A specificity map for the PDZ domain family

Abstract

PDZ domains are protein-protein interaction modules that recognize specific C-terminal sequences to assemble protein complexes in multicellular organisms. By scanning billions of random peptides, we accurately map binding specificity for approximately half of the over 330 PDZ domains in the human and Caenorhabditis elegans proteomes. The domains recognize features of the last seven ligand positions, and we find 16 distinct specificity classes conserved from worm to human, significantly extending the canonical two-class system based on position -2. Thus, most PDZ domains are not promiscuous, but rather are fine-tuned for specific interactions. Specificity profiling of 91 point mutants of a model PDZ domain reveals that the binding site is highly robust, as all mutants were able to recognize C-terminal peptides. However, many mutations altered specificity for ligand positions both close and far from the mutated position, suggesting that binding specificity can evolve rapidly under mutational pressure. Our specificity map enables the prediction and prioritization of natural protein interactions, which can be used to guide PDZ domain cell biology experiments. Using this approach, we predicted and validated several viral ligands for the PDZ domains of the SCRIB polarity protein. These findings indicate that many viruses produce PDZ ligands that disrupt host protein complexes for their own benefit, and that highly pathogenic strains target PDZ domains involved in cell polarity and growth.

Figure 1. PDZ Domains Are Highly Specific across Multiple Ligand Positions

A total of 72 PDZ domains (each with greater than ten peptides) corresponding to 2,998 ligands were analyzed to assess specificity for each ligand position. Specificity was measured using the SP score, which ranges from zero (least specific) at a given ligand position to one (most specific). Bars are colored as follows: all PDZ domains (black), human (grey), or worm (white). (A) Specificity profile for a representative PDZ domain (C34F11.9a-1) with SP scores shown above each ligand position. (B) Fraction of PDZ domains exhibiting significant specificity (SP > 0.2) at each ligand position. (C) The mean SP value at each ligand position. (D) The distribution of total SP (SP ^t) summed over all ligand positions.

Figure 2. Specificity Map Classifies the PDZ Domain Family

All 82 PDZ domains studied were clustered to create a specificity map, which was used as a guide to manually define PDZ specificity classes. Of the 82 domains, 73 are assigned to one of 16 classes, labeled to the right of each domain name. For consistency with the established PDZ domain classification system [16], each class is denoted by a numeral based on the specificity for position −2, followed by a letter to account for specificity across the rest of the binding site. C. elegans domains are highlighted in yellow. Sets marked with identical Roman numerals in parentheses are homologous PDZ domains in human/worm orthologs. Domains that exhibit unique specificities not part of any class are denoted by asterisks (*). The 16 classes are defined by the following C-terminal motifs: 1a (φ[K/R]XSDV); 1b (Ω[R/K]ET[S/T/R/K]φ); 1c (φφETXL); 1d (ETXV); 1e (TWΨ); 1f (ΩΩTWΨ); 1g (φφφ[T/S][T/S]ΩΨ); 1h (φφ[D/E][T/S]WΨ); 2a (FDΩΩC); 2b (WXΩFDV); 2c (WΩφDΨ); 2d (φφX[E/D]φφφ); 2e (φφφφ); 2f ([D/E]φΩφ); 3a (WΩ[S/T]DWΨ); 4a (ΩφGWF); φ, hydrophobic (V, I, L, F, W, Y, M); Ω, aromatic (F, W, and Y); Ψ, aliphatic (V, I, L, and M); and X, nonspecific.

Figure 3. Distinct Specificities of PDZ Domain Binding Sites

The specificity profiles of 72 PDZ domains reveal eight, seven, eight, seven, five, and three distinct specificities for ligand positions 0, −1, −2, −3, −4, and −5, respectively. At each position, distinct specificities are shown (magenta) with either the single-letter amino acid code or symbols, as follows: +, positive charge; −, negative charge; φ, hydrophobic (V, I, L, F, W, Y, and M); Ψ, aliphatic (V, I, L, and M); and Ω, aromatic (F, W, and Y).

Figure 4. Sequence Determinants of PDZ Domain Specificity

Heat map summary of the effects of mutations on the specificity of ERBB2IP-1. Each row represents one mutant, ordered by PDZ domain binding-site position (labeled to the right of each set of rows), and each column represents one ligand position. Mutations were chosen to represent the diversity of amino acids in 82 natural PDZ domains for which we have phage data. To minimize potential destabilization caused by structurally deleterious mutations, selections were performed at 4 °C, and under these low stringency conditions, the wild-type specificity profile, shown at top left, was somewhat less specific than that at room temperature (Figure 2). The mutation listed to the left of each row, at the PDZ domain position listed at the right according to a structure-based nomenclature [55], causes a change in specificity, shown in each row. The blue-to-red gradient indicates increasing difference relative to wild type, normalized per column with significant differences highlighted in green (greater than one standard deviation away from the mean difference over the column). Selected mutant profiles are highlighted (depicted as sequence logos to the left of the corresponding row), with significant specificity changes in the logo boxed in red. Structures of ERBB2IP-1 with a bound peptide ligand [36] are shown with mutated positions depicted as spheres. Red side chains indicate ligand positions for which specificity is altered by mutations at PDZ positions shown as red spheres.

