Experimental Procedures

Animals and Materials

Male Sprague-Dawley rats weighing approximately 300 g were purchased from Taconic, Inc. (Germantown, NY) and were kept three per cage in plastic, wire-top cages in an animal facility supervised by veterinarians. They were kept in a 24-h light/dark cycle and were provided food and water ad libitum. All efforts were made to minimize animal suffering.

Triton X-100 was obtained from New England Nuclear (Boston, MA). The scintillation Ready Fluor III cocktail was purchased from Beckman Instruments, Inc. (Fullerton, CA). Metrizamide was from Accurate Chemical and Scientific Corp. (Westburg, NY). Dextrose and paraformaldehyde were obtained from Fisher (Pittsburgh, PA). Imipramine (IMI) was a gift from Ciba Pharmaceutical Co. (Summit, NJ). Compound 48/80 (C48/80) was purchased from Sigma Chemical Co. (St. Louis, MO). Toluidine blue was obtained from Aldrich Chemical Co. (Milwaukee, WI). Glutaraldehyde was purchased from EM Sciences (Fort Washington, PA). Cacodylate HCl buffer, tannic acid, osmium tetroxide (OsO₄), propylene oxide, epoxy resin, hardener, and uranyl acetate were obtained from Polysciences, Inc. (Wallington, PA). N^α-mH was purchased from Calbiochem Corp. (La Jolla, CA). Thioperamide (Th) was a gift from Dr. M. Moskowitz (Massachusetts General Hospital, Harvard Medical School, Boston, MA). All other chemicals used were of analytical grade.

Perfusion System

A four-well, customized perfusion chamber was used to allow simultaneous investigation of four different conditions (Lambracht-Hall et al., 1990). The volume of each well was 1 ml and was optimal for the collection of sufficient brain perfusate to detect 5-HT and histamine. The top and bottom parts of the perfusion chamber were sealed with rubber o-rings and were held in place with four adjustable bolts. Air vents at the top of the wells were used to purge entrapped air and were also sealed by screws. The Plexiglas chamber wells were separated with coverslips from steadily circulating water maintained at 37°C. The temperature of the perfusion solutions was also held at 37°C by a standard circulating water bath. The brain slices were perfused by a four-channel microprocessor pump (MCP 2500; Buchler, Saddle Brook, NJ). To avoid artificial changes in the concentration of the released mediators by occasional air bubbles in the tubing, all solutions were first passed through four in-line air traps. The perfusion solution then entered at the top of each well, bathed the tissue sitting on a round Teflon mesh filter (1 mm pore size), and exited at the bottom of the chamber to a four-line simultaneous fraction collector (1500 RTR VII; ISCO Lincoln, NE).

Preparation of Brain Samples

Rats were decapitated, and the skull was opened through the midline. The brain was rapidly removed and put into cold oxygenated Krebs-Ringer-bicarbonate buffer (KRB; Lambracht-Hall et al., 1990) containing 118.0 mM NaCl, 4.85 mM KCl, 1.15 mM KH₂PO₄, 1.15 mM MgSO₄, 25.0 mM NaHCO₃, 2.5 mM CaCl₂, and 11.1 mM dextrose, at pH 7.4. The thalamic and hypothalamic areas were isolated and cut manually into slices of approximately 1.5 × 1.5 × 0.5 mm using disposable microblades in a cold room. The slices were then introduced together into cold oxygenated KRB at 4°C and were used as described below.

