Membrane-bound histamine N-methyltransferase in mouse brain: possible role in the synaptic inactivation of neuronal histamine

Authors


Address correspondence and reprint requests to Lindsay B. Hough, Neuropharmacology and Neuroscience, Albany Medical College MC-136, 47 New Scotland Avenue, Albany, New York 12208, USA. E-mail: HoughL@mail.amc.edu

Abstract

In the CNS, histamine is a neurotransmitter that is inactivated by histamine N-methyltransferase (HNMT), a soluble enzyme localized to the cytosol of neurons and endothelial cells. However, it has not been established how extracellular histamine, a charged molecule at physiological pH, reaches intracellular HNMT. Present studies investigated two potential routes of histamine inactivation in mouse brain nerve terminal fractions (synaptosomes): (i) histamine uptake and (ii) histamine metabolism by HNMT. Intact synaptosomes demonstrated a weak temperature-dependent histamine uptake (0.098 pmol/min-mg protein), but contained a much greater capacity to metabolize histamine by HNMT (1.4 pmol/min-mg protein). Determination of the distribution of HNMT within synaptosomes revealed that synaptosomal membranes (devoid of soluble HNMT) contribute HNMT activity equivalent to intact synaptosomes (14.3 ± 2.2 and 18.2 ± 4.3 pmol/min-tube, respectively) and suggested that histamine-methylating activity is associated with the membrane fraction. Additional experimental findings that support this hypothesis include: (i) the histamine metabolite tele-methylhistamine (tMH) was found exclusively in the supernatant fraction following an HNMT assay with intact synaptosomes; (ii) the membrane-bound HNMT activity was shown to increase 6.5-fold upon the solubilization of the membranes with 0.1% Triton X-100; and (iii) HNMT activity from the S2 fraction, ruptured synaptosomes, and synaptosomal membranes displayed different stability profiles when stored over 23 days at − 20°C. Taken together, these studies demonstrate functional evidence for the existence of membrane-bound HNMT. Although molecular studies have not yet identified the nature of this activity, the present work suggests that levels of biologically active histamine may be controlled by an extracellular process.

Abbreviations used
AChE

acetylcholinesterase

COMT

catechol-O-methyltransferase

DAO

diamine oxidase

EST

expressed sequence tag

GPI

glycosylphosphatidylinositol

HNMT

histamine N-methyltransferase

KRH

Krebs-Ringers-HEPES

MTase

methyltransferase

SAM

S-adenosyl-l-methionine

tMH

tele-methylhistamine

UTR

untranslated region.

Histamine has several physiologically distinct roles in both the periphery (e.g. gastric acid secretion and inflammation) and the CNS (e.g. neurotransmission and neuromodulation; Hough 1999). Histamine mediates its effects through a variety of histamine receptors (H1, H2, H3, and H4 receptors) which trigger signaling cascades through G proteins (Hough 2001). Following histamine release and receptor activation, neuronal histamine must be inactivated to prevent receptor desensitization.

The neurotransmission of biogenic amines such as dopamine and norepinephrine is inactivated by specific transporters that facilitate the re-uptake of these transmitters into nerve terminals (Hoffman et al. 1998). One laboratory has found low levels of sodium-dependent histamine transport in primary cultures of chick embryonic astrocytes (Huszti et al. 1990) as well as in rat cerebral endothelial cells (Huszti et al. 1995). In addition, histamine uptake was also reported to occur in bone marrow cells (Corbel and Dy 1994), although this uptake bears pharmacological resemblance to the histamine H3 receptor (Corbel and Dy 1996). However, no neuronal transporter for histamine has been identified and the levels of histamine uptake identified in other cell types have been suggested to be too small to account for the rapid inactivation of histamine (Hoffman et al. 1998).

Histamine can be metabolized by two pathways via the enzymes diamine oxidase (DAO, EC 1.4.3.6) and histamine N-methyltransferase (HNMT, EC 2.1.1.8), but the former functions predominantly in the periphery (Schwartz et al. 1991). HNMT, thought to be the important mechanism for histamine metabolism in the brain, facilitates the transfer of a methyl group from S-adenosyl-l-methionine (SAM) to histamine to form tele-methylhistamine (tMH). It has been repeatedly shown that the levels of tMH closely follow histamine turnover (Hough and Domino 1979; Oishi et al. 1989; Ligneau et al. 1998; Barnes et al. 2001). Furthermore, inhibition of HNMT leads to an increase in neuronal histamine, suggesting that this enzyme plays a critical role in histamine inactivation (Hough et al. 1986; Itoh et al. 1991).

Subcellular (Kuhar et al. 1971) and subsynaptosomal (Snyder et al. 1974) fractionation studies have shown HNMT to be a soluble enzyme. In addition, immunocytochemical studies have recently shown this enzyme is localized to the cytosol of neurons and endothelial cells within the CNS (Nishibori et al. 2000). Hence, it is unclear how extracellular histamine (a charged molecule at physiological pH; Green 1967), can be metabolized by intracellular HNMT. To address this problem, we have presently characterized histamine transport and HNMT activity in mouse whole-brain synaptosomes and discovered a membrane-bound form of this enzyme.

