0022-3565/04/3092-845–852$20.00 THE JOURNAL OF PHARMACOLOGY AND EXPERIMENTAL THERAPEUTICS Vol. 309, No. 2 Copyright © 2004 by The American Society for Pharmacology and Experimental Therapeutics 60806/1144303 JPET 309:845–852, 2004 Printed in U.S.A.

The Fenfluramine Metabolite (ϩ)-Norfenfluramine Is Vasoactive

Wei Ni, Melissa W. Li, Keshari Thakali, Gregory D. Fink, and Stephanie W. Watts Department of Pharmacology and Toxicology, Michigan State University, East Lansing, Michigan Received October 1, 2003; accepted January 29, 2004

ABSTRACT ϩ ϩ The anorexigen ( )-fenfluramine was used for treatment of 5-HT2A receptors and independence of ( )-norfenfluramine- obesity until the association of use with valvular heart disease induced contraction from stimulation of ␣-adrenergic receptors

and primary pulmonary hypertension. (ϩ)-Fenfluramine has and the sympathetic nervous system was validated by demon- Downloaded from been found in Chinese and Korean slimming pills. The hepatic strating 1) unchanged contraction to (ϩ)-norfenfluramine in ar- metabolite of (ϩ)-fenfluramine, (ϩ)-norfenfluramine, has affinity teries from chemically denervated rats; 2) a minimal effect of ␣ for 5-hydroxytryptamine (5-HT)2A and 5-HT2B receptors. We the 1-adrenergic receptor antagonist prazosin (100 nM) on tested the hypothesis that (ϩ)-norfenfluramine contracts arte- contraction; and 3) antagonism by [6-methyl-l-(1-methylethy)- rial smooth muscle in a 5-HT receptor-dependent manner and ergoline-8␤-carboxylic acid 2-hydroxy-1 methylpropyl ester acts as a pressor in the conscious rat. Isometric contraction maleate] LY53857 [6-methyl-1-(1-methylethy)-ergoline-8␤-car-

ϩ ␮ jpet.aspetjournals.org experiments showed that ( )-norfenfluramine (10 nM, 100 M) boxylic acid 2-hydroxy-1 methylpropyl ester maleate], a 5-HT2 but not (ϩ)-fenfluramine nor the isomer (Ϫ)-norfenfluramine receptor antagonist without ␣-receptor affinity. (ϩ)-Norfenflu- caused concentration-dependent contraction in arteries [Ϫlog ramine (10–300 ␮g/kg i.v.) caused a dose-dependent increase ϭ Ϯ EC50 (moles per liter), thoracic aorta 5.77 0.09; renal in mean arterial blood pressure in conscious rats, the maximum artery ϭ 6.29 Ϯ 0.02; mesenteric resistance artery ϭ 5.70 Ϯ of which could be virtually abolished by ketanserin (3 mg/kg i.v.)

0.06]. Contraction was dependent on the 5-HT2A receptor be- but not prazosin (0.2 mg/kg i.v.). Our findings demonstrate for cause ketanserin (10 nM) rightward shifted (ϩ)-norfenfluramine the first time that (ϩ)-norfenfluramine is vasoactive and has the response curves (aorta ϭ 16-fold, renal artery ϭ 26-fold, and potential to increase blood pressure. resistance artery ϭϾ100-fold). Dependence on activation of at ASPET Journals on June 8, 2015

The highly effective anorexigen (ϩ)-fenfluramine (Redux) valvulopathy. Recently, the Food and Drug Administration was approved as an individual agent by the Food and Drug warned that two Chinese diet pills contain fenfluramine (http:// Administration for long-term use in the medical management of www.fda.gov/bbs/topics/NEWS/2002/NEW00826.html). Un- obesity in 1996. American health care practitioners used (ϩ)- doubtedly, there are still individuals ingesting (ϩ)-fenflura- fenfluramine either alone or in combination with a noradrener- mine; thus, it is important for potential serious side effects of gic appetite suppressant, phentermine (combination of fenflu- this drug to be highlighted. ramine and phentermine, “fen-phen”); the combination was (ϩ)-Fenfluramine is a 5-hydroxytryptamine transporter never approved by Food and Drug Administration (www. substrate and potent 5-HT releaser (Garattini, 1995). The fda.gov/cder/news/phen/phenfen.htm). The effectiveness of the hepatic deethylated metabolite of (ϩ)-fenfluramine is (ϩ)- combination (fen-phen) in reducing body weight was similar to norfenfluramine, which also possesses the ability to release use of each individual drug alone, but lower doses of each drug 5-HT. Important for this study is the knowledge that (ϩ)- could be used with the intent of reducing side effects. However, norfenfluramine has been described as a 5-HT receptor ago- (ϩ)-fenfluramine was pulled from the market in 1997 because nist. Radioligand binding experiments show (ϩ)-fenflura- Ͼ ␮ of the incidence of pulmonary hypertension (Abenhaim et al., mine has a weak affinity (Ki 2 M) for the 5-HT2A, 5-HT2B, ϩ 1996) and aortic valvular disease (Connolly et al., 1997). Phen- and 5-HT2C receptor, whereas ( )-norfenfluramine has high ϭ termine was not withdrawn because it was not associated with affinity at the 5-HT2B receptor (Ki 11.2 nM), 5-HT2C re- ϭ ceptor (Ki 324 nM), and lesser affinity at the 5-HT2A This study was supported by National Institutes of Health Grant ϭ receptor (Ki 1516 nM) (Rothman et al., 2000). 5-HT2A HL58489.Article, publication date, and citation information can be found at http://jpet.aspetjournals.org. receptors mediate contraction to 5-HT in many arteries and DOI: 10.1124/jpet.103.060806. veins, especially conduit vessels (Martin, 1994). The 5-HT2B

