A Bead Aggregation Assay for Detection of Low-Affinity Protein-Protein Interactions Reveals Interactions between N-Terminal Domains of Inositol 1,4,5-Trisphosphate Receptors

Interactions between proteins are a hallmark of all cellular activities. Such interactions often occur with low affinity, a feature that allows them to be rapidly reversible, but it makes them difficult to detect using conventional methods such as yeast 2-hybrid analyses, co-immunoprecipitation or analytical ultracentrifugation. We developed a simple and economical bead aggregation assay to study low-affinity interactions between proteins. By coating beads with interacting proteins, the weak interactions between many proteins are sufficient to allow stable aggregation of beads, an avidity effect. The aggregation is easily measured to allow quantification of protein-protein interactions under a variety of controlled conditions. We use this assay to demonstrate low-affinity interactions between the N-terminal domains of an intracellular Ca2+ channel, the type 1 inositol 1,4,5-trisphosphate receptor. This simple bead aggregation assay may have widespread application in the study of low-affinity interactions between macromolecules.


Introduction
Interactions between protein domains, whether within or between proteins, are universal features of cellular physiology [1][2][3][4][5]. These interactions often occur with low affinity, reflecting the need for rapid changes in the interactions to facilitate dynamic regulation of cellular activities [4][5][6]. These features, the low affinity of the interactions and their regulation by changes in cytosolic environment, present considerable challenges when attempting to explore them by methods such as yeast two-hybrid (Y2H) [7], phage-display (PD) [8], immunoprecipitation (IP) [9] or analytical ultracentrifugation (AUC) [10]. Additional limitations of these methods include the need for specialized expertise (Y2H, PD) or equipment (AUC), or large amounts of material (IP, AUC), their inability to discriminate between direct and indirect interactions (IP, AUC) and the difficulty of replicating cytosolic conditions during the assay (Y2H, AUC).
Inositol 1,4,5-trisphosphate receptors (IP 3 R) are intracellular Ca 2+ channels. They are ubiquitously expressed in animal cells and mediate release of Ca 2+ from the endoplasmic reticulum (ER) in response to extracellular signals that stimulate formation of IP 3 [11]. As with all ion channels, activation of IP 3 R proceeds via re-arrangements of interactions between protein domains within the oligomeric channel. For IP 3 R, these conformational changes are initiated by IP 3 binding to the IP 3 -binding core (IBC, residues 224-604) of each of the four IP 3 R subunits, and then proceed via re-organization of intramolecular interactions between the IBC and suppressor domain (SD, residues 1-223) [12,13]. These conformational changes are proposed to disrupt contacts between the N-terminal regions of the four subunits and to culminate in opening of the Ca 2+ -permeable pore [13]. The versatility of IP 3 -evoked Ca 2+ signalling is increased further by the spatial organization of IP 3 R within ER membranes [14] and their association with a diverse array of additional proteins that regulate their activity [15,16]. The importance of interactions between protein domains for regulation of IP 3 R is clear and so is the need for simple, convenient assays to address them.
In this study, we developed an economical and rapid bead aggregation assay to study low-affinity interactions between proteins. Proteins are immobilized at high density on small beads so that individually weak interactions between protein partners on different beads collectively contribute to a multivalent interaction that causes stable aggregation of the beads. The assay is simple, inexpensive and rapid, and it requires only small amounts of proteins and a standard light microscope. It allows direct interactions between proteins to be quantified even when they occur with low affinity.

Results and Discussion
An assay for analysis of low-affinity interactions between macromolecules We sought to develop a simple and economical assay to analyze low-affinity interactions between proteins or other weakly interacting partners. We envisaged that by immobilizing proteins on the surface of a bead, many individually weak head-to-head interactions between proteins on different beads might collectively provide an interaction with sufficient avidity to cause the beads to become stably associated ( Figure 1A). We first tested the ability of such a bead aggregation assay to report interactions between molecules by immobilizing complementary anti-parallel DNA on two different populations of beads ( Figure 1B). Biotinylated double-stranded DNA (DNA-F) with a sticky end was immobilized on streptavidin (SVA)-coated beads and then incubated with a second population of beads coated with biotinylated dsDNA with complementary sticky ends (dsDNA-F9) (see Methods). Beads coated with either biotin-dsDNA-F or biotin-dsDNA-F9, where DNA can only weakly selfdimerize (with a predicted equilibrium dissociation constant, K D , of ,36 mM), did not aggregate. However, mixing the two populations of beads coated with complementary DNA (capable of forming dimers with a predicted K D of ,63 fM) caused them to aggregate ( Figures 1C and 1D). This was quantified by measuring the ratio of aggregated to single beads (see Methods). The results demonstrate that interactions between complementary partners can be quantified by this simple bead aggregation assay ( Figure 1D). The optimized protocol used to quantify aggregation of beads in subsequent assays is shown in Figures 2A-2D.

