Research ArticleMOLECULAR BIOLOGY

Stitching the synapse: Cross-linking mass spectrometry into resolving synaptic protein interactions

See allHide authors and affiliations

Science Advances  19 Feb 2020:
Vol. 6, no. 8, eaax5783
DOI: 10.1126/sciadv.aax5783
  • Fig. 1 Overview of XL-MS workflow and results.

    (A) Schematic workflow of XL-MS and its applications. ER, endoplasmic reticulum. (B) Pie chart showing the number of cross-links identified between cytoplasmic and extracellular regions of proteins (dataset 1). (C) Sunburst plot showing the annotation in synaptic functions of the cross-linked proteins identified [biological processes SynGO terms (11)]. Inner rings are parent terms of more specific child terms in the outer rings, color coded according to enrichment Q value. Notably, a wide and significant coverage of synapse-specific proteins was found distributed across both pre- and postsynapse functions. (D) Boxplot showing the distribution of protein abundances in different categories: Entire synapse, all proteins identified in standard proteomic analysis; All cross-linked proteins, proteins identified in our XL-MS dataset 1; Interprot. only, proteins involved in only interprotein cross-links; Intraprot. only, proteins involved in only intraprotein cross-links; and Protein with both, proteins involved in both intra- and interprotein cross-links. The median abundance for cross-linked proteins is three times higher than the synaptic proteome and 10 times higher for proteins involving both intra- and interprotein links, showing the influence of protein abundance in the detection of cross-links. Number of proteins and the median abundance are indicated in each box. Protein abundance data were obtained from (12). (E) Distribution of Cα-Cα distances of cross-links from eight selected protein complex structures (fig. S2, A to C).

  • Fig. 2 Conformational modeling of Camk2 kinase domains.

    (A) Cross-link mapping on the extended (PDB 5U6Y), the compact (PDB 3SOA), and the autoinhibited conformation of Camk2 (shown as autoinhibited dimer of kinase domains, PDB 2BDW and fig. S2D). Two highlighted monomers are shown in green and orange, respectively. For clarity, cross-links are mapped onto the monomer in green only. (B) Distribution of Cα-Cα distances of cross-links from Camk2 mapped onto the extended state (PDB 5U6Y). The hatched pattern indicates cross-links exceeding DSSO maximal distance restraint. (C) Illustration of domain flexibility and alternative conformational states of Camk2. In all conformational states, six monomers of Camk2 (half of the complex) are presented. Proposed models of a partial compact state (based on PDB 5U6Y and 3SOA) and a transient intermediate state between the extended and compact state (snapshot of morph shown in movie S1) are depicted. Cross-links between K258 (a major cross-linking site in the kinase domain, shown as blue sphere) and other lysines (red spheres) are highlighted. Cross-links are represented as the Euclidean distance between Cα atoms of the cross-linked residues, indicated in red (if below DSSO maximal distance restraint) or in magenta (if exceeding DSSO maximal distance restraint).

  • Fig. 3 Characterization of XL-based protein interaction network.

    (A) Core component and separate modules of the combined network (dataset 1). Nodes represent proteins, and edges show the interactions identified by XL-MS. Nodes are color coded based on functional clusters generated by unsupervised edge-betweenness clustering and annotated by GO enrichment analysis. Disconnected modules were grouped using DAVID Gene Functional Classification, and each group was subsequently annotated. Edges in red indicate protein-protein interactions (PPIs) reported in the literature (high confidence in at least one database of STRING, InWEB, and BioGRID). Sizes of the nodes are proportional to the protein abundances (log2 scale). Widths of the edges are proportional to the number of unique Lys-Lys cross-links. Insert represents the enlarged cluster annotated as intrinsic component of postsynaptic density membrane, with the edges labeled according to the number of samples in which the PPI was identified. Network representations of individual experiments are provided in figs. S3 to S6. dATP, 2′-deoxyadenosine 5′-triphosphate. (B) Distribution of the number of cross-linked proteins pairs annotated within the same GO group in the XL-based network (labeled in orange) and a randomly rewired network (100 iterations, labeled in light blue). (C) Distribution of the number of cross-links found between protein pairs present in at least one database (DB) with high confidence (labeled in red) or not present in any of the three databases (i.e., STRING, InWEB, and BioGRID; labeled in black). (D) Boxplot showing the distribution of protein abundances in different categories: xlink, cross-linked proteins identified in this study; and public DB, protein interactions present in at least one public database with high confidence (i.e., STRING, InWEB, and BioGRID). The number of proteins and median abundance are indicated in each box. Protein abundance data were obtained from (12).

  • Fig. 4 Peptide array analysis of selected cross-linked proteins.

    (A) Cross-link maps of selected SNARE proteins (dataset 1). Protein domains are depicted based on UniProt database and previous publications [*, from (31)]. Identified cross-links located in unique and shared peptides of Stx1a and Stx1b are represented as solid and dotted lines, respectively. (B) Results of Stxbp1 and control peptide array experiments. Examples of specific, nonspecific, and no binding signals are depicted. Arrowheads indicate antibody positive controls. Quantification and peptide sequences (two independent and two technical replicates) are described in fig. S9B and table S4. (C) Comparison of binding regions determined by peptide array (boxes colored in red) and XL-MS (edges). The number of samples in which the cross-links were identified is indicated on the edges (dataset 1). For peptide array assays, full-length proteins are shown as circles, and the protein used to generate peptide sequences in the arrays (Stx1b) is represented as sequence bars. Yellow edges match arrows with the same color in (D). (D) Cross-link mapping of Stxbp1 interactions. Stxbp1 (green) and Stx1 (blue) cross-linked sites were mapped onto the Stxbp1-Stx1 complex structure (PDB 3C98; Stx1 in closed conformation). Cross-links are indicated in red (if below DSSO maximal distance restraint) or in magenta (if exceeding DSSO maximal distance restraint). Yellow sticks show the two lysine residues of Stx1 (Lys45 and Lys55) found outside of the binding region with Stxbp1 as defined by the peptide array. Black edge indicates cross-linked site of Snap25.