ABBREVIATIONS: 5-HT, 5-hydroxytryptamine; PSS, physiological salt solution; PE, phenylephrine; 6-OHDA, 6-hydroxydopamine 6-OHDA; LY272015, 6-methyl-1,2,3,4-tetrahydro-1-[3,4-dimethoxyphenyl)methyl]-9H-pyrido[3,4-b]indole hydrochloride; SNS, sympathetic nervous sys- tem; NE, norepinephrine; RX8821002, 2-methoxyidazoxan; LY53857, 6-methyl-1-(1-methylethyl)-ergoline-8␤-carboxylic acid 2-hydroxy-1 meth- ylpropyl ester maleate; UK14304, 5-bromo-N-[4,5-dihydro-1H-imidazol-2-yl]-6-quinoxalinamine. 845 846 Ni et al. receptor was cloned in rat stomach fundus in 1992 (Foguet et Myograph Protocol al., 1992; Kursar et al., 1992) and plays an important role in The whole intestine was placed in a silastic-filled petri dish in cold mediating arterial contraction in deoxycorticosterone ace- PSS and pinned down. Second and third order arteries (200–300 ␮m, tate-salt hypertensive rats (Watts, 1998). Thus, we hypothe- inner diameter) were carefully dissected from the vein and placed in sized that (ϩ)-norfenfluramine may activate 5-HT receptors cold PSS. Two tungsten wires (California Wire, Grover Beach, CA) and be a vasoactive substance. were threaded through the lumen of the artery. One wire was Few studies have tested whether (ϩ)-fenfluramine can al- mounted to a micrometer and the other to a microtransducer. All ter smooth muscle contractility (Desta et al., 1998), and it has arteries were examined with an intact endothelium. Baths, filled never been determined whether (ϩ)-norfenfluramine con- with PSS, were warmed by a water-circulated jacket around the bath and are aerated with 95% O ,5%CO. Tissues equilibrated for 30 tracts smooth muscle. We investigated whether (ϩ)-norfen- 2 2 min before an optimal passive tension of 400 mg (determined previ- fluramine causes a concentration-dependent contraction in ously; Watts, 2002) was applied using the micrometer. Tissues equil- three isolated arteries and whether the contraction depends ibrated another 30 min with frequent buffer changes before chal- on 5-HT receptors, sympathetic nerves [a site of potential lenge with a maximal concentration of PE (10 ␮M). Experimental uptake of (ϩ)-norfenfluramine], or ␣-adrenergic receptors. protocols are as described above. We then investigated whether and the mechanism by which (ϩ)-norfenfluramine could alter mean arterial blood pressure Denervation Experiment Protocol in conscious normal rats, with the intent of connecting our Sympathetic neuronal denervation was induced by 6-hydroxydo- data derived from in vitro experiments to the whole animal. pamine (6-OHDA) injections (McCafferty et al., 1997). Rats were

treated with four doses of 6-OHDA over 7 days (50 mg/kg on days 1 Downloaded from and 2 and 100 mg/kg on days 6 and 7; 0.1% ascorbic acid in physio- Materials and Methods logical saline as vehicle i.p.). Rats were euthanized (pentobarbital, 60 mg/kg i.p.), and arteries were removed on day 8. Contractility exper- Animal Use iments were performed on the aorta, renal artery, and mesenteric The normal male Sprague-Dawley rats (0.225–0.300 kg; Charles resistance arteries from vehicle-treated and denervated rats. River Laboratories, Inc., Portage, MI) were used in these experi- ments. Glyoxylic Acid Protocol jpet.aspetjournals.org Aorta and mesenteric resistance arteries were removed and placed Isolated Tissue Bath Protocol in normal PSS. After fat was trimmed off, arteries were immersed Thoracic aorta and renal artery were removed and placed in nor- into glyoxylic acid (2%) for 5 min. Blood vessels were mounted on a mal physiological salt solution (PSS) containing 130 mM NaCl, 4.7 microscope slide and placed in an oven (100°C) for 5 min. The slides ⅐ were removed and blood vessels mounted in mineral oil and cover- mM KCl, 1.18 mM KH2PO4, 1.17 mM MgSO4 7H2O, 1.6 mM ⅐ slipped. Vessels were viewed using a fluorescence microscope (Nikon CaCl2 2H2O, 14.9 mM NaHCO3, 5.5 mM dextrose, and 0.03 mM

