Magnetization transfer of fluoxetine in the human brain using fluorine magnetic resonance spectroscopy.

Fluorine magnetic resonance spectroscopy ((19)F MRS) measurements of fluoxetine and metabolite concentration in the human brain underestimate true drug levels because of a bound, MRS-"invisible" pool of drug molecules. Magnetization transfer (MT) spectroscopy may be a useful technique for characterizing this bound pool of fluoxetine in the brain. Six subjects on consistent daily doses of fluoxetine underwent (19)F MT spectroscopy on a 1.5-T scanner using a train of three preparation pulses at -3000 Hz off resonance with 0.5 W of peak power deposition in tissue. One subject was scanned at multiple time points after initiation of drug therapy. Magnetization transfer signal contrast was quantified using VARPRO-based time domain fitting software. Magnetization transfer signal contrast was quantifiable with mean MT signal depression of 12.5% (SD = 5.0, n = 6). An inverse relationship between brain concentration and the MT signal contrast of fluoxetine was found (r = -.82, Spearman coefficient =.007). This study is the first in vivo application of (19)F MT spectroscopy and the first to demonstrate a quantifiable MT effect for a psychotropic medication in the human brain. Findings suggest that fluoxetine is substantially bound in the brain and that individual differences, inversely related to brain concentration, can be detected in the magnitude of MT contrast.

Characterization of human brain pharmacokinetics using a two-compartment model.

We developed a two-compartment pharmacokinetic model to systematically characterize 19F magnetic resonance spectroscopy (19F MRS) data on the concentration time course of psychotropic compounds measured in human brain. Using this model, brain volume of distribution and clearance were calculated for fluvoxamine as an index compound. Our interest in formalizing a multicompartment model was motivated by recent advances in the field of brain spectroscopy that allow the noninvasive characterization of brain uptake and elimination half-lives of fluorinated psychotropic compounds. Differences between central compartment single-dose and steady-state half-lives and the peripheral elimination half-life at steady state were used to formulate the model. Application of the model is illustrated using previously published data on the elimination half-lives of fluvoxamine from plasma and brain at steady state, along with the literature values for single-dose plasma elimination half-life. Applying the model, brain volume of distribution (1.12 L/kg +/- 0.2 SEM) and clearance (1.01 L/hour +/- 0.12 SEM) were calculated for fluvoxamine. The bioavailability of fluvoxamine to the brain from plasma was 1.85 +/- 0.23 SEM. The underlying multicompartment pharmacokinetics of fluvoxamine were reflected by the difference between brain and plasma elimination half-lives from steady state. This method to derive pharmacokinetic parameters using 19F MRS measurements of drug concentration in brain can be applied to characterize the pharmacokinetics of other fluorinated psychotropic compounds.

Brain elimination half-life of fluvoxamine measured by 19F magnetic resonance spectroscopy.

OBJECTIVE: This study used fluorine-19 magnetic resonance spectroscopy (19F MRS) to characterize the elimination of fluvoxamine from the human brain after abrupt drug discontinuation. The elimination half-lives of fluvoxamine in brain and plasma were determined to assess their interdependence and the relationship of brain half-life to the clinical practice of drug holidays and reports of acute withdrawal symptoms. METHODS: Six subjects completing clinical treatment with fluvoxamine were enrolled in the study. Spectroscopic quantification of whole brain fluvoxamine concentrations and chromatographic determination of plasma fluvoxamine levels were performed serially for up to 10 days after drug withdrawal. Psychiatric evaluation to assess withdrawal symptoms was also done at each scanning session. RESULTS: Elimination of fluvoxamine in the brain and plasma was optimally described by first-order kinetics; the mean elimination half-lives were 58 hours and 26 hours, respectively. The mean ratio of fluvoxamine brain elimination half-life to plasma half-life was 2.4. Three of the six subjects experienced mild to moderate withdrawal symptoms between the third and fifth days of the study, which corresponded to between one and two brain half-lives of fluvoxamine. CONCLUSIONS: The brain elimination half-life for fluorinated psychotropic compounds can be measured noninvasively by 19F MRS. The elimination half-life of fluvoxamine was found to be substantially longer for the brain than for plasma. The time course of withdrawal symptom onset and the rationale for drug holidays with fluvoxamine appear to be well explained by the brain elimination half-life.

