Abstract
Migraine can be triggered by systemic administration of the nitric oxide (NO) donor nitroglycerin (NTG) and by abrupt falls in plasma oestradiol. Calmodulin-dependent protein kinase II (CamKII) present in superficial dorsal horns is thought to play a role in sensitization of central nociceptors, a phenomen present in migraineurs. We therefore examined in rats the expression of CamKII in the caudal trigeminal nucleus (TNC) after subcutaneous NTG (10 mg/kg) and its modulation by oestrogen. In male rats and in ovariectomized females, after 4 h NTG increased significantly CamKII expression in the superficial layers of TNC, but not in the upper thoracic spinal cord. NTG had no effect on CamKII expression in oestradiol-treated ovariectomized animals. Thus NTG, i.e. NO, selectively enhances CamKII in the rat TNC and oestradiol blocks this effect. These data may help to understand the mechanisms by which NO triggers migraine attacks and oestrogens influence migraine severity.
Introduction
In migraineurs, systemic administration of nitroglycerin (NTG) causes a typical attack without aura after a delay of several hours (1, 2). A biphasic headache can also be triggered by NTG in patients suffering from chronic tension-type headache (3). The precise mechanisms by which NTG, and thus nitric oxide (NO), triggers the second headache in each of these disorders are not known, but the delay of action does not favour the direct vascular effect of NO as a culprit.
Besides enhancing the expression of inducible nitric oxide synthase in dural macrophages (4), systemic NTG activates, among others, second-order nociceptors in the caudal spinal trigeminal nucleus (5), where most trigeminovascular nociceptive afferents project (6). In these neurons, NTG also increases the expression of neuronal nitric oxide synthase (nNOS) (7), which suggests that NO stimulates trigeminal Aδ and C afferents and, via second-order nociceptors, may induce a self-amplifying process possibly responsible for central sensitization. This effect is inhibited by pretreatment with acetyl-salicylate, but not with sumatriptan, indicating that prostanoids are involved (8).
Gonadal hormones, in particular oestradiol, modulate the clinical expression of migraine. Women are three to four times more frequently affected by this headache disorder. Migraine attacks can be triggered by the sudden fall of oestrogen plasma concentrations in the premenstrum (9) and they may disappear during pregnancy or after menopause, when plasma concentrations of oestrogen are stable (10, 11). The neurobiological mechanisms which underlie these modulatory effects of oestrogen on migraine remain speculative. In mice it has been shown that oestrogens decrease nociception, notably in the trigeminal system (12).
Calmodulin-dependent protein kinase II (CamKII) is found throughout the central nervous system and regulates calcium signalling in synaptic transmission (13). CamKII is capable of autophosphorylation, which increases its activity (14) and is necessary for long-term potentiation induction in the hippocampus (15). It binds to N-methyl-D-aspartate (NMDA) receptors in rats and its α subunit increases ion currents through α-amino-5-hydroxy-3-methyl-4-isoxazole propionic acid (AMPA) and NMDA receptors (16).
CamKII is also present in the superficial layers of the spinal cord, where it plays a role in central sensitization. There, its native and phosphorylated forms increase in the rat after capsaicin administration regulating AMPA receptors; CamKII inhibitors can abolish this process elicited by capsaicin (17). Subcutaneous formalin injections enhance CamKII expression in the spinal dorsal horns of mice (18). Taken together, these data indicate that CamKII plays a crucial role in nociceptive processing and sensitization at the level of the spinal cord dorsal horn.
We therefore decided to study, in rats, the effect of systemic NTG administration on the expression of CamKII in the superficial laminae of the spinal C1–C2 portion of trigeminal nucleus caudalis (TNC) and to examine whether this effect can be modulated by oestradiol.
Methods
Animals
The procedures in this study followed the guidelines of the International Association for the study of Pain and the European Communities Council (86/609/ECC). They were approved by the Ethics Comittees of the Faculty of Medicine, Universities of Liège and Szeged. Twenty-four male and 48 female Wistar rats (250–350 g) were used. The animals were raised and maintained in standard laboratory conditions with tap water and regular rat chow available ad libitum on a 12 h−12 h dark–light cycle.