Materials and methods

Animals

Male Swiss-Webster mice (30–45 g, Taconic Farms, Germantown, NY, USA) were housed in groups of five or six with food and water freely available. They were maintained on a 12-h light–dark cycle (lights on at 07.00 h, lights off at 19.00 h). All experiments were conducted between 09.00 and 15.00 h. All animal procedures were approved by the Institutional Animal Care and Use Committee of Albany Medical College.

Preparation of crude synaptosomes

Mice were anesthetized with Nembutal [100 mg/kg, intraperitoneally (i.p.), Abbott Laboratories, Chicago, IL, USA] and were decapitated 5 min later. Whole brains were removed, weighed, and homogenized in 10 mL of 0.32 m sucrose containing 5 mm HEPES (pH 7.5) in a glass homogenizer with 10–15 strokes of a Teflon pestle (a single brain per homogenizer). The homogenate was transferred to a 30-mL centrifuge tube and centrifuged at 1000 g for 10 min. The P1 pellet was discarded and the S1 supernatant was transferred to a new tube and centrifuged (12 000 g for 20 min with the brake set to 1). The resulting S2 supernatant (microsomes and soluble enzymes) was assayed in some cases. The resulting P2 pellet was re-suspended in a Krebs–Ringers–HEPES solution (KRH; 118 mm NaCl, 4.8 mm KCl, 2.5 mm CaCl2, 1.2 mm MgSO4, 20 mm HEPES; adjusted to a pH of 7.4 with NaOH) and washed twice with 2 mL of KRH (unless specified otherwise). With the exception of Fig. 1(e), this crude synaptosomal fraction was used for all studies.

Figure 1.

Histamine uptake and methylation in mouse whole brain synaptosomes. (a) Synaptosomes were pre-incubated for 15 min at either 4°C (▴) or 37°C (▾) before [3H]histamine (0.9 µm) was added. Reactions were continued at the same temperature for the times shown (abscissa) and then stopped by filtration. Histamine uptake (37°C−4°C, ●) is also shown. Results (pmol/mg protein) are the mean± SEM of triplicate determinations from two experiments. (b) HNMT activity (◆) was determined by incubating intact synaptosomes with 0.9 µm histamine and 9.8 µm[3H]SAM at 37°C for various times as shown. Results (pmol/mg protein) are the mean ± SEM of triplicate determinations from two experiments. The temperature-dependent uptake of [3H]histamine (●, data of Fig. 1a) is also plotted for comparison. HNMT activity was significant at each time point (p < 0.01) when compared to histamine uptake at the same time. (c) The effect of tMH (30 and 100 µm, T30 and T100, respectively) and metoprine (10 and 30 µm, M10 and M30, respectively) on HNMT activity (left) and histamine uptake (37°C−4°C as in Fig. 1a, right) wasmeasured. Synaptosomes were incubated for 5 min at 37°C with either 0.9 µm histamine and 9.8 µm[3H]SAM for HNMT activity (left) or 0.9 µm[3H]histamine for histamine uptake (right). *p < 0.001 indicates a significant difference from control HNMT activity. (d) Intact synaptosomes were incubated with 9 µm histamine and 9.8 µm[3H]SAM for 15 min, the assay was stopped on ice and the samples were centrifuged (12 000 g for 20 min). Supernatants (S) were either separated from pellets (P) to determine the [3H]tMH in each fraction separately, or supernatant and pellets were extracted together (S + p) to determine total [3H]tMH formed in the assay. Fractions each received 75 µL of potassium borate (pH 11) and [3H]tMH was extracted as described in Methods for the HNMT assay. Total cpms detected for S, P, S + P, and enzyme blank (Blank) are shown. (e) Synaptosomes were prepared as a crude P2 (left) or a purified (right) fraction as described in the Methods section and were either left intact (Int) or were hypotonically ruptured (Rup) before incubation with 9 µm histamine and 9.8 µm[3H]SAM at 37°C for 15 min. HNMT activity (pmol/min-mg protein) is shown. Data are from two experiments performed in triplicate. *,+p < 0.001 indicate significant differences from intact HNMT activity (Int) in crude and purified fractions, respectively. #p < 0.01 indicates a significant difference from the crude P2 ruptured fraction (Rup). Error bars are sometimes smaller than the data points (a–e). Tubes contained 0.2–0.5 mg of protein (a–e).

Preparation of purified synaptosomes

The preparation of Percoll (Amersham Pharmacia, Piscataway, NJ, USA) solutions and subsequent purification of synaptosomes was based on Nagy and Delgado-Escueta (1984). The P2 pellet (described above) was re-suspended (one brain per 1.35 mL) in 0.32 m sucrose containing 5 mm HEPES (pH 7.5) and 0.5 mL of this homogenate was then diluted in 8.5% Percoll/sucrose isoosmotic solution to a concentration of 7.5% Percoll. The 7.5% layer was layered onto the 10%/16% Percoll gradients and centrifuged at 15 000 g for 20 min with the brake set to 0. Purified synaptosomes (corresponding to Band C; Nagy and Delgado-Escueta 1984) were collected from the gradients, and placed on ice until assay.