LABOPHOT) and UV illumination (G-1A; excitation, 546/10 nm; at ASPET Journals on June 8, 2015 CaNa2EDTA (pH 7.2). Vessels were trimmed of fat and cut into helical strips (aorta, 1 ϫ 10 mm; renal artery, 1 ϫ 5 mm). We used barrier filter, BA580). The absence of a fluorescent network of stain- endothelium-intact arteries in our experiments. ing of catecholamines in arteries confirms the effectiveness of Tissues were attached to a fixed, stainless steel rod at one end and 6-OHDA in causing sympathetic denervation. Method is according to to a force transducer (FT03; Grass Instruments, Quincy, MA) at the Ruijtenbeek et al. (2000). Only pictures of small resistance arteries other. Baths were filled with PSS, warmed to 37°C, and aerated with are shown as validation of this protocol, because no sympathetic 95% oxygen and 5% carbon dioxide. Each strip was placed under network was readily discernible in the aorta. optimum resting tension (previously determined; rat aorta, 1500 mg; In Vivo Experiments renal artery, 500 mg) and allowed to equilibrate for 1 h with frequent buffer changes. Tissues were then challenged with a maximal con- Catheters were constructed of polyvinyl chloride with silicone ␣ Ϫ5 rubber tips and advanced to the abdominal aorta and vena cava via centration of the 1-adrenergic agonist phenylephrine (PE 10 mM) to initiate a maximal contraction and washed repeatedly until tone the left femoral artery and vein in rats anesthetized with pentobar- returned to baseline. To examine the status of the arterial endothe- bital (50 mg/kg i.p.). The ends of the catheters were tunneled sub- lium, tissues were contracted with a half-maximal concentration of cutaneously to the head where the catheters were stabilized to the PE (10Ϫ8–10Ϫ7 mM), and once the contraction plateaued, the mus- skull by using jeweler’s screws and dental acrylic. Catheter ends carinic agonist acetylcholine (10Ϫ6 mM) was administered. We ob- were passed through a stainless steel spring attached to a plastic served a relaxation to acetylcholine greater than 60% of the PE swivel, through which infusions were given (venous end) and arterial (10Ϫ8–10Ϫ7 mM)-induced contraction as endothelium intact. Tissues end was connected to a pressure transducer (TRN050; Kent Scien- were again washed until baseline was reached, and then one of the tific Corp. Litchfield, CT). Upon regaining consciousness, rats were following protocols was followed. housed singly in stainless steel cages in a climate-controlled room Protocol 1: Testing of Fluramines as Contractile Agonists. with a 12-h light/dark cycle. At least 1-week recovery was allowed Concentration-response curves to fluramines were performed in a before experiments were begun. cumulative manner. Each concentration incubated a minimum of 3 (ϩ)-Norfenfluramine (10–300 ␮g/kg i.v.) was given in a cumula- min; when contraction reached a maximum, the next higher concen- tive manner at 6-min intervals. Mean arterial blood pressure and tration of agonist was added. Contraction to agonist was normalized heart rate were monitored before, during, and for 6 min after each by maximal contraction to PE. injection with a computerized DigiMed system. Antagonist or vehicle .Protocol 2: Testing of Effect of Antagonists on (؉)-Norfen- was given 30 min before (ϩ)-norfenfluramine fluramine-Induced Contraction. Vehicle or antagonist was added Statistics to the bath. Antagonists incubated with the tissues for1hatwhich time the cumulative response to (ϩ)-norfenfluramine in the presence Contractile data are expressed as ϮS.E.M. and reported as a of vehicle or antagonist was examined. In some experiments using percentage of the maximal contraction to PE (10Ϫ5 mM). Unpaired t LY53857, vehicle (0.1% dimethyl sulfoxide) or LY53857 (10 nM) was tests were performed and a p value Յ0.05 was considered statisti- ϩ added once the maximum contraction to ( )-norfenfluramine was cally significant. Agonist EC50 values were calculated using a non- established, and reduction in contraction after 30 min was recorded. linear regression analysis using the algorithm [effect ϭ maximum Norfenfluramine Is Vasoactive 847

ϩ response/1 (EC50/agonist concentration)] in the program Graph- Pad Prism. Apparent antagonist dissociation constants (KB values) were calculating using the following equation: log(dr Ϫ 1) ϭ log[B] Ϫ log KB, where dr is the EC50 value of agonist in the presence of the antagonist divided by the EC50 value of agonist in the absence of the antagonist, and [B] is the concentration of the antagonist tested. Blood pressure is reported as a change in mean arterial blood pres- sure and a repeated measures analysis of variance was used to ascertain statistical differences between groups.

Chemicals Acetylcholine chloride, (ϩ)-fenfluramine, 5-hydroxytryptamine hydrochloride, ketanserin tartrate, LY53857, phenylephrine hydro- chloride, prazosin hydrochloride, RX821002, UK14304, 6-OHDA, and glyoxylic acid were purchased from Sigma-Aldrich (St. Louis, MO). LY272015 was a generous gift from Eli Lilly and Company (Indianapolis, IN) and (ϩ)-norfenfluramine was graciously provided by SRI International (Menlo Park, CA). Downloaded from Results Response of Normal Rat Arteries to the Fluramines. Figure 1A compares response of endothelium-intact aorta from normal rats to (Ϫ)-norfenfluramine, (ϩ)-norfenflura- mine, and (ϩ)-fenfluramine. Only (ϩ)-norfenfluramine con- tracted the aorta in a concentration-dependent manner with jpet.aspetjournals.org Ϫ Ϯ ϩ a log EC50 value of 5.77 0.09 mM, whereas ( )-fenflura- mine and (Ϫ)-norfenfluramine were virtually inactive. Simi- larly, (ϩ)-norfenfluramine (Fig. 1B) but not (ϩ)-fenfluramine (Fig. 1C) contracted endothelium-intact renal arteries (Ϫlog ϭ Ϯ EC50 6.29 0.02 mM) and mesenteric resistance arteries Ϫ ϭ Ϯ ϩ ( log EC50 5.70 0.06 mM). The potency of ( )-norfen- fluramine-induced contraction in these three arteries was at ASPET Journals on June 8, 2015 relatively similar, but there were notable differences in the maximal contraction. In preliminary experiments in the rat aorta, removal of the endothelial cell layer did not affect ϩ Ϫ ϭ ( )-norfenfluramine-induced contraction ( log EC50 5.62 Ϯ 0.08 mM, max ϭ 75.89 Ϯ 4.29). Thus, in all experi- ments we have kept the endothelial cell layer intact to most closely parallel the in vivo situation. -Inhibition of (؉)-Norfenfluramine-Induced Contrac tion in Aorta and Pressor Response in Normal Rats by

