FORSCHUNG 141 (,HIMIA 47 (1993) Nr." (Mai) Chim;a 47 (/993) 141-/47 © Neue SchIVeizerische Chemische Gesellscllllft /SSN 0009-4293 Benzodiazepines, the Story of the Antagonist Flumazenil and of the Partial Agonist Bretazenil Walter Hunkeler: Born 1938 in Altishofen. Walter Hunkeler* Lucerne. 1955-58 apprenticeship as laborato- ry assistentatS;egfr;edAG inZofingen. 1960- 63 study of chemistry at the Institute of Tech- nology of the State of Berne. Burgdorf. and Abstract. The story offlumazenil and bretazenil is a typical example of a serendipitious 1965-69 at the University of Berne. Doctoral drug discovery. In 1979 benzodiazepine antagonists were unknown. No one was thesis under the direction of Prof. H. Sclwlteg- looking for them, but they were discovered nevertheless. Ro 15-1788 was selected for ger. Postdoctoral research at the ETH-Zurich with Prof. A. Eschenmoser. Since 1974 at F. clinical trial. Today 'this compound has the generic name f1umazenil, the trade name Hoffmann-La Roche AG in Basel. Anexate, and is registered in 70 countries. Itis the first specific benzodiazepine receptor antagonist for clinical use. Minor structural modifications of benzodiazepine antago- nists led to partial agonists, compounds with powerful anxiolytic and anticonvulsant properties. Compared with full agonists they have markedly reduced sedative and alcohol-potentiating effects. Bretazenil is in clinical trial; it exhibits promising anxio- other stems from the triazolobenzo-di- lytic activity especially in patients with panic attacks. In open clinical studies it has azepines, which were synthesized and in- shown remarkable antipsychotic activity. Structure-activity relationships and spin-offs vestigated at about the same time at Take- are discussed. da Chemical Industries in Japan and at the Upjohn Company in the United States [4][5]. These are more potent than previ- 1. The Chemical Pedigree of know this, when they started synthesizing ous benzodiazepines. FlumazeniJ (20) simpler analogues of anthramycin, since Alprazolam (4), to take one example, the existence of the BZR had not yet been is more potent in terms of anticonvulsant Chlordiazepoxide (1, Librium) is the established. A numberoftheir compounds and anxiolytic effects. The question that first anxiolytic with the structure of a showed some anticonvulsant and anti-con- now arose was whether annelation by a benzodiazepine. It was synthesized in 1955 flict activity in animal tests. Thus, 5 was tri azo I0 ring would also work in a different and submitted for testing by Leo H. Stern- selected as a candidate for anxiolytic eval- series, e.g., with the Lederle anthramycin bach in 1957 at F. Hofjinann-La Roche, uation in a limited clinical trial in man, but derivative? Would this result in a more Nutley, USA [1]. The time elapsing be- it failed to produce sufficient anxiolytic potent substance? Compound 6 was syn- tween the first pharmacological testing activity. Their results were published [3]. thesized to answer these questions: and introduction on the market was only This is one branch of the pedigree; the In December 1978, the two imida- two and a half years. In 1963, diazepam (2, Valium), the best known and best investi- gated benzodiazepine reached the market place. At this time, when benzodiazepines \ 0 YN'N were a major focus of attention in Nutley, ,~i:-" o~o one group there, Leimgruber, Batcho, and "y Schenker was working on antibiotics. They ~ /, ~ /, ~y;) CIM;0 " isolated anthramycin (3) and elucidated 0 0 its structure [2]. To their great surprise, Chlordiazepoxide Diazepam they discovered that it, too, was a 1,4- benzodiazepine. 0 0 Anthramycin (3) was later found to H3C N ,... , (' 0••.•.•••.••.• have no affinity for the benzodiazepine y " A receptor (BZR). Chemists working at Led- ~ I ~ rt"{) N erie at that time were not in a position to "1 CC;H CIM; 0 "'~~, : I ".6 ~ 0 0 CONH,o ~ ::~" 0 ~ *Correspondence: Dr. W. Hunkeler F. Hoffmann-La Roche. Ltd. Anthramycin Alprazolam PRPN ]5/235 CH-4002 Basel FORSCHUNG 142 CHIW,\ ~71IQQJ) Nr. SIMail zoesters 7 and 8 were synthesized and rine, had with 6.4 nM an even greater Table I. Billdillg Activities OfCol/1f1ol/luls submitted for testing together with the affinity for the BZR than diazepam. As a SYllthesi~ed 0/1 the Way to Fluma-;,ellil (20) triazolo compound 6. All turned out to be consequence of this preliminary result, inactive in the usual tests for benzodi- compounds without the CI-atom in the omp und Inhibition f 'II'lhu/cpam azepmes. phenyl ring were synthesized. Thus, the blllding i/ll'ifrrJ, IC'Ill' in nM Luckily for us, the benzodiazepine re- first three compounds in this new series