19F magnetic resonance spectroscopy investigation in vivo of acute and steady-state brain fluvoxamine levels in obsessive-compulsive disorder.

OBJECTIVE: The purpose of this study was to investigate the pharmacokinetics of fluvoxamine in the human brain by using fluorine-19 magnetic resonance spectroscopy (19F MRS) and to assess the relationships among fluvoxamine brain levels, fluvoxamine plasma levels, and clinical efficacy. METHOD: Eight subjects with DSM-IV obsessive-compulsive disorder were entered into a prospective, open-label treatment trial of fluvoxamine. 19F MRS measurements of whole brain drug and metabolite concentrations and spin-lattice (T1) relaxation times were performed serially in seven subjects for up to 5 months. A psychiatric determination of clinical response and a blood sample for plasma fluvoxamine measurement were obtained at each 19F MRS session. RESULTS: The subjects achieved steady-state brain concentrations of fluvoxamine within 30 days after consistent daily dosing, as determined by stabilization of brain fluvoxamine concentrations. The mean brain-to-plasma ratio at steady state was 24 to 1. Brain fluvoxamine T1 values from 140 to 230 msec were observed. All but one subject experienced substantial improvement in symptoms. The one nonresponder exhibited several-fold higher plasma and brain fluvoxamine levels. CONCLUSIONS: Brain fluvoxamine levels were substantially higher than plasma levels. Steady-state brain levels correlated to plasma levels but not to dose. Systematic assessment of treatment response in relation to brain or plasma fluvoxamine level was not feasible because of the marked and rapid clinical response during open-label treatment. These data suggest that fluvoxamine attains brain steady-state levels substantially faster than fluoxetine, with corresponding clinical implications.

Anti-muscarinic toxins from Dendroaspis angusticeps.

Toxins from the venom of the African green mamba, Dendroaspis angusticeps, fulfill a major need for selective ligands for some of the five genetically defined subtypes of muscarinic acetylcholine receptors (m1-m5). Two toxins have been found that are highly selective antagonists for m1 and m4 receptors (m1-toxin and m4-toxin, respectively). Two other toxins (MT1 and MT2) bind with high affinity to both m1 and m4 receptors, and are agonists. Components of the venom also modify the binding of radiolabeled antagonists to m2 receptors, but an m2-selective toxin has not yet been isolated, m1-Toxin can bind to m1 receptors at the same time as typical competitive antagonists, suggesting that this toxin binds to the N-terminal and outer loops of m1 receptor molecules, rather than within the receptor pocket where typical agonists and antagonists bind. The binding of toxins to the outer parts of receptor molecules probably accounts for their much higher specificity for individual receptor subtypes than is seen with smaller ligands. Toxins are useful for identifying, counting, localizing, activating and blocking m1 and m4 receptors with high specificity.

DNA synthesis in adult feline ventricular myocytes. Comparison of hypoxic and normoxic states.

Adult mammalian ventricular myocytes are terminally differentiated cells, and the prevailing perception has been that DNA synthesis and repair are not active. We tested the hypothesis that there is potential for DNA synthesis and repair by studying the ability of whole-cell extracts from adult myocytes to incorporate [alpha-32P]dCTP into damaged plasmids. Left ventricular myocytes were isolated from adult cat hearts by collagenase dissociation. Cells were maintained in room air (control extract, CE) or made ischemic (IE) with N2 displacement of O2 and extracted for total protein. The nicked form of the plasmid was produced by exposure to an Fe3+/ascorbic acid free radical generating system. Both IE and CE degraded the supercoiled form of the plasmid and incorporated [alpha-32P]dCTP into the nicked (32P/DNA mass; CE = 2.2, IE = 3.0) and linear forms (32P/DNA mass; CE = 28.7, IE = 25.2). Exposure of plasmids to UV light did not inhibit incorporation of label. Inhibition studies with the cell extracts suggested a participation of polymerase delta in myocyte DNA repair/synthesis. Myocyte extract was as active as extract from rapidly growing COS cells at incorporating labeled nucleotides into plasmid DNA. The ability of intact myocytes to incorporate [alpha-32P]dCTP into endogenous DNA was measured in isolated cells made permeable with saponin. Studies were done in room air or N2. Permeable cells incorporated [alpha-32P]dCTP into nuclear DNA, but maximal specific activity of DNA was observed at 15 minutes with ischemia and at 60 minutes with room air control cells (ischemia, 1.34 +/- 0.5, 0.86 +/- 0.33, 0.60 +/- 0.04; air, 1.0, 1.28 +/- 0.20, 1.87 +/- 0.38, at 15, 30, and 60 minutes, respectively). These data indicate that mammalian adult ventricular myocytes can actively repair and/or synthesize both exogenous and endogenous DNA. A DNA synthetic response to cellular damage may have important pathological and clinical implications.