5-HT and Histamine Release from Perfused Brain Tissue

[³H]5-HT (specific activity, 25.4 Ci/mmol [³H]5-HT; 5 × 10⁻⁷ M) was added to the pooled brain slices, which were then incubated in 37°C for 15 min. The slices were then washed four times with 40 ml of KRB containing 10⁻⁶ M IMI to prevent 5-HT reuptake (Lambracht-Hall et al., 1990). The presence of 10⁻⁶ M IMI had no effect on peritoneal MC secretion as previously reported (Lambracht-Hall et al., 1990). Approximately equal amounts of tissue by weight (455 ± 35 mg) were randomly distributed into the four individually perfused wells. The brain slices were initially perfused with KRB (pH 7.4) at a flow rate of 5 ml/min for 40 min to reduce nonspecific background; the flow rate was then reduced to 1 ml/min. All perfusion solutions were continuously saturated with 95% O₂/5% CO₂. After 12 min of slow wash, the tissue was perfused for 36 min with different dilutions of N^α-mH or Th to investigate any effect on MC morphology or mediator release. The slices were then stimulated for 24 min with C48/80 (100 μg/ml), the classic MC secretagogue that has been previously shown to require this high concentration for a noncytotoxic effect on CNS MCs (Lambracht-Hall et al., 1990), followed by a 20-min wash with KRB. Subsequently, the tissue was perfused for 24 min by modified KRB containing 40 mM KCl and 82.8 mM NaCl to cause maximal neuronal depolarization of endogenous histamine and 5-HT (Atack and Carlsson, 1972). The perfusion was finished by KRB wash for 12 min. Fractions of perfusate were collected every 4 min, and aliquots of 0.5 ml were mixed with 1.5 ml of scintillation fluid, vortexed, and counted for radioactivity reflecting serotonin release in a β scintillation counter. Other aliquots were used for determination of histamine release.

C48/80 was deemed selective for stimulation of MCs because 1) stimulation of brain slices with C48/80 did not affect subsequent stimulation with high K⁺ (Lambracht-Hall et al., 1990); 2) C48/80 did not trigger release of neuropeptides expected to be released from neurons (Marathias et al., 1991); 3) C48/80 induced 5-HT release from thalamus (Goldschmidt et al., 1985) and hypothalamus (Pollard et al., 1976), areas rich in MCs, but not from cerebellum, which does not contain MCs (Lambracht-Hall et al., 1990); 4) 5-HT release from neurons was dependent on extracellular calcium (Verdiere et al., 1975; Subramanian and Mulder, 1976), whereas secretion from stimulated MCs was independent, as has been described for the action of C48/80 on peritoneal MCs (Subramanian and Mulder, 1976; Lambracht-Hall et al., 1990); 5) 5-HT secretion by C48/80 from thalamic-hypothalamic slices was reduced by cromolyn, which inhibits MCs but not neurons (Lambracht-Hall et al., 1990); and 6) treatment with capsaicin, which depletes peptidergic sensory neurons, had no effect on C48/80-induced brain 5-HT release (Marathias et al., 1991).

Histamine Assay

A commercially available histamine radioimmunoassay (RIA) kit (Kabi-Pharmacia, Piscataway, NJ) was used under conditions in which histamine could be detected. Because of high cross-reactivity of the primary antihistamine antibody with the N^α-mH added in some experimental conditions, histamine was measured only when N^α-mH was not used, such as in control and Th-treated samples. The level of detection of the radioenzymatic assay (20 pg/ml) and that of HPLC (<50 pg/ml), both of which could distinguish histamine from N^α-mH, were not sufficiently sensitive to detect histamine in brain slices.

Electron Microscopic Procedures

The perfused thalamic and hypothalamic slices or purified homogeneic peritoneal MCs were fixed in modified Karnovsky’s medium containing 3% glutaraldehyde, 2% paraformaldehyde, and 0.5% tannic acid in 0.1 M cacodylate HCl buffer (pH 7.3) for 4 h at room temperature followed by 12 h at 4°C. The next day, the fixative was decanted, and the tissue was washed five times over the period of 6 h in 0.1 M cacodylate HCl buffer (pH 7.3). Then, the samples were osmicated with 1% OsO₄ in 0.1 M cacodylate HCl buffer (pH 7.3) for 2 h at room temperature. The tissue was then washed with cacodylate HCl buffer and dehydrated in a graded series of alcohol at 4°C, the last one consisting of absolute alcohol, which was changed four times; 100% propylene oxide was added and changed twice. For tissue impregnation, the Epon was prepared as follows: 100% propylene oxide was added to an equal volume of epoxy resin mixed with hardener. Infiltration was performed for 12 h at room temperature. Embedded tissue was then put in 100% epoxy resin with hardener and placed into a 56°C oven for 48 h. Next, 1-μm-thick sections were stained with 0.1% Toluidine blue (pH 2.5) and MCs were first identified by light microscopy. Areas of interest were selected and sections for electron microscopy were cut at 1000 Å and mounted on copper mesh grids. The sections were then stained with 7.5% uranyl acetate in 50% alcohol and lead salts and were examined with a Philips 300 Transmission Electron Microscope (Dimitriadou et al., 1990).

