Matthew J.
Leonard
,
Anthony R.
Lingham
,
Julie O.
Niere
,
Neale R. C.
Jackson
,
Peter G.
McKay
and
Helmut M.
Hügel
RMIT University, 1 Bowen street, Melbourne 3001, Australia. E-mail: Matthew.Leonard742@gmail.com
First published on 6th March 2014
RU58841 is active against baldness and is commercially available. The previously reported synthesis uses phosgene, three discrete inert atmosphere steps and three steps that require flash chromatography. Our synthesis uses no phosgene, only one inert atmosphere step and does not require flash chromatography. This is achieved by stepwise construction of the hydantoin moiety around the amino group of 3-trifluoromethyl-4-cyanoaniline and ring closure to give a 2-nitropropane leaving group. On a small scale we achieved an overall yield of 33%.
They are a staple scaffold displayed in many compounds with functional end use such as cosmetics, shampoos and skin lotions. Important medicinal compounds that incorporate the hydantoin moiety include phenytoin (antiepileptic) and nilutamide (anticancer)1 (Fig. 2). The chiral hydantoin compound (+)-hydantocidin (Fig. 2) is extracted from Streptomyces hygroscopicus and is a potent herbicide.2
Given the common occurrence of hydantoins in a diverse range of compounds, alternative hydantoin syntheses are potentially useful. Recent additions include access to 5,5-disubstituted hydantoins by treatment of nitriles with an organometallic reagent (RLi or RMgX) followed by KCN/(NH4)2CO3 (ref. 1) and use of carbodiimides to prepare intermediates that form hydantoins by ring closure and rearrangement.3
In the 1970s, T. Battmann and co-workers at the French pharmaceutical company Roussel-Uclaf explored hydantoin compounds to treat a number of ailments. They prepared RU58841 (Fig. 1), which was targeted as a prostate cancer treatment,4 but biological testing with rats produced hair growth. Roussel-Uclaf prepared homologues of RU58841 (Fig. 3) including RU58642 (a hydantoin) and RU56187 (a thiohydantoin) which were both found to be active non-steroidal anti-androgens.5
While RU58642 had a stronger anti-androgen effect than RU58841, its side effects made it unusable in patients. RU58642 has since found use in biochemical research into the androgen receptor.6,7
Research interest in anti-androgen pharmaceuticals thrived in the following decades. In 1994 Battmann et al. published biological evaluation of both RU58841 and RU56187 (ref. 5) and their synthesis of RU58841.4 They prepared the hydantoin by first reacting 3-trifluoromethyl-4-cyanoaniline with phosgene to produce an isocyanate intermediate (Fig. 4). The 1994 paper4 reported that the dose required for hair regrowth was only one third of that required to give systemic side effects making RU58841 viable as a topical anti-androgen.4 Further work by De Brouweri et al. confirmed the effect of RU58841 as an anti-androgen and baldness treatment by observing hair re-growth on bald human scalp segments that had been grafted to mice.8 As well as empirical studies, the biochemistry of RU58841 and other hydantoins have been evaluated for their effect as non-steroidal testosterone inhibitors.9
As the synthesis of RU58841 by Battmann et al. (Fig. 4) uses phosgene (a toxic gas which requires a high level of expertise with handling), alternative preparations have been sought. In 2006 Hügel et al. prepared RU58841 by N-arylation of 5,5-dimethylhydantoin.10 They achieved the RU58841 hydantoin synthon in 55% yield by NaH/halogen aryl-coupling of 5,5-dimethylhydantoin with 3-trifluoromethyl-4-cyanoaniline but the yield lowered to 30% when carried out on a 10 g scale. Replacing this NaH coupling with copper acetate promoted boronic acid coupling gave a 79% yield of the same synthon.10 When this method was tried using other anilines the yields were limited to ∼50% and yields above 60% could only be achieved when the aniline being attached to the hydantoin had either nitrile or methoxy substituents.11
To improve on the RU58841 syntheses by Battmann et al. and Hugel et al., in lieu of aryl-coupling we now present an approach which uses the same start material as Battmann et al.4 (3-trifluoromethyl-4-cyanoaniline) but builds the hydantoin moiety around the aniline so as to avoid the carbonylation step which requires either the highly toxic gas phosgene or phosgene equivalents reagents which are difficult to upscale and still somewhat toxic. This provides a more flexible synthesis than the boronic acid coupling approach by Hugel et al. as it is transferable to the preparation of a wider variety of other hydantoin target compounds.
