*2.1. The Structure of TRIOBP-1*

TRIOBP-1 consists of two major structured regions: a predicted Pleckstrin homology (PH) domain near the N-terminus and coiled coil domains that make up the C-terminal half of the protein (Figure 1b). These are separated by a linker region of approximately 100 amino acids, referred to as the "mid

domain" [13], which is predicted to be intrinsically disordered. Finally, TRIOBP-1 has an optionally translated disordered region at its extreme N-terminus, which is targeted to the nucleus of the cell [12]. This results from the existence of two different potential start codons, 59 amino acids apart from each other, and means that full length TRIOBP-1 can be either 593 or 652 amino acids in length [12]. The 593 amino acid long version of TRIOBP-1 was the first TRIOBP protein to be described, under the name TARA, for TRIO Associated Repeat on Actin [3] (also referred to as TAP68 [14]). This appears to be the more abundant species in many cell culture systems. The 652 amino acid long version may, however, be the principle TRIOBP-1 species in the human heart [15].

The presence of a PH domain near the N-terminus is strongly predicted [3], and it was confirmed that this region of TRIOBP-1 forms a compact folded domain [12]. Its structure has never been studied experimentally, but based on homology with other proteins, it seems to be a fairly typical PH domain with two extended unstructured loops sticking out of it (Figure 2). These loops consist predominantly of polar and charged amino acids. The second, and larger, of these loops is highly conserved in mammals (Figure 1c). The function of the PH domain is currently unknown, however, it likely acts as a protein–protein interaction domain. No interaction of TRIOBP-1 with phosphoinositides has been published, although this cannot be formally discounted.

**Figure 2.** A structural homology model of the PH domain of TRIOBP-1 (amino acids 64–172 of 652 amino acid TRIOBP-1). (**a**) Images of the model, with the first loop region displayed. (**b**) Amino acid sequence of the PH domain, with the first and second loop regions indicated. Coloring corresponds to the secondary structures seen in the molecular images. (**c**) Image of the model including a low quality prediction of the strength of the second loop region. Model generated using MODELLER 9.20 [16], based principally on the structure of the PH domain of DAPP1 (PDB ID: 1FAO), which includes sequence analogous to the first loop region. Shorter sections including the second loop were modeled with additional templates (PDB ID: 2DYN, 2D9Y, 3GOC, and 5YUG). Alignments were generated using CLUSTAL Omega 1.2.4 [8], and then optimized manually. Of the 20 models generated, the one with the lowest objective function score was visualized using YASARA 18.4.24 [17].

The C-terminal half of TRIOBP-1 is highly structured, having long been predicted to consist of coiled-coil (CC) domains (Figure 3a) [3]. Recent predictions suggest there to be approximately six CCs within this section of TRIOBP-1, which separate into two distinct domains: a larger central CC domain and a smaller C-terminal CC domain (Figure 3b) [12,14]. These two domains, and the central CC domain in particular, appear to be involved in many of the functions of TRIOBP-1 in the cell (Figure 3c). While the C-terminal CC domain is monomeric when expressed in isolation, the central CC domain forms an elongated hexamer, seemingly through distinct homodimeric and homotrimeric interactions (Figure 3d) [12]. The central CC domain is therefore responsible for the known oligomerization [22] of the full length TRIOBP-1 protein.

**Figure 3.** The structure of the coiled-coil regions of TRIOBP-1. All parts of this figure are to scale with each other. (**a**) Locations of predicted coiled-coils (CC). Solid filled boxes represent high confidence predictions, striped boxes represent lower confidence predictions, derived from PSIRPED [9]. CCs are colored based on their predicted inclusion in the central CC domain (blue) or C-terminal CC domain (red). Amino acid numbering from both the 593 amino acid and 652 amino acid TRIOBP-1 proteins are shown. Labeling of CCs is based on Bradshaw et al. [12] and differs from the numbering used by Katsuno et al. [18], who do not count the putative coiled-coil labeled here as CC1 in their numbering. Level of amino acid conservation is displayed using the same calculation and heat map as in Figure 1c. (**b**) Locations of constructs representing the central and C-terminal CC domains from two publications [12,14]. (**c**) Locations of regions of TRIOBP-1 involved in protein–protein interactions and functions [3,12,14,19,20]. Note that some proteins bind more than one region of TRIOBP-1. The only proteins so far reported to bind to TRIOBP-1 outside of these CC regions is TRIO, which binds to the mid domain between the central CC region and the PH domain [13]. The locations of two known phosphorylated residues and their associated kinases are also shown [14,21]. (**d**) Locations of fragments of TRIOBP-1 and the oligomeric states they adopt when expressed in isolation in vitro [12].

In addition to the main TRIOBP-1 species, which are approximately 70 kDa, smaller species have also been detected by western blotting, ranging in size from 45–60 kDa [15,23]. Based on the specificity of the antibodies used, these would be expected to represent the C-terminal 400–540 amino acids of TRIOBP-1, that is, the coiled-coil domains and variable amounts of the unstructured linker region, but not a complete PH domain. An additional splice variant, *TRIOBP-2*, has also been sequenced (annotated in genome assembly hg38), which encode the N-terminal sections of TRIOBP-1 including the PH domain and parts of the central coiled-coil domain. However, to date, this has not been thoroughly characterized.
