Total Syntheses of Chelidonine and Norchelidonine Via an Enamide-Benzyne-[2 + 2] Cycloaddition Cascade.

Abstract

Total syntheses of chelidonine and norchelidonine featuring an enamide-benzyne-[2 + 2] cycloaddition initiated cascade is described. The cascade includes a pericyclic ring-opening and intramolecular Diels-Alder reaction.

Our lab recently unveiled a de novo cascade of pericyclic ring-openings of amido-benzocyclobutanes and N-tethered intramolecular Diels-Alder [IMDA] cycloadditions initiated through an enamide-benzyne-[2 + 2] cycloaddition [1→2→3→4 in Scheme 1].¹ This tandem cascade possesses the unique feature of not only linking together the prevailing benzyne chemistry^2-4 with enamides that have become a highly versatile and accessible functional group,^5-7 but also accentuating the less developed thermally driven [2 + 2] cycloaddition reaction manifold,^8-10 while exploiting the powerful Oppolzer-type N-tethered IMDA strategy.^11-15 Accordingly, an application of this cascade in the synthesis of benzophenanthridine alkaloids [see 5 in the box] was pursued. In particular, we have been focusing on (+)-chelidonine 6 and (+)-norchelidonine 7, which despite being known for well over a century,^16-21 remain an excellent proving ground for showcasing synthetic methods.^22,23 Our strategy would be based on the above cascade using benzyne precursor 9 and enamide 10. We report here total syntheses of (±)- Our chelidonine and (±)-norchelidonine.

Scheme 1.

An Enamide-Benzyne-[2 + 2] Cascade to Chelidonine.

Synthesis of enamide 10²⁴ could be expeditiously achieved as shown in Scheme 2 from the commercially available aldehyde 11, featuring Sonogashira coupling,²⁵ reductive amination,²⁶ and Cu(I)-catalyzed amidation of vinyl bromide.²⁷ For comparisons in the later benzyne-[2 + 2] cycloaddition, we also desilylated 10 to access enamide 14 with an unsubstituted alkyne. The benzyne precursor, silylaryl triflate 9, was prepared from sesamol 15 in 2 steps.²⁸

Scheme 2.

Synthesis of Enamide 10.

We proceeded in the key enamide-benzyne-[2 + 2] cycloaddition with some trepidation because we had failed related cycloadditions using enamide tethered to an alkyne motif such as 10 during our method development.¹ Following Kobayashi's fluoride-based conditions²⁹ for in situ generation of benzyne 16 from silylaryl triflate 9, we initially explored TBAT [tetra-n-butyl-ammonium triphenyl-difluorosilicate].

As shown in Table 1, after screening through a few solvents [entries 1-3], 1,4-dioxane proves to be the optimal solvent, leading to the desired amido-benzocyclobutane 17 in 80% yield at rt, albeit it took 96 h [entry 3]. It is noteworthy that the benzyne-[2 + 2] cycloaddition is completely chemoselective in favor of the enamide motif as long as the alkyne is substituted with TIPS. On the other hand, when using enamide 14 with terminally unsubstituted alkyne, the cycloaddition was not clean. The crude NMR suggests that the acetylene had likely attacked the benzyne.

Table 1.

The Enamide-Benzyne-[2 + 2] Cycloaddition.

entry	9: equiv	FΘ source [equiv]	solvent	time [h]	yield [%]a
1	3.0	TBATb [5.0]	CH2Cl2	48	trace
2	3.0	TBAT [5.0]	THF	72	43
3	3.0	TBAT [5.0]	1,4-Dioxane	96	80
4	3.0	KF [5.0]	THFc	96	11
5	2.0	CsF [4.0]	CH3CN	15	79
6	3.0	CsF [5.0]	1,4-Dioxane	48	trace

^aIsolated yields.

^bTBAT: Tetra-n-butyl-ammonium triphenyl-difluoro-silicate.

^c18-Crown-6 [6.0 equiv] was used.

