Biological function is mainly carried out by a dynamic population of proteins and peptides which may be used as markers for disease diagnosis, prognosis and as a guide for effective treatment. The study of proteins is called proteomics and it is generally performed by two-dimensional gel electrophoresis and mass spectrometric methods. However, gel-based proteomics is methodologically restricted from the low mass region, which includes important endogenous peptides. The study of endogenous peptides, peptidomics, is complicated by protein fragments produced post-mortem during conventional sample handling. Nanoflow liquid chromatography and MS, together with improved methods for sample preparation, have been used to semi-quantitatively monitor endogenous peptides in brain tissue. When rapidly heat-denatured brain tissue was analysed, these methods enabled simultaneous detection of hundreds of peptides and the identification of several endogenous peptides not previously described in the literature. In an application of the MPTP (1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine) model for Parkinson's disease, the expression of the small protein PEP-19 was compared with controls. The levels were found to be significantly decreased in the striatum of MPTP-treated animals.
Proteomics is the large-scale study of proteins. It was originally defined as the protein complement expressed by a genome . However, a proteome is a highly dynamic entity that is constantly changing in time and differs from cell to cell. This is well illustrated in the example of the butterfly and its caterpillar in which every somatic cell of two appearing individual organisms have identical genomes . Considering this, the proteome definition nowadays includes all protein isoforms and modifications and the specification of the cell type(s) at a certain time [3,4].
After translation, it is believed that up to 80% of all proteins are subject to a range of more than 300 different types of post-translational modifications [5–10]. Via processing events, i.e. proteolytic cleavage or by addition of modifying groups, e.g. phosphorylation, glycosylation, acetylation and ubiquitination, the protein's properties are covalently changed to give it novel commissions or provide it with additional information . Post-translational modifications may be definite, e.g. proteolysis of peptide precursors to form hormones and neuropeptides [10,11], or of transient nature, e.g. phosphorylation regulation of kinase cascades [12–14]. Other processing events associated with protein turnover/aging degradation also contribute to increased complexity, primarily as protein fragments in the lower mass regions. This problem is sometimes observed in connection with the use of post-mortem tissue or deficient sample preparation protocols and is caused by either enzymatic or non-enzymatic proteolysis .
In 1931, von Euler and Gaddum [15a] discovered the neuropeptide Substance P. More than two decades passed before the first peptide hormones, vasopressin and oxytocin, were structurally characterized . In the early 1970s, the term neuropeptide was introduced by de Wied  to describe an endogenous peptide synthesized in nerve cells and involved in nervous system functions.
Today, it is known that neuropeptides are present in several types of cells and have a vast range of functions [18,19]. As peptide hormones, signal transmitters/modulators, these polypeptides of 3–100 amino acids influence physiological processes such as regulation of reproduction, growth, feeding, circadian rhythms and affective states [18,20]. They can be abundantly produced in large neural populations or in trace levels from single neurons .
A number of neuropeptide precursors contain one or more sequences of both related and unrelated peptides as illustrated by the processing example of POMC (pro-opiomelanocortin)  (Figure 1), and their differential generation varies in a tissue-specific manner. POMC is expressed both in the anterior and intermediate lobes of the pituitary. In the anterior lobe, the peptides generated include ACTH (adrenocorticotropic hormone) and β-endorphin, whereas, in the intermediate lobe, these peptides are processed further to α-MSH (α-melanocyte-stimulating hormone) and to acetylated and shorter forms of β-endorphin. These mutually exclusive sets of peptides have completely different biological activities .
Outline of the major processing steps of POMC leading to melanocortins and other neuropeptides
The expression peptidome and peptidomics came into use by several groups in 2001 [24–27], which in analogy to the proteomics technology, aims at the simultaneous visualization and identification of all peptides expressed in a cell or tissue. This rather heterogeneous entity contains small proteins, peptide hormones, neuropeptides and transient fragments of protein degradation. Based on their physical size, the low-molecular-mass endogenous peptides in an organism, tissue or cell may be called a subproteome (Figure 2).
The mouse proteome and peptide subproteome
The levels of the endogenous peptides are likely to vary from one moment to another, reflecting information of the particular physiological status, i.e. sex, age and stress [28,29]. By comparing the peptidome in samples of, e.g., diseased tissue with those in normal tissue, differential expression patterns can be revealed. This may lead to the discovery of biologically relevant peptides as biomarkers for diseases or indicators of certain pathological/pharmacological events [30–33]. Biomarkers offer a widespread applicability in medicine and drug development by serving as diagnostic flags for diseases, or as indicators of drug efficacy and adverse events. Thus peptide expression patterns may contribute to more individualized therapy [34–36].
Peptides obviously have physicochemical properties different from those of proteins. Their small size and high motility together with low ability to bind different stains make them impractical to focus and visualize in gels [27,37]. To address the inability of traditional proteomics methods to characterize the peptidome, an alternative method was developed which was used to analyse the small protein and peptide content in brain tissue [38,39]. The method and detection system for these analyses needed to fulfil several analysis criteria: (i) it must be sensitive and specific to allow small sample amount analyses (i.e. milligram quantities from sub-regions of mouse brain tissue); (ii) it had to allow for semi-quantitative comparisons of patterns among samples by simultaneous detection of the polypeptides in one experiment, and (iii) it should make further analyses and identifications of polypeptides that differ in abundance above normal biological variation possible.
