Extracellular Protein Microarray Technology for High Throughput Detection of Low Affinity Receptor-Ligand Interactions

Summary

Here, we present a protocol to screen extracellular protein microarrays for identification of novel receptor-ligand interactions in high throughput. We also describe a method to enhance detection of transient protein-protein interactions by using protein-microbead complexes.

Abstract

Secreted factors, membrane-tethered receptors, and their interacting partners are main regulators of cellular communication and initiation of signaling cascades during homeostasis and disease, and as such represent prime therapeutic targets. Despite their relevance, these interaction networks remain significantly underrepresented in current databases; therefore, most extracellular proteins have no documented binding partner. This discrepancy is primarily due to the challenges associated with the study of the extracellular proteins, including expression of functional proteins, and the weak, low affinity, protein interactions often established between cell surface receptors. The purpose of this method is to describe the printing of a library of extracellular proteins in a microarray format for screening of protein-protein interactions. To enable detection of weak interactions, a method based on multimerization of the query protein under study is described. Coupled to this microbead-based multimerization approach for increased multivalency, the protein microarray allows robust detection of transient protein-protein interactions in high throughput. This method offers a rapid and low sample consuming-approach for identification of new interactions applicable to any extracellular protein. Protein microarray printing and screening protocol are described. This technology will be useful for investigators seeking a robust method for discovery of protein interactions in the extracellular space.

Introduction

The method reviewed here describes the printing of a collection of extracellular proteins in a microarray format, followed by a method for screening of a target of interest against this library. We have identified protein multimerization as a crucial step for detection of interactions characterized by low binding affinities. To enhance detection of these interactions, we describe a protocol based on multimerization of the query protein of interest using microbeads.

Secreted and cell surface-expressed proteins (collectively termed extracellular proteins) along with their interacting partners are key regulators of cellular communication, signaling and interaction with the microenvironment. They are, therefore, essential in regulating many physiological and pathological processes. Approximately a quarter of the human genome (≈5,000 proteins) encodes for extracellular proteins, which, given their significance and accessibility to systematically delivered drugs, represent key targets for drug development¹. Consequently, extracellular proteins represent more than 70% of the protein targets with known pharmacological action for approved drugs on the market, known as the "druggable proteome". Despite their importance and abundance, the extracellular protein-protein interaction (ePPI) networks remain remarkably underrepresented in the available databases. This is fundamentally due to the complex biochemical nature of the extracellular proteins, which precludes their characterization using most available technologies². Firstly, membrane proteins are difficult to solubilize, a process that often involves harsh washing conditions and detergents; secondly, extracellular proteins often lack relevant post-translational modifications such as glycosylation that are absent when these proteins are expressed in commonly used heterologous systems. Finally, interactions between receptors, such as co-receptors expressed on immune cells, are often transient and characterized by very low affinities (K_D in the ~1 μM to >100 μM range). Altogether, the nature of these proteins and their binding partners render most widely utilized technologies, such as affinity purification/mass spectrometry (AP/MS) or yeast-two-hybrid, unsuitable for detection of interactions in the extracellular space^2,3.

In an effort to overcome these technical challenges and accelerate the discovery of novel interactions for extracellular proteins, we have developed a high coverage extracellular protein microarray^4,5. Microarrays offer the advantage of generating high-density surfaces with small amounts of sample, and are generally amenable to high throughput studies. Protein microarray-based studies have previously provided relevant insights into protein interactions for several model organisms, albeit mainly focusing on cytosolic interactions or on specific protein families^6,7,8. In contrast, limited work has been done to investigate extracellular protein interactions using this technology. We have developed a protein microarray method to enable studies of ePPIs by building a comprehensive and highly diverse library of purified secreted proteins and single transmembrane (STM) receptors expressed as recombinant extracellular domains (ECD) fused to common tags for affinity purification⁴. The success of the protein microarray screens relies heavily on the establishment of a high quality protein library. For expression of both the library and query protein, mammalian cells or insect cells were preferentially chosen as heterologous expression systems, to ensure proper addition of post-translational modifications such as glycosylation or disulphide bonds. SDS-PAGE, size exclusion chromatography and multi-angle laser light scattering are techniques commonly utilized to assess recombinant protein quality. The protein library is then spotted onto epoxysilane slides and stored at -20 °C for long-term use. Protein concentrations above of 0.4 mg/mL are recommended for the protocol described below. Therefore, low-expressing proteins may require a concentration step prior to sample printing and storage. Nevertheless, a main advantage of this technique is the small volume of protein required (50 μg of protein is sufficient to perform >2,000 screens), alongside minimal query protein consumption (20-25 μg per duplicates screen). Using the protocol and equipment described here, and provided libraries are available, results for individual query proteins can be generated within one working day.

