Traditionally, the analysis of neuropeptides was performed by Edman sequencing, in which the N-terminal amino acid is sequentially removed. The method was developed by Edman in 1950 and is used for neuropeptide sequencing in many different species—for example, NPY in the porcine brain [20] and gonadotropin-releasing hormone in the dogfish brain [21]. However, analysis by this method is slow and does not allow for the sequencing of neuropeptides containing N-terminal PTMs [22]. Immunological techniques, such as radioimmunoassay (RIA) and immunohistochemistry (IHC), have been used for measuring relative neuropeptide levels and spatial localization. RIA has been a popular tool to quantify neuropeptides in biological samples, such as NPY [23] and galanin in the rat brain [24]. Although RIA is a relatively sensitive technique capable of absolute quantification of peptide levels, it is not always specific for a single-peptide form because of its antibody-based detection method. In addition, unless specific antibodies are available for the different isoforms, RIAs do not distinguish between these modifications. IHC has provided essential information regarding the localization of neuropeptides within the complexity of the brain structure, improving our understanding of the distribution of peptides. However, IHC does not report the actual size or full identity of the peptide, including PTMs. In addition, these antibody-based methods require a priori knowledge of a potential neuropeptide in order to generate antibodies specific to the neuropeptides and only detect peptide sequences with known structures.

In contrast to immuno-based methods, mass spectrometry (MS) enables one to detect and identify the precise forms of neuropeptides without a priori knowledge of the peptide's identity, resulting in the identification of previously unknown neuropeptides. MS has been used to identify and characterize hundreds of endogenous peptides in various animals [25,26,27,28,29]. The high-throughput discovery of neuropeptides based on MS has made up the field of peptidomics (Fig. 2) [6,30,31,32,33]. While typical proteomic analysis involves the digestion of proteins using proteases, such as trypsin, to generate peptides that can be readily sequenced by tandem mass spectrometry (MS/MS), peptidomics focuses on the analysis of the native peptide forms, including PTMs, without using digestive enzymes. Therefore, it is important to identify and characterize as many neuropeptides as possible, because each peptide can have its own biological function. While MS-based peptidomics can be utilized to characterize a large number of neuropeptides simultaneously, it can be also applied to the quantification of individual neuropeptides.

Although MS allows the identification of a large number of neuropeptides in a high-throughput manner, sample preparation prior to MS is a critical step that affects peptide identification coverage in peptidomics experiments. When the animal is sacrificed and the brain tissue of interest is obtained, proteases rapidly degrade larger proteins into smaller peptides that fall into the mass range of neuropeptides. The presence of degraded products derived from highly abundant proteins often complicates the MS analysis and interferes with the identification of less abundant neuropeptides. Different techniques, including focused microwave irradiation before decapitation, microwave irradiation after decapitation, and boiling the tissue immediately after decapitation, have been used in order to inactivate the proteases responsible for the high background of protein degradation fragments [32,34,35,36,37,38]. In addition to the procedures of protease deactivation, peptide extraction and subsequent treatment are other steps that affect the number of identified neuropeptides. The use of different solvents and procedures can affect the peptide profiles detectable in brain tissue. Che et al. [31] compared different extraction conditions for the recovery of neuropeptides from mouse hypothalamus. They found that sonication and heating in water (70℃ for 20 min), followed by cold acid and centrifugation, enabled the efficient extraction of many neuropeptides without the formation of the protein degradation fragments seen with hot acid extractions. Bora et al. [39] and Lee et al. [37] also showed that multiple stages of peptide extraction, including boiling water, acidified acetone, and acetic acid, were able to maximize the number of identified neuropeptides from their respective analyses of supraoptic nucleus and suprachiasmatic nucleus (SCN) samples in the rat brain.

After sample processing, the extracted peptides are subjected to MS analysis. A mass spectrometer consists of an ionization source, a mass analyzer separating the ions based on their mass-to-charge ratio (m/z), and a detector recording the number of ions at each m/z value. Two commonly used types of ionization are electrospray ionization (ESI) and matrix-assisted laser desorption/ionization (MALDI). While each approach has its own advantages, MALDI is more tolerant to the presence of salts. Thus, MALDI is often applied to direct tissue profiling or molecular ion imaging from tissue sections. Direct tissue sample analysis by MALDI-based MS is usually performed by a simple sequence of steps, which include placing the tissue of interest on the MALDI plate, applying a droplet of matrix, and irradiating the co-crystallized tissue to cause desorption/ionization of the peptides [30]. MALDI imaging provides valuable information pertaining to the spatial localization of neuropeptides [40]. Although MALDI-MS can be interfaced with separation strategies, such as liquid chromatography (LC) and capillary electrophoresis, ESI-MS can be coupled more easily with the separation methods, because the ions are produced as an aerosol from the liquid phase with a high voltage in ESI. The introduction of a nano-electrospray, based on capillary LC, coupled to MS, significantly increases the sensitivity in LC-ESI-MS, and furthermore, the preferred formation of multiple-charge states in ESI allows for more comprehensive MS/MS fragmentation data.

