Analytic framework for peptidomics applied to large-scale neuropeptide identification

Abstract

Large-scale mass spectrometry-based peptidomics for drug discovery is relatively unexplored because of challenges in peptide degradation and identification following tissue extraction. Here we present a streamlined analytical pipeline for large-scale peptidomics. We developed an optimized sample preparation protocol to achieve fast, reproducible and effective extraction of endogenous peptides from sub-dissected organs such as the brain, while diminishing unspecific protease activity. Each peptidome sample was analysed by high-resolution tandem mass spectrometry and the resulting data set was integrated with publically available databases. We developed and applied an algorithm that reduces the peptide complexity for identification of biologically relevant peptides. The developed pipeline was applied to rat hypothalamus and identifies thousands of neuropeptides and their post-translational modifications, which is combined in a resource format for visualization, qualitative and quantitative analyses.

Figure 1. Comparison of peptidomics sample preparation methods.

(a) Workflow using microwave-irradiation, heat-stabilization or protease inhibitor perfusion before heat stabilization. Endogenous peptides from rat hypothalamus were extracted by homogenization in an urea-buffer and filtration using a 30-kDa cutoff filter. Peptides were desalted and concentrated on C₁₈-STAGE tips and analysed by LC-MS/MS. (b) Hierarchical clustering of identified neuropeptides. (c) Venn diagram of identified neuropeptides. (d) Analysis of neuropeptide length distributions. Box-plot analysis of neuropeptide length represented as number of amino-acid residues. Illustrations were generated using images from Servier.com.

Figure 2. Analytical framework for analysis of endogenous peptides.

(a) Optimized sample preparation protocol for experimental isolation of endogenous peptides from the hypothalamus before mass spectrometric analysis. (b) Computational data analysis was done in the context of orthologous protein groups enabling comparison between species and incorporation of previously annotated peptides from multiple databases and online resources to focus the analysis on a subset of peptides. (c) Generation of LPVs is schematically shown as removal of redundancy and merging of overlapping peptides. Illustrations were generated using images from Servier.com.

Figure 3. Neuropeptide feature analysis.

(a) Gene ontology enrichment comparing the identified peptidome to its corresponding hypothalamic proteome. (b) Logo plots for the C- and N-terminal regions flanking peptides split by pro-hormone precursor group membership. (c) Logo plots for the C- and N-terminal regions flanking LPVs split by pro-hormone precursor group membership. (d) Logo plots for the C- and N-terminal regions flanking amidated LPVs split by pro-hormone precursor group membership. (e) Overview of number of identifications and their modification state. In total, 14,416 unique peptides sequences were found, 20% were found in orthologous protein groups that contained a pro-hormone precursor.

Figure 4. Functional analysis of neuropeptide phosphorylation sites.

(a) Ki (shown as negative log, pKi) calculated from IC50 values determined under equilibrium conditions in competition with ¹²⁵I-NDP-α-MSH on MC1, 3, 4 and 5 receptors, respectively. (Cheng–Prusoff equation *P<0.05, ***P<0.001). Values are mean±s.e.m. (b) Representative figure of α-MSH and phosphorylated α-MSH-stimulated cAMP production in intact BHK cells expressing the human MC4 receptor. (c) Logo plots of the phosphorylation sites revealed a [ST]xE motif for the neuropeptide protein precursor group. In contrast, a [ST]P motif was found for the remaining phosphopeptides. (d) Occupancy of phosphorylation sites that was observed at least three times. Boxplots depicts variation observed across 32 replicates. Each line contains the Uniprot gene name and phosphorylated position in parentheses.