Total Synthesis of Monosporascone and Dihydromonosporascone †

The first total synthesis of monosporascone is presented. The five-step synthesis developed includes a silver acetylide-acid chloride coupling, domino Diels–Alder-retro-Diels–Alder reaction, and an intramole-cular Friedel–Crafts acylation, and provides the natural product in 57% yield overall. Selective reduction of monosporascone also afforded the related metabolite dihydromonosporascone.


Introduction
The naphtho [2,3-c]furandiones (isofuranonaphthoquinones) comprise a relatively small group of secondary metabolites, with a wide variety of biological activities, isolated from fungal, botanical, bacterial and insect sources.In 2005, when this class of compounds was comprehensively reviewed, 1 there were 17 natural products possessing the isofuranonaphthoquinone ring-system, and a similar number of partially reduced congeners.Since that time a single new member has been discovered: 1 (Fig. 1), which is moderately cytotoxic to a range of cancer cell lines and non-malignant human foreskin fibroblasts. 2sofuranonaphthoquinones continue to attract the attention of synthetic chemists, with recent syntheses making use of silver(II) and manganese(III)-mediated radical cyclisation of 1,4-naphthoquinone derivatives; [3][4][5] a double conjugate addition of a hydroxymethyldihydronaphthoquinone monoketal to propiolate esters; 6 oxidative skeletal rearrangement of a naphtho [1,2-b]furan-5-ol, applied to the synthesis of bhimamy-cin B (2); 7 a sequence involving consecutive [2 + 2 + 2] alkyne cyclotrimerisation, Ullman, Claisen, and ring-closing metathesis reactions; and, in the synthesis of 1, key Friedel-Crafts reactions. 8Tsunoda and co-workers recently completed an efficient total synthesis of the cytotoxic aphid pigment furanaphin (3), in a total of eight steps and 23% yield, using a key boron trifluoride-acetic acid-mediated Fries rearrangement. 9n a continuation of our interest in naphtho[2,3-c]furandiones and related compounds, 1,[10][11][12][13][14] we targeted monosporascone (4) for total synthesis (Scheme 1).Monosporascone and its dihydro derivative 5 were first isolated from the fungus Gelasinospora pseudoreticulata, and hence originally named GP-A and GP-B, respectively. 15Both compounds were shown to inhibit the pharmacotherapeutically important enzyme monoamine oxidase.Monosporascone (4) was named after the fungus it was subsequently isolated from -Monosporascus cannonballus 16 the causative agent of root rot and vine decline in commercial melon species.
Monosporascone is the only known isofuranonaphthoquinone with oxygenation only at the 5-and 7-positions, and thus presents a unique synthetic challenge.In addition, there are a number of related biologically active metabolites with the same substitution pattern that could conceivably be derived from monosporascone (Scheme 1), in some cases very succinctly.These considerations were the impetus behind the work described herein.

Results and discussion
The initial approach to monosporascone was based on our previous synthesis of the 5,8-dihydroxy analogue 12 (Scheme 2).In that instance the double Friedel-Crafts acylation of hydroquinone dimethyl ether (10) with furan-3,4-dicarbonyl chloride (11), 19 with concomitant demethylation, provided 12 cleanly and in excellent yield. 10 Application of this methodology to resorcinol dimethyl ether (13) gave complex mixtures with AlCl 3 and no reaction with SnCl 4 , with no sign of monosporascone (4) or its methyl ether 14 detected in any attempt.With AlCl 3 at least, presumably the first acylation at the doubly-acti-vated 4-position of 13 proceeds as expected to give 18 (Scheme 3).This is supported by the reaction of 13 with 3-furoyl chloride (15), which in the presence of SnCl 4 gave 16 in excellent yield (Scheme 2).The site for subsequent cyclisation in 18, however, is now strongly deactivated to electrophilic aromatic substitution by the ortho-carbonyl and further (weakly) deactivated by the two meta-methoxy groups.As a result cyclisation does not occur and side reactions ensue.With SnCl 4 as the Lewis acid, it is more difficult to explain why 15 reacts cleanly while 11 does not react at all.However, 1,4-dimethoxybenzene (10) was also unreactive with 11 under these conditions.
In any case, the failure of this initial foray required a rethink.Since it appeared that cyclisation of putative intermediate 18 was not possible, we chose to investigate the reverse approach, where the initial event in the construction of the central ring was bond formation at C5 of resorcinol dimethyl ether (or a derivative), allowing cyclisation onto the position activated by both ortho and para methoxy groups (Scheme 3 right).Although the precedent in Scheme 2 suggested that this approach should work from ketone 19, in parallel we also pursued the variant in which the furan is tethered by an activating alkyl bridge, as in 20; that is, via the naphtho[2,3-c]furan-4(9H)-one, with the view to install the carbonyl group 20 of monosporascone at a later stage.
