Abstract
Introduction
The field of neuropeptides had a slow start, with Leman describing substance P and Mutt describing vasoactive intestinal peptide (VIP) during the 1970s. The discovery of further peptides associated within the central nervous system (CNS) and the peripheral nervous system (PNS) accelerated during the early 1980s, largely because of the development of methods to determine the amino acid sequences in peptides and to produce specific antibodies effectively. There was an almost exponential explosion of published papers on the neuropeptides during the 1980s; it was followed by a parallel explosion of papers on neuropeptide receptors five to 10 years later.
Our research at that time had for a decade been directed towards studies of the cardiovascular system, with a strong focus on the cerebral circulation (1). It was widely believed that one nerve had one signal messenger (under the law of Cannon) Hence, the cerebral circulation was supplied with sympathetic nerves, storing noradrenaline, and parasympathetic nerves, storing acetylcholine, a finding verified in both laboratory animals and humans (2). In 1976, we were the first to demonstrate the VIP-containing nerve fibres in the feline cerebral arteries and to show that VIP was a strong vasodilator (3). The subsequent years were followed by studies of various parts of the circulation and the heart. The description of substance P, neuropeptide Y (NPY)and gastrin-releasing peptide in the cerebral circulation (4,5) was soon followed by the description of fibres storing neurokinin A, calcitonin gene–related peptide (CGRP), peptide histidine isoleucine or methionine (PHI/PHM), pituitary adenylate cyclase–activating peptide (PACAP) and galanin in the perivascular nerves. The studies were complemented with selective lesion studies, tracing experiments, and functional tests both in vitro and in vivo. The studies revealed the presence of these neuropeptides in several species, including humans, although with regional variations (1,5,6).
Analysis of neuropeptides in blood, cerebrospinal fluid and tissues
In order to verify the existence of a vasoactive peptide in a tissue, it was necessary to know the complete neuropeptide sequence. This information made it possible for neuroscientists to establish radioimmunoassay (RIA) methods to evaluate the functional and clinical importance of the neuropeptides. Below, we will discuss the complex nature of neuropeptide biology and our experience, including pitfalls in neuropeptide analysis in clinical material, with special reference to CGRP, given its recognized important role in migraine pathophysiology. Finally, we will indicate some novel potential methodological tracks for future investigation.
In tissue samples the case was rather straightforward: removal of the tissue from laboratory animals or in conjunction with human surgical procedures yielded stable levels of the peptide of interest if the samples were treated under optimal conditions. Electron microscopy revealed the localization of the neuropeptides in synaptic vesicles in nerve terminals using the difficult immunogold method (7). After removal, the tissue specimens were immediately frozen and kept at –80°C until they were extracted in boiling acid. This handling method prevented isolated peptides from being broken down, cleaved or digested by tissue-bound peptidases. This was verified using high-performance liquid chromatography (HPLC) of the material, followed by radioimmunoassay (RIA) of each fraction (8). However, the situation was different when collecting tissue specimens from diseased subjects at autopsy. For example, in cerebral vessels in individuals who had died of stroke there were three peaks in the HPLC and RIA analyses (9). In such circumstances, what do we then measure?
Neuropeptide immunoreactivity can still be determined in these specimens, but whereas antibodies of a polyclonal antiserum recognize different parts of the molecule (the amino acid sequence used for immunization) the peptidases have broken down the complete peptide molecule into shorter fragments. If RIA is done without HPLC, which by its very nature results in up to 100 fractions, the mixture of different peptide fragments would not be noted or appreciated. We have argued that the measured peptide fragments also are part of the biology and hence use the term neuropeptide immunoreactivity to demonstrate our understanding of this issue.
In studies of biological fluids such as blood or cerebrospinal fluid (CSF), the situation is more difficult since subsequent to its release from the synaptic terminals the peptide must pass the synaptic gaps and the surrounding cells and tissues, and must diffuse into the blood or CSF, where they are exposed to a large number of peptidases, which change the structure and quantity of various peptides. In the early studies, we and others used selective catheter insertions (with the aid of the radiology departments) to sample close to the organ of interest or in the vicinity of a suspected tumour. Such studies revealed, for example, VIPomas and CGRP-storing carcinomas, and other studies revealed raised neuropeptide levels in hypertension and congestive heart failure, associations with various forms of dementia and stroke and the release of neuropeptides in exercise tests. Clinical studies were carried out in order to understand the conditions of neuropeptide involvement in diseases. It was learnt early that sampling meticulously, especially taking care to decant plasma immediately, followed by cold centrifugation and then with rapid deep-freezing, all facilitated accurate peptide estimations. With procedures such as these, we and others uncovered quite a number of exciting stories of neuropeptide participation in various clinical conditions.
