Cranial dural permeability of inflammatory nociceptive mediators: Potential implications for animal models of migraine

Abstract

Background

Application of inflammatory mediators to the cranial dura has been used as a method to activate and sensitize neurons in the meningeal sensory pathway in preclinical behavioral studies of headache mechanisms. However, the relatively high concentrations and volumes used in these studies raise the question of whether the applied agents might pass through the dura to act directly on central neurons, thus bypassing the dural afferent pathway.

Methods

We used a radiolabeling approach to quantify the meningeal permeability of two of the inflammatory mediators, 5-HT and PGE₂, when applied to the cranial dura as part of an inflammatory mixture used in preclinical headache models.

Results

Both agents could be detected in samples taken four hours after dural application in the cerebrospinal fluid (CSF) and, in measurements made only for PGE₂, in the central nervous system (CNS) as well. Based on our measurements, we made estimates of the CSF and CNS levels that would be attained with the higher concentrations and volumes of 5HT and PGE₂ that were exogenously applied in previous pre-clinical headache studies. These estimated levels were comparable to or larger than normal endogenous levels, potentially large enough to have physiological effects.

Conclusions

The finding that the cranial meninges are permeable to the two tested inflammatory mediators PGE₂ and 5-HT raises some uncertainty about whether the behavioral changes observed in prior pre-clinical headache studies with these as well as other agents can be attributed entirely to the activation of dural nociceptors, particularly when the agents are applied at concentrations several orders of magnitude above physiological levels.

Introduction

The origin of intracranial headaches, in particular migraine, is thought to involve the activation of sensory primary afferent neurons that innervate the intracranial meninges and their related large blood vessels (1). Anatomical studies in animals have provided evidence that the sensory innervation of the intracranial meninges is supplied by neurons with cell bodies in the trigeminal and upper cervical dorsal root ganglia (2–4). Electrophysiological studies showed that this innervation, in particular of the dura mater, displays nociceptive properties in common with nociceptive sensory neurons that innervate other tissues of the body, including activation and sensitization by inflammatory mediators (5–8). These studies found that local administration to the dura of a combination of endogenous mediators found in inflammatory exudates, including prostaglandin E₂ (PGE₂), serotonin (5-HT), histamine (HA) and bradykinin (BK), at a pH of 5.0 is indeed a powerful stimulus that promotes a long-lasting excitation and hypersensitivity to mechanical stimuli in these neurons (6, 7, 9). These findings provided support for the notion that a local inflammatory process, which produces alterations in the chemical milieu of the meninges and possibly also the cerebral cortex, could lead to the activation and sensitization of cranial meningeal nociceptors, thus contributing to the ongoing throbbing pain of migraine (6, 10).

The pain of some headaches, including migraine, is often referred to the periorbital region of the trigeminal dermatome, and can also be accompanied by cutaneous hypersensitivity, or allodynia, in this region (11, 12). These sensory processes are thought to be mediated by central trigeminal dorsal horn neurons that receive convergent input from dural and periorbital skin afferents (13, 14). The development of cephalic allodynia, in particular, has been suggested to be mediated through a nociceptive process that involves an initial inflammatory activation of intracranial meningeal afferents and the ensuing central sensitization of dura- and skin-sensitive dorsal horn neurons (14).

The notion that local application of a mix of inflammatory mediators, or an “inflammatory soup” (IS), to the cranial meninges of animals promotes peripheral and central nociceptive changes that could explain the intracranial and extracranial sensory changes observed during migraine has prompted the use of such dural IS application in numerous behavioral assays with endpoints reminiscent of the sensory and affective changes that occur during migraine. These studies have shown that dural IS application evokes decreased feeding behavior (15), the development of facial and extracephalic allodynia (16–19), increased motivated behavior to seek pain relief (20) and the suppression of exploratory behavior, together with an increase in resting behavior (21). One feature of these behavioral models of migraine is that the concentration of the inflammatory mediators included in the IS that was applied to the dura mater was higher than those shown initially to activate and sensitize trigeminal dural afferents. Electrophysiological studies have demonstrated the persistent activation and sensitization of dural afferents with IS containing 10 µM of PGE₂, 5-HT and BK together with 100 µM of HA (6), whereas the IS in the various behavioral studies contained a 10–20 ×higher concentration of HA and PGE₂ together with a 100–200 × higher concentration of 5-HT and BK (16, 17, 20, 21). The use of concentrations greatly in excess of those needed for nociceptor activation raises the question of whether the diffusion of these agents from the site of dural application, even at very low levels, might be capable of exerting additional effects. In addition, the agents are not removed by wash in these behavioral experiments, and so diffusion can continue throughout the time course of the experiments (ranging from several hours to several days), or until the agents are biologically degraded.

