Abstract
Aim: To use an animal model to test whether migraine pain arises peripherally or centrally.
Methods: We monitored the spontaneous and evoked activity of second-order trigeminovascular neurons in rats to test whether traffic increased following a potential migraine trigger (cortical spreading depression, CSD) and by what mechanism any such change was mediated.
Results: Neurons (n = 33) responded to stimulation of the dura mater and facial skin with A-δ latencies. They were spontaneously active with a discharge rate of 6.1 ± 6.4 discharges s−1. Injection of 10 µg lignocaine into the trigeminal ganglion produced a fully reversible reduction of the spontaneous discharge rate of neurons. Neuronal discharge rate returned to normal by 90 min. Lignocaine reduced the evoked responses of neurons to dural stimulation to 37% and to facial skin stimulation to 53% of control. Induction of CSD by cortical injection of KCl increased the spontaneous discharge rate of neurons from 2.9 to 16.3 discharges s−1 at 20 min post CSD. Injection of 10 µg lignocaine into the trigeminal ganglion at this time failed to arrest or reverse this increase. Injection of lignocaine prior to the initiation of CSD failed to prevent the subsequent development of CSD-induced increases in discharge rates.
Conclusions: These results suggest that there is a continuous baseline traffic in primary trigeminovascular fibres and that CSD does not act to increase this traffic by a peripheral action alone − rather, it must produce some of its effect by a mechanism intrinsic to the central nervous system. Thus the pain of migraine may not always be the result of peripheral sensory stimulation, but may also arise by a central mechanism.
Introduction
The pathophysiology of migraine and, more particularly, its origin remain “a riddle, wrapped in a mystery, inside an enigma” (1). Recent reviews of the pathophysiology are numerous (2⊟11) but one cannot find a definitive answer in them. One cannot even find pathology in migraineurs between or even during an attack. In particular, a peripheral pathology which could lead to the pain of migraine has been especially elusive. One of us (11) has recently reviewed this lack of apparent pathology and suggested a number of experiments that might address the issue of whether this is “absence of evidence” or “evidence of absence”.
The idea that migraine might originate inside the brain (neural theory) was first clearly articulated by Liveing in 1873 (12). Evidence for it was sparse and the accumulation of more evidence was a fruitless task for nearly a century. One of the things which led to a reawakening of the neural theory was Lashley's observation (13) that the aura of migraine correlated with a moving cortical phenomenon, later thought by many (14) to be akin to Leao's cortical spreading depression (CSD) (15). In the intervening years, the vascular theory of migraine was widely accepted – and it still is by some (16). The neural theory does not demand a peripheral site of origin of migraine pain, but the vascular and neurovascular theories do. In the vascular theory, the pain is supposed to arise from pathology of cranial vascular tissue. This pathology generates traffic in the peripheral sensory nerves which supply the craniovascular vessels. In the related neurovascular theory, the pathology is supposed to arise in the sensory innervation of the vessels, rather than the vascular tissue (17). It has so far proved impossible to demonstrate purported pathologies satisfactorily and, if they exist, they must be extremely subtle or cryptic. Any such pathology ought to be the generator of a signal in trigeminal peripheral fibres, but this too has not been shown to occur in migraine, although an increase in trigeminal ganglion discharge has been shown to occur after CSD in rats. Many interventions which are known to provoke migraine in humans can increase trigeminal traffic, but this is only weak circumstantial evidence that such increases in traffic always occur in migraine as a result of them (11). CSD, the probable phenomenon underlying the aura in migraine with aura (18), has been reviewed by Charles and Brennan very recently (19).
This lack of pathology was reviewed in 2010 (11) and a proposed programme of experiments to confirm or refute it via human and animal experiments was put forward. The most definitive test would be to monitor sensory traffic in the trigeminal nerve in humans before, during, and after a “spontaneous” migraine: “Absence of increased traffic during headache would settle the issue, but presence of it might not because of the possibility of antidromic traffic generated by events in the sensory terminals at the first trigeminal synapse. Less definitive animal studies might be useful – does CSD produce c-fos activation in neurons of the trigeminal ganglion for instance, or does it result in increased traffic in the sensory ganglion and root” (11).