5-HT2A/2C Receptor Antagonist Ketanserin. To deter- mine the receptor type mediating (ϩ)-norfenfluramine-in- Fig. 1. A, effect of fluramines on contraction of endothelium-intact aorta from normal Sprague-Dawley rats. B, effect of (ϩ)-norfenfluramine on duced contraction, contraction in aorta was examined in the contraction of endothelium-intact mesenteric artery and renal artery. C, presence of 5-HT and ␣-adrenergic receptor antagonists. Ket- effect of (ϩ)-fenfluramine on contraction of endothelium-intact mesen- anserin (Fig. 2A; 10 nM) competitively shifted the (ϩ)-nor- teric artery and renal artery. Points represent means and vertical bars fenfluramine concentration-response curve rightward 16-fold the S.E.M. for the number of animals indicated in parentheses. in aorta from normal rats. The apparent dissociation con- ϭ Ϯ stant calculated from this shift (pKB 9.27 0.8) is consis- fundus (52.5-fold), prazosin (100 nM) abolished a PE-induced Ϫ5 tent with antagonism of the 5-HT2A receptor. Antagonists for contraction in rat aorta (percentage of 10 M of PE-induced ϭ Ϯ other relevant heptahelical receptors, e.g., the 5-HT2B recep- contraction, control maximal 111.8 4.9, prazosin-incu- ␣ ϭ Ϯ Ͻ ␮ tor antagonist LY272015 (10 nM), 1-adrenergic receptor bated maximal 3.05 1.7; p 0.05) and RX821002 (1 M) ␣ antagonist prazosin (100 nM), and 2-adrenergic receptor inhibited a UK14304-induced aortic contraction (percentage antagonist RX821002 (1 ␮M), did not significantly alter the of 10Ϫ5 mM PE-induced contraction, control maximal ϭ potency or maximal response elicited by (ϩ)-norfenfluramine 50.4 Ϯ 4.5, RX821002-incubate maximal ϭ 7.56 Ϯ 1.9; p Ͻ (Table 1). These antagonists were tested in concentrations 0.05). ϩ that are largely unable to interact with the 5-HT2A receptor Ketanserin (10 nM) inhibited ( )-norfenfluramine-induced and were effective at their specific receptor as suggested by contraction in renal artery (Fig. 2B) and mesenteric resis- reported Ki values in the Table 1. We validated that these tance artery (Fig. 2C). Notably, ketanserin produced an in- antagonists were used in appropriate concentrations in con- hibition that was quantitatively greater in resistance arter- trol experiments. Specifically, LY272015 (10 nM) rightward- ies (Ͼ100-fold shift) as compared with that observed in the shifted 5-HT-induced contraction of the isolated rat stomach renal artery or aorta. These results suggest that the predom- 848 Ni et al.

Fig. 2. Effect of 5-HT2A receptor antagonist ketanserin on (ϩ)-norfenfluramine-induced contraction in rat aorta (A), rat renal artery (B), and rat mesenteric resistance artery (C). The endothelium was intact in all arteries. D,

effect of 5-HT2A receptor antagonist ketan- serin on (ϩ)-norfenfluramine-induced pressor response in normotensive conscious rats. Points represent means and vertical bars the S.E.M. for the number of animals indicated in

parentheses. Asterisk (*) indicates statisti- Downloaded from cally significant differences from control re- sponse (p Ͻ 0.05). MAP, mean arterial blood pressure. jpet.aspetjournals.org at ASPET Journals on June 8, 2015 TABLE 1 Effects of heptahelical receptor antagonists on (ϩ)-norfenfluramine-induced contraction in aorta from normotensive rats n ϭ 4.

Ϫ Maximum Drug Target (Ki) log EC50 Contraction

nmol M % PE contraction Vehicle 5.77 Ϯ 0.09 82.54 Ϯ 2.53 a Ϯ Ϯ LY272015 (10 nM) 5-HT2B receptor (0.75) 5.75 0.01 91.61 2.96 ␣ b Ϯ Ϯ Prazosin (100 nM) 1-Adrenergic receptor (0.01) 5.64 0.01 82.23 3.79 ␮ ␣ c Ϯ Ϯ RX821002 (1 M) 2-Adrenergic receptor (0.95) 5.55 0.02 85.54 3.03 a Cohen et al. (1996). b Jones et al. (1987). c Naselsky et al. (2001). inant receptor mediating (ϩ)-norfenfluramine contraction in sor response were dependent on the sympathetic nervous arteries from normal rats is a 5-HT2A receptor but that there system, we performed experiments on sympathectomized is either a difference in the coupling of 5-HT2A receptors or rats. Figure 3A shows the ablation of the sympathetic ner- ketanserin may cause mild ␣-adrenergic antagonism in re- vous system (SNS) in mesenteric resistance arteries from sistance arteries. rats that received 6-OHDA. The fluorescence network is com- Before discriminating between these possibilities, we ex- posed of sympathetic nerve fibers as induced by the glyoxylic amined the effect of (ϩ)-norfenfluramine on blood pressure in acid reaction product formed with biogenic amines in small the conscious rat. Given intravenously, (ϩ)-norfenfluramine arteries from vehicle-treated rats (vehicle); we do not show increased blood pressure in a dose-dependent manner in pictures for aorta here because no discernible network was conscious normal rats without a concomitant change in heart observed. Arteries from 6-OHDA-treated rats lost most of the rate (data not shown). This pressor response was markedly nerve innervation, as seen by the loss of a fine white network reduced by ketanserin (3 mg/kg; Fig. 2D) and again raised of fibers (Fig. 3A). Importantly, we did not observe a signif- the possibility that (ϩ)-norfenfluramine was interacting with icant difference in contraction induced by (ϩ)-norfenflura- the ␣-adrenergic system/sympathetic nervous system. mine in aorta from normal rats and sympathectomized rats Effect of Chemical Sympathectomy on (؉)-Norfen- (Fig. 3B), and (ϩ)-norfenfluramine-induced contraction in fluramine-Induced Contraction in Aorta. To determine arteries from denervated rats was similarly antagonized by whether (ϩ)-norfenfluramine-induced contraction and pres- ketanserin compared with controls (Fig. 3B). (ϩ)-Norfenflu- Norfenfluramine Is Vasoactive 849 Downloaded from jpet.aspetjournals.org