At the age of 2 months, the female animals (n = 48) were ovariectomized under Nembutal anaesthesia and half of them had a 5 mm long Silastic capsule (3.18 mm o.d. and 1.57 mm i.d.; Dow Corning, Midland, MI, USA) filled with a 1 : 1 mixture of cholesterol (Sigma Chemical Co., St Louis, MO, USA) and 17β-oestradiol (Fluka, Buchs, Switzerland) inserted subcutaneously in the interscapular region. The capsules maintain oestradiol plasma levels in a range typical of those found in female rats at early proestrus (19).
Nitroglycerin administration
At the age of 3 months, half of the animals in all three groups [12 males, 12 ovariectomized females (ovx), 12 ovariectomized females treated with oestradiol (ovx + E2)] received a subcutaneous injection of NTG (prepared from Nitrolingual spray; Pohl-Boskamp GmbH, Hohenlockstedt, Germany) at a dose of 10 mg/kg. The other half received an injection of the vehicle (gift from Pohl-Boskamp GmbH) in the same location. Four hours after NTG or vehicle injection, the rats for the immunohistochemistry (eight in each group) were deeply anaesthetized with pentobarbital (Nembutal 80 mg/kg; Sanofi-Synthélabo, Paris, France) and transcardially perfused with 100 ml physiological saline followed by 500 ml 4% paraformaldehyde in phosphate-buffered saline (PBS). The cervical (C1–C2) spinal cords, as well as the thoracic (Th1) cord from all animals, were removed and postfixed overnight for immunohistochemistry. The animals for Western blotting (four in each group) were perfused only with physiological saline and the spinal cords were removed as above and processed for Western blotting.
Immunohistochemistry
After cryoprotection (30% sucrose overnight), 30 µm thick transverse cryostat sections were serially cut and collected in 16 wells containing cold PBS. Each well received sections from sequential 500 µm long tissue portions over the entire rostro-caudal extent of the C1–C2 and the Th1 spinal segments. After pretreatment with 0.3% H2O2, the free-floating sections were rinsed several times in 0.1
Western blotting
The dorsal portions of spinal cord segments were homogenized in cold Tris–HCl buffer (50 m
Data analysis
An observer blinded to the experimental procedures counted CamKII-immunoreactive (Ir) cells in laminae I–III of the C1–C2 and Th1 segments of the spinal cord, on three different series of immunostained sections in each animal. In each series the distance between the individual sections was approximately 500 µm along the rostrocaudal axis. The cell counts and relative Western blot optical densities were analysed with
Results
On microscopic examination of immunostained transverse sections, CamKII immunoreactivity was found in neurons of the TNC and in the neuropil of lamina II. As mentioned in Methods, we focused on immunoreactive neurons. CamKII-Ir cells were abundant in the superficial layers of the caudal spinal trigeminal nucleus. The number of cells was not significantly different between the various rostro-caudal levels, nor between sides of the TNC.
After vehicle injection there was no significant difference in the number of CamKII-Ir cells in the TNC superficial laminae I–III between male rats, ovx and ovx + E2 females (Figs 1a,c,e and 2).

Calmodulin-dependent protein kinase II immunoreactivity on transverse sections of the upper cervical spinal cord in males (a,b), ovariectomized (ovx) (c,d) and ovariectomized and treated with oestradiol (ovx + E2) females (e,f). Nitroglycerin administration (b,d,f) compared with vehicle (a,c,e) increases the number of immunoreactive cells in the superficial layers in males and ovx females (b,d) but not in ovx + E2 rats (f). Scale bar = 50 µm.

Histogram showing the number of calmodulin-dependent protein kinase II-immunoreactive (CamKII-Ir) cells in superficial laminae I–III of C1–C2 segments in the three animal groups 4 h after subcutaneous injection of vehicle (□) or nitroglycerin (▪) (mean + SEM, n = 8 per group). In males and ovariectomized (ovx) animals, but not in ovariectomized and treated with oestradiol (ovx + E2) animals, there is a significant increase in the number of CamKII-Ir cells (∗P < 0.05).
In contrast, 4 h after subcutaneous NTG administration there was a significant increase in the number of CamKII-Ir cells in males (Fig. 1b vs. 1a) and in ovx animals (Fig. 1d vs. 1c) compared with vehicle injections, but no change was found in ovx + E2 rats (Fig. 1e,f).
At the Th1 level the number of CamKII-Ir cells was overall smaller (sections not shown) and there was no significant difference between animal groups or between treatment conditions (Fig. 3).