This paper reports some early findings from such an experimental programme.
Methods
Experiments were approved by the University of New South Wales (UNSW) Animal Care and Ethics Committee and conformed to its guidelines. The experiments described here used methods, materials, and protocols which we have previously described in full (20⊟23).
Experiments were conducted in 16 male or female Sprague Dawley rats (250⊟350 g) sourced from ARC Perth and housed in the UNSW Biological Resources Centre. Rats were anaesthetized with intraperitoneal injections of urethane 1.5 g kg−1. Polyethylene catheters were inserted into the femoral artery and vein to allow continuous measurement of blood pressure and heart rate and to administer drugs and fluids. Rats were intubated but allowed to breathe normally. Core body temperature was monitored by a rectal thermistor and maintained at 37–38°C by a servo-controlled heating blanket. All physiological parameters were recorded on a Narcotrace chart recorder and, during experimental protocols, by software and written to a hard disk.
Animals were mounted in a stereotaxic frame. The trigeminal nucleus was exposed by a partial C1 laminectomy and partial occipital craniotomy. The dura was reflected away from the brain stem.
A 1-mm diameter hole was drilled in the parietal bone at a location 4 mm caudal to bregma and 2 mm lateral to the midline to allow access to the trigeminal ganglion (24). A 2-mm diameter hole was drilled just rostral and 4 mm lateral to bregma to allow measurement of cortical blood flow by laser Doppler methods. A 2.5 × 2-mm rectangular hole was made in the parietal bone just rostral to lambda and 4 mm lateral to the midline to allow electrical or mechanical stimulation of the dura mater.
Single-barreled pipette electrodes with a central tungsten wire were used to record discharges of neurons in the trigeminal nucleus. Electrodes were positioned orthogonal to the surface, from 1 mm rostral to 3 mm caudal to the obex and 2.0–4.5 mm lateral to the midline, their tips on the dorsal surface of the brainstem. Bacteriological agar (2% in saline, 40°C) was poured onto the exposed brainstem and allowed to set to reduce disruption of recording caused by movement related to cardiac pulsations and respiration. A piezoelectric microdrive was used to advance the electrode into the brainstem, 5 µm at a time, to a depth no greater than 1700 µm.
Cutaneous receptive fields (RFs) were classified according to standard criteria defined by Hu and Sessle (25) as nociceptive specific (NS: responding only to pinch with toothed forceps or deep pressure with a surgical probe), wide dynamic range (WDR: responding to pinch or deep pressure and light touch or brush with a small surgical probe) or low threshold mechanoreceptors (LTM: responding only to brush or light touch). The centre of each cutaneous RF was stimulated mechanically with an automated von Frey device, as previously described (26). Intermittent single von Frey stimuli were applied by activating the device from a stimulator which was in turn driven by the data-acquisition software. The effect was to substitute electrical stimulation with calibrated von Frey stimulation in the data-acquisition protocols. The dura mater was stimulated electrically with chlorided bipolar silver ball electrodes which delivered stimulus-isolated, supramaximal single shocks (40–150 V, 0.1–1.5 mA) of 250-µs duration every 5 s.
CSD was initiated with a 1-µl injection of 1 M KCl in water. A 27-gauge needle with a sleeve to limit penetration to a depth of 0.3 mm was quickly jabbed through the dura and into the cortex at the site of the electrical stimulus and the KCl ejected from a 10-µl micro-syringe. The needle was withdrawn and the injection site flushed with normal saline to remove excess KCl.