Electron Microscopic Morphometry

Brain MCs from each experimental group were photographed at a magnification ranging from 9600× to 16,000×. Each photograph contained a single MC at a magnification that allowed for individual evaluation of each secretory granule present. The total number of granules was also noted for the calculation of percentages represented by each group. Granules were categorized as intact or activated according to the following criteria:

Intact.

Intact granule contents are uniform in density and lack any substructure or vacuoles.

Activated.

Empty granules have little or no residual density present with empty vesicles (“honeycomb”). Partially empty granules have irregular density of granule contents with vacuoles or empty vesicles and contain “fibrillar material,” which is granular contents of varying amount and density, usually less than that of normal granules, without homogeneous density.

Prints were examined by an investigator blinded to the experimental code and were scored into four categories as described above. Scores were then segregated into their appropriate experimental treatment group, and results are presented as percent of total.

Peritoneal MC Purification and Amine Release

MCs were obtained from the peritoneal cavity of the same rats, the brain of which was used for the perfusion experiments by lavage in HEPES-buffered Locke’s solution containing 150 mM NaCl, 5 mM KCl, 2 mM CaCl₂, 5 mM HEPES, 1 g/liter BSA, and 1 g/liter dextrose (pH 7.2). MCs were then purified (purity > 90%) over 22.5% metrizamide (350g for 10 min), washed three times in Locke’s solution (200g for 5 min), and then labeled with [³H]5-HT (2 μCi/10⁶cells/2 ml) at 37°C for 45 min. After two washes, the cells were resuspended in Locke’s solution. Aliquots of 10⁵cells/100 μl were then added to different concentrations of the H₃ receptor ligands to be tested for 90 min, and the cells were stimulated by 0.1 μg/ml C48/80 in the experiments with N^α-mH and 0.05 μg/ml in the experiments with Th for an additional 10 min at 37°C as needed. The cells were then centrifuged at 400g for 10 min, and 0.5 ml of supernatant was added to 3 ml of scintillation fluid. The pellets were solubilized with 0.5 ml of 2% Triton X-100 and added to 3 ml of scintillation fluid as well. Radioactivity was determined in a β scintillation counter. All four tissue samples were weighed at the end of every experiment. Histamine was determined by the same RIA described above.

Viability

Viability was assessed according to the amount of lactic dehydrogenase (LDH) present in the perfusate using a commercially available kit (Sigma Chemical Co., St. Louis, MO).

Statistical Analysis

For mediator release from brain slices, the area under the curve (in counts per minute × minute) was calculated using trapezoidal areas between the experimental points. All data entry, formatting, calculations of area under the curve, statistical analysis, and graphing were done using a custom-designed program based on Excel with a Macintosh II cx computer. The data (in arbitrary units of counts per minute × minutes) were adjusted to eliminate differences in weight by dividing each perfused tissue sample by the average weight (in milligrams) value of all controls and all other samples taken together. Comparisons of results from experiments regarding 5-HT release from purified peritoneal MCs and from ultrastructural observations were made between the means of each condition using ANOVA. A value of p < .05 is considered statistically significant.

Results

Effects of N^a-mH and Th on Rat Brain MC Ultrastructure.