Fig. 3. A, glyoxylic acid fluorescence in small mesenteric resistance arteries from vehicle or 6-OHDA-treated rats. The white network is sympathetic

ϩ at ASPET Journals on June 8, 2015 nerve fibers. B, effect of the 5-HT2A/2C receptor antagonist ketanserin (10 nM) on ( )-norfenfluramine-induced contraction in aorta from 6-OHDA- ϩ treated rats. Dashed lines are data representing ( )-norfenfluramine-induced contraction in normal rats. C and D, effect the 5-HT2 receptor antagonist LY53857 (10 nM) had on maximum contraction to (ϩ)-norfenfluramine in renal artery (C) and mesenteric resistance arteries (D). Points/bars represent means and vertical lines the S.E.M. for the number of animals indicated in parentheses. Asterisk (*) indicates statistically significant differences (p Ͻ 0.05) between control/vehicle and artery incubated with antagonist.

TABLE 2 Ϫ ϩ Potency ( log EC50) and maximum contraction elicited by ( )-norfenfluramine in arteries from vehicle- and 6-OHDA-treated rats n ϭ 3to7.

Vehicle 6-OHDA

Ϫ %PE Ϫ %PE log EC50 Contraction log EC50 Contraction

MM Thoracic aorta 5.79 ϩ 0.31 83.9 Ϯ 4.7 5.85 Ϯ 0.13 87.4 Ϯ 4.90 Renal artery 6.31 Ϯ 0.10 120.3 Ϯ 5.6 6.44 Ϯ 0.13 114.9 Ϯ 19.9 Mesenteric resistance artery 5.40 Ϯ 0.13 52.3 Ϯ 8.6 5.62 Ϯ 0.09 84.2 Ϯ 18.8 ramine-induced contraction in the renal artery and mesen- little ␣-adrenergic receptor affinity (Cohen et al., 1988). Fig- teric resistance arteries of denervated rats was not altered ure 4 depicts that in aorta and in mesenteric resistance compared with control (Table 2), and maximal contraction to arteries, LY53857 significantly reduced (ϩ)-norfenflura- ϩ ( )-norfenfluramine was significantly reduced by the 5-HT2 mine-induced contraction. receptor antagonist LY53857 (10 nM; Fig. 3, C and D). Thus, The ability of (ϩ)-norfenfluramine to interact directly with it is unlikely that (ϩ)-norfenfluramine-induced contraction ␣-adrenergic receptors was investigated in mesenteric resis- depends on the peripheral SNS. tance arteries. At a concentration that abolishes contraction ␣ ؉ Effect of LY53857 and Prazosin on ( )-Norfenflura- to a maximal concentration of the 1-adrenergic receptor mine-Induced Vasoreactivity. To further rule out the agonist PE (10Ϫ5 M), prazosin (100 nM) reduced contraction ability of (ϩ)-norfenfluramine to stimulate contraction by to (ϩ)-norfenfluramine in arteries from control rats only at directly interacting with ␣-adrenergic receptors in small ves- the lower concentrations of (ϩ)-norfenfluramine (Fig. 5A). ϩ sels, and to confirm the involvement of the 5-HT2A receptor, The potency of ( )-norfenfluramine was not significantly re- ϩ Ϫ ϭ Ϯ we examined the effects of LY53857 (30 nM) on ( )-norfen- duced by prazosin ( log EC50 control 5.40 0.13 M; fluramine-induced contraction. This antagonist does not dis- prazosin ϭ 5.05 Ϯ 0.17 M; p ϭ 0.08). Importantly, prazosin ϩ criminate between 5-HT2A and 5-HT2B receptors, but it has exerted only a minimal inhibition on ( )-norfenfluramine- 850 Ni et al.

Fig. 4. Effect of the 5-HT2 receptor antagonist LY53857 (10 nM) on (ϩ)-norfenfluramine-in- duced contraction in endothelium-intact rat aorta (A) and mesenteric resistance arteries (B). Points represent means and vertical bars the S.E.M. for the number of animals indicated in parentheses. Asterisk (*) signifies statistically differences (p Ͻ 0.05) from responses of control tissues. Downloaded from induced contraction in mesenteric resistance arteries that Discussion were from 6-OHDA denervated rats (Fig. 5B), a tissue in which direct interaction with ␣-adrenergic receptors could The correlation between body weight and blood pressure is most simply be examined. Again, the potency of (ϩ)-norfen- significant and real in the human (Chiang et al., 1969; Stam- fluramine was not significantly reduced by prazosin (con- ler et al., 1978; for review, see Montani et al., 2002). Flecht- trol ϭ 5.62 Ϯ 0.08 M, prazosin ϭ 5.41 Ϯ 0.12 M; p ϭ 0.10). ner-Mors et al. (1988) reported that short-term treatment Finally, prazosin (0.2 mg/kg i.v.; Oates, 1979) only modestly with (ϩ)-fenfluramine in obese postmenopausal women de- jpet.aspetjournals.org reduced the pressor response to (ϩ)-norfenfluramine in con- creased blood pressure and heart rate. By contrast, long-term scious rats (Fig. 5C). use of anorexigen agent (ϩ)-fenfluramine to control body at ASPET Journals on June 8, 2015