Stainless steel 0.31-mm (30 G) needle tubing connected to a 10-µl syringe via polyethylene tubing was advanced under stereotaxic control into the centre of the trigeminal ganglion (generally at 9–10 mm below the brain surface), according to the trigeminal ganglion atlas of Schneider et al. (24). Correct placement was confirmed by jaw movement when the needle entered the ganglion and by the presence of dye in the ganglion at the end of the experiment.
Cortical blood flow (as a measure of underlying cortical neuronal activity) was monitored with a pencil probe (P-431, TSI, diameter 0.15 cm), connected to a single channel laser-Doppler perfusion monitor (Laserflo BPM403, Vasamedix, St Paul, Minnesota, USA) (22).
Data acquisition and processing
Signals from the microelectrode were passed to a Neurolog recording system and filtered of line noise via a Hum Bug filter, then passed to oscilloscopes, a window discriminator and to either custom data-acquisition programs as previously described (20⊟22) or to a Cambridge Electronic Designs 1401 interface driven by Spike 7 software. In the latter case the analogue signal was recorded continuously, subjected to cluster analysis, and saved to disk to allow later re-analysis. For protocol 2 c, in addition to recording responses in the form of window discriminator output, the raw multi-neuronal signal was matched to previously fitted templates (Spike 7 software) for up to seven separate neurons at each recording site. Signals (analogue, window discriminator, and triggers) were processed online to produce post-stimulus response histograms (PSTH) and peri-event (rate) histograms; these were sent to a printer and X-Y plotter and stored on disk for further analysis.
Statistical methods
To test for significance of changes in evoked responses and basal discharge rates, we used Student's t-test, the critical ratio test (27), and ANOVA with a Bonferroni correction. For peri-event histograms, conclusions of the critical ratio test were validated by applying a two-sample Kolmogorov–Smirnoff test to the cumulative frequency distributions of the basal and response discharge rates.
Histology
Electrolytic lesions were made at the bottom of a “plunge”, or at recording sites, by passing 5⊟10 µA DC cathodal current for 10⊟15 s through the recording electrode. At the end of each experiment, the brain, brainstem, and upper cervical spinal cord were removed, fixed in 10% phosphate buffered formalin, and cut into 50-µm sections on a freezing microtome and the sections stained with cresyl violet. Recording sites and the receptive fields associated with them were recorded on micrographs from the stereotaxic atlas of Paxinos (28). At the end of the experiment, methylene blue was injected into the ganglion through the same injector used for injection of lignocaine. After removal of the brain, the ganglion was inspected for staining.
Protocols
Two series of tests were carried out; not all neurons were subject to all three protocols.
Protocol 1: effect of lignocaine injection into the trigeminal ganglion on stimulus-evoked responses
All but one rat (n = 15) and all but three recorded neurons (n = 20 including replicate runs for two neurons) were tested for responsiveness in the control portions of this protocol to establish that the recorded neurons responded to stimulation of dura mater and skin in a stable manner. Dural and skin stimuli were applied alternately in each test. Fifty stimuli were delivered in all, resulting in two PSTHs of 25 responses each. For each stimulus, the total number of neuronal discharges falling within a prespecified time latency window was counted. The mean number of discharges per stimulus and its standard deviation were calculated from the cumulative data for each collection of 25 sample responses. At least six pairs of such histograms were acquired at 5-min intervals (control). Then, for 10 of these neurons, 5 µl of 2 mg ml−1 lignocaine (total amount 10 µg) was injected slowly into the trigeminal ganglion and further pairs of histograms were acquired at 5, 10, 15, 20, 30, 45, 60, 75, 90, and 120 min following such injection or until responses returned to baseline. All responses were expressed as a percentage of the mean of the six pre-injection control responses and plotted against time since the injection.
