Neurochemical Research. Vol. 22, No. 2, 1997. pp. 113-121

Characterization of Brain p-Carboline-2-N-Methyltransferase, An Enzyme That May Play a Role inIdiopathic Parkinson's Disease

Debra A. Gearhart,1-3 Edward J. Neafsey,1-2 and Michael A. Collins1'3'4

(Accented September 9, 1996)

The activity of p-carboline-2-N-methyltransferase results in the formation of neurotoxic N-meth-ylated P-carbolinium compounds. We have hypothesized that these N-methylated (J-carboliniumcations may contribute to the development of idiopathic Parkinson's disease. This report describesexperiments undertaken to optimize assay conditions for bovine brain (i-carboline-2-N-methyltrans-ferase activity. The activity of (3-carboline-2-N-methyltransferase is primarily localized in the cy-tosol, has a pH optimum of 8.5-9, and obeys Michaelis-Menten kinetics with respect to itssubstrates, 9-methylnorharman (9-MeNH) and S-adenosyl-L-methionine (SAM). Kinetic constants,KM and Vmjx, with respect to 9-MeNH, are 75 uM and 48 pmol/h/mg protein, respectively. TheKM for SAM is 81 (iM and the Vmux is 53 pmol/h/mg protein. In addition, enzyme activity isinhibited by S-adenosyl-L-homocysteine (SAH) or zinc, and is increased 2-fold in the presence ofiron or manganese. Enzyme characterization is a prerequisite to the purification of this N-meth-yltransferase from bovine brain as well as comparison of its activity in human brain from controland Parkinson's disease individuals.

KEY WORDS: Parkinson's disease; p-carbolines; S-adenosyl-L-methionine; N-methyltransferase; N-methyl-ation; bovine brain.

INTRODUCTION

The enzyme fi-carboline-2-N-methyltransferase(PC-2-NMT) converts P-carbolines ((JCs) into poten-tially neurotoxic 2-N-methylated cationic metabolitessuch as 2-methylnorharmanium and 2-methylharmaniumions (1,2). In tissue homogenates from guinea pig, rat,bovine, and normal human brain, various (3Cs are N-methylated at both the pyridyl (2N-) and indole (9N-)nitrogens by putative S-adenosyl-L-methionine-depend-

1 Neuroscience Graduate Program.2 Department of Cell Biology, Neurobiology and Anatomy.3 Department of Molecular and Cellular Biochemistry, Loyola Uni-

versity Chicago Medical Center, Maywood, Illinois.4 Address reprint requests to: Dr. Michael A. Collins, Department of

Molecular and Cellular Biochemistry, Loyola University ChicagoMedical Center, 2160 South First Avenue, Maywood, Illinois 60153Phone: (708) 216-4560; fax: (708) 216-8523.

ent N-methyltransferases (1-3). Substrates of pC-2-NMT include simple PCs such as norharman andharman that are widespread in the environment; sourcesinclude plants, tobacco smoke, industrial waste, alco-holic beverages, soy sauce, and pyrrolyzed tryptophan-rich foods such as grilled meat (4-10). In addition,endogenous formation of (3Cs may occur through a se-ries of reactions beginning with tryptophan or other in-doleamines (11,12). (3C-2-NMT substrates, norharmanand harman, are present in mammalian tissue, includingbrain (13-19), and both have been measured in humancerebrospinal fluid (20). Interestingly, plasma levels ofnorharman and harman, when compared to controls, ap-pear elevated in individuals diagnosed with Parkinson'sdisease (21).

We postulate that (3C-2-NMT and the products ofits catalytic activity, N-methylated p-carbolinium ions(Me-BCs+), may play a role in the neuropathogenesis of

1130364-3190/97/0200-0113S12.50/0 <D 1997 Plenum Publishing Corporation

114

idiopathic Parkinson's disease (PD) (22,23). This hy-pothesis was originally based on the structural similaritybetween Me-0Cs+ and l-methyl-4-phenyl-pyridiniumion (MPP+). In the central nervous system of human andnonhuman primates, MPP+ selectively destroys the do-paminergic nigrostriatal system of the basal ganglia, re-sulting in a parkinsonian condition (24-28). Consistentwith our hypothesis, Me-fiCs+ have several toxic prop-erties analogous with MPP+. Specifically, Me-pCs+ pro-duce nigral lesions and reduce striatal dopaminefollowing unilateral injection of the compounds into thesubstantia nigra of rats (29,30); inhibit mitochondrialrespiration at Complexes I and II of the electron trans-port chain (31,32); inhibit dopamine transport in ratbrain synaptosomes (33); and are toxic to PC12 cells inculture (34). Like MPP+, Me-3Cs+ are toxic to rat pri-mary mesencephalic cell cultures, although toxicity isnot as selective as that of MPP+ (35,36). Matsubara hasreported the presence of Me-pCs* in the frontal cortexand substantia nigra of normal human brain (19,37) andthat there is a significantly higher quantity of Me-pCs+

in the lumbar cerebrospinal fluid of persons afflictedwith idiopathic PD compared to control individuals (20).

We have characterized the following properties ofbovine brain PC-2-NMT: subcellular localization; ap-parent kinetic constants, Michaelis constant (KM) andmaximal velocity (Vmax) with respect to its cosubstrates9-methylnorharman (9-MeNH) and S-adenosyl-L-methi-onine (SAM); dependence of activity on protein concen-tration; pH optimum for enzyme assay; activity as afunction of assay time; inhibition by S-adenosyl-L-ho-mocysteine (SAH), a competitive inhibitor of SAM-de-pendent methyltransferases; and the effect of variousmetals on its activity. Delineation of these properties isessential to the development of a sensitive and repro-ducible enzyme assay for use in future studies, includingthe purification of pC-2-NMT from bovine brain andcomparison of J3C-2-NMT activity in parkinsonian andnormal human brain.

EXPERIMENTAL PROCEDURE

Reagents. (3-Carboline substrate, 9-methylnorharman, was syn-thesized in our laboratory using two different methods: according toMatsubara (2) or a modification of the method described by Rubottom(38). Unlabeled S-adenosyl-L-methionine (toluenesulfonate salt) andtritiated S-[methyl-3H]-adenosyl-L-methionine (60-85 Ci/mmol) werepurchased from Sigma Chemical and Dupont NEN, respectively. Allother reagents were of appropriate purity for their intended use andwere utilized as purchased.