Histogram showing the number of calmodulin-dependent protein kinase II-immunoreactive (CamKII-Ir) cells in superficial laminae I–III of the Th1 spinal cord segment in the three animal groups 4 h after subcutaneous injection of vehicle (□) or nitroglycerin (▪) (mean + SEM, n = 8 per group). There is no significant difference in either of the groups.
The results of Western blotting were in line with those of immunohistochemistry. We identified the bands representing the CamKII protein. In male and ovx animals, which had received NTG 4 h before (Fig. 4, lanes 2 and 4), the density of the CamKII protein band in the dorsal portion of the C1–C2 segments was increased compared with vehicle-injected rats (Fig. 4, lanes 1 and 3). In the ovx + E2 group the CamKII band was comparable after NTG (Fig. 4, lane 6) or vehicle injection (Fig. 4, lane 5). In the Th1 segments there was no apparent difference in either of the groups (blots not shown). Densitometric analyses of the blots confirmed that CamKII expression in the dorsal C1–C2 segments was significantly enhanced after NTG administration in male and ovx rats, but not in ovx + E2 animals (Fig. 5) and that there was no difference between NTG and vehicle injections in the dorsal part of the Th1 spinal segments in either group (Fig. 6).

Western blotting of calmodulin-dependent protein kinase II (CamKII) (50 kDa) and β-actin (43 kDa) in dorsal portions of C1–C2 spinal segments in males (lanes 1, 2), ovariectomized (ovx) (lanes 3, 4) and ovariectomized and treated with oestradiol (ovx + E2) (lanes 5, 6) animals. Compared with the vehicle (1, 3, 5), nitroglycerin administration (2, 4, 6) enhances the CamKII band in the male (2) and ovx groups (4), but not in the ovx + E2 group (6).

Histogram showing the optical densities of calmodulin-dependent protein kinase II (CamKII) Western blots in the C1–C2 segments of the three animal groups 4 h after subcutaneous injection of vehicle (□) or nitroglycerin (NTG) (▪) (mean + SEM, n = 4 per group). Data are expressed as proportions of β-actin band densities. In male and ovariectomized (ovx) rats there is a significant increase in optical density in the CamKII band after NTG (∗P < 0.05), whereas this increase is absent in ovariectomized and treated with oestradiol (ovx + E2) rats.

Histogram showing the optical densities of calmodulin-dependent protein kinase II (CamKII) Western blots in the Th1 segments of the three animal groups 4 h after subcutaneous injection of vehicle (□) or nitroglycerin (▪) (mean + SEM, n = 4 per group). Data are expressed as proportions of β-actin band densities. There is no significant difference between the two animal groups.
Discussion
To the best of our knowledge, this study demonstrates for the first time that systemic administration of NTG enhances CamKII immunoreactivity in laminae I–III of the spinal portion of the TNC. Previous studies have demonstrated increases of c-fos (5) and nNOS expression (7) in TNC after NTG administration at a dose of 10 mg/kg. The latter may be a molecular basis for a self-amplifying process and sensitization, as described clinically during migraine attacks (21). The effect of the NO donor on nNOS expression seemed to be selective for the trigeminal system, as no effect was detected in upper thoracic segments (7). We found the same trigeminal selectivity for the NTG-induced CamKII changes.
NO probably activates the trigeminal system via an effect on the peripheral nociceptive afferents, since capsaicin pretreatment, which destroys these small afferent fibres, abolishes the NTG-induced c-fos activation of secondary trigeminal nociceptors (22). NTG itself has different effects on nociception depending on the dose administered (23, 24). In the rat, an intravenous infusion of glycerylnitrate at low doses (2–50 µg/kg per min) does not cause c-fos activation in the TNC (25) or an increase of jugular vein calcitonin gene-related peptide (CGRP) levels (26), despite its capability to enhance stimulus sensitivity in the trigeminal system (27). In humans, microgram doses are used intravenously or sublingually to trigger migraine attacks (2). There may thus be species differences in the dose–response relationship of NTG. Interesting evidence to add to this discussion is the recent study by Nunez et al. showing that at low concentrations NTG may not release NO, but another compound (28). Higher doses of NTG, such as those used in our study, might thus be able to induce a more robust NO-related activation of trigeminal nociceptors. As far as the time course of the NTG-induced effects is concerned, its brain concentrations and those of cyclic guanosine monophosphate rise significantly 2 h after subcutaneous administration. Fos expression peaks at 1 h in neurons belonging to vasoregulatory areas, but only at 4 h in TNC neurons (29). For this reason, we chose the 4-h delay in our study. The precise time course of CamKII expression in TNC needs, however, to be determined in a dedicated study.