Perfusion of brain slices with KRB alone (control) resulted in ultrastructural changes of about 14% of MC secretory granules in 9% of studied MCs (Table 1and Fig. 1). These control cells showed the typical appearance of connective tissue MCs (CTMCs; Fig. 1) with numerous round, homogeneous, electron-dense granules and an unlobulated nucleus with peripherally clumped chromatin (Fig. 1). Stimulation with C48/80, however, resulted in activation of 54% of secretory granules in 42% of MCs (Table 1 and Fig. 2A). The morphology of the granules in MCs activated by C48/80 was quite striking and was distinct from classic degranulation, with typical compound exocytosis seen in homogeneic peritoneal MCs stimulated by 0.1 μg/ml C48/80 (Fig. 3D). Instead, these activated brain MCs showed altered granular content characterized by partially filled or empty multivesicular secretory granules often with a “honeycomb” structure (Fig. 2A). This appearance was very similar to that of MCs pretreated by both Th and N^α-mH before activation with C48/80 (Fig. 2B). Such changes were only minimally present in unstimulated brain tissue (Fig. 1), suggesting that they were not due to tissue incubation or preservation.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Transmission electron micrograph from a control brain MCs adjacent to an arteriole (vl) vessel lumen. Note numerous, intact electron dense granules. n, nucleus; g, intact granule; e, endothelium; er, erythrocyte; p, pericyte. Original magnification, 13,800×.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Transmission electron micrographs of brain MCs after (A) stimulation with C48/80 alone (100 μg/ml) or (B) pretreatment with N^α-mH (10⁻⁶ M) and Th (10⁻⁶ M) before C48/80 (100 μg/ml). Note similar degree of activation in both cases. Straight arrows indicate empty or partially empty granules; curved arrows indicate maximally affected granules containing amorphous fibrillar material. N, neuronal axons; g, intact granule; e, endothelium; vl, vessel lumen. Original magnification: A, 9600×; B, 16,800×.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Transmission electron micrographs of purified peritoneal MCs after (A) control, (B) N^α-mH (10⁻⁶ M) alone, (C) Th (10⁻⁶ M) alone, (D) C48/80 (0.1 μg/ml) alone, (E) stimulation with C48/80 after pretreatment with N^α-mH (10⁻⁶ M), or (F) after pretreatment with Th (10⁻⁶ M). Original magnification: 8850×.

Pretreatment with the H₃ receptor agonist N^α-mH (10⁻⁶ M) significantly reduced (p < .05) granule activation from 54% with C48/80 to 19% in 11% of MCs (Table 1, Fig.4A), whereas the antagonist Th (10⁻⁶ M) significantly increased (p < .05) such activation to 68% (Table 1, Fig. 4B). Th alone (10⁻⁶ M) increased basal control brain MC activation from 14 to about 21% (results not shown). When both drugs (N^α-mH and Th) were used simultaneously at equimolar concentrations, activation due to Th was reduced to 37% (Table 1, Fig. 2B); neither drug was used in excess of the other to investigate the possibility of complete reversal of each other’s effects. Neither N^α-mH nor Th at 10⁻⁶ M had any effect on homogeneic peritoneal MC activation either on their own (Fig. 3, B and C) or in response to stimulation by C48/80 (Fig. 3, E and F).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Transmission electron micrographs of brain MCs stimulated with C48/80 (100 μg/ml) after pretreatment with (A) N^α-mH (10⁻⁶ M) or (B) Th (10⁻⁶M). Note the reduced activation in the presence of N^α-mH and the obvious potentiation of the appearance of granules that have a multivesicular, honeycomb-like appearance in the Th-treated sample. Straight arrows indicate empty or partially empty granules; curved arrows indicate affected granules with amorphous fibrillar material. n, neurons; g, intact granule; vw, vessel wall; vl, vessel lumen. Original magnification: A, 13,800×; B, 8850×.

5-HT Release from Brain Slices.

N^α-mH used on its own did not have any statistically significant effect on 5-HT release from brain slices in any of the concentrations used (Fig.5). On the other hand, Th used at 10⁻⁶ M without any other stimulus significantly (p < .00005) increased 5-HT release from about 42 to almost 64% (Fig. 5). The addition of Th to N^α-mH (10⁻⁶ M for both) reversed (p > .05 compared with control) this increase (Fig. 5). These results imply there may be some tonal inhibition of 5-HT release by endogenous histamine.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

5-HT release from brain slices perfused with KRB. ∗, statistically significant difference of that condition compared with the control. Control (n = 11), 10⁻⁶ M N^α-mH (n = 7); 10⁻⁴ M or 10⁻⁵ M N^α-mH, 10⁻⁶ M Th; 10⁻⁶ M N^α-mH plus 10⁻⁶ M Th (n = 6); and 10⁻⁷ M or 10⁻⁸ M N^α-mH (n = 5).

