Total Synthesis of Daphnepapytone A

Abstract

The total synthesis of the sesquiterpenoid daphnepapytone A starting from (R)-carvone is reported. The carbocyclic framework of daphnepapytone A was formed by a one-pot Pauson-Khand/desilylative oxidation reaction and a [2+2] photocycloaddition of synthetic oleodaphnone. During the synthesis, it was discovered that wavelength-dependent irradiation of oleodaphnone resulted in intramolecular photometathesis to form a dicyclopentadienone. Late-stage oxidation and a highly stereo- and regioselective reduction completed the synthesis

Graphical Abstract

Introduction

Guaianes are a diverse class of sesquiterpenes, including more than 300 members and possessing a wide range of biological activities.^1–4 Structurally, guaianes and guaiane-derived sesquiterpenoids share a carbon skeleton consisting of a saturated azulene core appended with two methyl groups and an isopropyl group (Figure 1a). Biosynthetic oxidative modifications give rise to the broad structural diversity and anti-inflammatory (1),⁵ latent HIV reversal (2),⁶ antimicrobial (3),⁷ anti-cancer (4),⁸ and other biological properties that make guaiane-derived terpenoids valuable synthetic targets.^9,10

Figure 1.

Guaianes and retrosynthesis of daphnepapytone A.

In 2022, Dai, Zhao, and coworkers isolated the guaiane-derived sesquiterpenoid daphnepapytone A (5) from the shrub Daphne papyracea and reported its inhibitory activity of the carbohydrate hydrolase enzyme α-glucosidase (IC₅₀ = 159.0 ± 2.1 μM).¹¹ Daphnepapytone A has an unusual tetracyclic structure with five contiguous stereocenters and a cyclobutane-bridged guaiane skeleton. Biosynthetically, daphnepapytone A is proposed to originate from a [2+2] cycloaddition and allylic oxidation of oleodaphnone (2), another guaiane sesquiterpenoid isolated from the Daphne genus.¹²

As an indication of the growing interest in daphnepapytone A, three groups (Li/She, Nay, Stoltz) independently reported syntheses of 5 during the course of our studies.^13–15 Notably, all three groups targeted 2 at some stage of their synthetic efforts in a biomimetic manner, though the Stoltz group successfully pursued an alternative route to the natural product. Our strategy similarly aimed to test the biosynthetic hypothesis by targeting 2 as an elaborated intermediate (Figure 1b). Its cyclopentenone moiety could be formed by an allenyl Pauson-Khand reaction (PKR)^16–19 and coupled with a one-pot desilylative oxidation reaction from silyl-protected allenol 6. Formation of the allenol moiety could be accomplished by Grignard addition of an allene precursor to aldehyde 7, synthesized in two steps from (R)-carvone (8) following a reported procedure.²⁰

Results and Discussion

Our initial synthetic approach toward daphnepapytone A (5) centered on a photochemical transformation inspired by the rearrangement of α-santonin to isophotosantonic lactone.^21,22 Using this approach, 9-deoxo-oleodaphnone (9) was synthesized in four steps from (+)-dihydrocarvone using a Robinson annulation, DDQ dehydrogenation, the santonin phototransformation, and dehydration (see Scheme S1 in Supporting Information for more details). Unfortunately, attempts to oxidize the allylic methylene group to the corresponding ketone (oleodaphnone, 2) with CrO₃ or SeO₂ resulted in oxidative cleavage of the seven-membered ring instead (Scheme 1). This limitation ultimately led us to pursue a Pauson–Khand-based strategy that would introduce the C9 ketone functionality earlier in the synthesis.

Scheme 1.

Oxidative cleavage of 9 during the synthesis of oleodaphnone (2).

Like the previously reported approaches to daphnepapytone A, synthesis of our Pauson-Khand substrate (6) began with the epoxidation of (R)-carvone under basic conditions, using H₂O₂ and NaOH, to deliver α,β-epoxy ketone 11 in quantitative yields (Scheme 2).^13–15 Eschenmoser fragmentation of 11 was accomplished by treating the α,β-epoxy ketone with p-toluenesulfonyl hydrazide to generate the hydrazone in situ, which fragments to aldehyde 7 under acidic conditions in a 47% yield.^{20, 23, 24} Addition of a Grignard reagent generated from 1-bromo-2-butyne to aldehyde 7 generated an inseparable 3:1 mixture of allenol 12 and alkynol byproduct 13 in an 84% combined yield. It is worth noting that the organoindium reagent generated from 1-bromo-2-butyne and indium affords the desired allenol 12 as the sole product in an increased yield (89–95% yield), as demonstrated by the Li/She, Nay, and Stoltz groups.^13–15

Scheme 2.

Synthesis of Oleodaphnone

One of our earlier synthetic routes investigated the possibility of doing a PKR on the ketone form (18) of allenol 12, avoiding the need for a protecting group (Scheme 3). In this route, racemic, vinylogous acyl triflate 17 was synthesized in three steps from (R)-carvone following a reported procedure (see Scheme S2 in Supporting Information for more details).²⁵ Ketone 18 was generated in 50% yield by Dudley-type fragmentation of 17 using the Grignard reagent derived from 1-bromo-2-butyne.²⁶ While we expected cyclization of ketone 18 using common PKR conditions to yield racemic oleodaphnone ((±)-2), we instead isolated furan 19. Based on similar examples of metal-catalyzed cycloisomerizations of allenones,^27,28 we proposed that coordination of the distal bond of the allene to the metal catalyst initiated 5-endo-trig cyclization by the ketone, resulting in 19. Thus, protection or reduction of the ketone would inhibit the cycloisomerization and enable the PKR.