  • Fig. 5 Analysis of auxiliary protein interactors of the AMPAR.

    Interaction space models (semitransparent volume; left) and docking models (right) of the interactions between AMPAR (PDB 5IDE) and its known interactors Olfm1 (PDB 5AMO), Frrs1l DOMON domain (modeled based on PDB 4ZEL), and Cacng2/8 (PDB 5VOT) were generated on the basis of the XL-MS data. Interaction space models were calculated using the DisVis Webserver, and docking models were generated with HADDOCK2.2. Cacng2 was structurally aligned from PDB 5VOT.

  • Fig. 6 Evaluation of the XL-MS approach on three biologically independent replicates for hippocampal synaptosomes.

    (A) Venn diagram showing the number of identified cross-linked protein pairs (from interprotein cross-links) in different replicates (dataset 2). (B) Boxplot showing the distribution of protein abundances for cross-linked protein pairs identified in one, two, and all three replicates. The number of proteins and median abundance are indicated in each box. Protein abundance data were obtained from (12). (C) Percentage of cross-linked protein pairs present in any PPI databases (i.e., STRING, InWEB, and BioGRID) identified in one, two, and all three replicates. The total number of cross-linked protein pairs identified is indicated above each bar. (D) Distribution of the fraction of direct neighbors (cross-linked proteins pairs) from the triplicate dataset that are in the same cluster in the initial network (Fig. 3A, labeled in red) and a randomly rewired network (100 iterations, labeled in gray). (E) Pie chart showing the percentage of protein pairs from the initial hippocampal synaptosome dataset identified in the triplicate network. Only proteins containing interprotein cross-links in both datasets were considered.

Supplementary Materials

  • Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/6/8/eaax5783/DC1

    Fig. S1. General evaluation of the XL-MS approach.

    Fig. S2. Mapping of cross-linking data onto high-resolution structure of several protein complexes.

    Fig. S3. Characterization of XL-based protein interaction network from hippocampus synaptosomes.

    Fig. S4. Characterization of XL-based protein interaction network from hippocampus microsomes.

    Fig. S5. Characterization of XL-based protein interaction network from cerebellum synaptosomes.

    Fig. S6. Characterization of XL-based protein interaction network from cerebellum microsomes.

    Fig. S7. Extended and detailed XL-based protein interaction network (extended Fig. 3A).

    Fig. S8. XL-based protein interaction network analysis.

    Fig. S9. Protein interaction interfaces and peptide array analysis (extended Fig. 4).

    Fig. S10. Evaluation of XL-MS approach on biologically independent replicates for hippocampal synaptosome (extended Fig. 6).

    Fig. S11. Complete XL-based protein interaction network from all seven cross-linking MS experiments.

    Table S1A. Complete list of cross-links identified in the two datasets.

    Table S1B. SynGO enrichment analysis of proteins identified in dataset 1.

    Table S1C. Cross-linked protein list.

    Table S2. Clustering and GO enrichment analysis of the proteins in the XL-based protein interaction network.

    Table S3. Human protein mapping and overlap of cross-linked lysine positions with protein interaction interfaces.

    Table S4. Sequences and signal intensities for the peptides included in the two independent replicates of peptide arrays (fig. S9B).

    Movie S1. Dynamic simulation of the three conformational states of Camk2.

  • Supplementary Materials

    The PDF file includes:

    • Fig. S1. General evaluation of the XL-MS approach.
    • Fig. S2. Mapping of cross-linking data onto high-resolution structure of several protein complexes.
    • Fig. S3. Characterization of XL-based protein interaction network from hippocampus synaptosomes.
    • Fig. S4. Characterization of XL-based protein interaction network from hippocampus microsomes.
    • Fig. S5. Characterization of XL-based protein interaction network from cerebellum synaptosomes.
    • Fig. S6. Characterization of XL-based protein interaction network from cerebellum microsomes.
    • Fig. S7. Extended and detailed XL-based protein interaction network (extended Fig. 3A).
    • Fig. S8. XL-based protein interaction network analysis.
    • Fig. S9. Protein interaction interfaces and peptide array analysis (extended Fig. 4).
    • Fig. S10. Evaluation of XL-MS approach on biologically independent replicates for hippocampal synaptosome (extended Fig. 6).
    • Fig. S11. Complete XL-based protein interaction network from all seven cross-linking MS experiments.

    Download PDF

    Other Supplementary Material for this manuscript includes the following:

    • Table S1A (Microsoft Excel format). Complete list of cross-links identified in the two datasets.
    • Table S1B (Microsoft Excel format). SynGO enrichment analysis of proteins identified in dataset 1.
    • Table S1C (Microsoft Excel format). Cross-linked protein list.
    • Table S2 (Microsoft Excel format). Clustering and GO enrichment analysis of the proteins in the XL-based protein interaction network.
    • Table S3 (Microsoft Excel format). Human protein mapping and overlap of cross-linked lysine positions with protein interaction interfaces.
    • Table S4 (Microsoft Excel format). Sequences and signal intensities for the peptides included in the two independent replicates of peptide arrays (fig. S9B).
    • Movie S1 (.mp4 format). Dynamic simulation of the three conformational states of Camk2.

    Files in this Data Supplement:

Stay Connected to Science Advances

Navigate This Article