While neuropeptides can be identified by comparing experimental peptide masses obtained from MS data against a list of known neuropeptide masses, they are typically identified with both MS and MS/MS fragmentation data. While collision-induced dissociation is widely used for peptide identification, other fragmentation methods, such as higher energy collision dissociation and electron transfer dissociation, have recently improved the identification capability of peptides and PTM localization [41,42,43]. The automated identification of neuropeptides based on their MS and MS/MS data is pursued by using proteomic-based search engines, such as Mascot, SEQUEST, X!Tandem, Peaks Studio, and ProSightPC [37,44,45,46,47]. Since peptidomics searches typically employ criteria, such as no enzymatic cleavage and various common modifications (e.g., N-terminal acetylation and C-terminal amidation), they often require a larger search space and longer search times for peptide identification. Several neuropeptide prohormone databases, including SwePep, have been constructed to facilitate neuropeptide identification, and cleavage prediction programs, such as NeuroPred, were developed to predict cleavage sites in prohormones and to provide the masses of the resulting peptides [48,49]. In addition to the database search strategies, neuropeptides can be also identified by de novo sequencing, especially when the species of interest does not have a sequenced genome. While peptide spectra are compared with theoretical peptides in the database search strategy, de novo sequencing, which is the direct assignment of the amino acid sequence from the MS/MS spectrum, can be pursued when the database is not available. Although de novo sequencing can be performed manually, the process is labor-intensive and time-consuming. There are several software packages that can perform de novo sequencing directly on MS/MS data, including Lutefisk (http://www.hairyfatguy.com/lutefisk/), Mascot Distiller (Matrix Science), and PEAKS (Bioinformatics Solutions, http://www.bioinformaticssolutions.com/) [30]. The peptide sequences obtained from de novo sequencing are often compared with the nonredundant database of the National Center for Biotechnology Information to establish the peptide identities with the Basic Local Alignment Search Tool (http://www.ncbi.nlm.nih.gov/BLAST). The BLAST search compares a partial neuropeptide sequence against the database of a closely related species.

MS-based peptidomic techniques allow for measuring relative levels of peptides in different conditions of the samples, as well as peptide identification. Quantitation of peptides based on MS is generally achieved either by stable isotope labeling or label-free approaches (Fig. 3) [50,51]. In stable isotope labeling, the relative levels of peptides in two different samples are typically examined by labeling the peptides in one sample with a light stable isotope and those in the other sample with a heavy stable isotope; the two samples are then combined and analyzed together. The relative levels of the peptides are calculated by the difference in the MS peaks (peak intensity or peak area) of the two samples. While the stable isotopes can be introduced to the peptides either metabolically or through chemical reactions, isotope labeling through chemical reactions enables one to obtain quantitative information from biological samples, such as brain tissues, for which metabolic labeling is not available. A number of different isotopic tags, including acetic anhydride, succinic anhydride, and trimethylammonium butyrate (TMAB), have been used for quantitative analyses of endogenous peptides in various samples and conditions [6,34,52]. For example, the TMAB tag was used to evaluate the relative changes of endogenous peptides in response to food intake or drug treatment [53,54,55] and to assess the role of protein convertase 1/3 and 2 in the processing of the peptides in mouse models [56,57]. In label-free methods, each sample is prepared separately and then subjected directly to individual LC-MS or LC-MS/MS runs, followed by comparisons of either the mass spectral peak intensities of the detected peptides or the total number of MS/MS spectra identified for a peptide, called spectral counting [51]. Since there is no limit to the number of samples that can be compared, label-free quantitation allows for the comparison of multiple samples or conditions. The approach has been used to measure the changes in the levels of neuropeptides present during the daytime and nighttime in the rat SCN [58,59] and to examine the expression changes of endogenous peptides involved in the embryogenesis of Japanese quail brains [60] and the nucleus accumbens of morphine-dependent rats [61]. In addition, quantitative peptidomic analyses based on peak intensities has enabled us to discover endogenous peptides associated with repeated exposure to amphetamine and with individual variations in sensitivity to the behavioral effects of cocaine in rat models [62,63].