Intracranial vascular innervation
Brain circulation is innervated by three main systems: sympathetic nerves, storing noradrenaline, adenosine trisphosphate (ATP) and NPY; parasympathetic nerves, storing acetylcholine, VIP, PACAP, PHI/PHM, and nitric oxide; and intracranial sensory nerves, storing CGRP, substance P, neurokinin A, gastrin-releasing peptide, dynophin, PACAP and nitric oxide (6,10). Clinical studies designed to uncover primary headache mechanisms started with fruitful collaboration; we had the background and decided to test the hypothesis that neuropeptides are involved in primary headaches in an optimally designed approach. It is now history; as expected there was no release of the sympathetic or parasympathetic markers NPY and VIP, respectively, in acute migraine attacks. VIP release was seen in two cases of severe migraine with facial symptoms and in all cases of cluster headache (11–13) and confirmed by others (14)
Calcitonin gene-related peptide research
CGRP was discovered when alternative processing of RNA transcripts from the calcitonin gene was shown to result in the production of distinct mRNAs encoding CGRP, a 37-amino-acid neuropeptide (15). Subsequently it was revealed that (i) the calcitonin gene localized in neural tissue differentially encodes the gene to generate CGRP; and (ii) antisera against synthetic rat CGRP demonstrates CGRP immunoreactivity in the central and peripheral nervous systems (16). Several key peptide research groups with different interests soon noted these findings and incorporated them into work on their organs of interest. The London group found that intradermal injection of CGRP in humans results in a long-lasting increase in dermal blood flow, resembling the flare of the triple response of Lewis (17). A CGRP-specific antibody was produced already, in 1983, which came to facilitate research on the CGRP group of peptides at the Lund University.
Our efforts were focused on the extremely potent peptide CGRP, and the first salient data were reported at a regulatory peptide congress in 1984 (4,18). CGRP-containing nerve fibres were found in the walls of intracranial vessels and cell bodies containing CGRP co-localized with substance P were found in the trigeminal ganglion; surgical removal of the trigeminal ganglion or destruction of the trigeminal nerve removing specifically the CGRP/substance P–positive fibres in intracranial arteries (8). This paper verified the RIA for characterization and analysis of CGRP by HPLC (8). It was clearly demonstrated by HPLC that the extracted material from blood vessels and trigeminal ganglia showed only one peak in RIA, corresponding to authentic αCGRP. Similar results were later seen for other types of tissues both in animal and humans. The peptide material is present in vesicles (varicosities) of the nerve terminals and thus protected from breakdown by the abundant presence of endopeptidases (19,20).
We had developed a sensitive myograph with which we were able to distinguish endothelial and non-endothelial vasomotor responses. With this tool we showed for the first time that CGRP and VIP were acting on smooth muscle cells to cause potent dilatation whereas substance P and acetylcholine required an intact endothelium to produce relaxation of cerebral arteries (21). In addition, we found that the relaxant response to CGRP and VIP were associated with activation of adenylate cyclase and the production of cyclic adenosine monophosphate (cAMP) (21). Our work was later extended to include detailed morphological and functional studies of the human cranial vasculature (22–24). Almost two decades later, it was demonstrated that the CGRP receptor complex (calcitonin-like receptor–receptor activity-modifying protein 1 [RAMP1]) is present in human cerebral and meningeal arteries (25) and connected to a receptor component protein (RCP), which activates intracellular adenylate cyclase, mediating the relaxant response (26). CGRP was seen to be a potent arteriolar dilator in all species studied, including humans. Functionally, it was observed to have a role in the ‘trigeminovascular reflex’ (27). This reflex was activated by other vasoconstrictors but it was not activated by vasodilators (28,29). Therefore, it was not surprising when it was shown that administration of nitroglycerine is not associated with elevated levels of CGRP (30): neither glyceryl trinitrate (GTN) nor NONOate infusion caused any CGRP release during or shortly after infusion, whereas administration of capsaicin resulted in strongly increased CGRP levels. Similar results have been found in humans (31).
We already knew that there are a number of difficulties in measuring neuropeptides in the circulation, with many potent endopeptidases, and with the methodology for analysis of neuropeptides (32). When the CGRP RIA was first available, problems with cross-reactivity became an issue. Whereas extracted materials from tissues showed a single peak in the HPLC fractionation, the analysis of plasma samples showed several peaks which corresponded to shorter fragments of the CGRP molecule; the endopeptidases rapidly start to break down the peptide into shorter fragments, some of which react with the antibodies in the RIA. A similar situation was apparent in CSF samples (33,34). This problem was discussed extensively by the various research groups that also set up CGRP assays at the time (35,36) It is evident from the different immunization procedures that polyclonal antisera contain different antibodies, each of which is directed towards a specific part of the intact CGRP molecule and hence has the possibility to cross-react with smaller fragments as well as with other peptides or peptide fragments that share the same amino acid sequence. This problem can be circumvented by running each sample in using HPLC and collecting only the peak that corresponds to authentic CGRP. From a practical point of view, however, this is not workable, because HPLC will not create a single small peak but most often a broad peak that still contains smaller parts of the CGRP molecule. In addition, the method is time-consuming, laborious and costly. It has no practical use for clinical studies and is used mainly to verify the assay.
