|Home | About | Journals | Submit | Contact Us | Français|
A system-wide analysis of cell signaling requires detecting and quantifying many different proteins and their posttranslational modification states in the same cellular sample. Here, we present Protocols for two miniaturized, array-based methods, one of which provides detailed information on a central signaling protein and the other of which provides a broad characterization of the surrounding signaling network. We describe a bead-based array and its use in characterizing the different forms and functions of β-catenin, as well as lysate microarrays (reverse-phase protein arrays) and their use in detecting and quantifying proteins involved in the canonical and noncanonical Wnt signaling pathways. As an application of this dual approach, we characterized the state of β-catenin signaling in cell lysates and linked these molecule-specific data with pathway-wide changes in signaling. The Protocols described here provide detailed instructions for cell culture methods, bead arrays, and lysate microarrays and outline how to use these complementary approaches to obtain insight into a complex network at a systems level.
Protein arrays have become powerful tools to investigate the status of signaling pathways in cells or tissues. The ability to perform multiplexed assays on hundreds to thousands of samples enables time-resolved studies of cells stimulated or perturbed in different ways. The data from these studies can then be used to infer the structure of the underlying network. Protein array technology is well suited for these types of investigations because it provides a way to measure many different proteins in parallel while consuming very little material (1, 2). Over the past decade, two array platforms—bead-based arrays and lysate microarrays—have become well established in cell signaling research (Fig. 1). Both methods have been used to analyze signaling networks in a time-resolved fashion (3–6), and both methods offer multiplexing capabilities. In the case of bead arrays, a mixture of microspheres is used to detect and quantify different analytes in a sample. The beads are typically coated with capture antibodies specific to different analytes, and captured analytes are detected and quantified by using a mixture of fluorescently labeled detection antibodies (Fig. 1A). The identity of each bead is revealed by using an internal fluorescent color code. In the case of lysate microarrays, different samples are spotted onto a series of nitrocellulose-coated slides, and each slide is probed with a different antibody (Fig. 1B). In this case, the identity of each slide specifies the analyte and the location of each spot in the array specifies the sample. In both assays, posttranslational modifications can be detected by using posttranslational modification–specific antibodies.
One application of the bead-based assay is the acquisition of detailed information on a single protein. Because critical, highly connected nodes in signaling networks are often pleiotropic, it is important not just to quantify the abundance of the protein, but to obtain quantitative information on its different forms and on its interaction with other proteins. The specific state of a central signaling protein is often influenced by the surrounding network and, in turn, dictates downstream signaling. Thus, to understand the role of such a protein requires detailed information on not only the protein, but on its surrounding network as well. Here, we describe how to obtain such information in a time-resolved fashion, using, as an example, the response of hepatocarcinoma (HepG2) cells to stimulation with either a canonical Wnt ligand, Wnt3a, or a noncanonical ligand, Wnt5a.
In the case of Wnt signaling, the intracellular protein β-catenin is multifunctional, playing critical roles in both signaling and cell-cell adhesion complexes. β-catenin is also a proto-oncogene, and activating mutations in the gene that encodes β-catenin contribute to the genesis of common cancers, such as colorectal cancer and hepatocellular carcinoma (7–9). The different functions of β-catenin as a transcriptional coactivator and as a cell adhesion molecule are regulated by changes in protein abundance and phosphorylation state, both of which affect the ability of β-catenin to complex with other transcription factors or to interact with adhesion proteins, such as the cadherins (10–12). Increases in the abundance of cytoplasmic β-catenin and accumulation of the uncomplexed, transcriptionally active form of β-catenin are hallmarks of active β-catenin–dependent “canonical” Wnt signaling (13). Noncanonical signaling regulates cell polarity and cell movements and involves pathways, such as the planar cell polarity pathway, the Wnt to Jun N-terminal kinase pathway, or the Wnt to Ca2+ signaling pathway (14).
