Most yeast SH3 domains bind peptide targets with high intrinsic specificity

Abstract

A need exists to develop bioinformatics for predicting differences in protein function, especially for members of a domain family who share a common fold, yet are found in a diverse array of proteins. Many domain families have been conserved over large evolutionary spans and representative genomic data during these periods are now available. This allows a simple method for grouping domain sequences to reveal common and unique/specific binding residues. As such, we hypothesize that sequence alignment analysis of the yeast SH3 domain family across ancestral species in the fungal kingdom can determine whether each member encodes specific information to bind unique peptide targets. With this approach, we identify important specific residues for a given domain as those that show little conservation within an alignment of yeast domain family members (paralogs) but are conserved in an alignment of its direct relatives (orthologs). We find most of the yeast SH3 domain family members have maintained unique amino acid conservation patterns that suggest they bind peptide targets with high intrinsic specificity through varying degrees of non-canonical recognition. For a minority of domains, we predict a less diverse binding surface, likely requiring additional factors to bind targets specifically. We observe that our predictions are consistent with high throughput binding data, which suggests our approach can probe intrinsic binding specificity in any other interaction domain family that is maintained during evolution.

Fig 1. General mechanisms to obtain binding specificity in domain families.

A. Domains may use the interaction with an extended region that goes beyond the canonical binding site to obtain intrinsic specificity (1). For example, the Abp1p SH3 domain binds extended target peptides (17 residues) and was shown to possess high intrinsic binding specificity [9, 10]. Domains may also achieve intrinsic specificity through non-canonical recognition via an alternative binding surface far from the canonical one. For example, Pex13 is a peroxisomal membrane protein that contains an SH3 domain that binds Pex14p via the canonical binding surface, however, it also binds Pex5p through an alternative non-canonical surface [11, 12]. Furthermore, intrinsic specificity may be achieved through replacing the canonical binding site with a non-canonical one (2) that would lead to negative selection (3) with respect to proline-rich peptides that bind SH3 domains. For example, Fus1 peptide targets do not contain a canonical PxxP motif thus minimizing cross reactivity to proline containing peptides [13]. Some domains may have potential for contextual specificity using adjacent domains (4). For example, at least 2 of the 3 adjacent SH3 domains of Nck are required to bind their targets [14]. Spatial and temporal separation mechanisms may be another contextual specificity mechanism (6). For example, in vitro, Fyn SH3 domain and CD2BP2 both bind and compete with each other for the proline region in the target protein CD2. However CD2BP2 localizes to the cytosolic compartment where it interacts with CD2 in T-cells, while Fyn is present permanently in the lipid raft fraction unable to compete [15]. In some cases, both intrinsic and contextual specificity mechanisms may be used by a domain, such as the Pex13p example above (5). We note here that contextual specificity has been used elsewhere to mean the extended regions of SH3 domain binding peptides, outside their core binding motif [16]. This definition does not pertain to contextual specificity as discussed within this study. Figure adapted from [17] and [7]. B. An example of an extended peptide-domain interaction. The Ark1 peptide is represented in stick and the SH3 domain from Abp1 uses space-filling. The red region is surface I and the blue region is surface II. W36 is represented as green and is on the boundary of the two surfaces. Adapted from [18] (pdb code 2rpn).

Fig 2. Example sequence conservation analysis for orthologs of Abp1 SH3 domain.

The residues are colored according to the residue equivalence groups defined for entropy and PSSM calculations. The species names end with a number that refers to their taxonomic group (S1 Fig and S1 Table). The SC value is calculated as (paralog entropy)/(ortholog entropy). A standard numbering system [46] for the core 60 SH3 domain residues is indicated on the top row as well as the residue number in the full length S.cerevisiae protein (fifth row). The paralog entropy is calculated from an alignment of the 28 SH3 domains in S.cerevisiae.

Fig 3. Specific conservation values for the yeast SH3 domain family.

A. Alignment of the core 60 positions colored by ortholog SC values as a heat map (red high and yellow low SC values, with domains sorted alphabetically). The average SC value across the family is indicated for each position at the bottom of the table, along with the paralog positional entropy, surface labels and secondary structure. Dark Boxes indicate the 2 principal loop regions where high SC values are found. B. Specific conservation across the domain. The line is set at an SC value of 1.7, which is considered a potential threshold for significant specific conservation (where ortholog conservation is almost twice that of paralog conservation).

Fig 4. Summary of sequence conservation found in the yeast SH3 domain family.

Both SI and SII alignments on the right are heat map colored by either ortholog entropy (SI) or SC values (SII), where red represent high or significant values and yellow as non-significant. As such, we define high average SC values for SI, SII and other (all other residues except SI and II) when ≥ 1.7. We define and count significant additional insertions at the N- and C-terminus, RT-loop, N-Src loop and distal loop when ortholog positional entropy values are ≤ 3.3. Information about the specific binding peptides identified from published binding data [41] is also indicated in the following 3 columns. The “#_peptides” column is the total number of specific peptides where at least 1 species domain family has a binding proportion ≥ 0.5. The “BF_specific” column is the average binding fraction for the most specific (best) peptide across available species. The “Species” column contains 2 numbers, the first is the number of species where the most specific peptide has a binding fraction ≥ 0.5. The second is the number of species where binding data could be collected. Gaps in binding data, indicate the domain was difficult to purify for 2 or more species. Interestingly, from this dataset, known biological peptides targets are sometimes ranked higher for a given domain according to binding fraction values as opposed to binding intensity values (the method used by the authors of the high-throughput study). For example, the Ark1 peptide target (DKKTKPTPPPKPSHL) for Abp1 [10] ranks 2nd using binding fraction and 10th using intensity alone. In the case of the Pbs2 peptide (IVNKPLPPLPVAGSS) target for Sho1 [9] and the Cla4 peptide target (AHFQPQRTAPKPPIS) for Nbp2 [50] both intensity and binding fraction rank the peptides in top positions. Residues in the SI alignment that have a dark border are highlighted as being conserved and unique within the family.

Fig 5. SI and SII PSSM for yeast paralog alignment (28 domains) and example ortholog alignments for Fus1 (16 species) and Bud14 (29 species).

Total occurrence for each amino acid group for each position is indicated and colored as yellow (low) to red (high). Residues are grouped into SI (left) and SII (right). Dark outlined regions indicate most common preference for the family (≥ 20 occurrences). Overall, for SI there is a family preference for aromatic residues except the less conserved positions 9, 52 and 53. Notable exceptions include Fus1 that has cysteines at positions 37 and 54 (which are usually in the FWYH group). For SII, there is a loose family preference for polar/acidic residues except at position 49 where hydrophobic residues are found. The extent of conservation in the orthlog alignments in SI and SII vary, with a much greater variation seen in SII PSSMs. PSSMs for all domains (showing both complete domain sequence and only surface I/II) can be found in S4 File.

Fig 6. SI and SII family dendograms.

Clustering was based on SI (left) and SII (right) PSSMs. For SI dendogram, there is more significant clustering, which appears to concentrate domains that bind class I peptides into the red group and domains that bind class II peptides into the magenta group.