Gearhart, Neafsey, and Collins

Subcellular Fractionation. Bovine brain (including brainstem)was obtained fresh (Aurora Packing Co., Aurora IL), divided sagittallyinto equal halves, and frozen at -80°C for three months. Brain wasquickly thawed in a 40°C water bath and subsequently homogenizedin 2 volumes (2 ml/g tissue) of cold buffer "H1" using a chilledWaring blender at low speed (3 cycles X 30 seconds per cycle). Buffer"H1" contained 150 mM potassium chloride (KCI), 1 mM dithioth-reitol (DTT), 1 mM ethylenediaminetetraacetic acid disodium (EDTA),and 20 mM sodium phosphate dibasic (Na2HPO4) at pH 7.4. The hom-ogenate was filtered through two to four layers of gauze prior to cen-trifugation, Centrifugation was accomplished at 5-l0cC in a SorvallRC5B Superspeed centrifuge (SS-34 rotor) or a Beckman L8-M Ul-tracentrifuge equipped with a SW28 rotor. Centrifugation at 900 g (10minutes), 9000 g (15 minutes), and 100,000 g (one hour) affordedcrude nuclear, mitochondrial, and microsomal plus cytosolic fractions,respectively. Crude particulate fractions were diluted with 150 mMKG prior to inclusion in assays; crude cytosol was not diluted. Forexperiments other than subcellular fractionation, bovine brain (includ-ing brainstem) was obtained fresh (Aurora Packing Co., Aurora IL),divided sagittally into equal halves, and frozen at — 80°C for sevenmonths. Brain was quickly thawed in a 40°C water bath and subse-quently homogenized in 2 volumes (2 ml/g tissue) of cold "H2" usinga chilled Waring blender at low speed (3 cycles X 30 seconds percycle). Solution "H2" contained 150 mM KCI and 1 mM DTT. Thecrude cytosolic supernatant was obtained from filtered homogenatefollowing sequential Centrifugation at 15,000 g (30 minutes) and100,000 g (one hour). Cytosolic supernatant was stored at -80°C inpolypropylene tubes in 2-3 ml aliquots. Frozen aliquots were thawedas needed for use in (JC-2-NMT assays. For some studies, fresh bovinebrain was homogenized in buffer "H3" which contained 20 mMK2HPO4, 150 mM KCI, 1 mM DTT, 1 mM EDTA, 1 mM phenyl-methylsulfonylfluoride, and 1 ^ig/ml each leupeptin and pepstatin A.Cytosol was obtained and stored as described for buffer "H2". Com-position of the homogenation buffer had no obvious effect on pC-2-NMT stability or assay.

Determination of fiC-2-NMT Activity. This activity catalyzes thetransfer of the tritiated methyl moiety from SAM to 9-MeNH resultingin the formation of 2-(3H-methyl)-9-MeNH*(2,9-Me2NH>) and SAH(Fig. 1). The pSC substrate 9-MeNH was chosen over other (3Cs sinceits indole position (9N-) is already methylated, allowing for measure-ment of methylation occurring exclusively at the 2N-position of the(3C. The assay was carried out as previously reported by Matsubara(19) with modifications as detailed below. Enzyme assays were pre-pared in 1.5 ml polypropylene microcentrifuge tubes by mixing con-centrated stock solutions of 9-MeNH and SAM with whole brainhomogenate or subfraction. Stock solutions of SAM were prepared inwater, and 9-MeNH was dissolved in a buffer mixture consisting of50 mM sodium phosphate (Na2HPO4) and 50 mM bicine at pH 6 to9. Brain homogenate or subcellular fraction containing 0.014 to 3.33mg protein per assay tube was present in buffer "H1", "H2", or"H3". Stock solution volumes remained constant across studies,whereas stock solution concentrations and final pH varied dependingupon the experiment. Assay volume was 370 jxl comprised of the fol-lowing stock solution volumes: 50 |iL JH-SAM, 20 jil non-radioactiveSAM, 200 \aL 9-MeNH, and 100 |il enzyme solution. The assay wasgently vortexed and subsequently maintained at 37°C in a shakingwater bath for one hour, except in the time study.

Following incubation at 37°C the assay tube was placed on iceand protein precipitated by the addition of 200 |ll 0,5 N perchloricacid (HC1O4). The mixture was centrifuged to pellet precipitated pro-tein (7000 g for 5 minutes); supernatant was transferred to a glass

Characterization of Brain (J-Carboline-2-N-Methyltransferase 115

culture tube containing 1 ml 1% (w/v) ammonium hydroxide (NH4OH)and 100 ^tl 1 N potassium hydroxide. The protein pellet was vigor-ously resuspended in 500 (4.L 0.01 N HC1O4 and centrifuged (10,000g for 5 minutes); the resulting supernatant was combined in the glasstube with the initial supernatant. Contents of the glass tube wereslowly applied to a Sep-Pak* C18 Plus solid-phase extraction cartridge(Millipore-Waters part no. 20515, 360 mg). The cartridge was previ-ously pretreated with 5 ml each of methanol:5 N acetic acid (3:1, v/v),methanol, and 1% NH4OH (w/v) in that sequence. Slow applicationof all solutions to the cartridge (2 to 5 drops per second) was achievedusing a vacuum manifold. Following sample application, the extractioncartridge was washed with 8 ml of 1% NH4OH followed by 5 mlmethanol; these washes were discarded since they contained undesiredcompounds that contributed significantly to background radioactivityduring subsequent analysis. The final wash (5 ml), containing 3:1 (v/v)methanol:5 N acetic acid, promoted elution of 2,9-Me2NH+ from thecartridge into a clean glass culture tube. This eluate was evaporatedto dryness at 30 - 40°C using a Speed Vac concentrator (SavantInstruments Inc., Hicksville NY).

