Identification of novel peptide hormones in the human proteome by hidden Markov model screening

Abstract

Peptide hormones are small, processed, and secreted peptides that signal via membrane receptors and play critical roles in normal and pathological physiology. The search for novel peptide hormones has been hampered by their small size, low or restricted expression, and lack of sequence similarity. To overcome these difficulties, we developed a bioinformatics search tool based on the hidden Markov model formalism that uses several peptide hormone sequence features to estimate the likelihood that a protein contains a processed and secreted peptide of this class. Application of this tool to an alignment of mammalian proteomes ranked 90% of known peptide hormones among the top 300 proteins. An analysis of the top scoring hypothetical and poorly annotated human proteins identified two novel candidate peptide hormones. Biochemical analysis of the two candidates, which we called spexin and augurin, showed that both were localized to secretory granules in a transfected pancreatic cell line and were recovered from the cell supernatant. Spexin was expressed in the submucosal layer of the mouse esophagus and stomach, and a predicted peptide from the spexin precursor induced muscle contraction in a rat stomach explant assay. Augurin was specifically expressed in mouse endocrine tissues, including pituitary and adrenal gland, choroid plexus, and the atrio-ventricular node of the heart. Our findings demonstrate the utility of a bioinformatics approach to identify novel biologically active peptides. Peptide hormones and their receptors are important diagnostic and therapeutic targets, and our results suggest that spexin and augurin are novel peptide hormones likely to be involved in physiological homeostasis.

Figure 1.

Hidden Markov model (HMM) for the identification of peptide hormones. (A) State structure of the peptide hormone HMM with states indicated by letters and transitions between states indicated by arrows. States with numerical subscripts are single amino acid states, while states with the “n” subscript are multiple amino acid states whose length is determined by the transition probability between that state and other permitted states. N_n, B₁, H_n, B₂, C_n, and S_1–3, are N terminus, border, hydrophobic, C terminus, and cleavage site states, respectively, of the signal peptide feature. E_n, I_n, P_n, and T_n are extracellular, intracellular, peptide, and transmembrane states, respectively, while K_1–6 and F_1–6 are pro-hormone cleavage site states and G[K/R][K/R] is a simple sequence motif. START and STOP mark entry and exit points of the HMM. (B) Protocol for building and running the peptide hormone HMM. HMM states for individual sequence features were built by learning amino acid frequencies and transition probabilities from sets of proteins or motifs with known features (N = size of training set). Signal peptide states were built using a previously curated set of eukaryotic SWISS-PROT proteins; extra/intra/peptide/transmembrane states were built using selected sets of human SWISS-PROT proteins; and pro-hormone cleavage sites were built using a set of PC1/2 and furin sites from the MEROPS database. The peptide hormone HMM was assembled with the state-to-state transition constraints outlined in A. Finally, the HMM was used to assign states and scores to either a set of alignments of human, mouse, rat, and dog proteins from Ensembl or a set of human proteins from SWISS-PROT/TrEMBL.

Figure 2.

HMM successfully identified known peptide hormones. (A) Plot showing cumulative fraction of known peptide hormones (N_total = 77) (see Supplemental Table 1) identified among the top scoring 500 proteins; (solid line) aligned mammalian proteome as substrate; (gray line) human proteome as substrate. For the aligned proteome, 90% of known peptide hormones were found among the top 300 proteins. (B) Pie chart indicating percent composition of the top 300 proteins. Known peptide hormones, cytokines, growth factors, defensins, and other secreted proteins make up 61% of the proteins. Hypothetical and poorly annotated proteins make up 20% of the proteins and were submitted to further analysis to identify candidate novel peptide hormones.

Figure 3.

Primary structure of spexin and augurin. Sequence of spexin and augurin with the signal peptide underlined, pro-hormone cleavage sites boxed, and predicted processed peptide indicated in gray. Arrows indicate where the Flag antigen sequence (DYKDDDDK) was inserted to facilitate immunochemical detection of peptide products. Conservation among orthologs is shown below: (*) identity, (:) high homology; (·) low homology. The C-terminal glycine residue of the predicted spexin peptide is likely to be removed and the peptide amidated, a feature common to known peptide hormones. Both spexin and augurin peptides are enriched in aromatic amino acids.

Figure 4.

Colocalization of spexin and augurin with insulin in endocrine cells. Flag-tagged NPK, spexin, and augurin were transfected into rat pancreatic cells and fixed cells subjected to double immunofluorescence with Flag and insulin antibodies. (A) Neuropeptide K, (B) spexin, and (C) augurin show colocalization with insulin in small, cytoplasmic punctate structures. In all cases the majority of Flag immunoreactive puncta are also positive for insulin.

Figure 5.

Identification of spexin and augurin in cell supernatants. Vector control and Flag-tagged NPK, spexin, and augurin were transfected into rat pancreatic cells in culture, and cell supernatants were harvested and submitted to immunoblotting with a Flag antibody. (A) Supernatants from Flag-NPK and N-Flag-spexin transfected cells contained high (solid arrow) and low (open arrow) mobility bands that reflected processing of Flag-tagged products from these constructs. (B) Supernatant from N-Flag-Δspexin transfected cells contained a 4-kDa band, suggesting that the 6-kDa product seen for N-Flag-spexin was the result of cleavage significantly C-terminal to the spexin peptide. (C) Supernatant from C-Flag-spexin contained two bands (12 and 8 kDa), confirming C-terminal cleavage of spexin pro-peptide. A 12-kDa product is seen for both N-Flag and C-Flag spexin (open arrow), while a 13-kDa product is seen only in N-Flag-spexin and corresponds to incompletely N-terminally processed spexin (closed arrow). (D) The presence of two bands (10 and 8 kDa, solid arrows) in supernatant from Flag-augurin transfected cells probed with M1 Flag antibody demonstrated cleavage of augurin at the putative pro-hormone cleavage site as well as close to the C terminus of the pro-peptide. The same immunoblot probed with M2 Flag antibody revealed an additional low-mobility product, confirming cleavage at the predicted dibasic cleavage site immediately adjacent to the Flag tag (open arrow).

Figure 6.

Expression of spexin and augurin mRNA in mouse tissues. (A) In situ hybridization with antisense (αs) and sense (s) probes detected spexin mRNA in the submucosal layer (SML) of the esophagus and stomach fundus. (B) In situ hybridization with antisense (αs) and sense (s) probes detected augurin mRNA in the intermediate lobe (I) of the pituitary (A, anterior; P, posterior), glomerular layer (GL) of the adrenal cortex (Ctx, cortex; Med, medulla), choroid plexus (CP), and atrio-ventricular node of the heart (AVN) (A, aorta; V, ventricle).

Figure 7.

Spexin is a biologically active peptide hormone. (A) Representative muscle contractile response to 10 μM acetylcholine (ACh) and 1 μM spexin peptide (NWTPQAMLYLKGAQ-amide) in a rat stomach explant assay. Repeated administration of spexin peptide produced similar contactile responses. (B) Cumulative dose-response curve for contractile activity of spexin peptide on rat stomach explants (EC₅₀ = 0.75 μM, N = 6). Error bars indicate standard error.