However, the reaction time was clearly too long using TBAT-dioxane conditions. Thus, inorganic fluorides [entries 4-6] were also examined, and CsF in CH₃CN at rt turned out to be equally effective in providing 17, and more importantly, the reaction time was reduced to 15 h [entry 5]. It should be noted here that to avoid the competing removal of the TIPS group, we focused on low temperature reaction conditions instead of 110 °C adopted in our earlier communication.¹ As a result, an intriguing observation was made. That is these reactions appear to be much faster when using CH₃CN as solvent than 1,4-dioxane or THF.^10b

Armed with the desired amido-benzocyclobutane 17, we removed the TIPS group using TBAF as shown in Scheme 3. Subsequent heating of 18 in xylene at 120 °C afforded the tetracyclic benzophenanthridine 20 through a sequence of [2π+2σ]-pericyclic ring-opening and intramolecular Diels-Alder cycloaddition. An X-ray structure of 20 was attained to ascertain the integrity of this sequence [Figure 1], although 20 constitutes a formal synthesis of (±)-chelidonine, as it matched spectroscopically with Oppolzer’ intermediate.^22a-c

Scheme 3.

The Ring-Opening and [4 + 2] Cycloaddition Cascade.

Figure 1.

X-Ray Structure of Tetracycle 20.

Tetracycle 20 was found not to be very stable, as it was slowly converting to a new set of chemical resonances in CDCl₃ at rt. The new compound was ultimately assigned as the aromatization product 21, which could be envisioned from N-acyl iminium ion 22. It was later confirmed that 21 was a minor but persistent byproduct during the heating in xylene, and that an increasing amount of 21 was present when prolonged heating took place.

Most critically, we succeeded in a tandem process or one-pot formation of tetracycle 20. As shown in Scheme 4, treatment of enamide 10 with CsF at rt in CH₃CN followed by heating at 80 °C and subsequently at 120 °C using xylene as the solvent led to 20 in 65% yield overall. Removal of CH₃CN appeared to be critical prior to the addition of xylene and further heating at elevated temperature for the Diels-Alder cycloaddition. When directly heating the mixture at 120 °C without removing CH₃CN, an indistinguishable mixture was observed by crude proton NMR. Among the mixture, the ring-opened product such as 24 is likely present, presumably derived from the ring-opened zwitterionic intermediate 23, or directly from tetracycle 20 via a based-promoted elimination.

Scheme 4.

A Tandem Enamide-Benzyne-[2 + 2]–[4 + 2].

On the other hand, further purifying the crude desilylated benzocyclobutane 18 even through a short bed silica gel filtration proved to be unnecessary and provided no better yield. It is noteworthy that the overall process constitutes a four-bond and two-ring formation, thereby further accentuating the synthetic imminence of an enamide-benzyne-[2 + 2] cycloaddition manifold.

To complete our total syntheses, tetracycle 20 was subjected to BH₃-hydroboration-oxidation conditions to give a 1:1 separable isomeric mixture of alcohol 25-trans and 25-cis [Scheme 5]. Relative stereochemistry of the three contiguous stereocenters in 25-cis was unambiguously assigned through X-ray structure of 25'-cis, which has the Cbz protecting group removed and is essentially the C11-epimer of norchelidonine [Figure 2]. Unfortunately, we could not improve the diastereomeric ratio via other boranes such as 9-BBN, c-hex₂BH, and Et₂BH. Nevertheless, (±)-chelidonine 6 could be rapidly and efficiently synthesized from 25-cis through DMP-oxidation and alane reduction of both ketone and Cbz-urethane motifs. This would complete a facile 7-step total synthesis effort. Lastly, (±)-norchelidonine 7 was also completed through a sequence of DMP-oxidation, NaBH₄ reduction, and hydrogenation. Both synthetic samples spectroscopically matched the reported literature values.²²

Scheme 5.

Completion of the Total Syntheses.

Figure 2.

X-Ray Structure of 25′-cis [C11-epinorchelidonine].

We have described here total syntheses (±)-chelidonine [7 steps; 8.5% overall yield] and (±)-norchelidonine [8 steps, 8.7% overall yield], featuring an enamide-benzyne-[2 + 2] cycloaddition in a quadruple tandem cascade that also includes pericyclic ring-opening and intramolecular Diels-Alder cycloaddition. While Oppolzer's 1971 seminal total synthesis^22a inspired our efforts, the current total syntheses underscore both the power of enamides as synthetic building blocks, and the significance of benzyne chemistry, in particular, the benzyne-[2 + 2] cycloaddition in the efficient assembly of complex targets.