Reverse-Phase High Performance Liquid Chromatography (RP-HPLC). Activity of (3C-2-NMT was quantitated by RP-HPLC analysisof 2-(3H-methyl)-9-MeNH+ present in the Speed Vac residue. Residuewas dissolved in 200 |iL HPLC mobile phase by sonication and vor-texing. Samples were filtered through a 0.45 (i nylon membrane filter(Micro-Spin, Chrom-Tech Inc.) prior to injection (150 |il) onto theHPLC system. Isocratic HPLC analysis was carried out under ambientconditions on a Bondclone 10 u, C18 column (Phenomenex™, 300 mmX 3.9 mm). Mobile phase consisted of 17% (v/v) HPLC grade ace-tonitrile mixed with 83% (v/v) aqueous buffer. The buffer contained50 mM triethylamine in 250 mM sodium phosphate monobasic ad-justed to pH 3 with phosphoric acid (final mobile phase pH 3.4 ± 1).The HPLC system was comprised of a Shimadzu S1L-9A program-mable autosampler, a Beckman 110A pump (0.8 ml/minute), and anin-line radioactive flow detector (Radiomatic Instruments Flo-One\B).Ultima-Flo® AP (Packard Instruments) liquid scintillation cocktailwas used at a flow rate of 2.8 ml per minute, resulting in a scintillationfluid to mobile phase mixing ratio of 7:2. The radioactive flow detectorwas equipped with a 2.5 mL flow cell and an automatic stream splitter.It was necessary to bypass the radioactive flow detector by splittingthe HPLC column effluent to waste until about ten minutes into therun in order to reduce interference due to tritiated materials (e.g., 3H-SAM) present in the sample. Radioactive peak area corresponding to2,9-Me2NH* (retention time 12-13 minutes) was calculated using Ra-diomatic Flo-One\p system software, and resulting integrated peakarea was reported as total cpm under the 2,9-Me2NH* curve.

Determination of pH Optimum for /3C-2-NMT Assay. Bufferscomposed of 50 mM bicine (pK, 8.3) plus 50 mM Na2HPO4 (pKa, 7.2)spanning the range pH 6.6-9.0 were utilized to prepare 9-MeNH stocksolutions (1 mM); final 9-MeNH concentration in the assay was 540|iM. Duplicate samples were prepared to contain 0.6 mg protein frombovine brain cytosol (in "H2") and SAM at a final concentration of105 )iM (3.6 (iCi) in addition to 9-MeNH stock at the indicated pH.Final assay pH was determined to be identical to the pH of the bufferused to prepare 9-MeNH stock solutions. SC-2-NMT activity, ex-pressed as pmoles of 2,9-Me2NH+ formed in one hour per mg protein(pmol/h/mg protein), was plotted as a function of assay pH.

Assessment of flC-2-NMT Activity as a Function of Protein Con-centration. Bovine brain cytosol in "H2" was concentrated using aCentricon® 30 concentrator (Amicon ®, Beverly MA). Concentratedcytosol was then serially diluted with 150 mM KCl to provide solu-tions containing various protein concentrations for use in the pC-2-

NMT assay. Assays were prepared to contain: protein (0.014-3.33 mgper assay), 9-MeNH (540 \M), and SAM (105 |iM, 4.5 nCi) at pH8.5. BC-2-NMT activity, represented as pmoles of 2,9-Me2NH* formedper hour (pmol/h), was plotted as a function protein concentration inthe assay.

The Effect of Length of Assay Time on ftC-2-NMT Activity. As-says contained bovine brain cytosol (0.687 mg protein in buffer"H3"), 540 uM 9-MeNH, and 105 \iM SAM (2.88 |O.Ci) at pH 8.5.Duplicate samples were maintained at 37°C for 0, 15, 30, 60, 120, and180 minutes prior to termination of assay. (3C-2-NMT activity, denotedas pmoles of 2,9-Me2NH+ formed per mg protein (pmol/mg protein),was plotted as a function of time at 37°C.

Determination of Apparent Kinetic Constants With Respect to theMethyl Donor, SAM. Assays for this study contained bovine braincytosol in "H2" (0.79 mg protein), 9-MeNH (540 \lM, pH 8.5), 3H-SAM (2.3 or 6.0 nCi), and SAM at a final concentration ranging from2.6-520 (xM. BC-2-NMT activity, expressed as pmoles of 2,9-Me2NH*formed per hour per mg protein (pmol/h/mg protein), was plotted asa function of SAM concentration. Plotting and curve-fitting were ac-complished using Kaleidagraph® data analysis/graphics applicationsoftware (Abelbeck Software). Nonlinear curve fitting was employedto fit the data to the Michaelis-Menten equation for the determinationof apparent kinetic parameters, the Michaelis constant (KM) and max-imal velocity (Vmax). This program uses the Levenberg-Marquardt al-gorithm to fit data to a user-defined nonlinear equation. Data were alsoanalyzed using a Hanes plot, a linear transformation of the Michaelis-Menten equation.

Determination of Apparent Kinetic Constants With Respect to thefi-Carboline, 9-MeNH. Assays for this study included bovine braincytosol in "H2" (0.79 mg protein), SAM (260 uM, 5.6 uCi), and 9-MeNH at a final concentration ranging from 5 |iM-l mM at pH 8.5.Data analysis was achieved as described above.

Effect of S-Adenosyl-L-Homocysteine (SAH) on (1C-2-NMTActiv-ity. The effect of SAH on (3C-2-NMT activity was determined in prep-arations containing bovine brain cytosol in "H3" (1.33 mg protein),540 uM 9-MeNH, 105 |lM SAM (2.77 |lCi), and SAH at pH 8.5.Cytosol was concentrated and endogenous SAH theoretically removedby ultradiafiltration using a Centricon ® 30 concentrator; ultradiafiltra-tion was carried out against 10 mM Na2HPO4 (pH 7) containing 150mM KCl. Assays were prepared in duplicate and contained SAH rang-ing from 1 |iM to 1 mM. Results are reported as percent of controlPC-2-NMT activity in the absence of added SAH. Only the linearrange of the curve is shown (1 (j.M-100 u,M) for ease in determinationof the IC50 using linear regression analysis.