Approach 1: via a diarylmethane (33)   Our first approaches to monosporascone (see also the next section) sought to take advantage of available dimethyl furan-3,4-dicarboxylate (21) (Scheme 4), the precursor to acid chloride 11.Thus, 21 was mono-saponified and chemoselective reduction of the carboxylic acid 22 with borane-dimethyl sulfide afforded the known primary alcohol 23, 21 which was also previously made in low yield by direct partial reduction of the diester 21 with DIBAL. 22Swern oxidation, as reported, 22 then provided the required 'semialdehyde' 24.Addition of the aryllithium 25 generated from 1-bromo-3,5-dimethoxybenzene to this aldehyde gave the expected carbinol 26 in rather disappointing yield.
Although the final step in Scheme 4 could almost certainly have been improved with further experimentation, the rather onerous synthesis of aldehyde 24 (six steps from furan and dimethyl acetylenedicarboxylate) led us to explore a more efficient route (Scheme 5).
Low temperature addition 13 of the lithium acetylide generated from ethyl propiolate (27) to 3,5-dimethoxybenzaldehyde (28) gave the expected secondary alcohol 29, which underwent a domino Diels-Alder-retro-Diels-Alder reaction 13,23 with 4-phenyloxazole (30) 24 providing the 3,4-disubstituted furan 31.Lewis or Brønsted acid-catalysed Friedel-Crafts ring closure at this juncture could, in principle, provide access to monosporascone (4) via racemic monosporascol A (6) (Scheme 1); however, we expected the benzylic alcohol to be incompatible with such conditions, and as such this was not attempted.Instead 31 was deoxygenated with trimethylsilyl iodide, 13,25 affording the diarylmethane 32 in excellent yield.Saponification then provided the carboxylic acid 33 quantitatively after acidification.Attempts to generate the corresponding acid chloride 34 with thionyl chloride led to complete degradation, even at low temperature.The reaction was successful with oxalyl chloride, however, and the acid chloride 34 was surprisingly stable, not hydrolysing during TLC, for example.
Based on the 1 H NMR spectrum of the crude product, the attempted intramolecular Friedel-Crafts acylation of 34 with AlCl 3 gave primarily what appeared to be a dialdehyde (although this was not properly identified), presumably arising from ring-opening of the furan.Surprisingly, based on precedent, 13 the use of the milder Lewis acid SnCl 4 with the isolated acid chloride 34 led to complete degradation, with no 35 detected.When this reaction was repeated with acid chloride generated in situ using PCl 5 , cyclisation was successful, but accompanied by chlorination of the benzene ring, as apparent from the mass spectrum of the product 36.Presumably the chlorinating agent is PCl 5 , or perhaps Cl 2 arising from its dis-proportionation.The regioidentity of 36 was established by a 1D NOESY experiment: irradiation of H6 led to enhancements in the signals for both methoxy groups.The results described above suggest that chlorination, either before or after ring closure, is required to stabilise the product under the reaction conditions.
Our other endeavours (carried out in parallel) had born fruit at this time so, while it is probably possible to elaborate 36 to monosporascone through judicious redox transformations, we made no attempt at this task.
Approach 2: via a diarylketone (37)   Our first venture in this area mirrored the approach outlined in Scheme 4. Addition of one equivalent of aryllithium 25 to diester 21 did give the desired ketone 37, but only in low yield and, not unexpectedly, accompanied by the corresponding tertiary alcohol arising from double addition (Scheme 6).An attempt to saponify the ester under standard conditions (NaOH, heat) lead to ring-opening of the furan, as apparent from the absence of relevant signals in the 1 H NMR spectrum of the crude product.The proclivity of isofuranonaphthoquinones to conjugate addition at the furan α-positions has been noted previously, 10 and presumably extends to other furans with electron-withdrawing groups at the β-positions.Fortunately, under milder conditions (LiOH, 0 °C), competing ring-opening was avoided, providing the carboxylic acid 38 in good yield after acidification.
An attempt was made to improve on the yield of the key carbonyl substitution reaction by use of an organocuprate intermediary 39, generated by transmetallation of aryllithium 25 with CuCN/2LiCl. 26However, reaction of one equivalent of 39 with bis-acid chloride 11, followed by hydrolytic workup, failed to provide any of the expected keto-acid 38, nor any other identifiable product.
We also investigated the analogous reaction of novel bicyclic anhydride 41, which, unlike the acid chloride 11, can only undergo mono-substitution with an organocuprate.Anhydride 41 was prepared by dehydrative cyclisation of furan-3,4-dicarboxylic acid (40). 19Whilst 41 passed elemental analysis, and the spectroscopic data supported the cyclic anhydride structure (e.g., an IR absorption at 1780 cm -1 ), we were initially thrown by the upfield 13 C NMR chemical shift of the carbonyl carbons (155.2 ppm).However, the carbonyl carbons of other strained anhydrides resonate at similar frequencies (e.g.malonic anhydride: 160.3 ppm 27 ), and the mesomeric effect of the furan oxygen would be expected to further shield the carbonyl carbons in 41.Nevertheless, to help confirm the structure, 41 was reacted with N-methylbenzylamine; indeed this gave rise to the expected amide 42.