Secretion of 5-HT from brain slices in response to 100 μg/ml C48/80, which is considered a specific MC secretagogue, increased to 67.4 ± 10.2 cpm/mg × min (p < .05). This amount was reduced to 48.6 ± 7.9 cpm/mg × min by 10⁻⁷ M N^α-mH (p < .0008), as shown in Fig.6. On the contrary, Th (10⁻⁶ M) significantly (p < .0007) increased the release of 5-HT in the presence of C48/80 to 102.2 ± 14.4 cpm/mg × min. This increase is more than (p < .05) the amount released by Th alone, which is almost equivalent to that of C48/80 alone, indicating that Th augmented C48/80-induced 5-HT release. When Th (10⁻⁶ M) was used along with N^α-mH (10⁻⁶ M) in the perfusion, the drugs largely neutralized each other’s effects (Fig. 6). It should be noted that the effect of N^α-mH was not dose dependent and was highest between 10⁻⁶ and 10⁻⁷ M.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6

5-HT release from brain slices after stimulation with 100 μg/ml C48/80 in the presence of different concentrations of N^α-mH. p values are listed comparing each experimental condition to the control plus C48/80. Control (n = 11); C48/80 alone (n = 7) was statistically significant (p < .05) compared with control; 10⁻⁶ M N^α-mH (n = 7); 10⁻⁴ M or 10⁻⁵ M N^α-mH, 10⁻⁶ M Th; 10⁻⁶ M N^α-mH plus 10⁻⁶ M Th (n= 4); and 10⁻⁷ M or 10⁻⁸ M N^α-mH (n = 3). The combined effect of N^α-mH and Th was not statistically different from that of C48/80 alone (p > .05).

Incubation of brain slices with 40 mM K⁺ to trigger neuronal depolarization increased (p < .05) 5-HT secretion from 43.0 ± 3.7 to 68.1 ± 20.9 cpm/mg × min. N^α-mH reduced neuronal 5-HT secretion in all concentrations used (p < .05) but maximally inhibited it to 39.6 ± 8.6 cpm/mg × min at 10⁻⁴ M (p < .003). Th alone powerfully increased neuronal 5-HT secretion to 109.2 ± 26.9 cpm/mg × min (p < .0007), again indicating tonal inhibition of neuronal secretion by histamine. Unfortunately, the addition of both drugs was not done at varying concentrations. The simultaneous presence of both N^α-mH (10⁻⁶ M) and Th (10⁻⁶ M) in the perfusate buffer largely neutralized each other’s effects (Fig.7).

• Download figure
• Open in new tab
• Download powerpoint

Figure 7

5-HT release from brain slices after stimulation with 40 mM KCl. P values are listed comparing each experimental condition to the control. Control (n = 11); 10⁻⁶ M N^α-mH (n = 7); 10⁻⁴ M or 10⁻⁵ M N^α-mH, 10⁻⁶ M Th; 10⁻⁶ M N^α-mH plus 10⁻⁶ M Th (n = 6); and 10⁻⁷ M or 10⁻⁸ M N^α-mH (n = 5).

Histamine Release from Brain MCs and Neurons.

The effect of N^α-mH on histamine release from brain slices was difficult to assess because of high cross-reactivity of the primary antihistamine antibody used in the RIA with the N^α-mH added in these experiments. Despite an apparent decrease of histamine release in the N^α-mH-treated samples, the standard deviation was still too high to make any meaningful statistical evaluation possible. Determination of histamine radioenzymatically or by HPLC was not sufficiently sensitive to detect the brain amine levels released in the perfusate (results not shown).

The increase of histamine released due to Th (n = 4), however, was not affected by cross-reactivity with the antihistamine antibody (Fig. 8). Th alone increased basal histamine release from 0.002 ± 0.001 to 0.006 ± 0.003 μg/ml × min/mg. The presence of 10⁻⁶M Th increased stimulation with 100 μg/ml C48/80 from 0.007 ± 0.002 to 0.012 ± 0.0034 ng/ml/mg × min/mg (p < .05). The same was true for stimulation by K⁺ (40 mM), where Th increased histamine release from 0.006 ± 0.002 to 0.017 ± 0.004 ng/ml × min/mg (p < .003). It was, therefore, evident that the H₃ receptor agonist reversed some tonal inhibition of histamine on its own secretion from both the C48/80 and high K⁺-sensitive sources.

• Download figure
• Open in new tab
• Download powerpoint

Figure 8

Histamine release from brain slices in the presence of 10⁻⁶ M Th stimulated by either 100 μg/ml C48/80 (black columns) or 40 mM KCl (gray columns). p values are listed comparing the results in the presence or absence of Th.