␣ ϩ Fig. 5. Effect of the 1-adrenergic receptor antagonist prazosin (100 nM) on contraction to ( )-norfenfluramine in mesenteric resistance arteries from control rats (A) and rats given 6-OHDA (B). C, effects of prazosin (0.2 mg/kg i.v.) on (ϩ)-norfenfluramine-induced increases in blood pressure of the conscious rat. Points/bars represent means and vertical lines the S.E.M. for the number of animals indicated in parentheses. MAP, mean arterial blood pressure. Asterisk (*) indicates statistically significant differences (p Ͻ 0.05) between control/vehicle and artery/rat incubated with antagonist. Norfenfluramine Is Vasoactive 851 weight has resulted in primary pulmonary hypertension norepinephrine transporter. In large arteries, we do not be- (Abenhaim et al., 1996), valvular disease (Connolly et al., lieve (ϩ)-norfenfluramine interacts directly with ␣-adrener- 1997), and systemic hypertension in some populations (Ma- gic receptors or releases NE to cause contraction because 1) ϩ ␣ ␣ badeje, 1974). We demonstrate here the ability of ( )-norfen- the 1-adrenergic receptor antagonist prazosin and 2-adren- fluramine, a metabolite of (ϩ)-fenfluramine, to be vasoactive ergic receptor antagonist RX821002 were unable to block primarily through activation of a 5-HT receptor. (ϩ)-norfenfluramine-induced contraction in normal aorta; Contraction to (؉)-Norfenfluramine. Our results show and 2) we did not observe a significant difference between the that it is (ϩ)-norfenfluramine (Fig. 1, A and B), and not contraction to (ϩ)-norfenfluramine in aorta from denervated (ϩ)-fenfluramine (Fig. 1C), that causes concentration-depen- rats and normal rats. Thus, it is unlikely that NE is released dent contraction in various arteries. Aorta and renal artery from efferent sympathetic nerve endings and involved in were used as conduit arteries and were examined in addition (ϩ)-norfenfluramine-induced contraction in aorta. to a true resistance artery from the mesentery. Use of these We next determined whether a similar independence ex- small resistance vessels is important because they contribute isted in arteries smaller than the aorta. (ϩ)-Norfenfluramine to determination of total peripheral resistance. Our data induced contraction in both renal arteries and mesenteric demonstrate that (ϩ)-norfenfluramine contracts both conduit resistance arteries from 6-OHDA denervated rats, indicating and resistance arteries (Fig. 1, A and B), so contraction is not that the sympathetic nervous system was not necessary. vessel-specific. Moreover, the contractility experiments were Importantly, contraction to (ϩ)-norfenfluramine in these ar- performed in arteries with intact endothelium, thereby most teries could be antagonized by LY53857, supporting the in- Downloaded from closely approaching the in vivo situation. volvement of a 5-HT2A receptor in contraction. Our findings Our data suggest that (ϩ)-norfenfluramine contracts arte- suggest a weak direct interaction of (ϩ)-norfenfluramine rial smooth muscle in a 5-HT2A receptor-dependent manner. with the adrenergic receptor. This is on the basis of the weak Using a series of amine receptor antagonists, we found that antagonism exerted by prazosin in mesenteric resistance ar- contraction to (ϩ)-norfenfluramine was significantly inhib- teries from control and 6-OHDA-treated rats and the fact ϩ ited by the 5-HT2A/2C receptor antagonist ketanserin (10 nM) that prazosin only modestly affected ( )-norfenfluramine- in aorta from normal rats (Fig. 2A). The same concentration induced increases in blood pressure. To our knowledge, phar- jpet.aspetjournals.org of ketanserin also inhibited (ϩ)-norfenfluramine-induced macological interaction of (ϩ)-norfenfluramine with ␣-adren- contraction in renal artery and mesenteric resistance artery ergic receptors has not been reported. Fig. 2, B and C). In theory, this concentration of ketanserin (؉)-Norfenfluramine as a Pressor Agent. As early as) ␣ ϩ has little effect on 1-adrenergic receptor activation (ketan- 1974, treatment with ( )-fenfluramine was related to blood ϭ serin at 5-HT2A receptor, Ki 0.39 nmol, Leysen et al., 1982; pressure increases (Mabadeje, 1974). In 1999, it was reported ␣ ϭ ϩ and ketanserin at 1-adrenergic receptor, Ki 72.4 nmol; that ( )-fenfluramine increased systemic blood pressure, and

Korstanje et al., 1986). The 5-HT2C receptor, another recep- this pressor response was due to an increase in systemic at ASPET Journals on June 8, 2015 tor for which ketanserin has significant affinity, has never vascular resistance (Michelakis et al., 1999). As the active definitively been found in periphery (Barnes and Sharp, metabolite of (ϩ)-fenfluramine, (ϩ)-norfenfluramine has a