The meninges display various levels of permeability to small molecules (22–25). Dura mater permeability was shown to be a simple diffusion process, with the degree of permeability for different molecules primarily dependent on molecular weight (26, 27). Such permeability provides the basis for epidural analgesia. Thus, it is possible that the penetration of durally applied inflammatory mediators into brain tissue and cerebrospinal fluid (CSF) due to cranial meningeal permeability might contribute, at least in part, to the effects seen in behavioral studies. In the present study, we used a radiolabeling approach to quantify the meningeal permeability to two of the inflammatory mediators, 5-HT and PGE₂, when applied to the cranial dura as part of an inflammatory mixture used in preclinical headache models, in order to investigate whether the mediators reach central nervous system (CNS) levels that could directly influence central neurons, thus bypassing the dural afferent pathway.

Material and methods

Animals and surgery

All experimental procedures were approved by the Animal Care and Use Committee of the Beth Israel Deaconess Medical Center, and were in compliance with the ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines, including randomization into the different treatment groups. No adverse events occurred during the data collection. A total of 17 male Sprague-Dawley rats (250–350 g, Taconic) were used for the study. Prior to the study, the animals were kept two per cage under a constant 12 h light/dark cycle at room temperature. Food and water were available ad libitum. For dural administration of the IS and radioactive mediators, the animals were anesthetized with urethane (1.5 g/kg i.p.) and placed in a stereotaxic frame. A surgical exposure was made of the dura overlying the cisterna magna and a 23 gauge needle attached to polyethylene tubing was inserted through the dura into the subarachnoid space and anchored to the dura with super glue, for withdrawal of CSF samples. An area of the cranial dura was exposed by making a craniotomy of approximately 1 mm in diameter in the frontal bone, centered approximately 1 mm rostral and 1 mm lateral to the bregma.

CSF measurements

Tritium measurement was used as a method to measure the transdural diffusion of inflammatory mediators into the CSF following the application of inflammatory mediators to the cranial dura. A mixture of inflammatory mediators commonly used in preclinical headache studies (i.e. IS) was made, and spiked 1:1 with either tritiated 5-HT or tritiated PGE₂ (1 µCi/µl, Perkin-Elmer; the PGE₂ was obtained as a special order). The final concentration of the mixture was: 5-HT, BK, HA (all at 1 mM) and PGE₂ (at 100 µM) made in synthetic interstitial fluid (SIF) consisting of 135 mM NaCl, 5 mM KCl, 1 mM MgCl₂, 5 mM CaCl₂, 10 mM glucose and 10 mM HEPES, at pH 5.0. The tritiated compounds were present in the final mixture at a concentration of 4.4 µM (5-HT) or 3.25 µM (PGE₂)_. The 1µl volume of this mixture that was applied to the dura (see below) therefore contained 4.4 picomoles of tritiated 5-HT or 3.25 picomoles of tritiated PGE₂. A 1 µl volume of the inflammatory mediator mixture was applied to the exposed frontal dura and covered with a glass coverslip to prevent evaporation. In the 5-HT experiments, a baseline CSF sample was withdrawn from the cisterna magna prior to dural application of the inflammatory mediators, and additional CSF samples were withdrawn at two and four hours following the dural application. At each of these time points, approximately 5 µl was first withdrawn to remove the CSF present in the needle and tubing (“dead space”), followed by withdrawal of a 10 µl sample. For the two- and four-hour time points, two consecutive 10 µl samples were withdrawn. In the PGE₂ experiments, no baseline sample was taken, and a single 35 µl CSF sample was withdrawn at the four-hour time point following dural application. To measure radioactivity, each CSF sample was added to a vial of scintillation fluid (Ultima Gold, Perkin-Elmer) and counted in a Beckman LS6500 liquid scintillation counter. Each sample was counted for 20 minutes. In order to estimate the amount of dilution that occurred in the diffusion across the dura relative to the concentration that was applied to the dura, standard dilutions were made from the tritiated compound (n = 3), and samples of equal volume to the CSF samples were added to scintillation fluid and counted in the same manner as the CSF samples.