5-HT and Histamine Release from Purified Rat Peritoneal MC.

The release of 5-HT from peritoneal MCs stimulated with C48/80 was not affected by either the H₃ receptor agonist N^α-mH (Fig. 9A) or the antagonist Th (Fig. 9B). The same was true for histamine release (results not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 9

The effect of H₃ receptor-active drugs on the release of 5-HT from rat peritoneal MCs induced by 0.1 μg/ml C48/80 (A) N^α-mH and (B) that of Th (n = 6).

Viability.

Viability was assessed by the ultrastructural appearance of the cells and the lack of significant amounts, compared with controls, of LDH in the perfusate.

Discussion

Brain MCs are almost exclusively perivascular, often in close contact with neurons (Dimitriadou et al., 1990), and were characterized as CTMC because they contained histamine, rat mast cell protease I, heparin, as well as mRNA for IgE-binding protein (Pang et al., 1996). Electron microscopic observations clearly indicated that the H₃ receptor-active agents affected brain MC secretion. MC intragranular changes after stimulation with C48/80 were increased in the Th-treated group but decreased when treated with N^α-mH. The ultrastructure of activated brain MCs was unique in that it was not accompanied by typical exocytosis with extrusion of granule contents outside the cell. Instead, affected secretory granules showed a multivesicular substructure with or without electron-dense content. Such ultrastructural appearance could not be an artifact of fixation because control brain MCs contained unaltered, typical, electron-dense granules. Activated brain MCs with a similar morphological appearance had previously been noted by us in response to C48/80 or carbachol (Dimitriadou et al., 1990), as well as by other investigators (Ibrahim et al., 1979). CTMC-like cells with heterogeneous electron-dense granules and lipid bodies are also found perivascularly. These neurolipomastocytes are distinct from other granular cells (Dimitriadou et al., 1990), and lipid bodies are frequently seen in immature MCs (Dvorak et al., 1993), where they may be important for arachidonate metabolism.

The H₃ receptor-active drugs used here had no effect on peritoneal MC activation, as judged by ultrastructural observations or amine secretion. Our results support previously reported findings showing H₃-active agents not to have any effect on rat peritoneal (Kohno et al., 1994; Stenton and Lau, 1997) and lung (Dimitriadou et al., 1994) MCs, as well as on human adenoidal MC histamine secretion (Bent et al., 1991). However, the possibility remains that H₃ agonists may affect secretion of other mediators from extracranial MCs, as was shown for TNF-α (Bissonnette, 1996). The present results show that 5-HT released from nonneuronal compartments by C48/80, presumably from brain MCs, was reduced by the H₃ receptor agonist N^α-mH, whereas both released 5-HT and histamine were increased by the H₃ receptor antagonist Th. These findings suggest that brain H₃ receptors regulate the release of histamine and 5-HT from both neuronal and nonneuronal sources. A similar inhibitory action on both neurotransmitters had previously been reported but only for neurons (Arrang et al., 1983; Schlicker et al., 1988; Oishi et al., 1990).

Other physiologically relevant secretogogues, such as acetylcholine or substance P, were not used here because their effects are not specific for MCs. Despite the high amount required to penetrate the brain slices due to its cationic nature, C48/80 had been shown to be a selective secretagogue for brain MCs because it did not affect neuronal secretion, was inhibited by cromolyn, and was not cytotoxic as judged by ultrastructural appearance and lack of LDH in the perfusate (Lambracht-Hall et al., 1990). Even though C48/80 was reported to induce release of monoamines from brain slices obtained from W/W^v MC-deficient mice (Lovenberg and Cubeddu, 1988), the amount of C48/80 used in that study was at cytotoxic levels and cell viability had not been determined. Secretion of 5-HT from brain slices was calculated as the area under the curve (Lambracht-Hall et al., 1990), whereas that from peritoneal MCs was expressed as percent of total. This difference in calculation of the results was necessitated because purified MCs, unlike brain slices, are the sole source of histamine or [³H]serotonin. In contradistinction, thalamic-hypothalamic MCs contribute about 50% of total brain histamine and much less of 5-HT; in fact, histaminergic neurons are known to exist in the rat hypothalamus (Panula et al., 1984). Early attempts at purifying brain MCs failed because they degranulated during the process of purification.