Protocol 2: effect of lignocaine, CSD, and combinations of lignocaine and CSD on the spontaneous rate of responsive neurons
In this protocol, cortical laser Doppler blood flow was recorded continuously as an index of cortical activation. Protocol 2 was divided into three parts, as follows:
Protocol 2a: effect of lignocaine injection alone into the trigeminal ganglion on the basal discharge rate of responsive neurons
Spontaneous activity, in the form of 4-s counts of window-discriminator output or Spike 7 templated counts, from a previously-tested responsive neuron (11 rats, 14 neurons tested in protocol 1) was recorded for 512s (128 samples), at which point 10 µg lignocaine was injected slowly into the trigeminal ganglion (as for protocol 1) and recording continued until no further changes in discharge rate of the neuron occurred (recovery from lignocaine).
Protocol 2b: “reversal protocol”: effect of lignocaine on effect of a prior CSD
Spontaneous activity, in the form of 4-s counts of window-discriminator output of a previously tested responsive neurons (six rats, 10 neurons) was recorded for 600s, at which point CSD was initiated with a cortical KCl micro-injection and recording continued until no further changes in discharge rate of the neuron occurred. At this point, 10 µg lignocaine was injected slowly into the trigeminal ganglion and recording continued until no further changes in discharge rate of the neuron occurred.
Protocol 2c: “prevention protocol”: effect of CSD on effects of a prior injection of lignocaine
Spontaneous activity, in the form of 4-s counts of window-discriminator output from a previously tested responsive neuron (four rats, six neurons) was recorded for 1200s post-lignocaine (i.e. 20 min, the time at which the effects of lignocaine were normally maximal under protocol 1). At this point, CSD was initiated with a cortical KCl injection and recording continued until no further changes in discharge rate of the neuron occurred.
Results
Neuronal properties
Neurons responsive to both dural and skin stimulation were found over a depth range of 0–1625 µm (median depth 500 µm). They responded to stimulation with initial and peak latencies of 10.0 ± 3.2 and 15.3 ± 4.2 ms (dura) and 11.3 ± 3.5 and 20.9 ± 6.17 ms (skin). These are in the A-δ range; there were no neurons with apparent C-fibre latencies and all neurons were all classified as WDR, including one neuron in the most superficial lamina. Receptive fields were mostly located in trigeminal divisions V1 and V2 especially peri-orbitally, though two were located on the lower jaw (division V3). A typical response and the associated PSTHs are shown in Figure 1. Neurons were spontaneously active with a discharge rate of 6.1 ± 6.4 discharges s−1 (mean ± standard deviation for all tests of all neurons admitted to protocol 2, n = 20). These neuronal properties are detailed in Table 1.
Portion of a neurogram (upper) and the resultant post-stimulus response histograms (lower) showing responses to stimulation of the facial receptive field (left) and the dura mater (right). The upper traces show events associated with a single stimulus trigger, window discriminator (WD) output, and the neuronal recording. The Y-axis on the post-stimulus response histograms represents total number of WD events over 25 stimuli. Neuronal locations and response properties for neurons examined for their responses to stimulation, cortical spreading depression, and lignocaine A/P, antero/postero coordinate; DREZ, dorsal root entry zone.
Protocol 1: effect of lignocaine injection into the trigeminal ganglion on evoked responses
Injection of lignocaine resulted in a reduction of responses to both dural and cutaneous stimuli, detectable within 5 min and maximal by 20 min following the injection. Responses did not return to normal for 60–90 min. Mean time‐response curves for both modes of stimulus are shown in Figure 2.
Effect of 10 µg lignocaine into the trigeminal ganglion on the responses of second-order trigeminovascular neurons to electrical stimulation of the dura mater and mechanical stimulation of the craniofacial skin (protocol 1). Responses are expressed as percentages of the mean of six control responses (discharges/stimulus) prior to the injection. Lignocaine injection inhibited responses of the neurons; maximum inhibition was reached at 15⊟30 min and inhibition persisted for up to 60 min. In the region 0–30 min, responses to both skin and dura stimulation were significantly reduced (Student's t-tests, on five points, with Bonferroni's correction, p < 0.05).