During our efforts to access the RU58841 synthon 6 without using phosgene we discovered that, unexpectedly, reaction of the tertiary bromide 1 with excess sodium nitrite in DMF at room temperature produced an 86% yield of the stable nitro compound 2. Such a reaction at a tertiary carbon appears to have no literature precedent. The generality of this synthetically useful reaction is being explored and will be reported soon.
Upon discovery that the bromine had been substituted with a nitro group, this appeared a safer and more convenient approach to access to the desired hydantoin 6 for RU58841 than the synthesis reported by Battmann et al. (Fig. 4). In keeping with our aim to use less toxic reagents that are easy to handle, this new scheme (Fig. 5) achieves the hydantoin 6 and thence RU58841 by way of ‘reacylation’ of 3, bromo-nitro substitution to furnish 5in situ and then a ring closure of 5 with 2-nitropropane acting as the leaving group. Thus we present a simple new hydantoin preparation exemplified by the synthesis of compound RU58841 in 33% overall yield.
Compound 2 is an α-nitroisobutyranilide. This class of compounds has been described before only once12 where they were prepared by a different and more complex route. Bromine-nitro substitution from exposure to nitrite ions is unexpected for compound 1, as SN2 reactions at tertiary carbons are virtually unknown,13 while SN1 reactions of nitrite normally give nitrite esters rather than nitro compounds.13 We believe neither mechanism to be operative here. This general reaction has been explored and will be discussed elsewhere.
This ring closure provides a new way to prepare N-aryl hydantoins. It also yields 2-nitropropane as a leaving group which is of interest as it has not (to the best of our knowledge) been reported as a leaving group before.
Two alternative mechanisms can be envisaged for this cyclisation both of which adhere to the Baldwin rules. The anilide nitrogen could perform an acyl substitution at the other carbonyl in a 5-exo-trig attack, with loss of 2-nitropropane anion. However the high steric hindrance about this amide carbonyl makes this seem unlikely. Alternatively, the 2-nitropropane anion could be lost first to create a much more electrophilic, much less hindered isocyanate, which would then be attacked in a 5-exo-dig geometry (Fig. 6).
This mechanism is supported by our observation that the α-nitroisobutyranilide 2 appears to be converted by injection into a conventional mass spectrometer into the corresponding isocyanate, with loss of 2-nitropropane. The dipolar/partial ionic nature of DMF may help to stabilize the amide and encourage the anilide nitrogen to proceed via the isocyanate mechanistic pathway. This represents a novel isocyanate preparation and will be further studied to be reported elsewhere.
IR spectra were measured on a Varian 1000 FTIR spectrometer as KBr disks (4000–400 cm−1) with images provided in the ESI.†1H and 13C NMR spectra were recorded on a Bruker 300 MHz spectrometer (images provided in the ESI†). Chemical shifts in 1H NMR spectra were reported in parts per million (ppm δ) relative to the TMS signal and were measured using the residual chloroform solvent signal set to 7.24 ppm. Chemical shifts in 13C NMR spectra were measured relative to the central peak of the deuterochloroform signal (δ = 77.5 ppm). Coupling constants were reported in Hz. Low-resolution mass spectra for monitoring and compound confirmation were carried out on a Varian CP-3800 GC connected to a Varian Saturn 2200 GC-MS-MS equipped with a 30 m SGE BPX5 Column and also on a Micromass Platform II electrospray using MassLynx software. High-resolution mass spectra for compounds 1–5 were carried out at Monash University on an Agilent 6220 accurate mass LC-TOF system equipped with an Agilent 1200 series HPLC column. High-resolution mass spectra for compound 6 and RU58841 were carried out at RMIT on a Waters GCT Premier HR-TOFMS equipped with an Agilent 7890 GC column. Crystallography was carried out on a Bruker APEX II DUO diffractometer.