RPC (reverse-phase chromatography) has become the prime technique for separating peptides. The generally low hydrophobicity of peptides enables them to dissolve in aqueous systems without the use of detergents [25,37]. However, the hydrophobicity is sufficient for the peptides to adsorb to the stationary medium (e.g. C18). A gradient of an organic modifier (e.g. acetonitrile) elutes the adsorbed peptides from the reverse-phase medium which makes RPC a straightforward application suitable for ESI (electrospray ionization) and MS coupling [40,41]. Online LC (liquid chromatography)–MS offers automation of desalting, concentrating and separation of the peptides before MS analysis, which improves reproducibility.
In LC, as in the ESI interface, miniaturization is essential. This is certainly the case in neuropeptide analyses, where the sample amounts often are small and the sensitivity is an important issue. Capillary nanoflow LC provides higher relative peak concentrations which is important for lowabundance species . In that way, more sensitive MS detection, better use of the MS dynamic range and reduced discrimination during ionization is achieved [43,44]. Nanoflow decreases further the limit of detection owing to an increased ion transmission over the LC–MS interface [40,45].
When injecting large sample volumes (∼5 μl) on to nanoflow LC columns (i.e. 75 μm diameter) on-column analyte focusing or a column switching set-up is necessary. An advantage of the latter approach is the possibility of higher flow rates during application and desalting of the sample. In this way, experiment time is reduced and the column is spared through a certain level of sample clean up on a rather expendable pre-column. However, in the case of neuropeptides, analyses are often subsequent to partial separation (e.g. centrifugation, filtration) and thus are relatively clean, but desalting is mandatory.
A method for analysis of peptides and small proteins in milligram amounts of brain tissue was developed . A simple aqueous extraction protocol combined with centrifugal filter separation and nanoflow RPC–MS made detection of an approx. 1500 eluting peaks from a single brain sample possible. The method enabled semi-quantitative comparisons of peptide profiles from different brain regions. However, among the 19 identified peptides, the majority proved to be fragments from haemoglobin and other abundant proteins. The post-mortem proteolytic actions during sample handling were sufficient to produce a large number of non-specific degradation products in high quantities. It is clearly difficult to perform peptidomic analyses on brain tissue without detecting these fragments, since similar results have been reported by several other groups. [46–49]. Under standard conditions for sampling and processing of the sample, non-specific degradation produce an abundant layer of protein fragments that effectively mask the presence of endogenous peptides, and possibly also degrade the endogenous peptides.
To make use of the high resolution and detection provided by MS applications, it is essential to use sample preparation protocols that minimize the post-sampling changes, e.g. fragmentation of proteins and peptides which interfere with the analysis [46,50–52]. These proteolytic events can, to some extent, be delayed or minimized by the use of protease inhibitor cocktails, transgenic animal models  or heat inactivation of snap-frozen samples [39,50,51]. However, when working rapidly with animal brain tissue under standard conditions of sampling and sample preparation, there will be time for enzymatic proteolysis during a delay of ∼90 s (or even longer for a protease inhibitor to reach its agents or targets). The majority of gel-based proteomic studies are not adversely affected by this delay when only a few per cent of each protein is degraded, but it is detrimental to peptidomic studies . The fragments produced create an abundant layer of protein fragments in the mass range of the endogenous peptides and effectively mask their presence, and possibly also degrade the endogenous peptides [38,39,50,51]. As mentioned above, the fragments produced create a large number of ex vivo peptides in the mass range of the exogenous peptides.
In vivo focused microwave fixation may be used to ensure high sample quality before MS analysis of brain peptides and small proteins [39,54]. It is important to stress that microwaving is one of the more humane methods of killing, in comparison with other common methods in terms of animal stress and rapidity. Unfortunately, this procedure is not applicable to frozen samples because of uneven heating due to poor efficiency and the generation of local and overheated hotspots. For this reason, an instrument for conductive heat transfer to frozen and fresh samples was evaluated (Denator AB, Gothenburg, Sweden). By rapidly raising the tissue temperature to 90°C, the delay until enzyme inactivation is avoided [50,55,56]. The proteolytic events are thereby inactivated by irreversible heat denaturation and, in contrast with freezing methods, they permit further dissection and processing of the sample . Instant heat inactivation improves the detection of neuropeptides and their modifications when using immunoassays [57–60], as well as when using MS [39,50,51,61]. An advantage of the conductive heat method compared with microwaving is its uniform heat transfer on frozen samples.