Protocol 2a: effect of lignocaine injection into the trigeminal ganglion on the basal discharge rate of responsive neurons
Inhibition of peripherally-evoked responses (protocol 1) was accompanied by a decrease in the basal discharge rate of responsive neurons and followed the same time course, with a maximum effect (67% reduction) at 20 min. A typical effect is shown in Figure 3 and mean responses are shown in Figure 4 (top).
Effect of 10 µg lignocaine into the trigeminal ganglion (Vg) on the discharge rate of a responsive trigeminovascular second-order neuron. Lignocaine, injected at time 512s (4-s bin no. 128, indicated by the arrow and the artefact) reduced the spontaneous discharge rate of this neuron (Kolmogorov–Smirnov test C = 6.12, p < 10−9), an effect which persisted for 30–60 min. The vertical axis represents the number of discharges/4-s bin. Results for protocol 2, showing the lack of effect of lignocaine on the changes in neuronal rate induced by cortical spreading depression (CSD). Top (protocol 2a): effect of injection of lignocaine into the trigeminal ganglion (Vg) on the discharge rates of 15 neurons; lignocaine reduced the discharge rate of these neurons (left bars) maximal in the 20–40-min rate (middle bars), with recovery nearly complete by 90 min (right bars). Middle (protocol 2b): effect of one or more waves of CSD on the discharge rate of 11 neurons (middle bars) and the effect of subsequent injection of lignocaine into the Vg (right bars); lignocaine failed to reverse the effects of CSD (time samples same as for top charts). Bottom (protocol 2c): effect of injection of lignocaine into the Vg on the discharge rates of six neurons (middle bars) and the subsequent effect of one or more waves of CSD (right bars); lignocaine failed to prevent the effects of CSD (time samples same as for top charts). Both the means of the absolute discharge rates (left) and the means of the percentages of control levels of discharge rate (right) are shown. Common scales are used insofar as is possible.
Protocol 2b: effect of subsequent injection of lignocaine into the trigeminal ganglion on the prior increased discharge rate of responsive neurons induced by CSD
Cortical microinjection of KCl generally led to the occurrence of at least one and sometimes as many as six waves of CSD, as evidenced by an increase in cortical blood flow. At a variable time after the successful initiation of CSD, the basal discharge rate of responsive trigeminovascular neurons began to increase and continued to increase for upwards of 20 min in some animals. In some animals, “new” neurons not previously spontaneously active appeared in the Spike 7 traces. Injection of 10 µg lignocaine during this period did not reduce the firing rate of neurons and, in two-thirds of instances, failed to retard the increase in discharge rate caused by CSD. A typical result is shown in Figure 5 and mean responses for control, post CSD, and post lignocaine are shown in Figure 4 (middle).
Effect of cortical spreading depression (CSD) and subsequent injection of lignocaine into the trigeminal ganglion on the discharge rate of a responsive neuron in the trigeminal nucleus. CSD was initiated at time 580 s by cortical pinprick (indicated by first arrow and small artefact). The consequent CSD wave passed under the laser Doppler probe at around time 980 s, representing a propagation velocity of 2–3 mm min−1. After an initial spike and a quiescent period, the discharge rate of this neuron steadily rose to reach a peak at 1400 s. At this point (second arrow), 10 µg lignocaine was injected into the trigeminal ganglion, but there was no subsequent change in the discharge rate of the neuron.
Protocol 2c: effect of prior lignocaine injection into the trigeminal ganglion on the subsequent increased discharge rate of responsive neurons induced by CSD
Injection of lignocaine into the trigeminal ganglion reduced the discharge rate of responsive neurons, as in protocol 2a. Cortical pin-prick/KCl injection applied 20 min following lignocaine injection produced from one to five waves of CSD. Discharge rates of responsive neurons increased in a similar manner (although more briskly) to those seen in protocol 2b. Mean responses for control, post CSD, and post lignocaine are shown in Figure 4 (bottom).