The reaction of organocuprate 39 with anhydride 41 did provide the desired keto-acid 38, but unfortunately in no better yield than the aryllithium/ester substitution reaction (step a).Once again, the problems associated with monosubstitution of a furan-3,4-dicarboxylic acid derivative led us to consider an alternative approach in which the furan ring is constructed later in the synthesis.Specifically, we hoped to capitalise on the success of the successful cycloadditioncycloreversion described in Scheme 5 but with the even better dienophile, keto-ester 43 (Scheme 7).
Since we had 29 in hand, the first synthesis of 43 was by oxidation of the benzylic/propargylic alcohol with MnO 2 .To our surprise, the major product of this reaction was not that of oxidation, but tautomerisationthe alkene 44.The cis-configuration of the product 44 is based on comparisons of the vicinal coupling constant of similar compounds in the literature.In isolation the value for 44 is equivocal at 12 Hz, but comparable to that for the phenyl ketone 45 (11.7 Hz) 28 and very different from the trans-isomer 46 (15.5 Hz). 29Such cis-selective "redox isomerisation" has been reported previously using sodium carbonate as catalyst, 30 and presumably the slightly basic MnO 2 is responsible for this side-reaction in the current work.Indeed, when the MnO 2 was pre-washed with acid the formation of alkene 44 was diminished, but not completely avoided.The desired ynone 43 was also found to be light sensitive, decomposing under ambient conditions and complicating separation from the alkene.Fortunately a more direct and efficient synthesis 31 was achieved by the reaction of silver acetylide 48 32 with acid chloride 47 33 (Scheme 8), affording an excellent yield of 43, which was used promptly in the next step.
As expected, the Diels-Alder-retro-Diels-Alder reaction of 43 with 4-phenyloxazole 30 proceeded at considerably lower temperature than that required for the less electron deficient dienophile 29 (see Scheme 5), giving furan 49 in excellent yield (Scheme 8).
Attempts to cyclise ester 49 directly with Eaton's reagent 34 or polyphosphoric acid (PPA) 35 led to no reaction or decompo- sition at higher temperatures.Saponification of 49 provided the carboxylic acid 38, but this was also unreactive with PPA 36 and Eaton's reagent, and partially decomposed with concentrated sulfuric acid. 37Similarly, no cyclisation occurred in refluxing trifluoroacetic anhydride. 38When the acid chloride 50 generated in situ using PCl 5 was treated with SnCl 4 , 13 only a trace of monosporascone methyl ether ( 14) was isolated, the major product appearing (based on the 1 H NMR spectrum) to result from ring-opening of the furan.In direct contrast to the earlier observations with 33/34 (Scheme 5), reaction of 38 with oxalyl chloride resulted in multiple products but, with neat thionyl chloride, quantitatively provided the acid chloride 50, which was stable enough to be fully characterised.To our great delight, treatment of this isolated acid chloride 50 with five equivalents of AlCl 3 , 12 with an extended reaction period to allow selective demethylation of the peri methoxy group, then afforded monosporascone (4) in good yield.The NMR spectra of the synthetic product were virtually identical with those reported for the naturally-derived material. 15 proof of concept that monosporascone can be a synthetic precursor to the related natural products depicted in Scheme 1, 4 was subjected to reduction with zinc in acetic acid, 39 providing dihydromonosporascone (5) in modest (but unoptimised) yield.The 1 H NMR spectrum of this material also matched the data reported for the natural product. 15

Conclusions
The first total synthesis of the isofuranonaphthoquinone natural product monosporascone (4) has been achieved in five linear steps and an overall yield of 57%, via a sequence of silver acetylide acylation, cycloaddition-cycloreversion and Friedel-Crafts acylation reactions.The brevity and efficiency of this route can provide quantities of monosporascone sufficient for further biological evaluation, and also elaboration to several biologically active natural products bearing the same framework and substitution pattern, as exemplified by the synthesis of dihydromonosporascone in one extra step.
Method B. A 2.0 M solution of BuLi in hexanes (125 μL, 0.251 mmol) was added to a solution of 1-bromo-3,5-dimethoxybenzene (49 mg, 0.24 mmol) in anhydrous THF (2 mL) at −78 °C under argon.After stirring for 30 min, the solution of the aryllithium 25 was added dropwise to a suspension of anhydrous CuCN (131 mg, 1.46 mmol) and LiCl (124 mg, 2.93 mmol) in anhydrous THF (3 mL) at −78°, whereupon the solution turned yellow.The reaction mixture was warmed to −40°for 20 min to ensure complete formation of the organocuprate, whereupon the solution turned blue.The solution was cooled to −78°and a solution of 41 (35 mg, 0.25 mmol) in anhydrous THF (1 mL) was added dropwise.The reaction mixture was allowed to warm to room temperature over 6 h then quenched with 1 M HCl (2 mL), diluted with ether (30 mL) and extracted with saturated aqueous NaHCO 3 solution (3 × 20 mL) whereupon a white precipitate formed.The precipitate was filtered and the aqueous filtrate was carefully acidified (1 M HCl, 0 °C) then extracted with EtOAc (4 × 25 mL).The extract was dried and evaporated to give 38 as a yellow glassy solid (13 mg, 20%), spectroscopically identical with the material described above.