N^α-mH inhibited neuronal 5-HT secretion in thalamic-hypothalamic slices over the concentration range of 10⁻⁶ to 10⁻⁸ M by a maximum of about 30%, which is comparable with the 25% inhibition reported for rat brain cortex (Schlicker et al., 1988). However, no dose-response relationship was evident for C48/80-induced, presumably brain, MC secretion, which was also reduced by about 25% at 10⁻⁶ M. This discrepancy may be due to the low affinity of N^α-mH for H₃receptors on brain MCs, to the presence of only a few H₃ receptors; to the possible existence of H₃ receptor subtypes (Leurs et al., 1996), possibly with low-affinity H₃ receptor (West et al., 1990); or to a non-receptor-mediated effect. The lack of a dose-response had previously also been reported for the inhibitory effect of the H₁ receptor antagonist hydroxyzine on histamine release from rat peritoneal MCs (Theoharides et al., 1985). H₃ receptors have also been functionally identified in the dura mater (Matsubara et al., 1992), where MC activation (Dimitriadou et al., 1991) and vascular permeability (Dimitriadou et al., 1992) by trigeminal nerve stimulation were also inhibited by presynaptic 5-HT receptor agonists (Buzzi et al., 1992). However, such inhibition did not occur in tongue MCs treated similarly (Buzzi et al., 1992), supporting a functional difference between intracranial and extracranial MCs.

We were not able to directly investigate the effect of N^α-mH on histamine release due to the high background resulting from cross-reactivity of the primary anti-histamine antibody with the N^α-mH added during the perfusion. Reversed phase HPLC and a radioenzymatic assay could distinguish histamine from mH, but the sensitivity was at least 10-fold lower than with the RIA kit and could not detect histamine in the perfusate. An apparent regulatory effect of H₃ receptor ligands on histamine release, however, was suggested by the fact that the H₃receptor antagonist Th increased brain histamine release stimulated with either C48/80 or high K⁺; this peramide also increased the morphological changes evident with C48/80. Th had no effect on peritoneal MCs as previously reported for adenoidal MC secretion (Bent et al., 1991). These results imply that Th may be releasing some tonal inhibitory action of brain histamine. The inhibitory effect of H₃ receptor agonists on the release of [³H]histamine from cerebral cortex slices loaded with [H³] histidine reported previously was possible because histaminergic neurons have high histidine decarboxylase activity (Pollard et al., 1976; Panula et al., 1984). Inhibition of [³H]histamine release by electrical stimulation (Van der Werf et al., 1987) or high K⁺ was also maximal at 10⁻⁶ M exogenous histamine as reported here. Exogenous histamine also inhibited [³H]5-HT release from electrically stimulated rat brain cortex slices (Fink et al., 1990) where Th increased monoamine release. The MCs in the thalamic-hypothalamic area, however, have low histidine decarboxylase activity and do not depolarize (Pollard et al., 1976). They would not, therefore, synthesize adequate radiolabeled histamine for detection after stimulation with C48/80. Moreover, C48/80 previously used to stimulate hypothalamic MCs apparently impaired synthesis of [³H]histamine from [³H]histidine (Verdiere et al., 1975), making radioactive histamine detection inappropriate for our experiments.

The importance of brain MCs was reviewed recently (Theoharides, 1996;Kines and Powell, 1997). Mediators secreted from perivascular meningeal MCs (Dimitriadou et al., 1992) could contribute to the pathophysiology of migraines and cluster headaches (Theoharides, 1983). For instance, MCs close to nerve fibers from the temporal artery showed ultrastructural changes indicative of secretion only in the affected side of patients with cluster headache (Liberski and Mirecka, 1984). Moreover, dura MCs degranulated in response to acute psychological stress (Theoharides et al., 1995), which is known to trigger or worsen migraines. Little is known about the distribution or function of H₃ receptors or human cells. One study suggested that much longer preincubation with H₃ agonists was required to inhibit human monocyte function compared with rat MCs (Bissonnette, 1996). H₃ receptor agonists might be useful as prophylactic agents by reducing the release of brain MCs and neuronal vasoactive molecules.