Preparation of intact and osmotically treated synaptosomes

Intact synaptosomes were prepared by re-suspending the washed P2 pellet in KRH buffer to approximately 8–15 mg/mL of protein. Ruptured synaptosomes were prepared by re-suspending the washed P2 pellet in 900 µL of Milli-Q water per brain and homogenizing with a Polytron homogenizer for 30 s. The homogenate was placed on ice for 1 min and then homogenized for an additional 30 s. The homogenate remained on ice for 15 min before 100 µL of a 10× KRH buffer per brain was added to give a final concentration of 1× KRH (protein concentrations were approximately 8–15 mg/mL). This ruptured preparation was subjected to two KRH wash steps (unless specified otherwise). Intact and ruptured P2 homogenates were kept on ice until assay. Percoll-purified synaptosomes were either left intact by dilution with KRH, or hypotonically ruptured by dilution and homogenization in Milli-Q water and reconstituted as above. Some fractions were treated with 1 m NaCl, 0.1% Triton-X 100 (Sigma, St Louis, MO, USA), or Protease Inhibitor Cocktail (Sigma) as described. Use of Protease Inhibitor Cocktail followed the recommended protocol, but was used in a 2× concentration.

Uptake of [3H]histamine into crude synaptosomes

The washed (2×) P2 pellet was resuspended in KRH to a final protein concentration of 1–2 mg/mL [3H]histamine (14.76 Ci/mmol, NEN, Boston, MA, USA) was added at a concentration of 0.9 µm in a total volume of 400 µL. Incubations were terminated by the addition of 2 mL of ice-cold KRH and filtered onto Whatman GFC paper (pre-soaked in 0.1% polyethylenimine) with a Brandel filtration apparatus. The filters were washed twice with approximately 3 mL KRH. Each filter was placed in a plastic scintillation tube and 5 mL of cocktail (Ecoscint, National Diagnostics, Atlanta, GA, USA) was added. Samples were allowed to re-equilibrate for 12 h and were then counted in a scintillation counter. Blanks, consisting of samples containing all reagents except biological membranes, gave values between 200 and 400 cpm, which were subtracted from total cpm of all samples to obtain histamine uptake at 4°C and 37°C. Total cpm was always at least three-fold above filtered [3H]histamine blanks.

HNMT assay

The incubation of histamine and [3H]SAM with HNMT and subsequent extraction of [3H]tMH is based on a previously described method (Verburg et al. 1983). Plastic conical tubes contained [3H]SAM (1 µCi, 77 Ci/mmol, Amersham Pharmacia) along with unlabeled SAM (9.8 µm total, Sigma), histamine (9.0 µm unless specified otherwise, Sigma), and KRH in a total volume of 100 µL (where specified, tMH; Calbiochem, San Diego, CA, USA) or metoprine (Burroughs Wellcome Co., Research Triangle Park, NC, USA) were also included as HNMT inhibitors). All assays were performed in triplicate in a 37°C water bath for 15 min (unless specified otherwise). The HNMT reaction was stopped with the addition of 75 µL of 2.5 m potassium borate (pH 11) and the tubes were vortexed. To extract [3H]tMH, each tube received 4 mL of toluene/isopentanol (3 : 1), was capped and rapidly shaken for 5 min followed by centrifugation at 2440 g for 5 min. The organic layer (3.5 mL) was pipetted into a new conical tube containing 0.5 mL of 0.5 m HCl. The tubes were again rapidly shaken for 5 min and re-centrifuged. The organic layer was aspirated to waste and 1.5 mL of toluene/isopentanol (3 : 1) was added as a wash step. The tubes were rapidly shaken for 5 min and re-centrifuged. The organic layer was again aspirated to waste. Aliquots (0.3 mL) of the 0.5 m HCl layer were pipetted into a 20-mL glass scintillation vial and 15 mL of Aquassure (Packard, Meriden, CT, USA) was added. Scintillation vials were mixed by vortex and [3H]tMH was quantified by scintillation counting. In all experiments, enzyme blanks, consisting of samples containing all reagents except enzyme, gave values between 1500 and 3600 cpm, which were subtracted from total cpm of all samples to obtain enzymatic activity (see Fig. 1d for exception). Total cpm were always at least two-fold above enzyme blank, except in enzyme inhibitor studies (Fig. 1c). Substrate blanks (i.e. samples containing enzyme, but lacking histamine) were also determined in several experiments for both supernatant and membrane fractions. These values, found to be approximately 2% of total HNMT activity in both cases, were not subtracted from total cpm. In addition, 5 min of heating at 50°C inactivated both the soluble and membrane-bound HNMT enzymes by 96–97%.

Protein determination

The protein levels of mouse brain homogenates were determined using a BCA protein assay kit (Pierce, Rockford, IL, USA).