Discussion
The experiments reported here [part of a previously foreshadowed programme (11)] show that CSD – which could be either a trigger or a symptom of migraine (or both) ⊟ can generate sensory traffic in second-order sensory neurons independent of the generation of traffic in the first-order neurons which impinge upon them. These conclusions are fairly robust, but are subject to some limitations, summarized as: the experiments were conducted in anaesthetized rats; it is not known whether CSD produces pain in rats and; it is not known whether the neurons involved in this study are the same types of neurons as those involved in the perception of migraine pain in humans. We discuss these limitations below.
The nature of any link between CSD, neuronal activation, and migraine pain is still not clear. Past hypotheses have included activation of meningeal afferents by potassium (29) and induction of painful ischaemia (30). Such hypotheses require CSD to produce activation of trigeminovascular primary afferents. In our experiments, injections of lignocaine into the trigeminal ganglion at the height of the CSD response, in amounts that were able to interrupt stimulus-induced responses and to reduce basal discharge rate, were unable to either prevent or reduce the increase in discharge rate which followed CSD. The block of the basal discharges of these neurons by lignocaine is an indication that the basal discharge may be maintained in part by the same sensory afferents which mediate the evoked responses. Lignocaine is known to ameliorate migraine pain when infused intravenously (31) or intranasally (32).
In the experiments reported here, the induction of CSD was followed by an increase in the discharge rate of neurons which responded to trigeminovascular and facial cutaneous stimulation. We know from our own previous experiments (33) and those of others (34) that stimulus-evoked responses are also facilitated by CSD. Our observations are consistent with a number of reports which used the indirect c-fos method to evaluate the effect of CSD on the activity of trigeminovascular second-order neurons (35⊟37). Direct measurement of increased discharge rate of second-order trigeminovascular sensory neurons following CSD has not often been reported. Increased discharge rates occur in rats (34,38) but we were not able to demonstrate its occurrence in cats (39). This could be a species difference or it could be a methodological difference related to the method of CSD initiation – pinprick or pinprick with KCl injection (40). The c-fos measurements mostly reveal activation in superficial laminae of the trigeminal nucleus and may be due to activation of C-fibres. The neurons we and others have recorded from with electrophysiological techniques are usually in deeper laminae, especially lamina V and seem to be mostly driven by A-δ fibres. Whether the difference in locations and possible modalities of neurons revealed by these two different techniques is important, it is not easy to say. We can say, however, that descending inhibitory fibres appear to make presynaptic contacts with both types of fibres (41). Dorsal horn neurons which can be sensitized by increases in sensory traffic produced by dural inflammation are located mainly in lamina V (42), where we mostly recorded. Again, it is not easy to say whether the location matters when considering the two competing theories of “sensitization”.
The neurons from which we have recorded in these experiments are located in the same regions of the trigeminal nucleus and have characteristics and response properties consistent with those recorded by others (43⊟48) and ourselves (21,49,50) in past experiments. Those recorded here were located in the trigeminal nucleus proper, but it has been known for a long time that a considerable proportion of responsive neurons occur in the upper cervical spinal cord (51⊟53), the tail of the so-called trigeminocervical complex (54). Zhang et al. (34) have recently reported that neurons in the cervical spinal cord are also activated by CSD. In the experiments reported here, the measure of the real-time influence of CSD was a change in discharge rate. It is known from our previous work (33,55) and that of others (34) that evoked responses also peak at the time that “spontaneous” discharge rates peak. Such increases in responsiveness can also be seen following other putative migraine triggers such as flashing light (55). Such a peak was not sought in the experiment reported here because it was neither possible nor advisable to interrupt the acquisition of this data to acquire the necessary PSTHs at the critical times.