ceptor (BZR) had been discovered the already indicated that structure-activity 9 > 1000 year before [6][7]. Usingthe3H-diazepam relationship were different from those de- binding assay it became possible to deter- termined for the older benzodiazepines, 10 > 1000 mine the affinity of a compound for the such as 2. The fact that 7 and 8, with an BZR in vitro. The three compounds were affinity for the BZR, failed to show BZ- 11 > IO{){) tested in this assay, and the triazolo deriv- like activity was not very discouraging ati ve 6 that we had favored proved to have initially, because for metabolic or phar- 12 ;> 1000 no affinity for the BZR. However, the macokinetic reasons in vitro activity is imidazo-ester 8, with a CI-atom in the often not accompanied by in vivo activity. 13 u same position as in diazepam, had an IC50 Therefore, synthetic work in the series of of 62 nM. Compound 7, without the chlo- imidazobenzodiazepinones continued. 14 . .0 0 0 0 Rapid hydrolysisoftheester7, the lead (' /'.... compound, to the inactive acid 9 was a N!J OH ~'~ Ii ° possible explanation for inactivity of 7 in vivo. The amide 10 was also inactive ill 0;,6~ I: <J 0;;,," vitro. This was surprising since the classi- 0;;" cal benzodiazepine series contains a 0 0 35 ° number of highly potent imidazo amides. The lead compound 7 is chiral and has the 9 10 11 absolute configuration (S). The (R)-enan- tiomer 11 is inactive in the BZR binding test. Compound 12, containing the six- membered ring, was inactive but the result 0 0 0 with the N-methyl compound 14 showed that this ring is not required for increased rro~ affinity for the BZR. Later on, 13 with the four-membered ring was synthesized; it ~io~I h NJ ~-rif' 0;" has a high affinity for the BZR (TaMe I). 0 0 0 A considerable number of compounds synthesized in the first hal f of 1979 had 12 13 14 very high affinity for the BZR. In spite of this, the compounds displayed no or only borderline activity in I'il'o. The possibility Scheme J that these compounds might be antago- nists was discussed. Our suspicions were CI,CCH(OH), H NH H,NOH. HCI H AcOH, H,O, increased, when further testing revealed I~ Na,SO., H,O DN~/OH"-': ""N H,SO, H,SO, 0 D ' 0 I that the lead compound 7 also bound to the F ~ I ~ 95% F 89% F~ ~ 80% receptor in vivo, demonstrating that the 0 16 15 17 target, the BZR in the brain, was reached. In early September 1979 a rat was sedated o 1. POCI" CHCI, with diazepam (2), after which it received p-CH,·C.H,N(CH,), H 0 an injection of7. Even before the injection 2 2. CNCH2C02Et CH,NHCH,C0DMF, AcOH H BC;N)I "-': KOtBu,DMF was finished the rat got up and walked off. ~ New tests were immediately devised. In 75% F N\ 78% ,~r~ each situation the effects of diazepam o o Ro 15·1788 and other benzodiazepines were reversed 18 19 flumazenil 20 ANEXATE by 7. ,. NaH, CIPO(OEt), 2 CH2(CO,EI), ~ HC(OC,Hslo 2. Optimization: Flumazenil (20) (Scheme 1) HY'COOEt 1. NaNO" AcOH ~j-COOEt The lead compound 7 was not a pure N =) NaOEt,E!OH 2.H"Pd ()"-) antagonist. Some minor agonistic activity e.g. FBC; N was found, in aconflict test in rats. We \ F~N\ o o wanted a pure antagonist for therapeutic applications and as a scientific tool. In 21 22 23 October 1979 Ro /5-1788, f1umazenil (20) FORSCHUNG 143 CHIMIA ~71Iqq.l) Nr. ~ (Mail was synthesized, and, in January 1980, selected for clinical trial. 2.1. The Synthesis of Flumazenil (20) Starting with 4-fluoroaniline (15) the isatin 17 is synthesized via the Sandmeyer synthesis; isatin is then oxidized with per- acetic acid to the isatoic anhydride 18. Reaction with sarcosine in DMF leads to 24 25 the benzodiazepine-2,5-dione 19. This is converted to the iminochloride by reac- o tion with POCI3. In the key step the imida- zoester is built up by reaction with depro- rfll 0.•...•......... tonated ethyl isocyanoacetate [8]. Since ethyl isocyanoacetate is not very stable, an CIn;N~I :::,... N alternative synthesis based on the synthe- \ sis of midazolam was developed for large o scale-production. Tnthis synthesis diethyl- 26 27 malonate is used. The diester 21 is then transformed to the monoester 22 hy deethoxycarbonylation. Nitrosation and catalytic reduction lead to the amino com- Table 2. Influence of Substitution of the Phenyl Ring 011 the Pharmacological Acti\'ity pound 23. The final carbon atom is intro- Compounu Inhlbillon or H-UI<llCpam 1'1. pl:l1l duced by reaction with the orthoester. mal! ni,m 01 dialer,lIn 1- bmdin in ritru. I Sfl in 0\1 enelelralOlc. mi c. p.o .. r.DjIJ in m~g In 1981 our paper on the first selective antagonist of benzodiazepines [9) caused enormous interest in flumazenil in the 2.7 OA scientific community.