Data analysis

Where indicated, data were analyzed by analysis of variance (anova; with repeated measures for Figs 1b and 4) followed by Newman–Keuls post-hoc analyses with Statistica scientific software (StatSoft, Tulsa, OK, USA).

Figure 4.

Stability profiles of HNMT activity from different subcellular fractions. The S2 fraction (▴, 0.045–059 mg protein/tube), ruptured synaptosomal supernatant fractions (▪, prepared as described in Fig. 3a, 0.027–0.034 mg protein/tube), and the synaptosomal membrane fraction (●, prepared as described in Fig. 3b, 0.140–0.161 mg protein/tube) were incubated for 60 min at 4°C with 0.1% of Triton X-100 detergent and either assayed immediately for HNMT activity or stored at − 20°C and assayed on the day shown after the initial freeze (abscissa). The experiment was repeated in the presence and absence of protease inhibitors and the results from these two experiments (not different from each other) were averaged together. For each graph, the percent of day zero HNMT activity is represented (mean ± SEM, n = 6). Day zero HNMT activity in the absence of protease inhibitors were (mean ± SEM) 13.8 ± 0.2, 82.7 ± 1.1, and 55.9 ± 0.2 pmol/min-mg protein for synaptosomal membranes, ruptured synaptosomal supernatant fractions, and the S2 fraction, respectively. +p < 0.05, #p < 0.01, and *p < 0.001 indicate significant differences from the S2 fraction on the same day.

Results

Uptake versus methylation of histamine by synaptosomes

To assess the transport of histamine, [3H]histamine accumulation into synaptosomes was measured. Over various times between 2 and 20 min, a small but detectable temperature-dependent uptake of histamine was measured in intact synaptosomes (Fig. 1a). The uptake of histamine into synaptosomes reached 1.95 pmol/mg protein over 20 min.

Synaptosomes were then used to investigate the potential for histamine to be methylated by HNMT in these nerve terminals. Unexpectedly, intact synaptosomes were able to methylate histamine, with the activity reaching 28.9 pmol/mg at 20 min (Fig. 1b). Furthermore, when histamine uptake into intact synaptosomes was compared to the HNMT activity in an equivalent fraction, HNMT activity was nearly 15-fold greater than histamine uptake (Fig. 1b) providing a strong argument for the cellular role of HNMT.

The HNMT inhibitors tMH and metoprine were used to confirm that the activity being measured was due to HNMT. In addition, these inhibitors were used to block HNMT to ensure that histamine metabolism was not interfering with the ability to measure histamine uptake into synaptosomes (Fig. 1c). tMH inhibited HNMT activity by 83% and 84% at 30 µm and 100 µm, respectively, and metoprine inhibited HNMT activity by 88% and 94% at 10 µm and 30 µm, respectively. Neither tMH nor metoprine had any effect on histamine uptake over the same duration.

The greater HNMT activity compared to histamine uptake in synaptosomes prompted the investigation of the localization of labeled tMH after an HNMT assay. If [3H]tMH were found to be predominantly in the pellet fraction, then it would be good evidence that histamine was being methylated by soluble HNMT within synaptosomes. However, if [3H]tMH were discovered to be predominantly in the supernatant fraction, then it would support the hypothesis that histamine is not transported and subsequently methylated by soluble HNMT within synaptosomes, but rather is somehow methylated without entering synaptosomes. Therefore, [3H]tMH was separately extracted from the synaptosomal pellet and supernatant fractions following an HNMT reaction. Results (Fig. 1d) showed that 100% of the radiolabeled product was recovered in the supernatant fraction and none was found within the synaptosomal pellet (3H in pellet fraction was lower than 3H in enzyme blank). These findings suggested the possibility that histamine was being methylated by an extracellular reaction.

To ensure that the HNMT activity in the crude P2 fraction was specifically due to the nerve terminal fraction and not mitochondria or myelin (also present in a crude P2 preparation), synaptosomes were purified on a discontinuous Percoll gradient and HNMT activity of these purified synaptosomes was compared to the HNMT activity of a crude P2 synaptosomal preparation (Fig. 1e). Intact crude and purified synaptosomes contained similar levels of HNMT activity. In addition, to determine the total amount of HNMT present within synaptosomes under conditions where histamine and [3H]SAM were unobstructed by the lipid bilayer, synaptosomes were hypotonically ruptured and HNMT activity re-determined. When both purified and crude synaptosomes were ruptured, a much higher HNMT activity could be measured compared to intact synaptosome HNMT activity. Furthermore, ruptured purified synaptosomes resulted in slightly more HNMT activity (40.7 ± 2.3 pmol/min-mg protein) compared to ruptured crude synaptosome preparation (30.2 ± 3.1 pmol/min-mg protein), probably due to an enrichment of HNMT-containing membranes in the purified preparation. Comparisons of intact and ruptured fractions showed that approximately 25% of the total HNMT activity was detectable after the addition of extracellular substrates to intact preparations (Fig. 1e).