The relevance of testing for a descending influence on trigeminal sensation in migraine by using animal models, such as the one we have used here, has been greatly strengthened by the discovery of an apparently trigeminal-pathology-free model for migraine in rats (56). These animals display cyclic changes in sensation and behaviour consistent with those seen in, and reported by, migraineurs. Moreover, the effects of a number of migraine-provoking and migraine-treatment drugs in these animals is consistent with the effects of such drugs in humans. It would be interesting to repeat the experiments reported here in these animals. Animal behaviour experiments beg the question of whether animals experience what humans call migraine pain when “migraine-mimicking” conditions are imposed upon them. There is some evidence that they do (40) and some evidence that they do not (57). In freely moving rats, a single episode of CSD did not produce behavioural changes which could easily be interpreted as the result of pain (58). These authors could detect little behaviour attributable to pain, but a considerable change thought to be indicative of anxiety, a result that might be mediated by the amygdala (59). In rats, intracortical KCl injection induces CSD, produces cutaneous allodynia, and activates c-fos in the trigeminal nucleus, whereas pin-prick alone does not; injection of KCl into the overlying dura, however, produces allodynia and c-fos activation in the absence of CSD (40). The latter observation suggests that dural pain might be provoked directly, rather than via CSD, a possibility which was first suggested by Ingvardsen et al. (37,60). Clearly the effects of CSD are very dependent on methodological factors, on injection sites, on behavioural evaluation techniques and on the possible existence of multiple subcortical pathways.
It is generally agreed that responsive neurons in the trigeminocervical complex are second- or later-order neurons, with convergent inputs from a number of trigeminal, occipital, and cervical receptive fields including the dura mater, cranial blood vessels, the teeth, cornea, the skin, and the buccal cavity (61,62). As previously reported (43,47,51) most neurons in these experiments were spontaneously active with rates of a few discharges per second. It is not possible to say what “input” drives this basal discharge nor whether it is intrinsic to these neurons, but it would seem from our experiments that a good part of the drive must come from sensory fibres which pass through the trigeminal ganglion. In our experiments, some of this incoming traffic could possibly have arisen as the result of the surgery necessary for the experiments. We attempted to minimize this with the use of a long-acting local anaesthetic on wound edges and by minimizing the surgery itself. The basal discharge rates of neurons in this series of experiments are not significantly different from those in experiments where the surgery was more extensive. However, repetition of these experiments in the new rat migraine model (56) would put this supposition to the test.
Importantly, the present report shows that CSD-induced increased traffic in these second-order neurons is not influenced by block of conduction in first-order neurons in the trigeminal ganglion, whereas stimulus-induced responses are reduced. However, it is possible that CSD might have produced a rapid development of sensitization at the first sensory synapse prior to lignocaine injection and that this sensitization could have become “locked-in” and might not able to be influenced by a later reduction of sensory traffic by the injection of lignocaine into the trigeminal ganglion (63). To test this notion, we also injected lignocaine into the ganglion prior to the induction of CSD. The results were the same as when these two procedures were applied in the reverse order: an increase in discharge rate produced by CSD. This is further indication that the CSD-induced changes in neurotransmission at this synapse are not solely the consequences of increases in sensory traffic arriving from the periphery. We chose a time point of 20 min post-lignocaine injection, as this was the time at which responses to evoked stimulation reached a minimum. It would be interesting to test both earlier and (especially) later time points post lignocaine in order to establish whether the effect of lignocaine persisted beyond its presence in the ganglion (time–response curve), but these experiments were not essential to support our conclusions.