Bead aggregation assay applied to analysis of interactions between N-terminal fragments of IP 3 receptors
We next sought to apply the assay to analysis of protein-protein interactions using N-terminal domains of IP 3 R1 ( Figure 2E). These were chosen because cytosolic domains of IP 3 R are proposed to mediate interactions between IP 3 R [17][18][19], and binding of IP 3 to the NT (residues 1-604) initiates IP 3 R activation [12,13]. Biotinylated NT was immobilized on SVA-coated beads (see Methods). Silver staining confirmed that NT of the appropriate size was effectively immobilized and that no residual GST-NT (predicted size ,96 kDa) was detectable on the beads ( Figure 3A). NT-beads selectively bound 3 H-IP 3 with an affinity (pK D = 8.2460.27, n = 3, where pK D is the negative logarithm of the equilibrium dissociation constant, K D ) similar to that of NT in solution (pK D = 8.2860.09, n = 3) ( Figures 3B and 3C). These results suggest that NTs immobilized on SVA beads retain native conformations and accessible IP 3 -binding sites.
Beads without a surface-coating of biotinylated protein, or beads coated with either biotin-BSA or denatured biotin-NT did not aggregate, whereas beads with immobilized NT aggregated ( Figures 4A and 4B). Immobilization of denatured NT was confirmed by Western blotting ( Figure 4C). Although immobilized NT bound IP 3 appropriately ( Figure 3C), IP 3 had no significant effect on the aggregation of NT-beads ( Figures 4D  and 4E). These results suggest that interactions between native NT are sufficient to cause bead aggregation, but the interactions are unaffected by IP 3 binding under the conditions used for these analyses.

The suppressor domain of type 1 IP 3 receptor mediates interactions between N-termini
The suppressor domain (SD, residues 1-223) is essential for IP 3 R activation [20,21] and binding of IP 3 to the IP 3 -binding core (IBC, residues 224-604) rearranges the relationship between the IBC and SD [12,13]. We therefore assessed the interactions between isolated SD and IBC by immobilizing each on SVA beads. The NT, IBC and SD fragments were biotinylated with similar efficiency (Figures 5A and 5B). Furthermore, quantita-tive comparisons of the density of 3 H-IP 3 binding sites (B max ) derived from equilibrium competition binding to the NT and IBC ( Figure 5C) and quantification of the amounts of biotinylated proteins ( Figure 5A) established that there was no statistical difference in the stoichiometry of IP 3 binding (mol IP 3 /mol protein) for the two fragments ( Figure 5D). These results establish that the fragments used are likely to be similarly immobilized on SVA beads and, at least for the IBC and NT (the only fragments that bind IP 3 ), that each is similarly bioactive. Aggregation of SDbeads was similar to that of NT-beads, whereas IBC-beads did not aggregate ( Figures 5E and 5F). GST can itself dimerize [22] and might therefore have contributed to bead aggregation if residual GST-tagged SD or NT were present in the protein samples after purification. This is unlikely as none of the protein samples contained detectable GST-tagged fragments ( Figure 3A and Figure 5G). These results suggest that interactions between the NT are mediated by the SD ( Figure 5H).
The bead aggregation assay developed in this study is an economical and rapid method for detecting interactions between two ligands that should be generally applicable to analysis of lowaffinity interactions between macromolecules. The assay is best suited to qualitative assessment of such interactions and the effects of varying incubation conditions in parallel experiments. We have not attempted to define the absolute detection sensitivity of the assay, but we have shown that it resolves statistically significant differences between different immobilized ligands and that it succeeds in detecting specific interactions between the NT domains of IP 3 R that are too weak to detect by immunoprecipitation [23]. The structure of the NT docked into a low-resolution structure of native IP 3 R [13] indicates that SD-SD interactions within an IP 3 R tetramer are unlikely. Instead, our results suggest that NT domains of IP 3 R may interact in a head-to-head fashion, consistent with electron microscopy [19]. A physiological consequence of such head-to-head interactions may be the formation of ER stacks, which have been suggested to require the NT of IP 3 R1 [18]. Interactions between the NT of IP 3 R may also contribute to clustering of IP 3 R [24,25], and influence interactions of IP 3 R with modulatory proteins [15,16].