A major challenge in detecting protein interactions in the extracellular environment arises from their characteristically weak or transient nature, which precludes identification by most commonly used methodologies. Increasing the binding avidity greatly improves sensitivity for detection of weak protein interactions^9,10,11. Based on this principle we developed a method to multimerize the query proteins (expressed as Fc fusion) using protein A-coated beads^4,5. To avoid any potential inactivation of the query protein by random labeling, we instead label an irrelevant human immunoglobulin G with Cy5 and add it along with the query protein to the protein A beads, thus eliminating any artifacts due to the direct conjugation of a dye to the protein of interest. Given the micromolar affinities of several co-receptor pairs, the multivalent complexes greatly enhance signal to noise ratio, compared to Fc-fusion query proteins screened as soluble proteins⁴.

In summary, the goal of this protocol is to describe the preparation of microarray slides containing a pre-existing extracellular protein library for identification of receptor-ligand interactions. We review the steps for slide printing, followed by a protocol for screening of a protein of interest against the extracellular protein library. Moreover, we describe a method for enhanced detection of ePPIs based on microbeads to achieve increased avidity of the protein under study. The extracellular protein microarray technology described here represents a fast, robust and effective approach for screening and detecting novel ePPI with low false-positive ratios, and by utilizing only microgram quantities of the query protein under investigation. This technology has fueled multiple studies that have provided relevant insights into previously unknown cellular functions and signaling pathways for a variety of receptors^12,13, including viral immunoregulators¹⁴, and can be utilized to de-orphanize any extracellular protein of interest.

Access restricted. Please log in or start a trial to view this content.

Protocol

1. Generation of a Library of Extracellular Human Proteins

1. Compile a list of cell surface receptors or secreted proteins of interest to build the protein microarray library. Specific protein families (for example, the immunoglobulin superfamily) or proteins selectively expressed in particular cell types can be selected for the study.
2. For cell surface receptors, determine the extracellular domain (ECD) boundaries by identifying the signal peptide and transmembrane regions using software tools. Some of the relevant tools, freely available online, are referenced in the Table of Materials^15,16,17,18.
3. Synthesize the ECD for the genes of interest and clone into the relevant vector. Secreted proteins or the ECD of STM receptors fused to a number of common affinity tags can be purified from the conditioned media of cells transfected with the appropriate vector, or from baculovirus-insect cell expression systems.
   Note: Mammalian cell-based systems (such as HEK/293 or CHO cells) are recommended to maximize the likelihood of proper protein folding and addition of relevant posttranslational modifications such as glycosylation.
4. Purify the proteins by standard affinity purification methods. Previous efforts have described automated or semi-automated procedures suitable for purification of hundreds of proteins, which can be scaled to generate the set of proteins of interest^9,19,20,21.
   Note: SDS-PAGE, size exclusion chromatography (S), or multi-angle laser light scattering (MALLS) are recommended to assess any non-covalent aggregation and to control for the overall quality of the protein preparations.
5. Adjust proteins to 200 or 400 μg/mL whenever possible, and dilute stocks with 80% glycerol for long-term storage in cryogenic vials at -20 °C. These represent master stocks and should be accessed only when necessary.
6. Transfer aliquots of each protein (100 μL) to 96-well plates (stock plates), seal with adhesive foil, and store at -20 °C until microarray slide preparation. Stock plates are generated to minimize freeze-thaw cycles and ensure protein stability.