HNMT distribution within synaptosomes

To further investigate the mechanism of histamine inactivation, a detailed analysis of HNMT within synaptosomes was performed. Five sequential washes of the synaptosomal pellet were performed on ice. Each wash consisted of re-suspending the pellet in ice cold KRH and re-centrifugation. The supernatant of each wash step was then assayed for HNMT activity. The five washes reduced supernatant HNMT activity from 85 to 5 pmol/min-tube (Fig. 2a). HNMT activity was then measured in hypotonically ruptured synaptosomes, intact synaptosomes, and the supernatant of intact synaptosomes that were pre-incubated for 15 min at 37°C. As found previously, intact synaptosomes demonstrated only a fraction of the total HNMT activity detected by rupture (Fig. 2b, note break in ordinate). Surprisingly, about one-half of this intact HNMT activity (23 pmol/min-tube) was localized in the supernatant fraction as shown by the pre-incubated synaptosome supernatants (Fig. 2b). This increase in the supernatant HNMT activity compared to the last wash step (5 pmol/min-tube) was likely due to a change in the temperature when synaptosomes were taken from ice and incubated at 37°C. To determine the extent to which only intact synaptosomes contributed HNMT activity, the supernatant activity was subtracted from the intact activity (Int–Syn Sup, Fig. 2b), to reveal that intact synaptosomes contributed about 18 pmol/min-tube of HNMT activity.

Figure 2.

HNMT activity in intact and ruptured mouse synaptosomes. (a) Synaptosomes were prepared and washed five times (abscissa) in 2 mL of ice-cold KRH. Supernatant fractions from each wash step were incubated with 9 µm histamine and 9.8 µm[3H]SAM at 37°C for 15 min. HNMT activity (pmol/min-tube, mean ± SEM, ordinate) is shown. (b) Following the fifth wash step, HNMT activity (pmol/min-tube, mean ± SEM) was determined in either ruptured (Rup) or intact (Int) synaptosomes (prepared as described in the Methods section) by incubation with 9 µm histamine and 9.8 µm[3H]SAM at 37°C for 15 min. In addition, to determine the extent of HNMT appearance in the supernatant fraction during an HNMT assay, an equivalent preparation of intact synaptosomes was incubated for 15 min in KRH at 37°C, centrifuged at 12 000 g for 20 min, and the supernatant (Syn Sup) wasassayed for HNMT content as described above for Rup and Int synaptosomes. As the intact fraction contained activity due to both synaptosomes and the supernatant fraction, the HNMT activity attributed to only the intact synaptosomes was determined by subtracting the synaptosomal supernatant HNMT activity from the intact synaptosome HNMT activity (Int – Syn Sup). Data are from two experiments performed in triplicate. Tubes each contained 0.13–0.25 mg protein.

The above findings led us to the hypothesis that a membrane-bound form of HNMT exists on the plasma membrane which directly metabolizes extrasynaptosomal histamine. Hence, to determine directly whether synaptosomal membranes contained HNMT activity, synaptosomes were hypotonically ruptured and washed several times to remove the soluble fraction of HNMT (Fig. 3a). HNMT activity (16.5 pmol/min, Fig. 3b) was detectable in the membrane fraction, following the incubation with histamine and [3H]SAM. To determine the extent to which this HNMT activity was due only to membranes, supernatant HNMT activity (i.e. enzyme which was released from membranes during the 15 min HNMT assay) was estimated separately and subtracted from membrane HNMT activity. Interestingly,the resulting estimate of true membrane HNMT activity (14.3 ± 2.2 pmol/min-tube, Fig. 3b) was not significantly different from the intact synaptosome activity (18.2 ± 4.3 pmol/min-tube) estimated in Fig. 2(b).

Figure 3.

HNMT activity in mouse synaptosomal membranes. (a) Synaptosomes were hypotonically ruptured and washed four times (abscissa) in 2 mL of ice-cold KRH. Supernatant fractions from the hypotonic rupture and wash steps were incubated with 9 µm histamine and 9.8 µm[3H]SAM at 37°C for 15 min. HNMT activity (pmol/min-tube, mean ± SEM, ordinate) is shown. (b) HNMT activity (pmol/min-tube, mean ± SEM) was then determined in synaptosomalmembranes (Mem) by incubation with 9 µm histamine and 9.8 µm[3H]SAM at 37°C for 15 min. In addition, to determine the extent of HNMT in the supernatant fraction during an HNMT assay, an equivalent preparation of membranes was incubated for 15 min in KRH at 37°C, centrifuged at 12 000 g for 20 min, and the supernatant (Mem Sup) was assayed for HNMT content as described above for Mem. HNMT activity attributed to only the synaptosomal membrane fraction was determined by subtracting the synaptosomal supernatant HNMT activity from the synaptosomal membrane HNMT activity (Mem – Mem Sup). In addition, washed synaptosomal membranes were solubilized during a 60-min incubation at 4°C with 0.1% of Triton X-100 detergent and HNMT activity was measured (Mem + TX). Data are from two experiments performed in triplicate. Tubes each contained 0.13–0.25 mg protein. *p < 0.001 indicates a significant difference from membranes without Triton X-100 treatment (Mem). (c) Synaptosomal membranes were prepared as in (a). Membranes were then resuspended in either KRH or 1 m NaCl and incubated on ice for 30 min and then centrifuged. Membranes were then resuspended in KRH ± 0.1% Triton X-100 and incubated for 1 h on ice. HNMT activity was determined in KRH washed synaptosomal membranes (Mem), KRH washed synaptosomal membranes + Triton X-100 (Mem + TX 100), 1 m NaCl washed synaptosomal membranes [Mem (NaCl)], and 1 m NaCl washed synaptosomal membranes + Triton X-100 (Mem (NaCl) + (TX 100). Data represent triplicate determinations from two experiments. In one of these experiments, the pellet was only washed once, but identical results were obtained. +p < 0.001 indicate significant differences between Mem + TX 100 and Mem and Mem (NaCl) and Mem. #p < 0.001 indicates a significant difference between Mem (NaCl) + TX 100, and Mem (NaCl).