CNS and trigeminal ganglion measurements

In the PGE₂ experiments, in addition to the withdrawal of CSF samples, brain tissue was harvested to measure tritium. Following the CSF withdrawal at the four-hour time point, the animals were euthanized by anesthetic overdose and the brain was immediately removed. The brain was divided into the following samples: cerebral cortex (ipsilateral and contralateral), subcortical tissue (ipsilateral and contralateral), cerebellum and brainstem. The cerebral cortex was further subdivided into rostral and caudal portions, where the rostral portion contained the region that was immediately subjacent to the dural application of inflammatory mediators. The upper cervical spinal cord (C1 and C2) was also removed, as a single block. The trigeminal ganglion on each side was also harvested. Each of the samples was weighed and then solubilized by mincing with a blade and immersion in 2.5 ml of Soluene 350 (Perkin-Elmer) in a scintillation vial on a shaker for several days. The vial was then filled with scintillation fluid and counts were made as described above for the CSF samples. Control samples of the same regions were taken from animals that had not received tritium application and solubilized as described above to obtain counts of the background activity of solubilized tissue. The CNS regions were roughly similar in size (approximately 0.22–0.26 g each), except for the brainstem and the upper cervical spinal cord, which were smaller (approximately 0.18 g and 0.08 g, respectively). The trigeminal ganglia samples were considerably smaller than the other samples (approximately 0.02 g each). The analyte level for each sample was calculated relative to the size of the sample (in pg/g).

In separate animals that did not receive dural tritium application (n = 2), the brain was removed and a total of 0.5µl of the tritiated PGE₂ was injected directly into the cerebral cortex. The samples of cerebral cortex were then processed and counted for radioactivity as described above, in order to calculate the number of counts (in DPM) resulting from a known quantity of tritiated PGE₂ after solubilization. These measurements from the direct cortical injection of tritiated PGE₂ gave a value of 0.010 pg/DPM. This value was used as a conversion factor for calculating the quantity of PGE₂ present in the tissue samples in the experimental animals. Comparison of the counts obtained from these samples to the counts obtained from adding an equal volume of the tritiated PGE₂ directly to scintillation fluid gave an estimate of 57% for the efficiency of the solubilization procedure in releasing the tritium into the scintillation fluid, which is within the expected range (28). Two types of calculations were performed for the amount of diffusion of the durally applied tritiated compounds across the dura: (1) calculation of a dilution factor relating the concentration in the CSF to the durally applied concentration, and (2) calculation of the absolute amount of the tritiated compound present in the CNS, expressed as pg/g, and also converted to concentration using 1.05 g/ml as a factor for converting mass to volume in CNS tissue (29). A further calculation was performed to extrapolate the levels that would be present in the CSF and CNS when these mediators are applied to the dura at the concentrations and volumes that have been used in pre-clinical headache studies.

Statistical analysis

Data are displayed as the mean ±SEM. Group comparisons were analyzed using unpaired t-tests. The level of significance was set at p < 0.05.