Effect of Various Metals on QC-2-NMT Activity. The effect of anumber of transition metals on SC-2-NMT activity was determined byadding chloride salts of metals or EDTA at a final concentration of 1mM. Activity was measured in bovine brain cytosol in "H2" or "H3"(0.721 mg protein) at pH 8.5; substrates were present at 540 |lM and105 uM (3.4 uCi) for 9-MeNH and SAM, respectively. Results arereported as percent of control BC-2-NMT activity in the absence ofadded EDTA or metal chloride. Data were obtained from two differentexperiments, each with its own control. Composition of homogenationbuffer ("H2" or "H3") did not appear to affect results between stud-ies, when results are expressed as percent of control. Since raw databetween studies could not be compared directly, only the data fromthe second study was subjected to statistical analysis (SPSS for Win-dows). Means were compared using one-way ANOVA followed bythe Tukey-HSD post-hoc test.

Protein Measurement. Protein was determined using a modifiedLowry method as described in Scopes (39).

116

Table I. Subcellular Localization of PC-2-NMT Activity in BovineBrain"

Subcellular %Fraction

CrudeNuclearMitochondrialMicrosomalCytosolic

RecoveredActivity

—26.50.70.5

72.3

Specific Activity(pmol/hr/mg protein)

4.11.41,63.7

21.9

"Details in "Experimental Procedure".

Fig. 1. piC-2-NMT assay. Site of tritium label in molecule markedwith *. Abbreviations: 9-methylnorharman (9-MeNH), 2-[3H-methyl]-9-methylnorharmanium ion (2,9-Me2NH+), S-adenosyl-L-methionine(SAM), S-adenosyl-L-homoeysteine (SAH).

Fig. 2. pC-2-NMT activity present in bovine brain cytosol (0.6 mgprotein) as a function of assay pH. Assay was carried out for 1 h at37°C with SAM (105 |iM, 3.6 pCi) and 9-MeNH (540 U.M), Pointsdenote the average of two individual assays.

RESULTS

Subcellular Distribution of &C-2-NMT Activity. Asshown in Table 1 bovine brain (3C-2-NMT activity re-

Gearhart, Neafsey, and Collins

Fig. 3. (3C-2-NMT activity in bovine brain cytosol as a function ofprotein concentration and time at pH 8.5. Assay was carried out for Ih or indicated time at 37°C with SAM (105 uM, 4.5 or 2.88 |iCi) and9-MeNH (540 |J.M). (A) dependence on protein concentration (B) de-pendence on assay time (0.687 mg protein).

sides predominantly in the cytosolic supernatant, al-though there is also notable nuclear activity. Thecytosolic subfraction also exhibits the highest specificactivity. Comparable results were obtained upon subcel-lular fractionation of guinea pig brain (data not shown).

pH Optimum for /3C-2-NMT Assay, Under the assayconditions utilized, (3C-2-NMT activity increased line-arly from pH 6.6 to approximately pH 8.0, where activ-ity reached a plateau that extended to pH 9.0 (Fig. 2).

Linearity of f$C-2-NMT Activity as a Function ofTime and Protein Concentration. Under assay conditionsin which product formation was much less than substrateconcentration, PC-2-NMT activity increased linearlywith respect to protein concentration (Fig. 3A) and time(Fig. 3B). Linear regression analysis of activity as afunction of total protein or time showed a highly linearrelationship (r = 0.996 and 0.980, respectively).

Apparent Kinetic Constants With Respect to Sub-strates, SAM, and 9-MeNH. The high correlation coef-

Characterization of Brain (3-Carboline-2-N-MethyItransferase 117

Fig. 4. (3C-2-NMT activity in bovine brain cytosol (0.79 mg protein)as a function of substrate concentration (pH 8.5). (A) 9-MeNH variedfrom 5 \lM to 1 mM, SAM constant (260 |iM, 5.6 |aCi) (B) SAMvaried from 2.6-520 |lM, 9-MeNH constant at 540 nM. Kinetic par-ameters (KM and Vmax) were determined following nonlinear curve-fitting of the data to the Michaelis-Menten equation.

ficient (r > 0.96) obtained following nonlinear curvefitting of the data to the Michaelis-Menten equation in-dicates that (3C-2-NMT obeys Michaelis-Menten kinet-ics with respect to its substrates when the non-variedsubstrate is fixed near saturation. Values for KM and Vmax

are reported as parameter value ± standard error of thatparameter value. Graphical analysis of the data calcu-lates KM and Vmax values of 75 ± 20 \iM and 48 ± 3pmol/h/mg protein with respect to 9-MeNH (Fig, 4A),Regarding SAM, estimates of KM and Vmax are 81 ± 21uM and 53 ± 4 pmol/h/mg protein (Fig. 4B). As shownin Figures 5A and 5B, linear transformation of the Mi-chaelis-Menten equation and data analysis in the formof a Hanes plot generates KM and Vmax estimates similar

Fig. 5. Linear transformation of Michaelis-Menten equation to a formfor plotting as a Hanes plot. Conditions as in Fig. 4 (A) 9-MeNH (B)SAM.

to those obtained following nonlinear regression analysisof the data.

S-Adenosyl-L-Homocysteine (SAH) Inhibits /3C-2-NMT Activity. The activity of bovine brain (3C-2-NMTis inhibited by SAH in a concentration dependent man-ner (Fig. 6). The IC50 for SAH, i.e., the concentration ofinhibitor producing a 50% reduction in PC-2-NMT ac-tivity compared to control, is 14.8 fiM.

Effects of Transition Metals on J3C-2-NMT Activity.As summarized in Table 0, pC-2-NMT activity wasslightly but significantly increased in the presence ofiron (p < 0.05) and manganese (p < 0.05) relative tothe control assay. Increased (3C-2-NMT activity in thepresence of calcium did not reach statistical significance.Nickel, magnesium and copper did not appear to affectPC-2-NMT activity, but zinc completely inhibited for-

118

Fig. 6. Inhibition of 0C-2-NMT activity by S-adenosyl-L-homocy-steine (SAH) \ uM to 100 uM. Assay conditions: pH 8.5, 540 uM 9-MeNH, 105 jiM SAM (2,77 (iCi), and 1,33 mg protein as bovine braincytosol. Control (3C-2-NMT activity was 29,4 pmol/h/mg protein andrepresents the activity from bovine brain cytosol in the absence ofadded SAH.