1999). Importantly, the nonselective 5-HT2 receptor antago- significantly longer half-life (18 versus 30 h) (http://www.ma- nist LY53857, which has significantly lower affinity for an ripoisoncenter.com/ctr/9707fenfluramine.html). Based on the adrenergic receptor compared with ketanserin (Cohen et al., results of our in vitro experiments and the long half-life of 1988), also antagonized (ϩ)-norfenfluramine-induced con- (ϩ)-norfenfluramine, we speculate that the (ϩ)-fenfluramine traction in aorta. Collectively, these data suggest that the pressor response could be caused, at least partially, by (ϩ)- ϩ 5-HT2A receptor is likely involved in ( )-norfenfluramine- norfenfluramine. Our in vivo experiments are consistent induced contraction. with this possibility. (ϩ)-Norfenfluramine (10–300 ␮g/kg i.v.) However, it was a concern that the antagonism exerted by caused a dose-dependent increase in blood pressure in con- ketanserin in the mesenteric resistance arteries was quanti- scious rats (Fig. 2D) without a concomitant change in heart tatively greater than that observed in either aorta or renal rate. The concentration of (ϩ)-norfenfluramine that elicits artery. We interpreted these data to suggest that either the contraction and pressor responses are in a dose range con-

5-HT2A receptors in the small resistance arteries were ex- sidered therapeutic in humans (20 mg t.i.d.; plasma levels quisitely sensitive to antagonism of 5-HT receptors and/or approximately 1.7 ϫ 10Ϫ6 M) (http://www.maripoisoncenter. that ketanserin might exert ␣-adrenergic antagonism in com/ctr/9707fenfluramine.html). Thus, the potential for (ϩ)- these small arteries. Because of these data and the fact that norfenfluramine to cause these cardiovascular effects in the (ϩ)-norfenfluramine has been described as a norepinephrine human is real. In addition, we investigated the sensitivity of transporter substrate (Rothman et al., 2003), it was crucial to (ϩ)-norfenfluramine-induced pressor responses to ketan- determine whether (ϩ)-norfenfluramine interacted directly serin and prazosin in vivo. Compared with untreated normal with ␣-adrenergic receptors or indirectly through release of rats, pretreatment with 3 mg/kg ketanserin nearly abolished NE. the pressor effect of (ϩ)-norfenfluramine (Fig. 2D), whereas Independence from SNS and Modest Dependence on prazosin exerted only a modest effect (Fig. 5C). An important ␣-Adrenergic Receptor. Our previous experiments raise point to make is that in different experiments in which small the question of whether (ϩ)-norfenfluramine activates arterial function has been investigated, all point to the sug- ␣ 5-HT2A and/or -adrenergic receptor directly to cause con- gestion that the 5-HT2A receptor on small arteries is exquis- traction or stimulates local neuronal 5-HT or NE release, itely sensitive to antagonism. These experiments include the which in turn contracts the arteries. Although few peripheral quantitatively greater antagonism exerted by ketanserin and arteries are innervated with serotonergic nerves, 5-HT and LY53857 in mesenteric resistance arteries. If one were to ϩ ( )-norfenfluramine have the potential to be taken up by estimate the pKB for ketanserin in resistance arteries by 852 Ni et al. assuming that a parallel response to 5-HT in ketanserin- plasma norepinephrine responses to dexfenfluramine in obese postmenopausal women. Am J Clin Nutr 67:611–615. incubated tissues was observed in Fig. 2C, the pKB value Foguet M, Hoyer D, Pardo LA, Parekh A, Kluxen FW, Kalkman HO, Stuhmer W, and would be close to 11, significantly different from what has Lubbert H (1992) Cloning and functional characterization of the rat stomach fundus serotonin receptor. EMBO (Eur Mol Biol Organ) J 11:3481–3487. classically been found for 5-HT2A receptors. This difference in Garattini S (1995) Biological actions of drugs affecting serotonin and eating. Obes large versus small artery 5-HT2A receptor pharmacology will Res 3 (Suppl 4):463S–470S. Jones SB, Smith JM, Jones AW, and Bylund DB (1987) Alpha-1 adrenergic receptor continue to be of interest. binding in aortas from rat and dog: comparison of [3H]prazosin and beta-iodo- Collectively, (ϩ)-norfenfluramine likely directly activates [125I]-4-hydroxyphenyl-ethyl-aminomethyl-tetralone. J Pharmacol Exp Ther 241: the 5-HT receptor, which mediates contraction in arteries 875–881. 2A Korstanje C, Sprenkels R, Doods HN, Hugtenburg JG, Boddeke E, Batink HD, and leads to an increase in blood pressure. In our experi- Thoolen MJ, and Van Zwieten PA (1986) Characterization of flufylline, fluprofyl- ments, we cannot discount the possibility that some of effects line, ritanserin, butanserin and R 56413 with respect to in-vivo alpha 1-,alpha2- ϩ and 5-HT2-receptor antagonism and in-vitro affinity for alpha1-,alpha2- and of ( )-norfenfluramine are centrally mediated and this is a 5-HT2-receptors: comparison with ketanserin. J Pharm Pharmacol 38:374–379. noted limitation of these studies. Nonetheless, our work dem- Kursar JD, Nelson DL, Wainscott DB, Cohen ML, and Baez M (1992) Molecular cloning, functional expression and pharmacological characterization of a novel onstrates for the first time that (ϩ)-norfenfluramine is vaso- serotonin receptor (5-hydroxytryptamine2F) from rat stomach fundus. Mol Phar- active and has the potential to increase blood pressure. macol 42:549–557. Leysen JE, Niemegeers CJ, Van Nueten JM, and Laduron PM (1982) [3H]Ketan- In summary, the major finding in this study is that (ϩ)- 3 serin (R 41 468), a selective H-ligand for serotonin2 receptor binding sites. Bind- norfenfluramine, a hepatic metabolite of the anorexigen (ϩ)- ing properties, brain distribution and functional role. Mol Pharmacol 21:301–314. Mabadeje AF (1974) Proceedings: fenfluramine-induced hypertension. West Afr fenfluramine, is vasoactive and a pressor agent. In normal J Pharmacol Drug Res 2:54. Martin GR (1994) Vascular receptors for 5-hydroxytryptamine: distribution, function rats, the 5-HT2A receptor is the primary receptor mediating