As an additional, independent test of the existence of membrane-bound HNMT, synaptosomal membranes were solubilized with 0.1% of the detergent Triton X-100 (Mem + TX 100, Fig. 3b). Under these conditions, HNMT activity was enhanced 6.5-fold by detergent treatment. In contrast, HNMT from the soluble S2 fraction was not affected by the same Triton X-100 treatment (data not shown). In addition, incubation of the P2 membranes with 1 m NaCl for 30 min, followed by centrifugation of the homogenate, and re-suspension of the pellet in KRH, did not result in a decrease in membrane-bound HNMT activity (Fig. 3c), as would be expected for a peripherally bound enzyme. In fact, washing synaptosomal membranes with 1 m NaCl significantly increased the activity of membrane-bound HNMT from 13.8 ± 1.4 pmol/min-tube to 43.9 ± 1.7 pmol/min-tube. However, this treatment did not increase the total amount of HNMT present in the preparation as determined by Triton X-100 treatment. These findings show that HNMT exists in a membrane fraction and suggest that it is this fraction which contributes to the extracellular methylation of histamine.

HNMT from different subcellular locations displays different stability profiles

A study of the stability of several mouse brain fractions containing HNMT was undertaken to assess possibleheterogeneity in forms of the enzyme. Specifically, soluble HNMT from the S2 fraction, soluble HNMT from hypotonically ruptured synaptosomes, and membrane-bound HNMT from synaptosomal membranes (solubilized in 0.1% Triton X-100), exhibited strikingly different stability profiles after storage at − 20°C and subsequent thawing and assay (Fig. 4). Soluble HNMT from the S2 fraction was stable over time in the presence of 0.1% Triton X-100. However, HNMT liberated from membranes by solubilization with Triton X-100 lost 73% of its activity after 23 days at − 20°C. In addition, the soluble HNMT from within synaptosomes did not show the same stability profile as the soluble HNMT from the S2 fraction, losing 92% of its activity after 23 days. The experiment was repeated in the presence of protease inhibitors (see Fig. 4 legend) with identical results.

Discussion

The removal of neurotransmitters from the synapse following neuronal release and receptor activation is critical in order to terminate cell signaling. For example, biogenic amines such as dopamine and norepinephrine are actively recycled back into neurons by specific re-uptake transporters (Hoffman et al. 1998), whereas acetylcholine is metabolized to choline and acetate by acetylcholinesterase directly within the synapse (Massoulie et al. 1999). For the neurotransmitter histamine, the route of inactivation has been unclear and was the focus of the present study.

The present findings, each discussed in further detail below, strongly suggest that neuronal histamine is inactivated by a membrane-bound form of a histamine-methylating enzyme. The evidence includes: (i) negligible levels of synaptosomal histamine transport; (ii) quantitatively significant histamine methylation by intact synaptosomes; (iii) extracellular localization of methylated product after this reaction; (iv) direct methylation of histamine by washed membranes; and (v) liberation of enzyme from membranes by detergent.

HNMT is the enzyme primarily responsible for the metabolism of histamine in the CNS (Hough 1999). Although earlier studies showed that purified synaptosomes contained HNMT (Snyder et al. 1974), it has never been shown how histamine methylation could occur in nerve terminals without the existence of a histamine transporter. For example, the transport of histamine in astrocytes, although small in magnitude, was suggested to be linked to intracellular methylation (Huszti 1990). Therefore, histamine transport in synaptosomes was the first priority of the present study.

Present results showing that mouse synaptosomes demonstrate a small degree of temperature-dependent histamine uptake (Fig. 1a) are similar to findings in chick glia (Huszti et al. 1990), rat astrocyte (Rafalowska et al. 1987; Huszti et al. 1994), and mouse bone marrow cells (Corbel and Dy 1994). However, in our laboratory, histamine uptake in mouse synaptosomes was not saturable and was not sodium- or chloride-dependent (unpublished observations), both of which are well-known characteristics of other biogenic amine transporters (Rudnick and Clark 1993). The small temperature-dependent effects observed may be due to differences in histamine diffusion or non-specific binding. Furthermore, the lack of effect of two HNMT inhibitors on the histamine ‘uptake’ (Fig. 1c) is convincing that histamine metabolism was not interfering with the ability to measure histamine uptake. Hence, in agreement with many previous investigators (Neame 1964; Honegger et al. 1974), we are unable to demonstrate a histamine transporter in nerve terminal fractions.

