Identification of Signaling Protein Complexes by Parallel Affinity Precipitation Coupled with Mass Spectrometry

Protein–protein interactions play a pivotal role in both inter- and intra-cellular signaling. Identification of signaling protein complexes can thus shed important new insights into cell communications. We developed a parallel affinity precipitation protocol to overcome the disadvantages of the tandem affinity purification procedure, such as the potential disruption of target protein conformation, subcellular localization or function by epitope tags, the potential need of large amounts of cell culture or generation of stable cell lines, and relatively long duration the two-step precipitation takes. This new simplified assay of protein interaction is quick, economic and specific. This paper describes the details in the design and method of the assay.


Introduction
We developed the parallel affinity precipitation (PAP) coupled with mass spectrometry assay for quick identification of proteins that interact specifically with Krüppel-like factor 8 (KLF8) in mammalian cells 1 by modifying the tandem affinity purification (TAP) method. TAP was originally developed for purification of protein complexes from yeast. [2][3][4] TAP soon found increasingly wide application for isolation of protein complexes from mammalian cells. [5][6][7] TAP is based on two-step sequential co-precipitations of protein complexes using antibodies or affinity beads specific for one of two different epitope tags one of which is linked to the amino terminus and the other to the carboxyl terminus of the target protein (or bait protein). After resolved on a protein gradient gel, the identities of the target protein co-precipitated proteins (gel bands) are determined by mass spectrometry. The two-step precipitations plus a protein cleavage and/or protein complex elution step was designed in TAP to minimize the inevitable non-specific pull-down of irrelevant proteins. This design, however, often requires large amounts of cell lysate (from up to several liters of cell culture) and generation of a cell line stably expressing the target protein to start with. Another disadvantage of TAP is that tagging both termini of a target protein sometimes interferes with the target protein conformation, function, and even cellular localization resulting in identification of false complexes associated with the target protein. This is particularly true in the case of proteins like KLF family proteins if used as target proteins. Given their conserved DNA binding domain at the very carboxyl terminus, a tag to the carboxyl terminus of a KLF family protein will likely alter their protein interaction profile, if not nuclear localization, and even function. For these reasons, we developed the PAP assay. In the PAP assay, the target protein is tagged on only one of the two termini, which is less likely to interfere with function, conformation, or subcellular localization of the target protein. Both a HA-tagged target protein and a Myc-tagged target protein are generated. After transiently overexpressed, these tagged proteins and their associated protein complexes are co-immunoprecipitated (co-IPed) in parallel using anti-HA and anti-Myc antibody-conjugated beads, respectively. After resolving on a protein gradient gel, only the bands that are shown in both the anti-HA and anti-Myc precipitates but absent in the precipitates from both of the mock controls are collected for mass spectrometry. We found that PAP assay requires significantly less cells/lysates and saves large amounts of time without compromising specificity of the complex isolation. Since whole cell lysates are used, this method is suitable for identifying the interaction between any proteins, be they membrane, cytosolic, or nuclear. The TAP and PAP assays are schematically illustrated side-to-side in Figure 1.

Flow Chart
Construct hemagglutinin (HA)-tagged and Myc-tagged target protein plasmids

1.
Construct two target protein expression vectors, one with an HA tag and the other with a Myc tag:

1.1
Design primers with appropriate restriction enzyme digestion sites, and amplify target protein cDNA from cDNA cloning vector;

1.2
Use appropriate restriction enzymes to digest the cDNA fragment, then purify and insert them into the expression vectors, pKH3 and pHAN, respectively. These two vectors share identical cloning sites making the plasmid construction convenient, particularly for generating N-terminal HA or Myc tagged proteins. However, if a C-terminal HA or Myc tagged protein is preferred, there are also some cloning sites upstream of the tag peptide that can be used;

1.3
Transfect each of the two vectors at 2 μg/well into HEK293T cells grown in falcon tissue culture 6-well plates. After an overnight incubation, collect the whole cell lysates (WCLs) for Western blotting to verify the expression of the HA-or Myc-tagged target proteins.