The human choline acetyltransferase gene encodes two proteins.

Both 2,500- and 6,000-nucleotide (nt) mRNAs are generated by alternative splicing of the primary transcript from the human gene for choline acetyltransferase (ChAT), the 68-kDa enzyme that synthesizes acetylcholine. In vitro translation of cRNA derived from a clone of the 2,500-nt mRNA produced a protein with ChAT activity demonstrating that this transcript encodes the human ChAT enzyme. An antibody directed against a unique amino acid sequence predicted from the 6,000-nt ChAT gene transcript identified a 27-kDa protein on immunoblots of human nucleus basalis proteins. This protein was further shown to cross-react with antibodies prepared against the 68-kDa human ChAT enzyme. Gel-filtration chromatography of human nucleus basalis proteins demonstrated that the 27-kDa protein does not have ChAT activity, which eluted as a single peak coincident with that of the 68-kDa enzyme. The 27-kDa protein was, however, shown to colocalize with the ChAT enzyme in cholinergic neurons of the human spinal cord using immunohistochemical techniques.

Retinoic acid induces cholinergic differentiation of NTera 2 human embryonal carcinoma cells.

Retinoic acid (RA), a natural metabolite of vitamin A, influences the survival and neurotransmitter phenotype of several classes of vertebrate neurons during development. We now report that RA induces a subpopulation of NTera 2/clone D1 (NT2) human embryonal carcinoma cells to differentiate into postmitotic cells with cholinergic properties (NT2-N cells). After growth for 6 days in the presence of RA (10 microM) low levels of the acetylcholine-synthesizing enzyme choline acetyltransferase (ChAT) were detected in NT2 cell cultures. ChAT activity in the NT2 cell cultures continued to increase for at least an additional 22 days to a final activity of 50 pmol ACh synthesized/min/mg protein. Immunohistochemical staining of RA-treated cultures demonstrated that only those cells with a neuronal morphology (NT2-N cells) expressed the human ChAT protein. Since such cells comprised a small proportion (approximately 20%) of the population, the ChAT activity per neuronal cell was estimated to approach 250-300 pmol ACh/min/mg protein. Cultures composed of > 95% NT2-N cells had significantly lower ChAT specific activities and this could be increased by either ciliary neurotrophic factor or leukemia inhibitory factor, but not by nerve growth factor. We conclude that NT2 cells provide a system in which to study the molecular events that underlie neurotransmitter choice during the differentiation of human cholinergic neurons.

Partial volume effects in volume-localized phased-array proton spectroscopy of the temporal lobe.

MRS techniques can aide in confirming the location of seizure foci in temporal lobe epilepsy. N-acetyl aspartate (NAA), creatine plus phosphocreatine, choline-containing compounds, and lactate are most often the clinically relevant metabolites in these studies. We examined the importance of partial volume effects from tissue heterogeneity in temporal lobe spectroscopy on the metabolite ratios. Our study shows that localized spectroscopy, using three different voxel sizes, centered on the anterior body of the hippocampus, produces significantly different values for the NAA to the creatine ratio. The spectroscopy was performed at 1.5 T using the PRESS pulse sequence and a phased-array coil system specifically designed for the temporal lobe. The data exhibits a clear trend of increasing NAA to creatine ratios with increasing voxel size. This trend demonstrates that partial volume effects can contribute to variation of NAA to creatine ratios in healthy subjects.

Proton magnetic resonance spectroscopy investigation of hyperventilation in subjects with panic disorder and comparison subjects.