Figure 5. Mutations Affecting PDZ Domain Specificity

ERBB2IP-1 (grey) is shown with a bound peptide ligand (WETWV_COOH; cyan) (PDB entry 1N7T) [36]. PDZ domain binding-site positions that were subjected to mutagenesis are shown as spheres. In each panel, PDZ domain positions at which mutations affected specificity for the indicated ligand position are colored red and other mutagenized positions are colored green. PDZ domain positions are labeled in black according to a structure-based nomenclature [55].

Figure 6. Specificity Profiles of Orthologous Domains Are Highly Conserved

All worm and human ortholog pairs with mapped PDZ domains in our dataset are shown. The domain architecture, as defined by SMART [33], is shown for each worm (top) and human (bottom) protein in an ortholog pair. The specificity profiles defined by peptide phage display are shown below or above the worm or human PDZ domains, respectively. The name and length of each protein is indicated on the left or right, respectively. The orthologous protein pairs are drawn to scale. The following protein pairs could be unambiguously identified as orthologs on the basis of common domain architecture and high sequence identity: (A) C34F11.9a/DVL2, (B) F54E7.3/PARD3, (C) Y54G11A.10/LIN7A, (D) C25F6.2a/DLG1, (E) W03F11.6/MLLT4, and (F) F17E5.1a/CASK.

Figure 7. PDZ Domain Sequence Identity Accurately Predicts Binding Specificity

(A) ERBB2IP-1 structure (grey) is shown with a bound peptide ligand (WETWV_COOH; colored) [36]. PDZ domain binding site positions are shown as spheres, and positions that were analyzed by mutagenesis are colored green. PDZ positions are labeled in black according to a structure-based nomenclature [55], and peptide positions are labeled in red. We defined the PDZ binding site as 17 residues that make contact with the ligand (closer than 4.5 Å) in at least one of nine different structures (PDB entries 1N7T, 2H2B, 2H2C, 1I92, 2HE2, 1BE9, 2GZV, 1IHJ, and 1N7F). (B) The relationship between binding-site sequence identity and specificity profile similarity. Each point represents a pair of PDZ domains from our mapped set. Red circles represent pairs assigned to the same class, as defined in our specificity map, and blue squares represent all other pairs. The lower-right quadrant, absent of data points, contains an example for one pair of PDZ domains (ERBB2IP-1 and LRRC7–1), which exhibit a specificity profile similarity of 0.95 and a binding-site sequence identity of 0.88 (sequence mismatches are shown in red).

Figure 8. Prediction of PDZ Domain Specificity

A network view of predicted PDZ domain specificities. Worm and human PDZ domains are shown as blue or pink nodes, respectively. Diamonds denote domains with experimentally phage-mapped specificity profiles, and circles denote domains with predicted specificity profiles. Lines connect domains with greater than 70% sequence identity in the binding site, and line width is proportional to sequence identity. Connected domains are predicted to have high specificity profile similarity scores (>0.83). Network was created using Cytoscape 2.5 [54].

Figure 9. Viral Proteins Interfere with Host Cellular Function by Targeting the PDZ Domains of SCRIB

(A) Many viral proteins bind SCRIB PDZ domains. Affinities were determined as IC₅₀ values for peptides representing viral C termini binding to SCRIB PDZ domains and the first PDZ domain of ZO-1 (TJP1–1) [20]. Ligand sequence positions that match the specificity profiles for SCRIB, TJP1–1, or both, are colored green, blue, or red, respectively. Orange and yellow indicate high-affinity (IC₅₀ < 10 μM) or moderate-affinity (IC₅₀ > 10 μM) interactions, respectively. Asterisks (*) indicate no detectable interaction (IC₅₀ > 500 μM). Double asterisks (**) indicate influenza A strain designations [13], rather than RefSeq accession numbers. (B) Loss of the late phase of T cell polarization induced by our designed synthetic peptide that targets SCRIB PDZ domains 1, 2, and 3. The receptor Crtam interacts with the PDZ domains of SCRIB to control cell growth and maintain polarity of T cells [40]. These effects are reversed by the addition of our designed peptide (Π-RSWFETWV, peptide 14) that binds with high affinity to the SCRIB PDZ domains, but not by a designed nonbinding peptide with mutations at the 0 and −2 positions (Π–RSWFEAWA, peptide 15). The symbol Π denotes the internalization sequence from the Antennapedia protein (RQIKIWFQNRRMKWKK), which has been shown to be internalized into cells [56]. Naive Crtam^−/− CD4 T cells were electroporated with plasmid DNA expressing Crtam or a mock DNA control. Cells were treated with synthetic peptides (1.0 μM) and stained for Talin, a marker for the leading edge of polarized T cells [40]. (C) Our designed SCRIB PDZ-binding peptide (peptide 14) triggers T cell proliferation. Cells were treated with plasmid DNA and peptides, as described in (B), and cellular proliferation was measured by the incorporation of [³H]-thymidine. Data are representative of three independent experiments. Error bars indicate the standard deviation (SD). The p-value was determined by statistical analysis performed with a control using the Dunnett method.