As mentioned in the Introduction, there is strong evidence that CamKII plays a key role in nociceptive processing and sensitization of central sensory neurons. Briefly, in the superficial layers of the spinal dorsal horns it is abundant both in neuronal perikarya and in the neuropil (17). The latter is explained by the fact that more than half of dorsal root ganglion cells are CamKII+, especially in the trigeminal ganglion (30). CamKII immunoreactivity in laminae I–II of the spinal cord is increased after subcutaneous injections of formalin (18) or capsaicin (17), or after intrathecal injections of substance P (31), and this increase can be blocked by CamKII inhibitors (17). The hyperexcitability induced by capsaicin in trigeminal ganglion neurons via inactivation of I(A) currents is also mediated in part by CamKII (32), as well as the activation of the vanilloid receptor 1 by phosphorylation in rat ganglion cells (33). It has been recently discovered that nociceptive stimuli up-regulate CamKII in the dorsal horns by peptidergic afferents (34), which are also crucial in migraine. In cultured neurons it has been shown that CamKII is able to decrease nNOS activity by phosphorylating this protein (35), raising the remote possibility that CamKII might counteract the NTG-induced nNOS activation. Moreover, in rats and mice calmodulin can activate various adenylyl cyclases which contribute to sensitization in the spinal cord (36). We now show that CamKII can also be activated in secondary trigeminal nociceptors by high doses of the NO donor NTG, which suggests that it may play a role in NTG-induced migraine headaches. Although the involvement of CamKII in migraineurs is not yet proven, these data suggest that this enzyme plays an important role in migraine pathogenesis.
Oestrogens are known to modulate nociception (37), including migraine attacks (9). We have shown that protracted oestrogen deprivation tends to have pronociceptive effects in the orofacial formalin model of pain in mice (12) and in the rat nitroglycerin model, where it blunts CGRP and 5-HT changes in the superficial laminae of the TNC (38). In both instances the observed abnormalities are reversed by oestradiol administration. In the hippocampus, where CamKII is known to be crucial for long-term potentiation (15), oestradiol can rapidly induce its neuronal expression (39, 40). Our results contrast with the latter, in so far as chronic administration of oestradiol has no detectable effect on baseline expression of CamKII in TNC, but is able to suppress its activation by nitroglycerin. Oestrogen receptors are present on spinal sensory ganglion neurons (41, 42) and in the spinal grey matter (43). Oestradiol may thus act at the genomic level, which would modulate the expression of CamKII and hence annihilate any detectable change in its immunoreactivity after NTG administration. Whether these findings are relevant to the hormonal influences on migraine remains speculative. However, combined with the clinical observation that female migraineurs are more sensitive to certain trigger factors during the perimenstrual period, they may suggest that the low oestrogen levels of the premenstrual and menstrual phases render female migraineurs also more prone to the attack-triggering effects of NTG. By the same token, one could draw a parallel between our data in rats, showing that chronic oestradiol treatment suppresses the selective activation of secondary trigeminal nociceptors by nitroglycerin, and the clinical observation that migraine attacks tend to be suppressed when sex hormone levels are high and stable, such as during pregnancy (10, 11).
In summary, nitroglycerin, a NO donor, is able to induce CamKII expression in the superficial layers of the TNC in the rat. This effect is annihilated by chronic high concentrations of oestradiol. Considering the known biological properties of CamKII, one may expect that its increased expression after NO enhances nociception in the trigeminal system. In contrast, the suppression of this activation by high oestradiol levels can be regarded as trigeminal antinociception. The acute NTG-induced and the chronic oestradiol-dependent changes both seem to be selective for the trigeminal sytem. They may thus be relevant to an understanding of the delayed NTG-triggered headache attacks in migraineurs and in patients suffering from chronic tension-type headache, and of the protective effect of stable plasma oestrogen levels. They may also give some hints on the molecular effects of NO donors and ovarian steroids in trigeminovascular pain syndromes, such as migraine, and open novel therapeutic perspectives.
Acknowledgements
This study was supported by grants no. 3.4523.00 and 3.4563.04 from the National Fund for Scientific Research (Belgium) to J.S. and by grants T-029979, T-043436 and M036252 of OTKA (Hungary) to L.V. A.P. was the recipient of a EURON Marie Curie Fellowship in 2003.