The methylation of histamine by intact synaptosomes (Fig. 1b) has not been previously reported. Coupled with a lack of histamine uptake (which was thought to be needed to perform the methylation), this finding suggested that histamine did not have to enter synaptosomes to be methylated. Strong direct evidence for this hypothesis was found by showing that labeled tMH was exclusively localized in the supernatant fraction following histamine methylation by intact synaptosomes (Fig. 1d). In addition, there is no evidence that SAM, a permanently ionized species, enters cells. Thus it was hypothesized that HNMT existed on the outside of synaptosomes where it could readily access extracellular histamine and SAM. The finding that the histamine-methylating activity of intact synaptosomes (18 ± 4.3 pmol/min-tube, Fig. 2b) can be accounted for in washed synaptosomal membranes (14 ± 2.2 pmol/min-tube, Fig. 3b), strongly suggests that a membrane-bound form of a histamine-methylating enzyme exists and may contribute to extracellular histamine metabolism.

The appearance of HNMT in the supernatant fraction after incubation of washed intact synaptosomes at 37°C (Fig. 2b) was unexpected. If HNMT associates with membranes during homogenization with sucrose, then the 37°C incubation may have liberated some of this enzyme which was not removed by washing at 4°C (Fig. 2a). This seems unlikely, as membranes washed the same way did not liberate the enzyme during an identical 37°C incubation (Fig. 3b). Synaptosomal integrity was investigated by measuring lactate dehydrogenase in the supernatant fraction following a 15-min pre-incubation of synaptosomes at 37°C and preliminary experiments did not show any evidence of synaptosomal disruption under these conditions (data not shown). An alternative explanation, that soluble HNMT within synaptosomes is liberated by 37°C incubation, seems possible, but the physiological relevance of this remains unclear. The hypothesis that intracellular HNMT is released by nerve terminals under physiological conditions has not yet been studied. However, we have performed experiments with unlabeled substrates to exclude the possibility that extracellular histamine and SAM can induce the appearance of extracellular HNMT, a hypothesis that might have accounted for the intact HNMT activity in synaptosomes (data not shown).

Although proteins can attach to plasma membranes by a number of different methods, two broad classes of ‘peripherally bound’ and ‘integrally bound’ proteins have been distinguished. Classical methods for distinguishing these types of proteins include high salt treatments that remove peripherally, but not integrally bound proteins, and detergents that solubilize both classes of proteins. HNMT was not reduced in the membrane fraction by incubation with 1 m NaCl (Fig. 3c). Interestingly, HNMT activity associated with membranes increased three-fold after high-salt washes, probably due to the liberation of other peripherally bound proteins that reduce substrate accessibility to HNMT. HNMT activity was increased by 6.5-fold after membrane solubilization with Triton X-100 (Fig. 3b), strongly suggesting the existence of an integrally bound histamine-methylating enzyme. For example, membrane-bound acetylcholinesterase (AChE, EC 3.1.1.7) solubilized in this manner showed a four-fold increase in activity (Dudai and Silman 1974). Early studies also found increases in brain HNMT after detergent treatment (Kuhar et al. 1971; Taylor and Snyder 1972), but the significance of the result was not clear at that time.

The activity of the newly characterized membrane-bound HNMT (15 pmol/mg-tube) is small when compared with the total enzyme levels in brain (278 pmol/min-tube). However, the former activity is more than sufficient to account for the in vivo rate of neuronal histamine inactivation. Specifically, mouse whole-brain histamine turnover occurs at a rate of 0.52 nmol/g tissue-hour (Oishi et al. 1989; Barnes et al. 2001). This is roughly equivalent to a tMH formation rate of 5.1 pmol/h-mg protein. Thus, 15 pmol/min-tube (equivalent to 135 pmol/h-mg protein of tMH formed by membrane-bound HNMT) is more than sufficient metabolic power to account for previous in vivo findings even if extracellular histamine and SAM do not reach micromolar concentrations.

Although the existence of membrane-bound HNMT has not been previously suggested, earlier studies found results which seem consistent with this form of histamine metabolism. For example, Baenziger and colleagues, in a series of papers, characterized the coupled transport and metabolism of histamine into methylimidazoleacetic acid in endothelial cells and skin fibroblasts (Haddock et al. 1987). This process, dubbed ‘histamine degradative uptake’, may have utilized membrane-bound HNMT, but was not identified assuch. Further studies are needed to determine if non-neuronal cells contain membrane-bound histamine-methylating activity.