Results

CSF measurements

In the 5-HT experiments (n = 6), the tritium counts in the CSF samples taken from the cisterna magna at two and four hours (two samples per time point) following the dural application were significantly higher than background (p < 0.01), and were nearly constant across the four samples in each animal. Comparison of the counts obtained from the CSF samples to those obtained from standard dilutions of the tritiated compound gave an estimate of approximately 2.5 x 10⁵ for the dilution of the applied tritium in diffusing through the dura to reach the CSF of the cisterna magna (taking into account the 1:1 dilution of the applied tritium) (Figure 1(a) and (b)). In the experiments with tritiated PGE₂ in place of tritiated 5-HT (n = 9), a similar result was obtained, with an estimated dilution factor of approximately 2.5 x 10⁵ (Figure 1(c) and (d); four-hour time point only). This dilution corresponds to a CSF concentration of 18 pM for 5-HT and 13 pM for PGE_2. This dilution factor would result in a CSF concentration of approximately 4–8 nM for 5-HT, and 0.4–0.8 nM for PGE₂, for the applied concentrations of 1–2 mM 5-HT and 0.1–0.2 mM PGE₂ that have been used in behavioral studies. Further factoring in the 10–20-fold larger volume (10–20 µl) that has been used in the behavioral studies extrapolates to an estimated CSF concentration in those studies of 40–160 nM 5-HT and 4–16 nM PGE_2.

Figure 1.

CSF measurements. Measurement of diffusion of tritiated 5-HT ((a), (b)) or PGE₂ ((c), (d)) into the CSF following application of 1 µl of a mixture of inflammatory mediators to a 1 mm diameter area of the cranial dura overlying the frontal cortex (final concentration of the applied solution: 5-HT, BK, HA, all 1 mM, and PGE₂, 100 µM). (a) Tritium measurements, in DPM (mean ± SEM), of 10 µl samples of CSF withdrawn from the cisterna magna prior to (n = 6) and at two and four hours following (n = 6, for each time point) dural application of inflammatory soup spiked with tritiated 5-HT. Two consecutive 10 µl samples were drawn at two and four hours following the dural application. (b) Measurements of the counts, in DPM (mean ± SEM), of 10 µl samples of standard dilutions of the tritated 5-HT (n = 3) for comparison with the measurements of the CSF samples in (a). (c) Same measurement as in (a), in experiments in which the inflammatory mediators were spiked with tritiated PGE₂. A single 35 µl sample was withdrawn four hours following the dural application (n = 9). (d) Measurements of the counts, in DPM (mean ± SEM), of 35 µl samples of standard dilutions of the tritated PGE₂ (n = 4), for comparison with the measurements of the CSF samples in (c). For both 5-HT and PGE₂, the mean of the tritium measurements of the CSF samples was extremely close to that of the 5 x 10⁵ dilution standard (dotted line). Since the tritiated compound was applied to the dura in a 1:1 dilution, we estimate that the diffusion through the dura into the CSF resulted in a further dilution of approximately 2.5 x 10⁵.

CNS and trigeminal ganglion measurements

In the PGE₂ experiments described above, in addition to the withdrawal of CSF samples, CNS tissue and trigeminal ganglia were also harvested for measurement of tritium at the four-hour time point (n = 8). The counts were significantly higher than control for each of the CNS regions (p < 0.01), but not for the trigeminal ganglia (p > 0.3). The total amount of tritiated PGE₂ present in the CNS tissue samples (brain plus upper cervical spinal cord) per animal was calculated as 7.8+/−2.9 pg (see Methods), or 3.9 pg/g of tissue. The measurements of the CNS samples showed a highly non-uniform distribution (Figure 2). The cerebral cortex subjacent to the dural application site (rostral cortex, ipsilateral) showed a much higher level than the other CNS regions (16.3+/−8.7 pg/g). From this peak, the level showed a sharp drop-off of approximately fourfold in the adjacent regions of cerebral cortex (caudal cortex, ipsilateral, 4.2 +/−4.7 pg/g; rostral cortex, contralateral, 3.6+/−2.5 pg/g). Moving further from the peak site, the level showed a further drop-off of approximately twofold in the caudal contralateral cortex (2.3+/−0.9 pg/g) and the ipsilateral subcortical block (2.0+/−0.8 pg/g). There was a slight further drop-off in the remaining CNS regions, which showed a nearly uniform level of 1.7–1.8 pg/g across a long rostrocaudal distance extending from the subcortical region to the upper cervical spinal cord (subcortical block, contralateral, 1.7+/−1.0; cerebellum, 1.8+/−0.9; brainstem 1.7+/−1.6; C1/C2 spinal cord, 1.8+/−0.8). The mean value of 3.9 pg/g for the entire CNS sample is equivalent to a mean CNS concentration of 12 pM, using a conversion factor of 1.05 g/ml (29). This concentration for the CNS is approximately the same as the concentration we calculated from our CSF samples (13 pM, previous section). Based on the roughly 300–1200-fold larger quantity (concentration and volume) of PGE₂ used in the preclinical headache studies, our measurements extrapolate to a CNS concentration in those studies of 3.6–14.4 nM.