OBJECTIVE: The purpose of this study was to investigate differential effects of hyperventilation on brain lactate in patients with panic disorder and comparison subjects as a possible mechanism for explaining previous observations of an excess rise in brain lactate among panic disorder subjects during lactate infusion. METHOD: Seven treatment-responsive patients with panic disorder and seven healthy comparison subjects were studied with proton magnetic resonance spectroscopy to measure brain lactate during controlled, voluntary hyperventilation over a period of 20 minutes. Hyperventilation was regulated with the use of capnometry to maintain end-tidal PCO2 at approximately 20 mm Hg during the period of hyperventilation. Blood lactate was measured prior to and at the end of hyperventilation. RESULTS: At baseline the two groups had similar brain lactate levels. Panic disorder subjects exhibited significantly greater rises in brain lactate than comparison subjects in response to the same level of hyperventilation. Blood lactate levels before and after 20 minutes of hyperventilation were not significantly different between groups. CONCLUSIONS: Controlled hyperventilation increases brain lactate and does so disproportionately in subjects with panic disorder. This increase in brain lactate may result from decreased cerebral blood flow due to hypocapnia, and individuals with panic disorder may have greater sensitivity to this regulatory mechanism.

m1-toxin.

The venom of the Eastern green mamba from Africa, Dendroaspis angusticeps, contains a number of toxins which block the binding of 3H-antagonists to genetically-defined m1 and m4 muscarinic acetylcholine receptors. Most of the anti-muscarinic activity of the venom is due to the presence of a newly-isolated toxin, "m1-toxin", which has 64 amino acids and a molecular mass of 7361 Daltons. At present m1-toxin is the only ligand which is known to be capable of fully blocking m1 receptors without affecting m2-m5 receptors. It binds very rapidly, specifically and pseudoirreversibly to the extracellular face of m1 receptors on cells, in membranes or in solution, whether or not the primary receptor site is occupied by an antagonist. Bound toxin can either prevent the binding and action of agonists or antagonists, or prevent the dissociation of antagonists. The toxin is useful for identifying m1 receptors during anatomical and functional studies, for recognizing and stabilizing receptor complexes, and for occluding m1 receptors so that other receptors are more readily studied.

Two mRNAs are transcribed from the human gene for choline acetyltransferase.

The product of the choline acetyltransferase (ChAT) gene is the enzyme that synthesizes the neurotransmitter acetylcholine. A 14.4-kb portion of the human ChAT gene contains 7 exons, which are estimated to comprise approximately one-third of the human protein coding sequence by comparison with porcine ChAT mRNA. Two of the exons were used to identify polyadenylated human ChAT gene transcripts on Northern blots. An exon with 84% identity to the region of porcine ChAT mRNA that codes for the amino terminus of the corresponding protein detected 6,000- and 2,300-nucleotide mRNAs in RNA isolated from human CHP134 neuroblastoma cells. Only the 2,300-nucleotide mRNA was detected by a second probe containing an exon with 96% identity to porcine ChAT mRNA in the domain that encodes amino acids 204-263 of the predicted porcine ChAT protein. Further evidence that two species of human mRNA are produced from the human ChAT gene was obtained from nuclease protection assays using an antisense RNA probe prepared from a human ChAT cDNA clone. Total RNA isolated from either CHP134 cells or adult human nucleus basalis protected 525- and 400-nucleotide fragments of this probe, confirming the presence of two species of RNA that differ by the inclusion of an internal exon. cDNA clones of each of these transcripts have been isolated. Their sequences suggest that the 2,300-nucleotide mRNA encodes enzymatically active human ChAT, while translation of the 6,000-nucleotide mRNA would be terminated prematurely by a shift in the reading frame. These results indicate that a complex pattern of transcription produces two mRNAs with different coding potentials from the human ChAT gene.

Nerve growth factor regulation of choline acetyltransferase gene expression in rat embryo basal forebrain cultures.

Nerve growth factor (NGF) increases the activity of choline acetyltransferase (ChAT), the synthetic enzyme for acetylcholine, in rat basal forebrain neurons both in vivo and in vitro. In poly(A)+ RNA isolated from cultures prepared from the embryonic (E15) rat basal forebrain, radiolabeled probes from the human ChAT gene detected a 3,700 nt and a less abundant 2,300 nt transcript. After growth in the presence of NGF, the abundance of both mRNAs was increased approximately twofold, paralleling the increase in ChAT enzyme activity. In vivo, the human ChAT probes detected a single 3,700 nt form of ChAT mRNA in both embryonic and adult rat basal forebrain. These results suggest that the NGF-mediated increase in ChAT activity in basal forebrain cultures is regulated at the transcriptional level.