2.1
Overexpress the HA-protein or Myc-protein by transfecting the pKH3target or pHAN-target vectors at 12 μg/well into HEK293T cells in falcon 100-mm tissue culture dishes; pKH3 or pHAN empty vectors are transfected as mock controls;

2.2
After a 48-hour incubation, collect WCLs in 1 mL/dish of the NP-40 lysis buffer supplemented with protease inhibitors on ice;

2.3
Add 50 μL of the NP-40 lysis buffer pre-washed anti-HA or anti-Myc antibody-conjugated agarose beads to 1 mL of the WCLs. Add the same amount of the beads to the mock control WCLs;

2.4
Incubate overnight at 4°C on a rotation rack. Centrifuge at 8000 g to collect the precipitates;

2.5
Wash the precipitates with the NP-40 lysis buffer 3 times at 4°C, then add the TruSep SDS Sample buffer (50-200 μL) and denature precipitated proteins at 95°C for 10 minutes.

3.
Resolve the precipitated protein samples by gradient Tris-HEPES gel electrophoresis:

3.1
Parallel-load (in the order of molecular weight marker, HA mock, HA-Target, Myc-Target, Myc mock, molecular weight marker; see Fig. 1H) the co-IPed protein samples (10-50 μL) into 4%-20% Tris-HEPES gradient mini gels previously described, 1 load dual color protein markers on both sides of the gel for molecular weight indicator;

3.2
Run the gel for 60 minutes at 100 volts or until the dye front gets close to the bottom of the gel.

4.1
Wash the gel for 5 minutes three times with distilled, deionized nano-Q water (ddH 2 O) at room temperature;

4.3
Rinse the gel briefly with ddH 2 O three times;

4.4
Wash the gel with 50% ethanol in ddH 2 O at 4°C for overnight to decrease the gel background;

4.5
Rinse the gel briefly with ddH 2 O three times at room temperature;

4.6
Stain the gel in Bio-Safe ™ Coomassie stain solution on a rocker by gently rocking for 1 hour;

4.7
Rinse the gel briefly with ddH 2 O three times at room temperature;

4.8
Wash the gel with ddH 2 O overnight at room temperature;

4.9
Wrap the gel with a piece of clean Saran wrap. Scan the gel for the image of the protein bands (note, never touch the gels with bare skin to avoid keratin contamination);

4.10
Keep the gel in 1%-2% acetic acid at room temperature.

5.1
Compare the HA-protein and Myc-protein lanes to select the protein bands of the same molecular weights that appear in both of the target lanes but are not visible or appear as much lighter bands in either of the mock lanes. Collect also the target protein bands for identity verification (Fig. 1).

6.1
Excise the bands of interest in both HA-protein and Myc-protein lanes for mass spectrometry analysis;

6.2
Cut the gel into small pieces and wash the pieces in 50% Acetonitrile, 100 mM NH 4 HCO 3 to remove residual SDS or Coomassie blue;

6.3
Shrink the gel pieces in acetonitrile and dry the gel pieces in Speed-Vac;

6.4
Swell the gel pieces in a digestion buffer containing 50 mM NH 4 HCO 3 and 12.5 ng/μL of trypsin (Sigma, Proteomics grade) in an ice-cold bath for 45 minutes;

6.5
Remove the extra supernatant;

6.8
Extract the cleavage products (peptides) by one change of 20 mM NH 4 HCO 3 ;

6.9
Spin, remove, and transfer the supernatant to a new Eppendorf tube; 6.10 Further extract the cleavage products for 3 more times using 5% formic acid plus 50% acetonitrile at the room temperature; 6.11 Pool the supernatants together; 6.12 Speed-Vac dry the supernatant until the volume is 1-2 μL.