2. Extracellular Protein Microarray Printing

1. Use a standard microarrayer for slide printing. The instrument utilized for the protocol described here has a capacity of 57 slides/run and uses a printhead holding 48 spotting pins. These pins generate spots of ~100 μm diameter separated by a spot-to-spot distance of ~350 μm and draws 0.25 μL per load. Under this configuration, up to 8,000 spots per slide can be accommodated.
2. Generate working plates (384 well plates, 10 μL sample/well) from the stock plates. Perform this step manually prior to slide printing using the microarrayer. Then, spot proteins with quill-type spotting pins onto epoxysilane slides at 60% humidity (to prevent dehydration of the protein spots).
   Note: Cy3-labeled Bovine serum albumin (BSA) can be spotted in duplicates between each protein sample (5 μg/mL in PBS/40% glycerol) to visualize the array for mask fitting (see section 4). Although recommended, this step is optional.
3. Subsequent to printing, remove the protein microarray slides from the humidified environment and block them overnight with 5% milk in PBST to inactivate the surface.
   Note: The preferred approach is to use an ultrasonic fogger to generate a fine mist of blocking solution that settles onto the slide surface.
4. Store slides at -20 °C in 50% glycerol to prevent freezing.

3. Preparation of Multivalent Bait Complexes

Note: Interactions between extracellular proteins are often characterized by low affinities. To enable detection of these interactions by increasing binding avidity, a multivalent approach based on capturing the query protein, expressed as Fc-tagged ECD, on protein A-coated microbeads was developed⁴.

1. Label the carrier IgG used for detection with Cy5 monoreactive dye and separate the free dye using desalting columns. Determine the dye to protein ratios by measuring ultraviolet absorbance at 280 and 650 nm.
   Note: Protein-dye ratios between 2.0 and 4.0 are normally used. It is recommended to spin the Cy5 conjugates at 100,000 x g for 15 min to remove potential soluble aggregates due to the labeling process.
2. Determine the optimal microbead-to-protein ratio by titration of protein A against a constant amount of the query protein. The minimal saturating volume of beads where no free Fc-tagged protein remains, as measured using a competitive assay determined by biolayer interferometry, is used for the screening.
3. Form the microbead-protein complexes by incubating the Fc-tagged query and the Cy5-IgG with protein A microbeads and mix on a tube rotator in PBS for 30 min at room temperature protected from light.
4. To form these complexes, use the query protein at a final concentration of 20 μg/mL. To calculate the molar ratio of the query protein and Cy5-IgG, divide the molecular weight of the query protein by the molecular weight of the IgG (150,000 Da) and multiply by 40 μg/mL to determine the concentration of Cy5-IgG needed.
5. Supplement samples with soluble protein A (1 mg/mL) immediately prior to incubation with the microarray slides (see 4.2 section below) to prevent binding of any free protein A beads to the Fc fusion proteins that may be present on the array.
   Note: The final reaction volume may vary depending on the hybridization station or incubation chamber utilized. The instrumentation described in the Table of Materials allows sample incubation using relatively small volumes (~200 μL per slide).

4. Extracellular Protein Microarray Screening and Processing.

Note: There are a number of manufacturers that provide automatic processing platforms. If a hybridization station is not available, the following steps can be performed manually, ensuring that there is sufficient volume of buffer to keep the slides submerged at all times.

1. Warm the slides at room temperature and rinse with PBS/0.1% Tween 20 (PBST) to remove residual glycerol before loading onto the hybridization station.
2. Use the following screening protocol:
   1. Wash with PBST for 1 min.
   2. Load 200 μL of 1 mg/mL protein A in PBS/5% milk and incubate for 30 min to prevent uncomplexed protein A microbeads from binding to Fc-tagged proteins that may be present in the microarray.
   3. Wash five times with PBST for 1 min.
   4. Load 200 μL of the query:microbeads complex in the presence of 1 mg/mL protein A and incubate for 30 min.
   5. Wash with PBST for 1 min.
3. After hybridization, rinse slides with water, place in individual 50-mL conical tubes, and dry by spinning at 900 x g for 5 min.
4. Finally, scan the slides with a microarray scanner appropriate to detect Cy3 (if BSA-Cy3 has been printed) and Cy5 fluorescence by exciting at 532 and 635 nm, respectively.
5. Perform data analysis using the accompanying microarray data analysis. The relevant array list (as a .gal file) is loaded into the data extraction software. At this stage, utilize the Cy3-BSA spots to find blocks and use the auto-fitting options in the software, followed by manual alignment of the features when necessary. Save files as .gpr files for further analysis.