Liberation of membrane-bound histamine-methylating activity by detergents, but not high salts, suggests the existence of an integrally bound enzyme. The only known form of HNMT in humans does contain a hydrophobic region extending from F158 to S176 predicted to be capable of spanning a lipid bilayer (PredictProtein, Columbia University). However, the recent crystallization of this enzyme revealed that this region of HNMT corresponded to α helix 6 and β sheet 5 of a 7 β sheet catalytic methyltransferase (MTase) domain common to most SAM dependent enzymes (Horton et al. 2001). Thus, if this form of HNMT spanned the membrane, then the catalytic domain of the enzyme would be disrupted. Furthermore, the known enzyme is not predicted to be glycosylphosphatidylinositol (GPI)-anchored (Eisenhaber et al. 1999) because it lacks a protein sorting signal, GPI anchor site, and hydrophobic C-terminal tail. Thus, the presently described membrane-bound activity is not likely to have the polypeptide structure of soluble HNMT.

If HNMT can anchor to membranes via an integral association, then a polypeptide isoform of the enzyme is likely to exist. Analogously, the membrane-bound variant of the soluble enzyme catechol-O-methyltransferase (COMT, EC 2.1.1.6) contains a 50 amino acid protein sorting signal preceding the soluble form of the enzyme (Lundstrom et al. 1991). This sorting signal serves to target COMT to membranes and then to anchor the enzyme in the lipid bilayer. The two forms of COMT are encoded within the same mRNA, but two separate AUG start sites dictate the translation of one protein or the other. Alternatively, membrane-bound AChE is created from an mRNA isoform with an alternate fifth exon that upon translation leads to a protein variation that can form a GPI anchor to membranes (Li et al. 1991). Hence, soluble and membrane-bound HNMT may originate from one gene containing transcriptional or translational variations, or possibly even from an entirely separate gene.

The current finding that the activities of soluble and membrane-bound HNMT from synaptosomal fractions rapidly decay over 23 days, while the HNMT activity from the S2 fraction remains stable, is consistent with the existence of multiple forms of HNMT (Fig. 4). In support of our findings, electrophoretic mobility studies suggested the existence of multiple forms HNMT in several species, including human (Axelrod and Vesell 1970). Additional studies have also found multiple forms of HNMT, but the molecular basis for these potential isoforms is unknown (Verburg and Henry 1986). It is possible that one of these previously reported forms of HNMT may contain a variation that facilitates membrane association, but further studies are needed to clarify this.

An alternative possibility to the existence of a membrane-bound HNMT isoform is that the known form of HNMT can somehow associate with membranes. The recent crystallographic report of HNMT (Horton et al. 2001) revealed that in addition to the MTase domain common to all SAM-dependent enzymes, HNMT had a subdomain similar to the β motif found in a bacterial SAM-dependent protein methyltransferase (CheR). The β motif of this protein has been shown to associate with the cytoplasmic portion of the bacterial chemotaxis receptor (Djordjevic and Stock 1998). This may suggest that the known form of HNMT can associate with specific receptors on the plasma membrane via this CheR-like subdomain. If so, it is not clear why detergents, and not salts, are needed to liberate histamine-metabolizing activity from membranes.

Other evidence that supports the idea that only one form of HNMT exists is the expressed sequence tag (EST) database at the National Center for Biotechnology Information (NCBI). A database search of well over 200 human ESTs, including the entire UniGene database for HNMT, revealed that nearly every EST aligns precisely with one or more of the six known exons of HNMT. A large number of variations do occur, but are exclusively localized to the 3′ untranslated region (UTR). These variations suggest that the 3′ UTR may extend up to 2579 nucleotides further than presently known and may explain previous studies of HNMT using Northern Blot probes that identified dual bands at 1.6 kb and 3.5 kb (Preuss et al. 1998). However, no ESTs point to a precise region within the 3′ UTR which might encode an HNMT isoform. Studies are presently underway in our laboratory to continue the search for HNMT isoforms.

The present results strongly suggest the existence of a new, membrane-bound histamine-methylating enzyme. This discovery, if confirmed, has several important implications. HNMT is thought to be an important regulator of histamine tone not only in brain, but also in pulmonary function (Yamauchi et al. 1994) and endothelial cell biology (Baenziger et al. 1994), yet the existence and/or significance of the new enzyme have not been addressed in these systems. The present work also suggests that the new enzyme might require extracellular SAM, or perhaps be oriented in the plasma membrane in such a way to use both extracellular histamine and intracellular SAM, possibilities which require further investigation. Indeed, the details of the utilization of SAM by other membrane-bound methyltransferases (e.g. COMT) are also completely unknown. Finally, if the histamine-methylating enzyme described presently is shown to be a new protein, then development of isoform-specific inhibitors of histamine metabolism could lead to important new classes of medications. For example, inhibition of brain HNMT has been shown to reverse amnesia (Malmberg-Aiello et al. 2000), lower pain threshold (Malmberg-Aiello et al. 1997) and decrease appetite (Lecklin et al. 1995). As known inhibitors of HNMT have side-effects which limit therapeutic utility (Duch et al. 1980), such new medications might be clinically useful.

Acknowledgements

This work was supported by grants (DA-03816 and DA-07307) from the National Institute on Drug Abuse. We thank Dr David Martin (Wadsworth Laboratories, New York State Department of Health, Albany, NY) for helpful discussions.

Ancillary