The analytical methods described here are designed to provide a holistic view of the complex interactions mediated by β-catenin and how these interactions influence its function (15, 16). Data obtained with these methods can then be used to train computational models of Wnt signaling, which provide insight into the structure of the network and how best to intervene pharmacologically (17–21). More generally, the dual approach described here could be used to gain insight into other complex pathways with central signaling proteins, such the DNA damage response network and p53, or the epidermal growth factor (EGF) signaling network and the EGF receptor.
We used human HepG2 cells to study the activation of the Wnt signaling pathway. Previous studies have shown that treatment with Wnt3a at final concentration of 100 to 200 ng/ml activates the β-catenin/TCF pathway (22, 23). The methods described can easily be adapted to other cell lines that respond to other stimuli. The optimal concentration of ligand should be determined in preliminary experiments based on cellular and molecular response.
Each experimental condition should be tested in biological triplicate. To generate comparable sample sets for bead array and lysate micro-array analysis, perform two identical cell culture experiments in parallel. Careful attention to confluence and plating conditions is required and these variables should be standardized to the greatest extent possible.
Proteins can be directly conjugated to carboxy-modified magnetic beads with EDC/NHS chemistry resulting in an amide bond. These beads can be used directly for bead array–based sandwich immunoassays or protein-protein interaction assays. If buffer agents like glycine, stabilizing agents like BSA, or other additives impede the covalent immobilization through an amide linkage, site-directed noncovalent strategies can be used. For instance, tag- or species-specific antibodies can be used for the selective capture of recombinant proteins or antibodies. We used antibodies specific for total β-catenin and E-cadherin as capture molecules in assays for the determination of β-catenin and E-cadherin–β-catenin complexes.
Some proteins cannot be covalently coupled to carboxy-modified beads because of the presence of unfavorable additives, such as carrier protein or glycine buffer, or sodium azide or low protein concentration. In this case, the antibody or the bait protein can be coupled to the bead through a noncovalent immobilization strategy by using tag- or species-specific antibody-conjugated beads. We used antibodies that recognize phosphorylated β-catenin or GST-tagged bait proteins (GST-ECT and GST-TCF4) to illustrate this method. The indirect immobilization process involves several washing steps to remove excess unbound bait molecules, which can interfere with other assay components or the sample.
Before the assay procedure, the cellular samples need to be adjusted to a consistent protein concentration. Depending on the cell type and the abundance of the protein of interest (in this case, β-catenin), 10 to 25 µg of total protein is usually appropriate for analysis. We suggest determining optimal amounts in separate preliminary experiments. A dilution series of cell culture lysates can be used to identify a suitable range of lysate amounts for detection of the proteins of interest.
Although we describe how to perform the process using a magnetic plate separator and manual washing in a 96-well plate, it is also possible to use a robot—for example, a magnetic bead transfer system (King Fisher, Thermo Fisher Scientific), which facilitates semiautomated washing and incubation (24).
The microarray assay is performed by using printed slides created from a “source plate,” which contains the cellular samples in the orientation matching the pin configuration of the microarrayer. The layout of the source plate depends on pin configuration of the microarrayer. We recommend that the design should allow the biological and technical replicates of each sample to be printed in one row (typically three biological replicates for each in duplicate). Because most antibodies exhibit a sigmoidal, rather than linear, relationship between analyte concentration and fluorescence, we correct for this nonlinearity by including twofold serial dilutions of a control lysate for each microarray, which are then used as an in-well calibration standard.
For this microarray assay, it is important that the lysate samples are cleared to remove viscous insoluble supernatant, which we achieve by filtering the samples using a filter plate.
The microarray slides are scanned in the 680-nm and 800-nm channels with an Odyssey imager.
For the assay to work, the protein of interest must be present in sufficient quantity to be detected. Therefore, it is important to use cells in which the pathway of interest or protein of interest is known to be present to serve as a positive control. Lack of signal in the positive control is an indication of technical problems with the assay rather than an indication that the protein is not present in sufficient quantity to be detected in the sample.