5. Data Analysis

1. Save data as GPR files and process in R using the limma package, commonly utilized for analysis of microarray data^4,5.
   1. For preprocessing, do background correction and within-slide normalization. Since each protein is printed in duplicated spots, combine both replicate measurements to create a single score for that protein in the library.
   2. Calculate scores for each slide and analyze results for the intersection of high-scoring candidates between slides.
   3. Use duplicate microarrays from separate print-runs to control for slide variability and use intersection plots representing data from both screens to call final hits.
   4. Finally, perform an additional level of filtering to exclude promiscuous proteins within the array. Such non-specific binders are identified as proteins exceeding a specific threshold hit frequency as defined by the user across independent screens.
      Note: This protocol is described in detail in Tom et al. 2013⁵. In our case the non-specific binder calling is based on the determination of the cumulative prey hit rate, and a data-driven elimination threshold of 5% is used.

Access restricted. Please log in or start a trial to view this content.

Results

A schematic of the workflow for the extracellular protein microarray technology is shown in Figure 1. Once the microarray slides containing the extracellular protein library are available, the screening of the protein of interest and data analysis can be completed within one day. Many physiologically relevant interactions between membrane-embedded receptors are characterized by very weak binding strengths (K_D in the micromolar range). To improve de...

Access restricted. Please log in or start a trial to view this content.

Discussion

A significant number of orphan receptors remain in the human genome, and novel interacting partners continue to emerge for extracellular proteins with previously characterized ligands. Defining the receptor-ligand interactions in human and model organisms is essential to understand the mechanisms that dictate cellular communication during homeostasis, as well as dysregulation leading to disease, and therefore inform new or improved therapeutic options. Nevertheless, detection of extracellular protein interactions by wide...

Access restricted. Please log in or start a trial to view this content.

Materials

Name	Company	Catalog Number	Comments
Ultra Pure MB Grade glycerol	USB Corporation	56-81-5	Protein storage
SeptoMark blocking buffer	Zeptosens	BB1, 90-40	Blocking buffer microarray slides
Bovine serum albumin	Roche	03-117-957-001	Slide control for mask fitting (optional)
Polypropylene multiwell plates	Greiner Bio One	82050-678	Protein storage
Polypropylene multiwell plates	Arrayit	MMP384	Slide printing
NanoPrint LM60, or similar contact microarrayer	Arrayit	NanoPrint LM60, or similar contact microarrayer	Slide printing
Micro spotting pins	Arrayit	Micro spotting pins	Slide printing
ZeptoFOG blocking station	Zeptosens	ZeptoFOG blocking station, 1210	Block slides after printing
Skim milk powder	Thermo Fisher	LP0031	Blocking solution
Epoxysilane-coated glass slide	Nextrion Slide E	1064016	Microarray slides
Glass holder and slide rack set	Wheaton	900303	Slide storage
Cy5 monoreactive dye	GE Healthcare	PA23031	Albumin labeling
Cy5 monoreactive dye	GE Healthcare	PA25001	Human IgG labeling
Pro-spin desalting column	Princeton Separations	CS-800	Remove free dye
Adhesive aluminum foil seal	AlumaSeal	F-384-100	Seal stock plates
Polypropylene cryogenic vials	Corning	430658	Master vials for protein library storage
Protein A microbeads	Miltenyi	120-000-396	Query protein multimerization
Human IgG	Jackson Immunoresearch	009-000-003	Irrelevant IgG for labeling
Protein A	Sigma	P7837	Microarray slide blocking
Hybridization station, a-Hyb or similar	Miltenyi	Hybridization station, a-Hyb or similar	Automated microarray processing (optional)
GenePix 4000B scanner or similar	Molecular Devices	GenePix 4000B scanner or similar	Slide scanning
GenePix Pro or equivalent data extraction software	Molecular Devices	GenePix Pro or equivalent data extraction software	Data processing
Signal P4.1	DTU Bioinformatics, Technical University of Denmark	online software	Prediction tool to determine presence and location of signal peptide cleavage sites
TMHMM 2.0 server	DTU Bioinformatics, Technical University of Denmark	online software	Prediction of transmembrane helices in proteins
Phobius	Stockholm Bioinformatics Center	online software	A combined transmembrane topology and signal peptide predictor
TOPCONS	Stockholm University	online software	Prediction of membrane topology and signal peptides