After instant proteolytic inactivation of the sample using microwaves, the subsequent MS analysis from 1 mg of brain tissue revealed a dramatic decrease in the total number of observed peptides . The identification results showed 39 peptides all believed to be derived from specific processing. It is, of course, impossible to infer biological activity of a peptide from its sequence alone. Further studies, e.g. receptor affinities, gene knocking, antisense blocking and functional studies by administration of each peptide to cellular systems or live animals have to be conducted. There are, however, a few signatures present in many active peptide sequences to look for. First, more than one active peptide is often processed from one precursor protein . Also, the processing of active peptides is frequently at specific sites carried out by dedicated peptidases . Finally, active peptides are commonly modified by C-terminal amidation [20,63]. Indeed, among the peptides identified, 20 were previously known classical endogenous peptides and 19 were previously uncharacterized peptides from precursors known to contain neuropeptides, and others not previously reported peptides with the characteristic processing sites (Figure 3).
Selected part of the elution profile of a nanoflow LC–MS experiment of the peptide subproteome of rat hypothalamus
Assigning identities to the detected endogenous peptides is presently a bottleneck in peptidomics [15,64]. Most tools and identification utilities are designed for identifying proteins from tryptic digests of proteomic samples. As the average size of a tryptic peptide is in the optimal range of most MS instruments, a few fragments from most proteins can be detected. The goal is to identify a specified number of fragments from one specific protein to deduce its identity; it does not matter if several of its fragments remain undetected. In the case of endogenous peptides in peptidomics studies, however, the objective is to characterize the primary structure and identify every present peptide in the sample, since there is no redundancy to rely on. To increase the challenge, these peptides are typically larger than their average tryptic counterparts, leading to more complex MS/MS spectra.
Imaging MALDI (matrix-assisted laser-desorption ionization) MS
In the technique of mass spectrometric tissue imaging [65,66] peptides and proteins in thin (10–20 μm) tissue sections are analysed in situ. The sections are typically coated with a raster of matrix before an ordered array of mass spectra is acquired from the matrix spots where each spectrum represents the local molecular composition at known x,y co-ordinates. Image profiles of selected peptides and proteins in the section are generated by extracting their corresponding m/z ranges. This approach not only requires less sample manipulation, but also informs of the spatial localization of the peptides and proteins in the tissue analysed. However, brain tissue has high concentrations of salt which make washing in ethanol necessary . There is a possibility of analyte dilution and dispersion in this washing step. Also, the spatial resolution is limited by the ability to apply tiny amounts of discrete matrix spots in a repeatable manner with sufficient accuracy . Applications of IMS (imaging MALDI MS) range from low-resolution peptide and protein profile images of selected areas in mouse brain  to single neural cell peptide profiling analyses  and high-resolution imaging of proteins  and drug metabolites in whole rat sections .
Differential expression analysis
Proteomics technologies provide a useful repertoire for disease-related applications such as diagnostics and therapy. The current collection of tools and their use is likely to expand to meet the need for rapid and accurate analyses . Particularly, the area of protein and peptide differential expression may lead to the discovery of novel biomarkers for monitoring of disease, predicting drug response, disease diagnosis and subtype classification .
The involvement of endogenous peptides in medicine either as drugs or as drug targets cannot be overestimated because of their wide spectrum of functions. An evident example, and maybe the first, is the use of the peptide insulin for therapeutic treatment of diabetes .
In an attempt to demonstrate the utility of proteomics, the peptidomic methods were used to identify neuroadaptations in an animal model of Parkinson's disease. After dopaminergic neural depletion in substantia nigra caused by the neurotoxin MPTP (1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine), the expression levels of PEP-19 in the striatum were investigated . The peptidomic analysis revealed a significant decrease of the N-terminally acetylated small protein PEP-19 compared with the control. Further analyses by IMS and in situ hybridization on coronal brain sections confirmed the reduction and also showed the spatial distribution of PEP-19 to be predominant in the striatum (Figure 4). This 6.7 kDa neuronally expressed polypeptide belongs to a family of proteins involved in calcium transduction through interactions with calmodulin. The IQ motif, which constitutes more than half of PEP-19, has the ability to interact with and modify the affinity of calmodulin to calcium, thus influencing the calmodulin-transduced calcium response [74,75].
IMS analysis of brain tissue sections in an animal model of Parkinson's disease
It is anticipated that the peptidomics methodology may well be used in the expanding field of biomarker discovery. The need for biomarkers which can be used for early detection of disease, for the identification of new targets for therapeutics or for more effective drug development through better monitoring of therapeutic effect and toxicity is obvious as the population ages and neurological diagnoses increase.
New Approaches for Elucidating Protease Biology and Therapeutic Opportunities: Biochemical Society Focused Meeting held at Royal Agricultural College, Cirencester, U.K., 4–6 January 2007. Organized by B. Austen (St George's, University of London, U.K.), R. Leatherbarrow (Imperial College London, U.K.) and C. Southan (AstraZeneca, Sweden).
Our research is sponsored by the Swedish Research Council (VR), grant numbers 11565, 2002-6116, 2004-3417, the Swedish Foundation for International Cooperation in Research and Higher Education (STINT) Institutional grant, the K&A Wallenberg Foundation, and the Karolinska Institutet Centre for Medical Innovations, Research Program in Medical Bioinformatics.