Materials and Methods
Expression and purification of IP 3 receptor fragments All IP 3 R fragments lack the S1 splice site, but they are numbered by reference to full-length rat IP 3 R1 containing the S1 splice site (GenBank accession number: GQ233032.1). N-terminal fragments were expressed with an N-terminal glutathione Stransferase (GST) tag followed by a PreScission cleavage site (LEVLFQGPLGS) to facilitate purification, and a C-terminal biotinylation sequence (EFGGGLNDIFEAQKIEWHE) for immobilization on SVA beads ( Figure 2E). Preparation of Nterminal fragments from constructs in the pGEX-6P2 plasmid (GE Healthcare, Cardiff, UK) was described previously [26]. The sequences of all constructs were verified.
The open reading frames of the tagged SD, IBC or NT were cloned into the expression vector pGEX-6P2 and transformed by heat-shock into AVB101 E. coli (Avidity, Aurora, Colorado, USA). These bacteria have an IPTG-inducible birA gene to allow expression of biotin ligase. Bacteria were incubated at 37uC for 16 h on LB-agar containing ampicillin (100 mg/mL). A resistant colony was then grown in 20 mL of ampicillin-containing LB medium in an orbital shaker (250 rpm, 37uC). After 12 h, 10 mL of the culture was diluted into ampicillin-containing LB medium (500 mL) and incubated (150 rpm, 22uC) until the optical density at 600 nm (OD 600 ) reached ,1 (,9 h). Protein expression was induced by addition of IPTG (0.5 mM), and the pellet (6000x g, 20 min) was collected after incubation for 20 h (150 rpm, 15uC). The pellet was washed with phosphate-buffered saline (PBS, 10 mL; 13.7 mM NaCl, 0.27 mM KCl, 1 mM Na 2 HPO 4 , 0.2 mM KH 2 PO 4 , pH 7.3, 4uC) and re-suspended in Tris-EDTA medium (TEM, 44 mL; 50 mM Tris-HCl, 1 mM EDTA, pH 8.3 at 4uC) containing protease inhibitor cocktail (Roche, Hertfordshire, UK; EDTA-free complete protease inhibitor cocktail, 1 tablet/50 mL). PopCulture (Merck Millipore, 10% v/v) and 2mercaptoethanol (1 mM) were added, and the suspension was incubated with lysozyme (100 mg/mL) and RNase (10 mg/mL) on ice for 30 min. The lyzate was sonicated on ice for 20 s and the supernatant was recovered (300006 g, 1 h). IP 3 R fragments were purified via the N-terminal GST tag. The supernatant (50 mL) was centrifuged to remove debris (60006 g, 30 min, 4uC), mixed with glutathione Sepharose 4B beads (1 mL of 50% slurry, GE Healthcare), incubated with gentle end-overend rotation (22uC, 30 min) and transferred to a PD-10 column (GE Healthcare). All subsequent steps were performed at 4uC. The beads were washed three times with PBS (5 mL) containing protease inhibitor cocktail, and three times with PreScission buffer (5 mL; 50 mM Tris-HCl, 1 mM EDTA, 150 mM NaCl, 1 mM dithiothreitol, pH 7.5). GST-PreScission protease mix (40 mL added to 460 mL of PreScission buffer; GE Healthcare) was then added to the beads. The column was sealed and incubated with gentle end-over-end rotation for ,4 h. The eluate containing purified biotinylated IBC, SD or NT was then collected and stored at 280uC. GST-tagged PreScission is retained by the glutathione beads.

Preparation of coated SVA beads and bead aggregation assay
The DNA sequences (Invitrogen) used were adapted from [27] (59-39, single-stranded sticky ends are underlined): biotin-dsDNA-F, biotin-ACCCTTCGCACAGTCAATCCAGAGAGCCCTGCCT-TTCATTACGATCATAACTTGG and biotin-dsDNA-F9, biotin-ACCCTTCGCACAGTCAATCCAGAGAGCCCTGCCTTTC-ATTACGACCAAGTTATGA. For analysis of DNA-induced aggregation of beads, streptavidin-coated magnetic T1 Dynal beads (SVA-beads, Invitrogen) were washed three times with TEM (25 mL) using a DynaMag-2 magnet (Invitrogen). The washed beads (10 pmol binding sites, ,300,000 binding sites/ bead) were then incubated with either biotin-dsDNA-F or biotin-dsDNA-F9 (20 pmol) for 30 min at 22uC with gentle rotation. For analyses of interactions between the DNA sequences, equal amounts of the two sets of coated beads were mixed and incubated in TEM (25 mL) for 30 min at 22uC with gentle rotation.
For analyses of interactions between N-terminal fragments of IP 3 R, washed SVA beads (10 pmol binding sites) in 25 mL of TEM were incubated with 20 pmol of NT-biotin, SD-biotin, IBC-biotin, BSA-biotin (Sigma-Aldrich) or denatured NT-biotin (85uC, 10 min) at 22uC with gentle rotation. After 30 min, coated SVA-beads were magnetically isolated, washed twice with TEM (50 mL) and re-suspended in TEM (25 mL). For imaging, SVA-   beads were diluted 10-fold in TEM, plated on glass-bottomed imaging dishes (MatTek, number 0 coverglass) and allowed to settle for 15 min. Differential interference contrast (DIC) images were captured using an iXon 887 EMCCD camera (512 pixels6512 pixels; Andor Technology, Belfast, Ireland) and an Olympus IX81 microscope with a x60 objective, and acquired using CellR imaging software (Olympus Europe, Hamburg, Germany).

Analysis of bead aggregation
The protocol for analysis of bead aggregation is summarized in Figure 2A. For each image, an intensity threshold was applied, which selects pixels that define beads ( Figure 2C). Ten single beads were randomly selected from each field and a one pixelthick outline was used to define the perimeter of each bead, from which its area was calculated (excluding the outline pixels). In subsequent analyses, particles with areas greater than the mean area of a single bead plus twice the standard deviation (the area threshold) were considered as aggregates. The aggregation ratio was then calculated from: