Protein context shapes the specificity of SH3 domain-mediated interactions in vivo

Abstract

Protein-protein interactions (PPIs) between modular binding domains and their target peptide motifs are thought to largely depend on the intrinsic binding specificities of the domains. The large family of SRC Homology 3 (SH3) domains contribute to cellular processes via their ability to support such PPIs. While the intrinsic binding specificities of SH3 domains have been studied in vitro, whether each domain is necessary and sufficient to define PPI specificity in vivo is largely unknown. Here, by combining deletion, mutation, swapping and shuffling of SH3 domains and measurements of their impact on protein interactions in yeast, we find that most SH3s do not dictate PPI specificity independently from their host protein in vivo. We show that the identity of the host protein and the position of the SH3 domains within their host are critical for PPI specificity, for cellular functions and for key biophysical processes such as phase separation. Our work demonstrates the importance of the interplay between a modular PPI domain such as SH3 and its host protein in establishing specificity to wire PPI networks. These findings will aid understanding how protein networks are rewired during evolution and in the context of mutation-driven diseases such as cancer.

Fig. 1. Definition of SH3-dependent PPIs in vivo and their functional impact.

a CRISPR–Cas9 SH3 editing approaches to study SH3 domains in living cells. For domain deletion (stuffing), SH3 sequences are replaced by a stuffer sequence encoding a linker using CRISPR–Cas9. SH3 dependency can then be tested. b PPIs of WT and SH3-deleted proteins. Colors represent different types of PPIs (assessed in quadruplicate). c Matrix similarity scores (MSS) of SH3-specific position weight matrices (PWMs) for prey corresponding to different types of PPIs from b. SH3-dependent PPIs are enriched for SH3-binding motifs relative to the Random, SH3-independent, and SH3-inhibited PPIs (p = 1.5 × 10⁻¹⁷, p = 0.0038 and p = 2.1 × 10⁻⁰⁸, Mann–Whitney test, one-tailed, n = 10,000 random protein sequences, n = 95 preys for SH3-independent PPIs, n = 116 preys for SH3-inhibited PPIs, n = 249 preys for SH3-dependent PPIs). For example, the proportion of high MSS SH3-binding motifs (95th percentile of the MSS distribution for random peptides) determined is 11/95 for SH3-independent, 5/116 for SH3-inhibited, and 75/249 for SH3-dependent partners. d PWM MSS for sequences of SH3-dependent preys (as in 1C), and for PWMs randomly assigned to SH3s (p = 0.25, Mann–Whitney test, two-tailed, n = 10,000 random assignments of an SH3 PWM to an SH3 prey, n = 249 preys for SH3-dependent PPIs). e Growth of gene- or SH3-deleted strains compared to WT under 51 stress conditions. For Bzz1, Bem1, and Sla1, the data shows deletion of all SH3s. f Same as panel e, but for individual or combinations of SH3 deletions. D represents SH3 replaced with a linker sequence. Strains were grown in 12 replicates (e, f analysis performed with n = 12 independent colony growth per strain). For every boxplot, the median is represented as a bold center line and hinges are for the 25th and 75th percentiles (first and third quartiles). Whiskers extend from the hinges to maximum 1.5 times the Q3–Q1 interquartile range. For c, d, outliers are represented as black dots. Source data are provided as a Source Data file. See also Supplementary Data 1 and Supplementary Data 2.

Fig. 2. SH3 domains are rarely sufficient to establish their endogenous specificity in their host protein.

a In vivo SH3 domain swapping. The stuffer DNA from an SH3-deleted strain is replaced by SH3s from other genes. The consequences of SH3 swapping on the host protein PPIs are illustrated. b PPIs of Abp1 with its endogenous SH3 domain or with an SH3 from other yeast proteins or from its human orthologs. Blue shades correspond to PPI interaction scores measured by DHFR-PCA (PCA score). A scaled cartoon representation of Abp1 is shown above the heatmap. Bold SH3 domains represent human orthologous, orange SH3s are controls, and green is for the SH3-deleted protein. Each PPI was assessed in quadruplicate. c Cophenetic correlation between Abp1 SH3 swapped interaction clusters (based on the results of the panel b, n = 4 independent DHFR-PCA colony growth per Bait–Prey pairs) with the SH3 domain sequence similarity clusters. The empirical p value (p = 0.005) was obtained from permutation. d Number of PPIs gained by Abp1 upon domain swapping. PPIs originally detected as dependent on the domain when present in its host protein are shown in blue (analysis performed with n = 4 independent DHFR-PCA colony growth per Bait–Prey pairs). e Sla1 PPIs with the SH3 from Abp1 inserted at each of the three SH3 positions. Each SH3 is either the WT Sla1 domain (WT|WT|WT) or is Abp1 SH3 (Abp1|WT|WT is Sla1 with its SH3-1 swapped with Abp1 SH3). PPIs were measured in quadruplicate. Source data are provided as a Source Data file. See also Supplementary Data 1.

Fig. 3. SH3 domain sequences affect both SH3-dependent and -independent PPIs.

a The stuffer DNA of an SH3-deleted strain is replaced by a library of single mutants. Categories of mutations based on their effect on PPI types are shown. b, c Impact of mutations on binding represented in terms of deep mutational scanning (DMS) score. b Tolerance to SH3 mutations of the SH3-dependent interaction Abp1-Hua2. Blue: low DMS score, i.e., reduced interaction strength, red: increased strength. The secondary structure of the SH3 is shown above. The black bars on the top represent the level of conservation across the 27 S. cerevisiae SH3s. Positions in bold represent residues in contact with the ligand (see also Supplementary Fig. 3H). c Same as panel b, but for the SH3-independent Abp1-Lsb3 PPI. The data shows the average of two biological replicates. d DMS scores for the two PPIs. DMS scores below the first percentile of the synonymous variant (dashed line) are defined as destabilizing the PPI. Four categories of mutants are shown: (1) destabilizing both PPIs, (2) specifically destabilizing Hua2 PPI or (3) Lsb3 PPI, and (4) neutral or slightly increasing binding. The minimum distance relative to a binding peptide is shown in blue (Ark1 peptide). e Minimum distance to the Abp1-bound peptide and (f) the relative solvent accessibility (RSA) for each category. Dest is for destabilizing mutations. For e, f, n = 349 neutral amino acid mutations, n = 98 amino acid mutations that destabilize Lsb3 only, n = 126 amino acid mutations that destabilize Hua2 only, n = 528 amino acid mutations that destabilize both Lsb3 and Hua2. For the boxplots, the median is represented as a bold center line and hinges are for the 25th and 75th percentiles (first and third quartiles). Whiskers extend from the hinges to a maximum of 1.5 times the Q3–Q1 interquartile range and outliers are represented as black dots. Source data are provided as a Source Data file.

Fig. 4. Shuffling SH3 positions alters Sla1 interactome in vivo and impacts its function in endocytosis.

a Domains are moved from one position to another in proteins containing multiple SH3s (domain shuffling). Possible outcomes on PPIs are presented. b Number of PPIs detected relative to WT Sla1 for SH3-shuffling or -deletion per SH3 position (measured in quadruplicates). c Function of Sla1 as an adaptor linking early proteins with the actin machinery in clathrin-mediated endocytosis. d PPIs of Sla1 SH3-shuffled or -deleted with clathrin-mediated endocytosis-related partners. The color code represents the strength of the PPIs (PCA score). Sla1 1|2|3 represents the WT protein. e Schematic of Sla1 foci assembled at the cell membrane and their movement toward the center of the cell during internalization before disassembly, in time. f Representative microscopy images of cells expressing different Sla1-GFP proteins at multiple time points. Foci from WT Sla1-GFP, the negative control D|D|D, and a Sla1-GFP shuffle (3|1|2) are shown. Green: first time frame; red: others. A merge with the starting point for each time frame is shown. g Average effective distance traveled by Sla1-GFP particles towards the centroid of the cell for the different proteins. Color transparency represents the proportion of events that are not completed yet. Plus signs (+) represent the moments in time when 95% of the foci have disassembled. h Average number of complete or incomplete events, i.e., that were not completed before the end of the image acquisition period. For panels f–h, the numbers of Sla1-GFP foci that were analyzed are n = 3286 for Sla1 1|2|3, n = 473 for Sla1 D|D|D, n = 686 for Sla1 2|1|3, n = 963 for Sla1 1|3|2, n = 652 for Sla1 3|2|1, n = 745 for Sla1 2|3|1 and n = 994 Sla1 3|1|2). Source data are provided as a Source Data file. See also Supplementary Data 1.

Fig. 5. Human NCK2 SH3 domain shuffling alters its interactome in cells and its ability to phase separate.

a–c AP-SWATH quantitative MS data of NCK2 PPIs (duplicates, PPIs with a SAINT analysis FDR < 1%, see “Methods”). a Scaled cartoon representation of NCK2 with its domains. Log2 of the average spectral counts (SWATH score) for the PPIs are shown. Baits were clustered based on the similarity of their PPI profiles. The color code represents the PPI strength (SWATH score). The bait nomenclature 1|2|3 is for WT NCK2 and D|D|D is for the triple SH3-inactive negative control (W38K/W148K/W234K). b, c NCK2 PPIs for partners with known binding sites on NCK2. The spectral count from the two biological replicates was compared to the WT NCK2 score (Ratio NCK2mut/NCK2WT). The log2 average ratio is shown (WT NCK2 ratio = 0). Error bar represents the mean plus and minus one standard deviation (n = 2 independent biological experiments). d Fluorescence polarization dissociation constants (K_D) for NCK2 full-length recombinant proteins with an SH3-1 direct (CD3Ɛ) or SH3-2 direct (PAK1) partner (triplicates, K_D error values represent plus and minus one SE). e NCK2 phase separation after 24 h of incubation. Soluble and phase-separated (pellet) proteins were quantified via Coomassie staining (typical replicate shown above). The proportion of proteins in the pellet compared to the total protein content (soluble + pellet) is shown for the seven replicates. Proportions of phase-separated proteins were compared to WT NCK2 (pairwise one-way ANOVA, p values: 2|1|3 = 0.90, 1|3|2 = 0.00067, 3|2|1 = 0.41, 2|3|1 = 0.60 and 3|1|2 = 0.0054). For the boxplot, the median is represented as a bold center line and hinges are for the 25th and 75th percentiles (first and third quartiles). Whiskers extend from the hinges to a maximum of 1.5 times the Q3–Q1 interquartile range. The black dots represent the different data points. Source data are provided as a Source Data file. See also Supplementary Data 3.