and classification. Pharmacol Ther 62:283–324. Downloaded from vessel contraction and pressor response, although a modest McCafferty DM, Wallace JL, and Sharkey KA (1997) Effects of chemical sympathec- ␣ direct interaction with 1-adrenergic receptors may occur. tomy and sensory nerve ablation on experimental colitis in the rat. Am J Physiol ϩ 272:G272–G280. Our findings reveal a new and important action of ( )-nor- Michelakis ED, Weir EK, Nelson DP, Reeve HL, Tolarova S, and Archer SL (1999) fenfluramine and confirm that drug design for body weight Dexfenfluramine elevates systemic blood pressure by inhibiting potassium cur- control in the future should avoid a pharmacophore similar to rents in vascular smooth muscle cells. J Pharmacol Exp Ther 291:1143–1149. Montani J-P, Antic V, Yang Z, and Dulloo Abdul (2002) Pathways from obesity to (ϩ)-norfenfluramine. hypertension: from the perspective of a vicious triangle. Int J Obesity 26 (Suppl

2):S28–S38. jpet.aspetjournals.org Naselsky DP, Ashton D, Ruffolo RR, and Hieble JP (2001) Rabbit alpha2- Acknowledgments adrenoceptors: both platelets and adipocytes have alpha2A-pharmacology. J Phar- macol Exp Ther 298:219–225. We are grateful to the pharmaceutical companies Eli Lilly & Co. Oates HF (1979) Relationship between the hypotensive and alpha-adrenoceptor (Indianapolis, IN) and SRI International (Menlo Park, CA) for do- blocking actions of prazosin. Arch Int Pharmacodyn Ther 237:282–287. nating the drugs. Rothman RB, Baumann MH, Savage JE, Rauser L, McBride A, Hufeisen SJ, and Roth BL (2000) Evidence for possible involvement of 5-HT(2B) receptors in the cardiac valvulopathy associated with fenfluramine and other serotonergic medi- References cations. Circulation 102:2836–2841. Abenhaim L, Moride Y, Brenot F, Rich S, Benichou J, Kurz X, Higenbottam T, Rothman RB, Vu N, Partilla JS, Roth BL, Hufeisen SJ, Compton-Toth BA, Birkes J, at ASPET Journals on June 8, 2015 Oakley C, Wouters E, Aubier M, et al. (1996) Appetite-suppressant drugs and the Young R, and Glennon RA (2003) (ϩ)-Fenfluramine and its major metabolite, risk of primary pulmonary hypertension. N Engl J Med 355:609–616. (ϩ)-norfenfluramine, are potent substrates for norepinephrine transporters. Barnes NM and Sharp T (1999) A review of central 5-HT receptors and their J Pharmacol Exp Ther 305:1191–1199. function. Neuropharmacology 38:1083–1152. Ruijtenbeek K, le Noble FA, Janssen GM, Kessels CG, Fazzi GE, Blanco CE, and De Chiang BN, Perlman LV, and Epstein FH (1969) Overweight and hypertension. Mey JG (2000) Chronic hypoxia stimulates periarterial sympathetic nerve devel- Circulation 39:403–421. opment in chicken embryo. Circulation 102:2892–2897. ␣ Stamler R, Stamler J, Riedlinger WF, Algera G, and Roberts RH (1978) Weight and Cohen ML, Schenck KW, and Kurz KD (1988) 5-HT2 receptor antagonists: 1 vs blood pressure. Findings in hypertension screening of 1 million Americans. JAm 5-HT2-receptor blocking properties in blood vessels. J Cardiovasc Pharmacol 11 (Suppl 1):S25–S29. Med Assoc 240:1607–1610. Cohen ML, Schenck KW, Mabry TE, Nelson DL, and Audia JE (1996) LY272015, a Watts SW (1998) The development of enhanced arterial serotonergic hyperrespon- siveness in mineralocorticoid hypertension. J Hypertens 16:811–822. potent, selective and orally active 5-HT2B receptor antagonist. J Ser Res 3:131– 144. Watts SW (2002) Serotonin-induced contraction in mesenteric resistance arteries: Connolly HM, Crary JL, McGoon MD, Hensrud DD, Edwards BS, Edwards WD, and signaling and changes in deoxycorticosterone acetate-salt hypertension. Hyperten- Schaff HV (1997) Valvular heart disease associated with fenfluramine- sion 39:825–829. phentermine. N Engl J Med 337:581–588. Desta B, Steed S, Ravel D, Laudignon N, Vanhoutte PM, and Boulanger CM (1998) Address correspondence to: Wei Ni, Department of Pharmacology and Acute and chronic effects of dexfenfluramine on the porcine pulmonary artery. Gen Toxicology, B445 Life Sciences Bldg., Michigan State University, East Lan- Pharmacol 30:403–410. sing, MI 48824-1317. E-mail: [email protected] Flechtner-Mors M, Ditschuneit HH, Yip I, and Adler G (1988) Blood pressure and