Table II. Effect of Transition Metals on (3C-2-NMT Assay"

Metal*

IronIron"ManganeseColbaltCalciumNickelMagnesiumCopperZinc

% of Control''

209 ± 0.7 (2)''231 ± 13 (2)''192 ± 13 (2)146121 ± 0 (2)116101101

0 ± 0 (2)

"Details in "Experimental Procedure".'As chloride salt, 1 mM, 2+ oxidation state.''3* oxidation state.'Values are mean ± SD for (n) replicates.'Significantly different than control (p < 0.05), ANOVA followed byTukey-HSD post-hoc test.

mation of 2,9-Me2NH+. 0C-2-NMT activity was 139%of control in the presence of the chelating agent EDTA(1 mM); the statistical significance of the EDTA effectwas not determined.

DISCUSSION

Fitting of kinetic data to the Michaelis-Menten (M-M) equation indicates that pC-2-NMT obeys Michaelis-Menten kinetics with respect to both 9-MeNH and SAMwhen one substrate is fixed and the other is varied. Pre-vious studies from this laboratory reported kinetic con-stants with respect to 9-MeNH of 17.8 ^M and 2.2pmol/h/mg protein, for KM and Vmax values, respectively(2)—results that differ from those reported here. Kinetic

Gearhart, Neafsey, and Collins

parameters with respect to SAM were not determined inprevious studies.

Two factors may contribute to the difference in ki-netic constants reported by Matsubara and those reportedhere: (a) SAM concentration utilized between studiesand (b) the method of data analysis utilized in each in-vestigation. Regarding the effect of SAM concentration,Matsubara determined kinetic parameters with respect to9-MeNH with SAM fixed at 8 (J.M. The present studyindicates that 8 jjM SAM is well below its KM of 81fiM. In fact, if 8 |iM SAM is inserted into the M-Mequation using our KM and Vmax values for SAM, thecalculated activity is about 5 pmol/h/mg protein, a valuecomparable to the Vnvjx described by Matsubara (2), Inour study, SAM concentration was fixed at 260 uM,equivalent to a concentration three times its KM; the lit-erature suggests (40) that a higher SAM concentrationis more appropriate for testing the adherence of (3C-2-NMT to M-M kinetics with 9-MeNH as the variable sub-strate. Increasing SAM concentration from 8 (J.M (theconcentration used by Matsubara) to 260 uM (the con-centration used in our study), while 9-MeNH is varied,increased the Vmax estimate approximately 20-fold.

In addition to the effect of SAM concentration dis-cussed above, data analysis methods employed betweenstudies may contribute to the discrepancy in the KM for9-MeNH reported by Matsubara (2) and the value re-ported here. Matsubara utilized a non-weighted Linew-eaver-Burk linear transformation (double-reciprocalplot) of the M-M equation in data analysis. Though com-monly used, this plot does not reflect experimental erroraccurately, a point discussed by Cornish-Bowden (41).In fact, when our kinetic data are fitted to the Linew-eaver-Burk equation, the estimated KM with respect to9-MeNH was nearly identical to the 17.8 |lM valueobtained by Matsubara. It is noteworthy that close scru-tiny of a Lineweaver-Burk plot of our data revealed amarked downward curvature of data points near the y-axis (data not shown). This observed deviation from li-nearity in the double-reciprocal plot prompted us toinvestigate other options regarding data analysis. Weconsequently employed nonlinear regression analysiswhich fits the data directly to the Michaelis-Mentenequation, i.e., without linear transformation of the orig-inal equation. We believe that our use of nonlinear re-gression analysis circumvents certain problemsassociated with the use of linear transformation and sub-sequent graphical analysis of nonlinear data. For a re-view of the applicability of nonlinear regression in theanalysis of nonlinear data (e.g., enzymatic data) thereader is directed to papers by Wilkinson (42) and Mo-tulsky (43). We also generated Hanes plots of our data,

Characterization of Brain (3-Carboline-2-N-Mcthyltransferase 119

following linear transformation of the M-M equation asreferenced in Cornish-Bowden (41), As shown in Fig-ures 5A and 5B, Hanes plots yield kinetic parametersthat are reasonably close to those obtained followingnonlinear regression analysis.

The Vmax of (3C-2-NMT with respect to both sub-strates is quite small. Since Vmax is derived from theproduct of enzyme concentration and k,,.,,, its small mag-nitude may indicate that BCs are poor substrates for thisN-methyltransferase (i.e., low turnover number) or thatthe amount of [JC-2-NMT present in brain tissue is quitesmall. This particular property of fJC-2-NMT, i.e., for-mation of neurotoxic products in relatively low amounts,would be consistent with our hypothesis that this enzymemay participate in the neuropathogenesis of Parkinson'sdisease, a disease that takes years to manifest symptomsrelated to the underlying neurodegeneration. We specu-late that the low activity of pC-2-NMT could result inthe generation of small quantities of neurotoxic Me-(iCs+, which over a period of many years cause the deathof a population of neurons that is sufficient to elicit theclinical characteristics of PD.

Linearity of 8C-2-NMT activity as a function oftime and protein concentration is fundamental to reliablemonitoring of enzyme activity during protein purifica-tion. Note that enzyme activity was not linear, but in-stead plateaued, with respect to time or proteinconcentration in preliminary experiments which utilizedSAM at 8 uM instead of 105 jaM (data not shown). Thelinearity of 0C-2-NMT activity through 180 minutes at37°C is also an indication that the enzyme is thermallystable under these conditions.

All SAM-dependent methyltransferases studied todate are inhibited by S-adenosyl-L-homocysteine (SAH)in a competitive manner (44,45). This is not surprisingsince products often inhibit the enzyme responsible fortheir formation, and SAH is a product of SAM-depend-ent methylation reactions. Inhibition by SAH varies de-pending upon the specific methyltransferase affected,with Ki's falling into the 100 nM to 12 jiM range. Fiftypercent inhibition of BC-2-NMT activity is observedwhen added SAH is present at about 14.5 uM, consistentwith the behavior of other SAM-dependent methyltrans-ferases (44).