Figure 2.

CNS measurements. Diagram (a) and plot (b) show the distribution of PGE₂ in the different CNS regions. The “X” in (a) marks the approximate position of the dural application site. Measurements (mean ± SEM shown) were significantly above background for each of the CNS regions (p < 0.01), but not for the trigeminal ganglia (NS—not significant).

Discussion

The primary goal of this study was to measure whether commonly used inflammatory agents that have been applied to the cranial dura of rats and lately also in mice (30), namely 5-HT and PGE₂, can pass through the dura and reach detectable levels in the CSF and CNS. In addition, we have attempted to estimate by extrapolation from these measurements the CSF and CNS levels that would be attained in behavioral studies that use dural application of higher concentrations and quantities of these inflammatory mediators as a headache-inducing stimulus, and to evaluate whether these estimated central levels would be expected to have behavioral or physiological effects, based on previous studies. For these purposes, we chose to make measurements, using radiolabeling, for these two agents, which were applied to the dura as part of a mixture of inflammatory mediators commonly used in pre-clinical headache studies that also included BK and HA. For measurement purposes, the inflammatory mixture was “spiked” with a small amount of the radiolabeled agent (either 5-HT or PGE₂). The rationale for applying this mixture rather than applying the radiolabeled agent alone was that the inflammatory effects might increase the permeability of the dura. The key finding of our study was that, in spite of the small quantity of the radiolabeled agents in the durally applied mixture, both of these agents, although differing in molecular weight and lipophilicity, could be detected in the CSF and, in additional measurements made only for PGE₂, in the CNS as well.

Extrapolation from the small quantity of the durally applied radiolabeled agents used in our experiments to the much higher concentrations and volumes used in pre-clinical headache studies gave estimates of the CSF levels that would be attained in those studies of 40–160 nM for 5-HT and 4–1.6 nM for PGE₂, and CNS levels of 0.5–20 ng/g for PGE₂ (for dural application of 1–2 mM 5-HT and 0.1–0.2 mM PGE₂ in a volume of 10–20 µl (16, 17, 19, 20, 31). These estimated values represent the increase that would result from the dural application, and thus would be added onto the endogenous levels that are normally present. As discussed below, these exogenous levels are comparable to or larger than the normal endogenous levels that have been measured previously, and are potentially large enough to have physiological effects.

Previous studies have reported normal baseline CSF levels of 1.4–2.6 nM (average 2 nM) for 5-HT (32, 33) and 0.28–0.6 nM (average 0.4 nM) for PGE₂ (34, 35). Endogenous brain levels of PGE₂ of 4.7 ng/g have been reported (ranging from 3.3–12.0 ng/g in different brain regions (36)); this is equivalent to a CNS concentration of 14 nM (using a conversion factor of 1.05 g/ml brain tissue density for converting brain mass to volume (29)). The exogenous CSF values that we estimated for the pre-clinical headache studies by extrapolating from our results are greater than these endogenous CSF levels by approximately 20–80-fold for 5-HT and 10–40-fold for PGE₂. The exogenous brain levels of PGE₂ that we estimated for the pre-clinical headache studies are roughly 25–100% of these endogenous brain levels. Thus, our results indicate that dural application of these agents in the quantities used in the pre-clinical headache studies would produce a very large increase above endogenous levels in the CSF, and a moderate increase above endogenous levels in the brain, during the first four hours of application.