To analyze β-catenin signaling, we measured free β-catenin, which is the transcriptionally active form, in HepG2 cells. HepG2 cells express a mutant form of β-catenin that lacks N-terminal regulatory sites and thus is constitutively active. For cells that exhibit normal Wnt pathway activity and require Wnt ligand for activation, free β-catenin in the absence of added Wnt ligand is undetectable. Therefore, we suggest preparing positive control samples from human embryonic kidney (HEK) 293 cells exposed to Wnt3a or a glycogen synthase kinase 3 (GSK3) inhibitor, as previously described (21).
Once it has been determined from the positive controls that the problem is technical, there are several potential causes of the lack of signal. Coupling efficiency of the noncovalent immobilization procedure can be an issue. Regarding the GST-tagged bait proteins used here for the analysis of β-catenin activity, this can be assessed by incubation with PE-labeled antibody that recognizes GST. Whereas the bead-coated phosphorylation-specific antibodies that recognize β-catenin can be detected with a PE-labeled antibody that recognizes rabbit IgG. Additionally, the covalent immobilization of antibodies that recognize E-cadherin or total β-catenin can be tested by using a PE-labeled antibody that recognizes goat IgG.
Moreover, degradation of the bait molecules coating the beads will result in an undetectable signal. To exclude loss of signal caused by degradation, bait molecules can be analyzed separately by using SDS-polyacrylamide gel electrophoresis (SDS-PAGE).
A high background signal on a reverse-phase array usually arises from a high degree of cross-reactivity by the primary antibody. If this is the case, the data are unreliable and should be excluded from the analysis, and alternative primary antibodies must be identified.
Low or no signal on reverse-phase arrays can arise if the concentration of the primary antibody is too low (or if the stock of antibody is too old). Increasing the primary antibody concentration and using newer detection reagents should improve the signal.
Bead-based arrays can be used in a focused way to obtain information on a specific signaling protein of interest (in this case, β-catenin) or a set of signaling proteins (24). Molecule-specific characteristics of the protein—such as abundance, posttranslational modifications, complex formation, and activity—can be detected in parallel (21). Overall, bead-based assay systems enable high sample throughput with low sample consumption (10 to 20 µg of total protein per sample) and are ideal for studies in which cells are analyzed under many stimulation conditions and many time points. In terms of assay development issues, capture and detection reagents can be tested rapidly, without the need for large quantities of material. Thus, additional analytes can be added to the existing multiplexed assay in a relatively straightforward fashion.
In general, bead-based assay systems are limited in the number of proteins that can be analyzed in parallel (typically <25). Because of cross-reactivity, the signal-to-noise ratio decreases with an increasing number of detection antibodies in the system. Thus, bead-based assays are most appropriate for performing focused studies that involve a relatively small number of analytes but a large number of samples.
Regarding the development of such assays, two antibodies for each analyte are required: One is the capture (or bait) molecule, and the other is the detection molecule. Therefore, the availability of antibodies is a strong limiting factor. Moreover, the antibodies used in such assays should recognize its target in a native form, a fact that additionally limits the number of suitable antibodies.
The bead array panel described here provides molecular snapshots of one protein. Conditions have been optimized especially for extracting native β-catenin, its modified forms, and the intact cadherin–β-catenin complex. In general, the appropriate detergent has to be investigated; salt concentration and inhibitor and buffer additives of the lysis buffer have to be optimized. This could be an issue if single assays are combined to a multiplex format, because the optimal extraction conditions for proteins are always slightly different. Thus, the extraction protocol for bead-based arrays detecting native proteins is a compromise and sensitivity of the singleplex can decrease in a multiplex format.
One of the primary advantages of the microarray format is that it is compatible with large-scale investigations. Thousands of biological samples can be arrayed on a single slide, to provide information on cells subjected to many different stimulation conditions, time points, and other perturbations. It is also compatible with extremely small sample sizes; a single microarray spot contains as much lysate as would be obtained from a single cell. As little as 5 µl of a 1 mg/ml solution of lysate is needed for printing, and this is sufficient to print hundreds of arrays. In addition, lysate arrays are not limited with respect to their level of multiplexing. Because each array is probed separately, additional proteins can be quantified simply by probing additional arrays with new antibodies. Finally, the denaturing conditions used in this assay permit the uniform detection of proteins, regardless of whether or not the epitope is masked by protein-protein interactions.