The manganese and iron effects, although not dra-matic, are intriguing since both metals are associatedwith parkinsonism (46). Manganese miners occasionallydisplay parkinsonian symptoms such as bradykinesia.Neuropathologically, the globus pallidus shows exten-sive cell loss but other basal ganglia components, in-cluding the substantia nigra may also be affected (47);however, clinically manganese poisoning is not classi-

fied as PD. In contrast, aberrant iron metabolism hasbeen suggested to play a role in Parkinson's disease(PD); persons afflicted with PD exhibit higher than nor-mal free iron (48) and decreased ferritin protein in thesubstantia nigra (49). Hypothetically, if iron is indeedelevated in the substantia nigra of persons with PD, thehigh iron concentration could conceivably increase BC-2-NMT activity in that brain region leading to local for-mation of neurotoxic N-methylated B-carboliniumcompounds and subsequent cell death.

It is possible that pC-2-NMT represents a previ-ously characterized enzyme for which f$-carbolines werenot evaluated as substrates. Saavedra et al. (50) de-scribed the distribution of a non-specific N-methyltrans-ferase present in the brain of several species, includinghuman postmortem brain tissue. That SAM-dependentenzyme utilizes tryptamine as well as several otheramines as substrates. Similar to (5C-2-NMT, the enzymeexhibited a KM with respect to SAM of 52 fiM and aVmax in the range 2-15 pmol/h/mg protein, dependingupon the tissue. The subcellular distribution and specificactivity of the non-specific N-methyltransferase de-scribed by Saavedra are similar to those that we reportfor BC-2-NMT,

Ansher and Jakoby (51,52) purified amine N-meth-yltransferases A and B from rabbit liver. The enzymesexhibited broad, overlapping N-methylating activities to-ward numerous amines, including several azaheterocy-cles. The azaheterocycles tested included derivatives ofpyridine and isoquinoline, but p-carbolines were nottested as substrates. It is noteworthy that both N-meth-yltransferases A and B were purified from cytosol andboth exhibited pH profiles with respect to desmethyli-mipramine and tryptamine that were similar to that ofPC-2-NMT toward 9-MeNH. Ansher et al. (53) alsofound these amine N-methyltransferase activities in hu-man, monkey, and rodent cytosolic brain extracts; sub-strates included 4-phenyltetrahydropyridine and4-phenylpyridine, potential precursors of neurotoxicMPP+. Specific activities toward these substrates were inthe range 0.17-23 pmol/h/mg protein, similar to the val-ues reported in the present study.

In addition, Naoi et al. (54) reported the presenceof a cytosolic N-methyltransferase activity in humanbrain homogenates that forms potentially neurotoxic N-methyl-l,2,3,4-tetrahydroisoquinoline from tetrahydrois-oquinoline (TIQ). The TIQ-methylating activity exhibitedkinetic parameters and a pH optimum similar to thoseof BC-2-NMT. Likewise, Matsubara identified a cyto-solic N-methyltransferase activity present in rodent brainthat catalyzes methylation at the 2N-nitrogen of tetra-hydro-fj-carbolines (55). It is conceivable that (3C-2-NMT

120

and any of the N-methyltransferases described above arethe same enzymes.

The cytosolic location of pC-2-NMT disagrees withthe primarily nuclear location of the enzyme previouslyreported from this laboratory, in which cytosolic activitywas undetectable (2), This discrepancy between studiescould be due to variation in the intensity of homogena-tion preceding subcellular fractionation. Since subcellu-lar enzymatic markers were not utilized to monitorfractionation, the cause of the discrepancy remains un-known. However, cytosolic (3C-2-NMT activity is con-sistent with the subcellular distribution of theN-methyltransferases previously described by Saavedra(50), Ansher (53), Naoi (54), and Matsubara (55).

The studies delineated in this report define experi-mental conditions to be utilized in subsequent investi-gations; specifically PC-2-NMT purification frombovine brain cytosol and measurement of its activity innormal and parkinsonian human brain. For the proposedstudies, cosubstrates 9-MeNH and SAM will be presentat 540 (AM and 105 jiM (4-6 jiCi) respectively; assayswill be maintained at 37°C for 60 minutes at pH 8.5. Inconclusion, we have described several properties ofbrain p-carboline-2-N-methyltransferase (fJC-2-NMT),the activity of which results in the formation of poten-tially neurotoxic N-methylated p-carbolinium com-pounds (Me-(3Cs+). We postulate that Me-pCs+ andPC-2-NMT may contribute to the development of idio-pathic Parkinson's disease.

ACKNOWLEDGEMENTS

This work was supported by NIH Grant NS23891. The commentsand criticisms of Dr. Richard Schultz and Dr. William Simmons areappreciated.

REFERENCES

1. Collins, M. A., Neafsey, E. J., Matsubara, K., Cobuzzi, R. J., andRollema, H. 1992. Indole-N-methylated B-carbolinium ions as po-tential brain-bioactivated neurotoxins. Brain Res. 570:154-160.

2. Matsubara, K., Neafsey, E. J., and Collins, M. A. 1992. Novel S-adenosylmethionine-dependent indole-N-methylation of 8-carbo-lines in brain particulate fractions. J. Neurochem. 59:511-518.

3. Gearhart, D. A., Collins, M. A., and Neafsey, E. J. 1995. Prelim-inary characterization of p-carboline-2-N-methyltransferase frommammalian brain. Abst. Soc. Neurosci. 21:1252. Annual Meetingof the Society for Neuroscience. San Diego, CA.

4. Adachi, J., Mizoi, Y., Naito, T., Yamamota, K., Fujiwara, S., andNinomiya, I. 1991. Determination of B-carboiines in foodstuffs byhigh-performance liquid chromatography and high-performanceliquid chromatography-mass spectrometry. J. Chromatog. 538:331-339.

Gearhart, Neafsey, and Collins

5. Bosin, T. R., Faull, K. F., and Barchas, J. D. 1988. Harman inalcoholic beverages: pharmacological and toxicological implica-tions. Alcohol. Clin. Exp. Res. 12:679-682.

6. Gross, G. A., Turesky, R. J., Fay, L, Stillwell, W., Skipper, P.,and Tannenbaum, S. 1993. Heterocyclic aromatic amine formationin grilled bacon, beef and fish and in grill scrapings. Carcinogen-esis. 14:2313-2318.