For the workup it should be noted that 3 is highly soluble in EtOAc. Another type of hydrogenation may be preferred to perform this reaction on large scale. Our workup was as follows: reaction mixture was vacuum filtered through celite to remove iron solids and the celite flushed with hot ethanol. Water–ethanol–2-propanol was removed by rotary evaporator and the solids dissolved in ethyl acetate–water. At pH < 7, compound 3 resides in the aqueous layer. Thus, at the initial pH ∼2, the ethyl acetate fraction was discarded, removing orange impurities. Conc. NaOH was added to the aqueous portion and a dark green/black Fe0 precipitate appeared at pH ∼5. The precipitate was removed by passing the mixture through celite. On addition of more NaOH, a white solid (3) precipitated. Ethyl acetate was added to dissolve this and the pH of the aqueous layer was brought to ∼10. When a little more Fe0 precipitate drifted to the bottom of the aqueous layer, this fraction and the solid were drawn off and discarded. The ethyl acetate fraction was then washed with another volume of deionised water before being dried with MgSO4 and again filtered through celite. The celite was rinsed with hot ethyl acetate. The solvent was removed from the combined ethyl acetate fractions by rotary evaporator to give 2.72 g (10.0 mmol) of 3 as clean white fluffy crystals, mp 113–115 °C; yield 75%; Rf = 0.11 in 1:
1 hexanes–EtOAc; IR (cm−1): 3388, 3362, 3277 & 3233 (N–H), 2993, 2972, 2934, 2231 (CN), 1699 (C
O), 1613, 1514, 1490, 1422, 1326, 1178 & 1138 (C–F), 1050, 909, 888, 858, 749, 734, 673, 558; 1H NMR (300 MHz, 75 mg: 0.5 mL DMSOd6): δ 1.34 (6H, s, CH3), δ 3.53 (br, s, NH), δ 5.05 (br, s, NH), δ 8.04 (d, ArH5, J 8), δ 8.18 (dd, ArH6, J 2, J 8), δ 8.43 (d, ArH2, J 2); 13C NMR (75 MHz, 75 mg: 0.4 mL DMSOd6): δ 29.2 (s, C-3A/B), δ 56.3 (s, C-2), δ 102.4 (s, C-4′), δ 116.8 (s, CN), δ 117.8 (q, C-2′, J 5), δ 123.1 (s, C-6′), δ 123.4 (q, CF3, J 136), δ 132.6 (q, C-3′, J 17), δ 137.2 (s, C-5′), δ 144.6 (s, C-1′), δ 179.1 (s, C-1); Neg ESI HRMS: calcd for C12H12N3OF3 (M − H): 270.0854, observed: m/z 270.0864.†
For crystallization, the solid was dissolved in dichloromethane and passed through a short column of silica gel, eluting with ethyl acetate. After evaporation of the solvent, the residue was further purified by extraction with boiling n-heptane. After cooling, the n-heptane was decanted off and the solid was recrystallized from m-xylene/n-pentane. Crystallization was slow and was completed overnight at 8 °C. The resulting crystals, (mp 121–123 °C) were pure enough for use in the next step, but contained co-crystallized m-xylene. To remove this, they were heated to 60 °C for 6 h under high vacuum, giving a white powder that could be converted to solvent-free crystals for crystallography (mp 107–109 °C) by very slow evaporation from toluene.
Mp 107–109 °C or 121–123 °C when co-crystallized with m-xylene; Rf = 0.78 in 1:
1 hexanes–EtOAc or 0.07 in 4
:
1 hexanes–EtOAc; IR (cm−1): 3401, 3312 (N–H), 2992, 2932, 2229 (CN), 1722, 1664 (C
O), 1611 (C
O), 1512, 1427, 1328, 1174 & 1132 (C–F), 1049, 882, 850, 555; 1H NMR (300 MHz, 32 mg: 0.4 mL DMSOd6): δ 1.51 (6H, s, C-3/CH3), δ 1.94 (6H, s, C-6/CH3), δ 3.38 (NH-2), δ 8.09 (d, ArH5, J 8), δ 8.15 (dd, ArH6, J 2, J 8), δ 8.34 (d, ArH2, J 2), δ 10.03 (NH-1); 13C NMR (75 MHz, 32 mg: 0.4 mL DMSOd6): δ 24.8 (s, C-3A/B), δ 31.7 (s, C-5A/B), δ 58.2 (s, C-5), δ 61.7 (s, C-2), δ 102.3 (s, C-4′), δ 116.8 (s, CN), δ 117.8 (q, C-2′, J 5), δ 123.2 (s, C-6′), δ 123.5 (q, CF3, J 136), δ 132.4 (q, C-3′, J 17), δ 137.3 (s, C-5′), δ 146.0 (s, C-1′), δ 171.1 (s, C-4), δ 174.7 (s, C-1); Neg ESI HRMS: calcd for C16H17N3O2F3Br (M−): 420.0359, observed: m/z 420.0362.†
R
f = 0.50 in 1:
1 hexanes–EtOAc; neg ESI HRMS: calcd for C16H17N4O4F3Br (M − H): 385.1124, observed: m/z 385.1133.