Both 5-HT and PGE₂ have potential sites of action in widespread regions of the CNS, and are involved in a wide range of physiological functions, including pain, which could have direct or indirect influences on the outcome measures examined in preclinical headache models. The role of central 5-HT in pain modulation is complex. While 5-HT has been implicated in endogenous analgesia, descending serotonergic pain facilitation has also been described (37). The sleep-promoting effect of central PGE₂ may be of particular significance in regard to the decrease in behavioral activity observed following dural application of inflammatory mediators (19, 21), although it is not clear from the existing data whether the central levels attained in such studies would be sufficient for the sleep-promoting effects (37). There is a large body of evidence for the involvement of central PGE₂ in mediating pain hypersensitivity (38, 39) at central levels that are comparable to the levels we calculated for the preclinical headache studies. Increases in CSF levels of PGE₂ are observed in conditions associated with pain or pain hypersensitivity, such as that induced by intrathecal administration of IL-1β or subcutaneous formalin injection (34). Intrathecal injection of 1 ng of IL-1β, a dose sufficient to induce pain hypersensitivity (40), resulted in PGE₂ levels of approximately 1.8 nM (34), while LPS, which induces a sickness reaction including hyperalgesia that is mediated by central IL1β, results in PGE₂ levels of 2.4–3.4 nM (41, 42). The central levels of PGE₂ that, according to our calculations, would be reached in preclinical headache studies would exceed the levels attained in these hyperalgesic states induced by IL-1β and LPS. The finding that central administration of COX inhibitors reduces or blocks pain behavior or pain hypersensitivity (34, 43, 44) further supports a role for central elaboration of PGE₂ in pain hypersensitivity. Hence, a central action of PGE₂ could also explain the development of the behavior that is commonly observed in preclinical headache studies (17, 20, 31). It would be of interest to examine whether the lower concentrations of durally applied inflammatory agents that have been used to activate dural nociceptors in electrophysiology studies would induce a more restricted distribution of allodynia, as such concentrations would reduce the possibility of central effects.

In addition to the two agents examined in the present study, the mixture that is commonly applied to the dura in pre-clinical headache studies also contains HA and BK. BK has a substantially higher molecular weight than the other molecules and so would be expected to exhibit more limited diffusion. However, histamine is the smallest molecule of the four agents in the mixture, and so might be expected to exhibit a significant degree of dural penetration. This is supported by evidence that the histamine that is released by dural mast cells as a result of craniotomy or scoring of the cranial bone enters the CNS in an amount that is sufficient to produce plasma extravasation in the subjacent cortex, indicative of a breakdown of the blood brain barrier (45). In addition to its cerebrovascular effect, histamine is known to have effects on a variety of central functions. (46, 47). As with 5-HT and PGE₂, these effects could directly or indirectly affect the outcome measures used in preclinical headache studies, although the arousing effect of central histamine would tend to counteract the reduction in activity that has been reported in some of these studies (21)