Promotion of neuronal survival in vitro by thermal proteins and poly(dicarboxylic)amino acids.

Evaluating molecules for their ability to promote survival and growth of neurons, we tested thermal proteins on cultures of dissociated fetal rat forebrain neurons. (Thermal proteins are polyamino acids formed when mixtures of amino acids with minimal proportions of glutamic or aspartic acid are heated.) Thermal proteins, added to low-density cultures in serum-free medium, stimulated neurite outgrowth and induced the formation of neuronal networks which survived for 6-10 days. Neurons in control cultures failed to grow and degenerated completely within 2-4 days. Effective concentrations (EC50) of thermal proteins ranged from 3 to 100 micrograms/ml. They were equally effective when present in the medium during the culture time or after precoating of the culture dishes. A single preparation which contained only aspartic and glutamic acid was effective, and similar survival promoting actions were then found for polyglutamic acid and mixed polyamino acids containing glutamic or aspartic acid. Thermal proteins and polyglutamic acid acted in a specific manner since, under the same experimental conditions, many control peptides, proteins and growth hormones failed to promote survival of neurons. Furthermore, their effects were antagonized by heparin, but not heparan sulfate nor chondroitin sulfate. These findings suggest that sequences of successive dicarboxylic amino acid residues are able to promote survival and neurite elongation of cultured neurons and that such sequences are responsible for the survival promoting action of thermal proteins. They invite the speculation that sequences of successive dicarboxylic amino acids, while occur in many proteins and show a high degree of evolutionary conservation, may have functional role in molecular recognition processes during neuronal development.

Evidence that both Ca(2+)-specific sites of skeletal muscle TnC are required for full activity.

To investigate the role of the Ca2(+)-specific (I and II) sites of fast skeletal muscle troponin C (TnC) in the regulation of contraction, we have produced two TnC mutants which have lost the ability to bind Ca2+ at either site I (VG1) or at site II (VG2). Both mutants were able to partially restore force to TnC-depleted skinned muscle fibers (approximately 25% for VG1 and approximately 50% for VG2). In contrast, bovine cardiac TnC (BCTnC), which like VG1 binds Ca2+ only at site II, could fully reactivate the contraction of TnC-depleted fibers. Higher concentrations of both mutants were required to restore force to the TnC-depleted fibers than with wild type TnC (WTnC) or BCTnC. VG1 and VG2 substituted fibers could not bind additional WTnC, indicating that all of the TnC-binding sites were saturated with the mutant TnC's. The Ca2+ concentration required for force activation was much higher for VG1 and VG2 substituted fibers than for WTnC or BCTnC substituted fibers. Also, the steepness of force activation was much less in VG1 and VG2 versus WTnC and BCTnC substituted fibers. These results suggest cooperative interactions between sites I and II in WTnC. In contrast, BCTnC has essentially the same apparent Ca2+ affinity and steepness of force activation as does WTnC. Thus, cardiac TnC must have structural differences from WTnC which compensate for the lack of site I, while in WTnC, both Ca2(+)-specific sites are probably crucial for full functional activity.

Complexities and sequence similarities of mRNA populations of cholinergic (NS20-Y) and adrenergic (N1E-115) murine neuroblastoma cell lines.

The clonal murine neuroblastoma cell lines NS20-Y and N1E-115 have been proposed as models for examining the commitment of neural crest cells to either the cholinergic or adrenergic phenotype, respectively. The validity of this model depends in part on the extent to which these two cell lines have diverged as a result of their transformed, rather than neuronal properties. In order to quantitate differences in gene expression between NS20-Y and N1E-115 cells, the mRNA complexity of each cell type was determined. An analysis of the kinetics of hybridization of NS20-Y cell mRNA with cDNA prepared from NS20-Y cell mRNA demonstrated the presence of approximately 11,700 mRNA species assuming an average length of 1900 nucleotides. A similar analysis using mRNA isolated from N1E-115 cells and cDNA prepared from N1E-115 cell mRNA demonstrated that the adrenergic cell line expressed approximately 11,600 mRNA species. The species of mRNA expressed by each cell line were resolved into high, intermediate, and low abundance populations. In order to determine whether mRNAs were expressed by the cholinergic, but not by the adrenergic cell line, NS20-Y cDNA was hybridized to an excess of N1E-115 cell mRNA. An analysis of the solution hybridization kinetics from this procedure demonstrated that the two cell lines do not differ significantly in the nucleotide complexity of their mRNA populations. The extensive similarity between the two mRNA populations suggests that only a small number of genes are expressed differentially between the two cell lines and supports their use as models for the differentiation of cholinergic and adrenergic neurons.

