption of the peptide spots, which will compromise the ability to obtain an interpretable signal. It is important that the “active” side of the arrays remain untouched by hands or pipette tips. Additionally, if the peptide array is sealed with a cover slip during the incubation period, considerable care must be taken in removing the slip to prevent damage to the array. Allowing the cover slip to self-disassociate in the primary wash solution prevents damage to the array.

Notes and Remarks

The conservation of many phosphorylation sites within orthologous proteins enables the rapid creation of peptide arrays for kinome analysis in species for which limited phosphorylation databases exist but genomic information is available. This bioinformatics approach eliminates the need to generate and characterize species-specific, phosphorylation-specific antibodies. Additionally, focusing on enzymatic activity (rather than product abundance) enhances the sensitivity of detecting changes in kinase activity. This approach also enables development of high-throughput tools that can simultaneously assess changes in a large number of preselected phosphorylation events. Such customization of arrays can be taken a step further by development of arrays focused on specific biological responses or cell functions. The bovine array used in the current investigation was biased for phosphoprotein targets known to be involved in innate immune signaling pathways.

The use of peptide arrays for kinome analysis is still in the early stages of development, and a number of technical and biological challenges remain to be addressed. For example, whereas the arrays are designed to monitor kinase activity as a marker of signal transduction activity, the net degree of phosphorylation of a particular target reflects the dual and opposing contributions of kinase and phosphatase activity. The presence of a panel of phosphatase inhibitors within the cellular extracts analyzed here negates the contribution of these enzymes to the net phosphorylation status of a particular target in order to simplify analysis of kinase activity.

In some cases, peptide substrates may not serve as ideal ligands for the corresponding kinase. This could manifest itself in a number of ways. There is the danger of assigning false positives as a result of phosphorylation of a particular peptide by a kinase other than the one that mediates cellular phosphorylation. For this reason, results obtained through peptide arrays should be verified by alternative methodologies that permit direct characterization of the phosphorylation status of the protein or functional characterization of the associated phenotype. That the peptides may be recognized by the correct protein kinase, but with lower efficiency than when the sequence is in the context of an intact protein, also presents a danger for false negatives where the activity of a kinase may be underestimated. For this reason, it is difficult to make quantitative comparisons of the levels of activities of different kinases on the basis of the extents of phosphorylation of their corresponding peptide substrates. It is more appropriate to compare the relative levels of activity of a particular kinase under different experimental conditions.

Peptide substrates on the array may be exposed to kinases to which they would not be physiologically exposed (because of higher-order organization within the cell or because of cell type–specific expression patterns). In these instances, phosphorylation of the peptide by that kinase represents a true experimental signal that has no physiological importance and should therefore be considered a false positive.

A species-extrapolated peptide array will experience the same limitations as a peptide array created for species with well-defined phosphorylation information, with one additional challenge. A paralog protein (one that is separated by gene duplication), rather than an ortholog (one that is separated by a speciation event), may produce matches of up to 100% sequence similarity for the phosphorylation motif of interest. Inclusion of such peptides on the array will produce false positive signals.

The intent in the design of the arrays is to provide a robust tool for kinome analysis that is customizable with respect to species of interest as well as to phosphorylation events involved in specific cellular processes. Although these arrays have proven to be reliable indicators of biological signaling, it is nonetheless most appropriate to use these arrays as a tool to identify modification events that can then be verified through functional assays or other independent methodologies.

References

View Abstract

Stay Connected to Science Signaling

Navigate This Article