Analysis of calcitonin gene-related peptide immunoreactivity in primary headaches
We hypothesized that methodology was crucial and required sampling from the jugular vein (the bulbus) in acute migraine to be successful. We had at that time the knowledge, derived from other types of clinical studies, that methodology was a key issue. From tissue studies we knew that RIA worked well but that the plasma sample had to be removed into pre-chilled vials with peptidase inhibitor on ice and immediately cool-centrifuged (in our case done at the place of removal of the sample in the clinic), the plasma decanted and immediately deep frozen, and as soon as possible freeze-dried. With this type of methodology in hand, we were confident that we could examine if there was a significant neuropeptide release in conjunction with primary headache attacks in patients who appeared at the hospital with a moderate-to-severe attack.
Our first study was designed to test the method on trigeminal ganglion stimulation in humans and to verify the result in cats. We took advantage of the fact that trigeminal neuralgia subjects were being treated with thermocoagulation and, as expected, a strong release of both CGRP and substance P was recovered in the jugular venous blood in both humans and cats (37). At this time we also sampled plasma from the cubital fossa and observed no significant rise in neuropeptide levels, a finding verified by others (38). The scene was set and followed by the demonstration that CGRP is released during migraine (11,13), in cluster headache attacks (12), in another case of trigeminal neuropathic pain with neurovascular features (39) and in chronic paroxysmal hemicrania (40). On the basis of these data, we suggested that the link between CGRP release and migraine would offer a new avenue for therapeutic development (41). The migraine results were first presented at the International Headache Congress in Sydney in 1991 and at subsequent conferences in the field. We were disappointed by the feeble support for the idea that CGRP had a key role in primary headaches; the research was mainly focused on neurogenic inflammation in the dura mater and agents involved in this mechanism, such as substance P (42). Further support for CGRP as a key player in the trigeminal system, however, came from immunohistochemistry, demonstrating that CGRP-containing neurons are most frequent in the human trigeminal ganglion (43). The findings have been supported by several groups since then (44). Then, in a single controversial report, stating that there is ‘no release of CGRP in migraine attacks’, much of the considerable understanding gathered during the early years of neuropeptide research was disregarded. The researcher visited patients at their home or workplace, recorded clinical data and then kept the jugular venous samples on ice until returning to the hospital. Consequently, the time between sampling and centrifugation was more than one to two hours. Becausee the half-life of CGRP in plasma is less than 10 minutes, it could be predicted that little CGRP would be left in the sample and a negative conclusion be reached (45).
The final proof of the concept that CGRP in the trigeminovascular system has a key role was the demonstration that a specific CGRP antagonist, olcegepant, was effective in acute migraine attacks and had fewer side effects than other treatments (46). This has now been underscored by two studies with an oral CGRP antagonist, telcagepant, revealing an efficacy similar to that of the triptans (47,48). The CGRP antagonists show strong selectivity towards the human CGRP receptor complex, consisting of the calcitonin-like receptor (CLR) and the RAMP1, which are parts of the CGRP receptor. This discovery opens a new path in primary headache treatment.
Comments and future perspectives
To return to the title of the paper, what are the pitfalls? In experimental studies of tissue samples, our data would suggest that when peptides are within the vesicles of the perivascular nerves, in tumours or in specific stores of cell populations they are protected from degradation. As soon as they appear in the tissue fluids, in CSF or in the circulation they are rapidly broken down by the peptidases. Today, we have identified a large number of those CGRP fragments. Our experience with peptidase inhibitors in the sampling vials is that they are not effective, and the best way to take care of the sample is to treat it meticulously after its removal.
The way of measuring CGRP today is slightly different; there are sandwich methods that have antibodies directed towards both the N- and C-terminals, hence the intact molecule is measured. This is clearly a step in the right direction. The uses of matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) to detect CGRP and its fragments suggest novel possible approaches. The main advantage for this new method over commonly used immunoassays is the high specificity of detection which overcomes problems caused by antibody cross-reactivity (49). Furthermore, there is the ability to examine complex samples without the need for extensive purification. MALDI-TOF MS can be used without any sample pre-treatment to study the in vitro processing of peptides in biological fluids, reflecting in vivo conditions during different pathological conditions. Detailed information about neuropeptide processing and knowledge of the chemical structure of the peptides could provide the basis for clinical assays and illuminate the pathophysiology of neurological disorders.