Mp 210–212 °C; yield 83%; UV max = 256 nm (ε = 16200); Rf = 0.25 in 1:
1 hexanes–EtOAc; IR (cm−1): 3337, 3121, 2983, 2936, 2242 (CN), 1789, 1725 (C
O), 1612 (C
O), 1504, 1440, 1398, 1282, 1182, 1135, 1049, 899, 855, 808, 762, 733, 658, 559, 441; 1H NMR (300 MHz, 20 mg: 0.4 mL CD3CN): δ 2.88 (6H, s, CH3), δ 6.85 (s, NH), δ 8.62 (dd, ArH6, J 2, J 8), δ 8.70 (d, ArH5, J 8), δ 8.75 (d, ArH2, J 2); δ13C NMR (75 MHz, 20 mg: 0.4 mL CD3CN): δ 25.3 (s, CH3 × 2), δ 59.7 (s, hyd-C-5), δ 108.8 (s, C-4′), δ 116.5 (s, CN), δ 123.7 (q, CF3, J 136), δ 124.9 (q, C-2′, J 5), δ 130.4 (s, C-6′), δ 133.4 (q, C-3′, J 17), δ 137.0 (s, C-5′), δ 138.2 (s, C-1′), δ 154.4 (s, hyd-C-2), δ 177.1 (s, hyd-C-4); GC-(EI)TOF-HRMS: calcd for C13H10N3O2F3 (M): 297.0725, observed: m/z 297.0713.
The crude yield was 92–98% when the DMF and water was removed by evaporation but for a more pure product the flask contents were cooled overnight at 0 °C and the solids collected for an 80% yield of RU58841 (996 mg, 2.70 mmol). Following instructions from Battmann et al.4 the product was recrystallized from diisopropyl ether. We found RU58841 hard to dissolve even in a large volume of diisopropyl ether. The addition of a similar volume of n-heptane as an anti-solvent was necessary followed by overnight standing at 8 °C to crystallize.
Mp 71–72 °C; yield 80%; UV max = 261 nm (ε = 15100); Rf = 0.07 in 1
:
1 hexanes–EtOAc or 0.18 in 1
:
3 hexanes–EtOAc or 0.42 in EtOAc; IR (cm−1): 3392 (OH), 3133, 2944, 2876, 2234 (CN), 1774, 1719 (C
O), 1612 (C
O), 1505, 1438, 1413, 1377, 1312, 1179 & 1133 (C–F), 1051, 894, 837, 763, 675, 555; 1H NMR (300 MHz, 20 mg: 0.4 mL CDCl3): δ 1.52 (6H, s, CH3), δ 1.64 (2H, m, CH2-3), δ 1.82 (2H, m, CH2-2), δ 2.12 (1H, s, OH), δ 3.39 (2H, t, CH2-1, J 6), δ 3.68 (2H, t, CH2-4, J 6), δ 7.89 (d, ArH5, J 8), δ 7.98 (dd, ArH6, J 2, J 8), δ 8.13 (d, ArH2, J 2); δ13C NMR (75 MHz, 20 mg: 0.4 mL CDCl3): δ 23.7 (s, CH3 × 2), δ 26.4 (s, al-C-2), δ 30.0 (s, al-C-3), δ 40.4 (s, al-C-1), δ 62.2 (s, hyd-C-5), δ 62.3 (s, al-C-4), δ 108.4 (s, C-4′), δ 115.3 (s, CN), δ 122.3 (q, CF3, J 136), δ 123.3 (q, C-2′, J 5), δ 128.2 (s, C-6′), δ 133.8 (q, C-3′, J 17), δ 135.6 (s, C-5′), δ 136.8 (s, C-1′), δ 153.2 (s, hyd-C-2), δ 174.9 (s, hyd-C-4); GC-(EI)TOF-HRMS: calcd for C17H18N3O3F3 (M): 369.1300, observed: m/z 369.1294.
C, H, N: calcd for C17H18N3O3F3: 55.28, 4.91, 11.38. Found: 55.19, 4.84, 11.28.
2 (CCDC 904089) was obtained after crystallization in DMF with the addition of water. The crystal structure revealed a 1:
1 co-crystallization of 2 with DMF.
3 (CCDC 893326) confirmed the structure. The experimental procedure gave a suitable crystal directly.
4 (CCDC 892388) this single crystal proved difficult to grow and involved first co-crystallizing with m-xylene, removal of solvent under high vacuum at 60 °C, then recrystallizing in toluene to give a single crystal for diffractometry.
Footnotes |
† Electronic supplementary information (ESI) available. CCDC 892388, 893326, 894556 and 904089. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c4ra00332b |
‡ It is possible to capture 2-nitropropane as a commercial by-product during this step as we have observed it on a small scale reactive distillation. However optimal conditions for simultaneous ring closure and capture of 2-nitropropane have not yet been developed. |
This journal is © The Royal Society of Chemistry 2014 |