7. Holmstedt, B. 1982. Beta-carbolines and tetrahydroisoquinolines:historical and ethnopharmacological background. Pages 3 13. inF. Bloom, Barcha, J., Sandier, M., and Usdin, F. (eds.), Beta-carbolines and tetrahydroisoquinolines, A. R. Liss, New York.

8. Rommelspacher, H., Damm, H., Strauss, S., and Schmidt, G.1985. Ethanol induces an increased excretion of harman in thebrain an urine of rats. Naunyn Schmiedebergs Arch. Pharmacol.327:107-113.

9. Rommelspacher, H. and Schmidt, L. 1985. Tetrahydroisoquino-lines and beta-carbolines: putative natural substances in plants andanimals. Prog. Drug Res. 29:415-459.

10. Felton, J. S. and Knize, M. G. 1990. Heterocyclic-amine muta-gens/carcinogens in foods. Pages 471-501. in C. Cooper andGrover, P. (eds.), Handbook of Exp. Pharmacol.

11. Susilo, R., Damm, H., Rommelspacher, H., and Hofle, G. 1987.Biotransformation of I-methyl-tetrahydro-B-carbolinc-l-carbox-ylic acid to harmalan, tetrahydroharman and harman in rats. Neu-rosci. Lett. 81:327-330.

12. Melchior, C. M. and Collins, M. A. 1982. The routes and signif-icance of endogenous synthesis of alkaloids in animals. Crit. Rev.Toxicol. 10:313-356.

13. Airaksinen, M. M. and Kari, I. 1981. fi-Carbolines, psychoactivecompounds in the mammalian body, part I: occurrence, origin, andmetabolism. Med. Biol. 59:21-34.

14. Bosin, T. R., Borg, S., and Faull, K. F. 1989. Harman in rat brain,lung and human CSF: effect of alcohol consumption. Alcohol. 5:501-511.

15. Collins, M. A. 1983, Mammalian alkaloids. Pages 329-358. in A.Brossi (eds.), The Alkaloids, Academic Press, New York.

16. Fekkes, D., Schouten, M. J., Pepplinkhuizen, L., Bruinvels, J.,Lauwers, W., and Brinkman, U. 1992. Norharman, a normal bodyconstituent. Lancet. 339:506.

17. Schouten, M., and Bruinvels, J. 1986. Endogenously formed nor-harman (beta-carboline) in platlet rich plasma obtained from por-phyric rats, Pharmacol. Biochem. Behav. 24:1219-1233.

18. Shoemaker, D. W., Cummins, J. T., Bidder, T. G., Boettger, H.G., and Evans, M. 1980. Identification of harman in rat arcuatenucleus. Arch. Pharmacol. 310:227-230.

19. Matsubara, K., Collins, M. A., Atsushi, A., Ikebuchi, J., Neafsey,E. J., Kagawa, M., and Shiono, H. 1993. Potential bioactivatedneurotoxicants, N-methylated P-carbolinium ions, are present inhuman brain. Brain Res. 610:90-96.

20. Matsubara, K., Kobayashi, S., Kobayashi, Y., Yamashita, K., Ko-ide, H., Hatta, M., Iwamota, K., Tanaka, O., and Kimura, K. 1995.Beta-carbolinium cations, endogenous MPP(f) analogs, in the lum-bar cerebrobrospinal fluid of patients with Parkinson's disease.Neurology. 45:2240-2245.

21. Kuhn, W., Muller, T., Grobe, H., Dierks, T., and Rommelspacher,H. 1995. Plasma levels of the )3-carbolines harman and norharmanin Parkinson's disease. Acta. Neurol. Scand. 92:451^154,

22. Collins, M. A. and Neafsey, E. J. 1985. B-carboline analogues ofN-methyl-4-phenyI-l,2,5,6-tetrahydropyridine (MPTP): endoge-nous factors underlying idiopathic Parkinsonism? Neurosci. Lett.55:179-184,

23. Collins, M. A, 1994. Potential parkinsonian protoxicants withinand without. Neurobiol. Aging. 15:277-278.

24. Burns, R., Chiueh, C., Markey, S., Ebert, M., Jacobowitz, D., andKopin, I. 1983. A primate model of parkinsonism: selective de-struction of dopaminergic neurons in the pars compacta of thesubstantia nigra by N-methyl-4-phenyl-l,2,3,6-tetrahydropyridine.Proc. Natl. Acad. Sci. USA. 80:4546-4550.

Characterization of Brain p-Carboline-2-N-Methyhransferase 121

25. Bums, R., Markey, S., Phillips, J., and Chiueh, C. 1984. The neu-rotoxicity of l-methyl-4-phenyl-l,2,3,6-tetrahydropyridine in themonkey and man. Can. J. Neurol. Sci. 11:166-168.

26. Chiueh, C., Markey, S., Burns, R., Johannessen, J., Jacobowitz,D., and Kopin, I. 1984. Neurochemical and behavioral effects ofl-methyl-4-phenyl-l,2,3,6-tetrahydropyridine (MPTP) in rat,guinea pig and monkey. Psychopharmacol. Bull. 20:548-553.

27. Davis, G., William, A., Markey, S., Ebert, M., Caine, E., Reichert,C., and Kopin, I. 1979. Chronic parkinsonism secondary to intra-venous injection of meperidine analogues. Psychiatry Res. 1:259—249-254.

28. Langston, J. W. 1995. MPTP as it relates to the etiology of Par-kinson's disease. Pages 367-4OO. in J. H. Ellenberg, Koller, W.C., and Langston, J. W. (eds.), Etiology of Parkinson's Disease,Marcel Dekker, Inc., New York.

29. Neafsey, E. J., Drucker, G., Raikoff, K., and Collins, M. A. 1989.Striatal dopaminergic toxicity following intranigral injection inrats of 2-methyl-norharman, a (3-carbolinium analog of N-methyl-4-phenylpyridium ion (MPP*). Neurosci. Lett. 105:344-349.