The major drawback of this method is that it requires highly selective antibodies. Sandwich-style immunoassays, such as those used in the bead-based assay system, require each protein to be recognized by two different antibodies with nonoverlapping epitopes and are therefore less subject to off-target signal. Reverse-phase arrays, in contrast, use only one antibody per analyte and are, therefore, more prone to interference from nonspecific antibody binding. Moreover, antibodies that perform well in Western blots do not necessarily yield quantitative data when used for lysate microarrays. It is therefore necessary to evaluate each antibody by direct comparison of data obtained by arrays and Western blots.
In this study, we used two complementary types of miniaturized proteomic methods to study canonical and noncanonical Wnt signaling. The combination of these methods enabled high-throughput collection of detailed information on the central signaling protein, β-catenin, and on the surrounding network.
We used a bead array panel (21) to study changes in several forms of β-catenin in response to Wnt3a and Wnt5a in HepG2 cells. A truncated form of β-catenin lacking residues 25 to 140 is abundant in HepG2 cells, along with a smaller amount of wild-type β-catenin (8, 25). Thus, they represent a model system in which to study a hyperactivated Wnt pathway. In the absence of added ligand, HepG2 cells had large amounts of free and total β-catenin, and both pools contained wild-type and truncated β-catenin (Fig. 1A). The E-cadherin–β-catenin complex was abundant in these cells, which confirms an intact E-cadherin–mediated cell-adhesion mechanism in HepG2 cells (26). In addition, we observed increasing amounts of total β-catenin, free β-catenin, and Ser675 phosphorylated β-catenin, but not that of Ser33/Ser37/Thr41 Ser33, and Ser552 phosphorylated β-catenin, upon Wnt3a treatment (Fig. 2). In response to Wnt5a, no changes have been observed with the bead array panel.
To link molecule-specific data with pathway-wide changes in protein abundance, we used lysate microarrays to assay 21 cell signaling proteins or phosphorylated proteins involved in Wnt signaling (Fig. 3B). Differences in the response of the cells with regard to the amount of total β-catenin, TCF, and axin2 were apparent in cells exposed to Wnt3a or Wnt5a (Fig. 3A). We also detected Wnt3a-mediated activation of the lipoprotein related protein 6 (LRP6) Wnt co-receptor in HepG2 cells (detected as an increase in receptor phosphorylation) after 30 min (Fig. 3B). An increase in the amount of total β-catenin was detectable after a 1-hour exposure to Wnt3a (Fig. 3A). We also observed an increase in the amounts of Axin1 and Axin2, which are negative regulators of the canonical Wnt pathway (27, 28) in response to Wnt3a. Thus, the reverse-phase array data revealed an inducible canonical Wnt pathway in HepG2, even though these cells harbor hyperactivated β-catenin.
In contrast, the noncanonical Wnt ligand Wnt5a generally induced opposing effects on Wnt pathway proteins compared to those induced by Wnt3a (Fig. 3A). However, the overall amount of β-catenin was unchanged (Figs. 2 and and3A).3A). Thus, we observed ligand-specific responses of HepG2 cells in response to activation of either the canonical or noncanonical pathway (Fig. 3C). Consistent with previously published reports (22), we observed that activation of noncanonical signaling by Wnt5a impaired canonical Wnt activity at the level of the receptor and failed to increase the abundance of β-catenin.
Funding: This work was funded by the German Federal Ministry of Education and Research [BMBF; grant FKZ 313081E (Systems Biology)] and NIH grants 54HG006097, R21 CA126720, and U54- HG006097. K.L. was supported through the Boehringer Ingelheim Fonds Travel Grant. T.S.G. is a Human Frontier Science Program Fellow.
Competing interests: G.M. is a founder, employee, and shareholder of Merrimack Pharmaceuticals; a founder, consultant, and shareholder of Makoto Life Sciences; and a scientific advisory board member of Aushon Biosystems. T.O.J. is a member of the scientific advisory board of Luminex Corp.