These results are at considerable variance from the recent results of Zhang et al. (18) who demonstrated that CSD can generate an increase in traffic in about 50% of first-order cell bodies in the trigeminal ganglion, an increase which they attributed to activation of meningeal nociceptors. Our results suggest to us that the increased traffic in second-order neurons seen here and in first-order neurons by Zhang et al. may not arise solely from increased incoming traffic in first-order neurons. Moreover, the increases in traffic seen by Zhang et al. were rather less frequent and more modest than those seen in our own or in their experiments on second-order neurons. It is hard to say whether the increase in traffic in the trigeminal ganglion seen by Zhang et al. was generated orthodromically, antidromically, or both. A test for orthodromic activation would be to determine if it still occurred in animals in which the tentorial nerve had been severed. “Antidromic” activation of trigeminovascular sensory nerves has been widely speculated upon and is thought to play a part in neurogenic neurovascular inflammation, the “triple response”, “axon reflex” in peripheral nerves, and perhaps vascular calibre changes (64). In the neurovascular theory of migraine, antidromic activity is considered to both arise and exert its effects peripherally by mechanisms similar to those responsible for the axon reflex (65). There is, however, no particular reason to exclude the central ends of these nerves as a generator of traffic. This could occur, for instance, by the withdrawal of descending inhibitory controls in the manner which we have suggested (66). It is, in fact, virtually required by our own theory, which supposes that the increases in second-order neuron activity arise from removal of descending hyperpolarizing control on the central ends of trigeminovascular sensory terminals.
The idea that trigeminal sensory neurotransmission might alter in the absence of changes in peripheral sensory traffic has been given strong support by the recent results of Stankewitz et al. (67). They demonstrated, through the use of functional magnetic resonance imaging techniques in humans (68,69), that the trigeminal nucleus is activated before a migraine attack and thus before any detectable rise in peripheral traffic and certainly before the pain develops. This supports a role for a central rather than a peripheral mechanism and is completely in accord with the findings presented here.
The evidence for a central (probably cortical) origin for migraine is accumulating rapidly and has been summarized recently by Sprenger (70). There are now also a number of studies which have demonstrated that the “migrainous cortex”, both interictally and during a migraine attack, can produce symptoms such as photophobia, which cannot be explained on the basis of an effect on peripheral sensory traffic.
The mechanism through which functional and structural changes in the cortex might lead to headache in humans and the observations in the new rat model is unknown. Attention is turning away from a peripheral sensory pathway to a descending pathway in the central nervous system. The behavioural changes (in both rats and humans), at the very least, seem to be unlikely to be accounted for by a peripheral mechanism. Subcortical structures which have been implicated in the elaboration of migraine symptoms include the amygdala (59) the striatum (71), the caudate nucleus (72), the thalamus (73), the periaqueductal grey matter (48), the nucleus raphe magnus (33), the hypothalamus (74), the locus coeruleus (75), the A11 dopaminergic group (76), and the trigeminal nucleus itself (77). Injection of lignocaine into the nucleus raphe magnus (55,78) or the periaqueductal grey matter (55) increases the responses to trigeminal second-order neurons to both peripheral stimulus and to CSD. Injection of the neuronal excitant glutamate into these two nuclei produces the opposite effect (55). CSD and light flash (a migraine trigger) decrease the discharge rate of neurons in these two nuclei (33,79) and increase the responses of trigeminal neurons to peripheral stimulation; in the case of light flash, this can hardly be accounted for by a peripheral mechanism (33). We hypothesize that an alternative intra-central nervous system pathway from the cortex, to the brain stem nuclei, to the trigeminal nucleus has a greater explanatory power (11,66) and we have begun to accumulate evidence about the existence of such a pathway (33,55).
Our results indicate that under the conditions of our experiments, there is probably a continuous traffic in the trigeminal sensory system, which helps to maintain the discharge rate of trigeminovascular second-order neurons. The discharge rate of these neurons can be greatly accelerated by CSD, but reduction of traffic in sensory nerves by ganglion block fails to ameliorate this increase. We infer from this that the acceleration in the discharge rate of sensory neurons produced by CSD may reside in part in some mechanism that does not involve a primary sensory input. We further infer that the increase in the activity of second-order neurons which (presumably) occurs during the pain of migraine is not necessarily contingent on an increase in the activity of sensory afferents to them as a sine qua non. This may have major importance in understanding how migraine pain arises and how to better treat it.
Footnotes
Acknowledgement
We are grateful to the reviewers of this report for suggesting the experiments carried out under protocol 2c.
Funding
This work was supported by the National Health and Medical Research Council (grant: 568713) and the Australian Brain Foundation.