Our central measurements of tritiated PGE₂ showed a highly non-uniform distribution, with a sharp peak in the region of the cerebral cortex below the site of dural application, and a rapid drop-off in adjacent regions. A similar highly localized cortical distribution was found following dural application of muscimol (25). This sharp peak in the PGE₂ distribution is evidence for a local accumulation in the CSF of the subarachnoid space immediately below the dural site, and subsequent local entry into the cortex at that site, either by diffusion across the pia or by CSF flow into the perivascular space of penetrating blood vessels, through the “glymphatic system” (48). Entry into the other CNS regions could occur by a combination of (1) intraparenchymal diffusion from this peak cortical zone into the immediately surrounding CNS regions, and (2) spread through the CSF of the subarachnoid space, where the spread is apparently rapid enough to reach the spinal cord within the four-hour time frame of our experiments, followed by entry into the CNS. The mean level of tritiated PGE₂ in the CNS samples, expressed as a concentration, was very similar to the concentration in the CSF. By contrast, the reported levels for endogenous PGE₂ are much higher in the brain than in the CSF, by a factor of approximately 35 (14 nM vs. 0.4 nM, as noted above). This large difference between the exogenous (tritiated) and endogenous PGE₂ in their relative distributions in the CNS versus the CSF can be understood in light of the origin of the PGE₂ and the predominant direction of CSF flow. The exogenous PGE₂ reaches the CNS entirely via the CSF, whereas the endogenous PGE₂ presumably originates primarily within the CNS, and endogenous levels in the CSF represent outflow from the interstitial space of the CNS parenchyma. The respective distributions of the exogenous and endogenous PGE₂ are evidence for a relatively effective movement of agents from the subarachnoid space into the brain parenchyma, and comparatively little movement in the opposite direction (48).

Whereas all of the sampled CNS regions showed counts above background, no tritiated PGE₂ could be detected in the trigeminal ganglia. This selective access to the CNS but not the trigeminal ganglia by dural application is the reverse of what is observed for the administration of agents by IV infusion, which accesses trigeminal ganglia but not CNS (49). Although the CSF would be expected to circulate to the subarachnoid space above the trigeminal ganglia, the entry of substances into the ganglia from the CSF would require penetration through the epineurium covering the ganglia. If the degree of penetration of the epineurium by PGE₂ were of a similar order of magnitude to the degree of penetration of the cranial dura, then the amount that would be expected in the ganglia samples would be below our detection limit by at least three orders of magnitude. It should be noted that the trigeminal ganglia samples were approximately fourfold smaller than the smallest of the CNS samples (spinal cord), and consequently would have required a proportionately greater increase per unit weight in order to attain a level that could be detected above background.

A number of methodological aspects of our preparation could potentially affect our dural permeability measurements. First, the craniotomy needed for the direct dural application of chemicals (both in our study and in the preclinical headache studies) inevitably alters the state of the dura (for example, by inducing dural mast cell degranulation (45, 50) and could potentially affect dural permeability. Such effects would also be present in the preclinical headache studies to which we are comparing our results. One difference between our study and the preclinical headache studies is the use of anesthetized versus awake, freely moving animals. It was beyond the scope of our study to investigate the possible effects of anesthesia and movement, although it may be noted that anesthesia actually facilitates the clearance of agents from the interstitial space and CSF (51), suggesting that in awake behaving animals the inflammatory mediators might be retained for a longer period of time in the parenchyma and CSF.

The present results, with two commonly used inflammatory mediators—5HT and PGE₂—have implications for the interpretation of previous findings with preclinical headache models and also for the design of future studies. The possibility of central effects from durally applied agents in prior studies raises some uncertainty about whether the observed behavioral changes can be attributed entirely to the activation of dural nociceptors. This is particularly true for those agents that were applied at concentrations several orders of magnitude above physiological levels. Behavioral effects observed with agents such as low pH or hypotonic solution (52–54), which differ from physiological fluids by smaller amounts, may be more safely attributed to a dural site of action. For future studies with such models it would be of interest to determine which of the observed behavioral changes can be evoked by the lower concentrations that have been shown to be effective for activating and sensitizing meningeal nociceptors (6, 9). In addition, the possibility of central effects will be increased in studies that use mouse instead of rat (30), due to a higher transdural permeability (27).

Key Findings

Application of the inflammatory mediators 5-HT and PGE₂ to the rat’s cranial dura, as is done in preclinical models of headache, results in entry of these agents into the CSF and CNS.

Estimates of the increase in CSF and CNS levels of 5-HT and PGE₂ that would be attained in the preclinical headache studies following exogenous administration are comparable to or larger than normal endogenous levels and potentially large enough to have physiological effects.

The finding that the cranial meninges are permeable to key proinflammatory mediators raises some uncertainty about whether the observed behavioral changes observed in prior studies can be attributed entirely to activation of dural nociceptors.