30. Neafsey, E. J., Albores, R., Gearhart, D., Kindel, G., Raikoff, K.,Tamayo, F., and Collins, M. A. 1995. Methyl-p-carbolinium an-alogs of MPP* cause nigrostriatal toxicity after substantia nigrainjections in rats. Brain Res, 675:279-288,

31. Albores, R., Neafsey, E. J., Drucker, G., Fields, J. Z., and Collins,M. A. 1990. Mitochondrial respiratory inhibition by N-methylatedB-carboline derivatives structurally resembling N-methyl-4-phen-ylpyridine. Proc. Natl. Acad. Sci. USA. 87:9368-9372.

32. Fields, J. Z., Albores, R. R., Neafsey, E. J., and Collins, M. A.1992. Inhibition of mitochondrial succinate oxidation—similari-ties and differences between N-methylated p-carbolines andMPP*. Arch. Biochem. Biophys. 294:539-543.

33. Drucker, G., Raikoff, K., Neafsey, E. J., and Collins, M. A. 1990.Dopamine uptake inhibitory capacities of B-carboline and 3,4-dih-ydro-B-carboline analogs of N-methyl-4-phenyl-l,2,3,6-tetrahy-dropyridine (MPTP) oxidation products. Brain Res. 509:125-133.

34. Cobuzzi, R. J., Neafsey, E. J., and Collins, M. A. 1994. Differ-ential cytotoxicities of N-methyl-S-carbolinium analogues ofMPP* in PC 12 cells: insights into potential neurotoxicants in Par-kinson's disease. J. Neurochem. 62:1503-1510.

35. Collins, M. A., Neafsey, E. J., Zeevalk, G., Albores, R., Kindel,G., and Tamayo, F. 1995. Environmentally-derived carboliniumanalogs of MPP* are toxic to dopaminergic neurons in vitro andin vivo. Pages 537-543. in I. Hanin (eds.), Alzheimer's and Par-kinson's diseases, Plenum Press, New York.

36. Collins, M. A., Neafsey, E. J., and Matsubara, K. 1996. |3-car-bolines: metabolism and neurotoxicity. Biogenic Amines. 12:171-180.

37. Matsubara, K. 1996. Occurrence of neurotoxic (i-carbolinium cat-ions in mammalian central nervous system. Biogenic Amines. 12:161-169.

38. Rubottom, G. M. and Chabala, J. C. 1974. N-alkylindoles fromthe alkylation of indole sodium in hexamethylphosphoramide: 1-benzylindole. Org. Syn. 54:60-62.

39. Scopes, R. K. 1994. Solutions for measuring protein concentra-tion. Pages 349-350. in C. R. Cantor (eds.), Protein Purification:Principles and Practice, Springer-Verlag, Inc., New York.

40. Dixon, M. and Webb, E. C. 1979. Enzyme Kinetics. Pages 79-85. in M. Dixon, Webb, E. C., Thorne, M. A., and Tipton, K. F.(eds.), Enzymes, Academic Press, Inc., New York.

41. Comish-Bowden, A. 1979. Introduction to enzyme kinetics. Pages25-31. in Fundamentals of Enzyme Kinetics, Butterworth Inc.,Boston.

42. Wilkinson, G. N. 1961. Statistical estimations in enzyme kinetics.Biochem. J. 80:324-332.

43. Motulsky, H. J. and Ransnas, L. A. 1987. Fitting curves to datausing nonlinear regression: a practical and nonrnathematical re-view. FASEB J. 1:365-374.

44. Lawrence, F. and Robert-Gero, M. 1990. Natural and syntheticanalogs of S-adenosylhomocysteine and protein methylation.Pages 305-340. in W. K. Paik and Kim, S. (eds.), Protein meth-ylation, CRC Press, Boca Raton.

45. Deguchi, T. and Barchas, J. 1971. Inhibition of transmethylationsof biogenic amines by S-adenosylhomocysteine. J. Biol. Chem.246:3175.

46. Irwin, 1. and Langston, J. W. 1995. Endogenous toxins as potentialetiologic agents in Parkinson's disease. Pages 182-186. in 1. H.Ellenberg, Keller, W. C., and Langston, J. W. (eds.), Etiology ofParkinson's Disease, Marcel Dekker, Inc., New York.

47. Mena, I. 1979. Manganese poisoning. Pages 217-237, in P. J.Vinken, Bruyn, G. W., Cohen, M. M., and Klawans, H. L. (eds.),Handbook of clinical neurology, vol 36: intoxications of the ner-vous system, part I, North Holland Publishing, Amsterdam.

48. Dexter, D. T., Well, F. R., and Agid, F. J. 1989. Increased nigraliron content in postmortem parkinsonian human brain. Lancet. 2:1219-1220.

49. Dexter, D. T., Carayon, A., and Vidailhet, M. 1990. Decreasedferritin levels in Parkinson's disease. J. Neurochem. 55:16-20.

50. Saavedra, J. M., Coyle, J. T., and Axelrod, J. 1973. The distri-bution and properties of the non-specific N-methyltransferase inbrain. J. Neurochem. 20:743-752.

51. Ansher, S. S. and Jakoby, W. B. 1986. Amine-N-methyltransfer-ase from rabbit liver. J. Biol. Chem. 261:3996-4001.

52. Ansher, S. S. and Jakoby, W. B. 1987. Amine N-methyltransfer-ases. Meth. Enzymol. 142:660-667.

53. Ansher, S. S., Cadet, J. L., Jakoby, W. B., and Baker, J. K. 1986.Role of N-methyltransferases in the neurotoxicity associated withthe metabolites of l-mcthyl-4-phenyl-l,2,3,6-tetrahydropyridine(MPTP) and other 4-substituted pyridines present in the environ-ment. Biochem. Pharmacol. 35:3359-3363.

54. Naoi, M., Matsuura, S., Takahashi, T., and Nagatsu, T. 1989. AN-methyltransferase in human brain catalyzes N-methylation of1,2,3,4-tetrahydroisoquinoline into N-methyl-1,2,3,4-tetrahydro-isoquinoline, a precursor of a dopaminergic neurotoxin, N-meth-ylisoquinolinium ion. Biochem. Biophys. Res. Comm. 161:1213-1219.

55. Matsubara, K., Collins, M. A., and Neafsey, E. J. 1992. Mono-N-methylation of tetrahydro-S-carbolines in brain cytosol: absenceof indole methylation. J. Neurochem. 59:505-510.