Isolation and sequence of a cDNA clone for rabbit fast skeletal muscle troponin C. Homology with calmodulin and parvalbumin.

The binding of Ca2+ to troponin C (TnC) regulates skeletal muscle contraction. We have isolated a full-length cDNA clone for fast skeletal muscle TnC from a neonatal rabbit skeletal muscle library and determined its nucleic acid sequence. The amino acid sequence deduced from this clone matches the previously reported amino acid sequence (Collins, J. H., Greaser, M. L., Potter, J. D., and Horn, M. J. (1977) J. Biol. Chem. 252, 6356-6362) except at the amino terminus. According to the nucleotide sequence, the first 2 residues of TnC are threonine-aspartic acid, which is the reverse of the order reported previously. The isolation of the adult form of TnC from a neonatal library suggests that there may be no developmental isoforms of fast TnC. The protein coding region of the fast TnC clone has 67% homology with the reported nucleotide sequence for chicken slow TnC (Putkey, J. A., Carroll, S. L., and Means, A. R. (1987) Mol. Cell. Biol. 7, 549-1553). The homologies between the nucleotide sequences of TnC, calmodulin, and parvalbumin provide evidence that all three proteins were derived from a common precursor molecule which had four Ca2+-binding sites.

A sulfhydryl group of the canine cardiac beta-adrenergic receptor observed in the absence of hormone.

Canine cardiac beta-adrenergic receptors contain a free sulfhydryl group in the adrenergic ligand binding site. [125 I]-Iodohydroxybenzylpindolol [( 125 I]-IHYP) binding to cardiac beta-receptors was inhibited 80% by treatment with 1 mM p-chloromercuribenzoic acid (pCMB). Occupation of the beta-receptors by an antagonist prior to treatment with pCMB prevented this effect suggesting that a sulfhydryl group is present in or near the ligand binding site of the cardiac beta-receptor. In the presence of agonists, the sensitivity of cardiac beta-receptors to pCMB was increased. Incubation of isoproterenol-occupied cardiac beta-receptors, resulted in a 57% inhibition of [125 I]-IHYP binding measured after extensive washing to remove bound agonist. The ability of isoproterenol to increase the reactivity of cardiac beta-adrenergic receptors supports the hypothesis that agonists produce a conformational change upon binding.

Inhibition of choline acetyltransferase by monoclonal antibodies.

Four monoclonal antibodies were obtained to rat brain choline acetyltransferase (CAT). The enzyme was purified 95,000-fold from rat brain by precipitation with acetic acid at pH 4.5, fractionation with 40 to 60% (NH4)2SO4, CM-Sephadex chromatography, and affinity column chromatography on agarose-hexane-coenzyme A. The enzyme preparation was applied to the affinity column in the presence of 10 mM acetylcholine to increase the affinity of CAT for coenzyme A; the enzyme then was eluted with 10 mM acetyl coenzyme A. Fusion of P3X63 Ag8 myeloma cells with spleen cells isolated from a BALB/c mouse that had been immunized with affinity-purified CAT with a specific activity of 29.4 mumol of ACh synthesized/min/mg of protein resulted in the isolation of four hybridomas synthesizing antibodies to CAT that inhibit the activity of the enzyme. Anti-CAT 1 or 2 inhibits CAT activity 100%. At the highest antibody concentration tested, anti-CAT 3 inhibited acetylcholine synthesis 80%. Hybridoma antibody-dependent inhibition of CAT activity was reversed by dissociation of immune complexes via dilution, demonstrating that antibody binding does not irreversibly alter the structure of the enzyme. When bound to [rabbit anti-mouse IgG . protein A Staphylococcus aureus] complexes, anti-CAT 1, 3, and 4 each were effective reagents for the precipitation of CAT activity from solution. Thirty-one to 53% of the precipitated enzyme was recovered following the dissociation of immune complexes. Anti-CAT 1, 2, and 3 inhibit CAT from 18-day chick embryo brain, NS20-Y mouse neuroblastoma cells, and rat brain.