A dynamic view of domain-motif interactions

Abstract

Many protein-protein interactions are mediated by domain-motif interaction, where a domain in one protein binds a short linear motif in its interacting partner. Such interactions are often involved in key cellular processes, necessitating their tight regulation. A common strategy of the cell to control protein function and interaction is by post-translational modifications of specific residues, especially phosphorylation. Indeed, there are motifs, such as SH2-binding motifs, in which motif phosphorylation is required for the domain-motif interaction. On the contrary, there are other examples where motif phosphorylation prevents the domain-motif interaction. Here we present a large-scale integrative analysis of experimental human data of domain-motif interactions and phosphorylation events, demonstrating an intriguing coupling between the two. We report such coupling for SH3, PDZ, SH2 and WW domains, where residue phosphorylation within or next to the motif is implied to be associated with switching on or off domain binding. For domains that require motif phosphorylation for binding, such as SH2 domains, we found coupled phosphorylation events other than the ones required for domain binding. Furthermore, we show that phosphorylation might function as a double switch, concurrently enabling interaction of the motif with one domain and disabling interaction with another domain. Evolutionary analysis shows that co-evolution of the motif and the proximal residues capable of phosphorylation predominates over other evolutionary scenarios, in which the motif appeared before the potentially phosphorylated residue, or vice versa. Our findings provide strengthening evidence for coupled interaction-regulation units, defined by a domain-binding motif and a phosphorylated residue.

Figure 1. Coupling between phosphorylation events and domain-binding motifs.

For each domain family (SH2, WW, PDZ and SH3), the bars denote the percent of motifs found to be phosphorylated either within or near them. Solid-colored and empty rectangular bars represent intra-motif phosphorylation and near-motif phosphorylation, respectively. All motifs are derived from the high reliability dataset, while phosphorylation events are derived from three data sets: LTP (low throughput evidence only), HTP (phosphorylation events based on evidence from high-throughput resources), and LTP+HTP (any type of evidence). Asterisks represent statistically-significant results (Methods, Table 1 and Table S1).

Figure 2. Phosphorylation events as double switches.

(A) A protein (black horizontal line) includes a segment that matches two sequence patterns: the first is typical for SH3 domain binding (green), and the second typifies SH2 domain binding (red). The non-phosphorylated form binds SH3 and not SH2 (upper), while phosphorylation inverts the binding preferences (lower). (B) Specificity switches within the PDZ domain family. A protein (black horizontal line) includes a segment that may bind distinct PDZ domains (upper). The non-phosphorylated form binds PDZ^a and not PDZ^b, while phosphorylation inverts these binding preferences (lower).

Figure 3. Dual sequence patterns used for the identification of potential double switches in human proteins.

Column titles include sequence patterns for motifs that bind SH3 or class I WW domains (in red), and row titles include sequence patterns for motifs that bind different types of SH2 domains, upon motif phosphorylation (in blue). Each table cell includes a merged sequence pattern that hints at a dual binding potential of the motif to both SH2 and SH3 (or WW) domains. The columns under class I WW and SH3-1 titles represent the strict analysis scheme. Sequence patterns were extracted from the ELM database . (i) An example for a dual motif. The PP.Y.N. sequence pattern is composed of the SH2^Grb2 Y.N. and the class I WW PP.Y patterns. (ii) Note that this sequence pattern encompasses seven positions.

Figure 4. Step-wise appearance of motifs and potential phosphorylation sites.

(A) The motif is older than the potential phosphorylation site. The human CDK inhibitor 1B (top line) includes an SH3-binding motif (RxxK, highlighted in red) and a proximal tyrosine that may affect the motif's interaction potential upon phosphorylation , (highlighted in cyan). The sequence pattern is conserved from C. elegans to human, but the tyrosine is conserved only between rat and human. This suggests that an old domain-binding motif has gained phospho-regulation in more recent organisms. Protein accessions are according to the Uniprot or Ensembl databases. (B) Potential phosphorylation site is older than the motif. The human Tau protein includes an SH3-binding motif (PxxP) and a proximal threonine that inhibits the motif's interaction potential upon phosphorylation . This phosphorylation was also shown to induce a conformational change that unlocks the closed form of the protein . The motif is conserved from X. tropicalis to human, while the potential phosphorylation site may have appeared earlier in evolution (present in D. melanogaster). This suggests that the domain-binding potential was established close to already functional phosphorylation.

Figure 5. Phylogenetic traces of PDZ interaction-regulation unit evolution.

This matrix summarizes the results for units of PDZ binding motifs and near-motif phosphorylation. The eukaryotic evolutionary tree is depicted above and left to the matrix (abbreviations below). The rows indicate the organism in which the motif probably appeared. The columns indicate the organism in which a potentially phosphorylated residue appeared. The order in which the motif and potentially phosphorylated residue appeared can thus be deduced from the matrix cells. For instance, the brown-framed cell represents the three cases in which the motif appeared in D. melanogaster and the potentially phosphorylated residue appeared in chicken. Accordingly, all cells below the diagonal (cyan) represent cases in which the potentially phosphorylated residue appeared after the motif. The diagonal cells represent cases in which the motif and the potentially phosphorylated residue appeared together. The cells above the diagonal represent cases in which the motif appeared after the potentially phosphorylated residue (red). Organism abbreviations: CHIMP- p. troglodytes, MOUSE- m. musculus, RATUS- r. norvegicus, BOVIN- b. taurus, CHICK- g. gallus, XENTR- x. tropicalis, DANRE- d. rerio, CIONA- c. intestinalis, DROME- d. melanogaster, ANOGA- a. gambiae, CAEEL- c. elegans, YEAST- s. cerevisiae, DICDI- d. discoideum, ARATH- a. thaliana and PLAFA- p. falciparum.

Figure 6. Frequency of various phylogenetic traces of motif-phosphorylation coupling.

The stacked-bar graph details the relative frequency of the three possible phylogenetic traces of the interaction-regulation units (for either intra-motif phosphorylation or near-motif phosphorylation sites): (i) co-appearance of the motif and the potentially phosphorylated residue in the same organism (grey), (ii) the motif appeared before the potentially phosphorylated residue (cyan) (iii) the potentially phosphorylated residue appeared before the motif (red). For each domain we tested if the distribution of the various scenarios deviates from random by a χ² test. Asterisks denote statistically significant results (based on Table S6).