7.
Protein identification by electrospray tandem mass spectrometry:

7.2
One end of the analytic column is pulled into a PicoFrit using a Sutter Instrument laser puller and placed directly into the QSTAR XL NanoSpray II source, limiting the peak dispersion after the column. The other end of the column is connected to the 10-port valve stream selector through a conductive Micro Tight union (Sigma-Aldrich) using 25 μm ID × 365 μm OD × 8 cm capillary tubing;

A Waters
CapLC system consisting of an autosampler, A, B, and C pumps and a stream selector is used for desalting and gradient formation. The pump outlet is connected to the 10-port valve through a splitting tee that reduces the flow from the pump from 9 μL/minute to 300 nL/minute using a length of restriction tubing made from fused silica;

7.4
Contact closure connection is made between the autosampler and the mass spectrometer;

7.5
Waters/Micromass Masslynx 3.5 software is used to program the CapLC system. The peptide mixture (50 μL) is injected and desalted on the trap column for 6 minutes and eluted onto the analytic column at 300 nL/minute by the application of a series of mobile phase B gradients (5%-10% B for 4 minutes, 10%-30% B for 61 minutes, 30%-85% B for 5 minutes, 85% B for 5 minutes, and 10% B for 6 minutes);

7.6
The mass spectrometer QSTAR XL is controlled through Analyst 1.

7.7
The MSMS data (wiff file) are converted to a Mascot generic file format using a Mascot "script" 1.6b27 for the database search; an overlap between two identity groups indicates highly potential interacting protein candidates.

Figure 1. Schematic illustration of TAP (A-C) and PAP (D-H) assays
In TAP, assume that the double tagged (HA and Myc in this case) target protein (#1, red) interacts specifically with two proteins (#2, brown; #3, yellow) in the target protein overexpressed cells (Target) (A). Control cells (mock) express the HA and Myc tags only (B). Also assume that non-specific interactions occur between protein #4 (green) and the HA tag, protein #5 (black) and the Myc tag, protein #6 (blue) and the anti-HA antibody (Y-shaped in blue), protein #7 (purple) and the anti-Myc antibody (Y-shaped in red), as well as protein #8 (brown) and the beads (solid circle in light blue). If the sequential precipitations are done first using the anti-HA antibody-conjugated beads followed by the anti-Myc antibodyconjugated beads, all the interacting proteins, specific or non-specific, would be precipitated by the first anti-HA IP except for proteins #5 and #7 which would be washed off prior to the second anti-Myc IP (C; lanes 1 and 3). The anti-Myc IP would then pull down the protein #8 only (C, lane 2) or the protein #8 together with the target protein and its specific interacting proteins #2 and #3 while leaving behind the non-specific proteins #4 and #6 (C, lane 4). In contrast, in PAP the two tags are added to the same end of the target protein (N-terminus in this case) to create two separate, single-tagged target protein constructs. Cells transiently over-expressing each of the target proteins (D, HA-Target and E, Myc-Target) or the tags only (F, HA-Mock and G, Myc-Mock) are used for a single step IP. The specific interacting proteins (#2 and #3) along with the target proteins would be pulled down by both antibodies only from the target cells (H, lanes 2 and 3) but not from the mock cells (H, lanes 1 and 4). Notes: The close distance between the Myc tag and the protein #2 binding site on the target protein (A) could jeopardize the interaction leading to failure to identify it. There is likely a need in TAP for transfecting two control vectors, one encoding the HA tag and the other encoding the Myc tag to make the mock cells (B). Protein #8 could be removed by preclearing IP using the antibody-free beads. In fact, the pHAN vector provides a Myc-His tag fusion. In this case, a single pHAN-target construct should suffice if specific precipitation using nickle-beads works as well. A protein band of a single molecular size does not necessarily mean a single protein identity in the band; more often than not, a single band contains more than one protein identity of the same molecular size. A target-protein binding protein could also weakly interact non-specifically with the tags, antibodies or beads, in this case the protein would be precipitated in a larger amount (thicker band) from the target cells than from the mock cells. Proteins interacting indirectly with the target protein are not drawn for fear to complicate the illustration. Abbreviations: IP, immunoprecipitation; PAP, parallel affinity precipitation; TAP, tandem affinity purification.