Scheme 3.

Dudley-type fragmentation of 17 to generate ketone 18 and an unexpected cycloisomerization of 18 under PKR conditions.

These preliminary investigations showed that oxidation of allenol 12 to a ketone followed by the allene-alkyne PKR would be unsuccessful, so protection of the allenol was necessary. Thus, the 3:1 mixture of allenol 12 and alkynol 13 was treated with TMSOTf to generate the TMS ethers 6 and 14 in 95% combined yield (Scheme 2). The desired TMS-protected allenol 6 could be separated as a 1:1 mixture of diastereomers for use in the PKR.

Additionally, the TBS-protected allenol 20 was synthesized in a similar manner to 6, using TBSOTf instead, for optimization of the allene-alkyne PKR reaction (Table 1).^16–19 An initial screening of three catalysts showed [RhCl(CO)₂]₂ to give the highest yield (entries 1–3) and was used for all further optimization. Yields increased with higher catalyst loadings (entries 3–5); however, only a 4% increase in yield was observed from 10 mol% to 15 mol%. Thus, 10 mol% was selected as the optimal catalyst loading. Higher temperatures (144 °C) with o-xylene as opposed to lower temperatures (110 °C) with toluene also showed a significant increase in yield (entries 4 and 6). Surprisingly, the TMS ether 6 gave higher yields than the corresponding TBS ether 20 using the optimized conditions (entries 6 and 7) and was used for the remainder of the synthesis.

Table 1.

Optimization of the PKR.

entry	cat.a	mol%	solvent	temp (°C)	Rb	yield (%)
1	A	120	CH2Cl2	rt	TBS	8
2	B	5	toluene	110	TBS	<5
3	C	5	toluene	110	TBS	41
4	C	10	toluene	110	TBS	51
5	C	15	toluene	110	TBS	55
6	C	10	o-xylene	144	TBS	62
7	C	10	o-xylene	144	TMS	74

^aCatalyst A: Co₂(CO)₈, B: [RhCl(CO)dppp]₂, C: [RhCl(CO)₂]₂

^bAll reactions carried out using 0.3 mmol of 6 and 20.

Due to the lability of the TMS protecting group, a one-pot PKR/desilylative oxidation reaction was employed to avoid problems with partial deprotection of the TMS group during purification of 15 after the PKR (Scheme 2). Initial experiments showed that acetic acid was sufficiently acidic to deprotect the TMS group. In light of this, it was envisioned that addition of AcOH to TMS ether 15 would generate the corresponding alcohol 16, which would be subsequently oxidized to 2 by DMP, producing AcOH as a byproduct to continue the cycle. In practice, trace amounts of acetic acid within DMP allowed the desilylative oxidation of 15 to reach completion in 90% yield after 18 h without adding any additional acetic acid. This was surprising, as previously reported desilylative oxidations of TMS-protected alcohols required additives or did not reach completion with secondary TMS ethers.^29–32 Regardless, the addition of 50 mol% of acetic acid reduced the reaction time to 5 h with a comparable yield of 92%. To further increase pot-efficiency, the solvent was removed after completing the PKR and the AcOH/DMP deprotection/oxidation cycle was carried out on the crude mixture to afford 2 in 45% yield from allene 6.

With oleodaphnone (2) in hand, the key [2+2] photocycloaddition was investigated. Initial attempts to produce cyclobutane adduct 22 by irradiating 2 with 300 nm light in 1,4-dioxane delivered a 1:5 mixture of 22 and dicyclopentadienone 23 (Figure 2a). The presence of 23 suggested a [2+2] cycloreversion of 22 was occurring,³³ but attempts to prevent this degradation of 22 by carrying out the reaction at 0 °C and using other solvents (dichloromethane and acetone) yielded similar results. However, switching to a less energetic wavelength, 370 nm, resulted in 22 as the sole product in 67% yield.

Figure 2.

Investigations into the [2+2] cycloaddition and photometathesis of 2.

To confirm that 23 was forming from 22 and that the [2+2] cycloreversion was wavelength-dependent, separate solutions of 22 in 1,4-dioxane were irradiated with either 300 nm or 370 nm light (Figure 2b). Dicyclopentadienone 23 was obtained in 74% yield within 6 h at 300 nm and was not observed within the 24 h reaction period at 370 nm.

To gain insight into the unexpected wavelength-dependent product outcome during irradiation of oleodaphnone (2), the light absorption characteristics of 2 and cyclobutane 22 were examined (Figure 2c). Most notably, oleodaphnone strongly absorbs at 300 nm with an extinction coefficient (ε) of 12,600 M⁻¹ cm⁻¹ and weakly absorbs at 374 nm (ε = 201 M⁻¹ cm⁻¹). Consistent with the experimental results, cyclobutane 22 displays no notable absorption features at or above 370 nm (ε < 30 M⁻¹ cm⁻¹) and a gradual increase in absorption with a decrease in wavelength, peaking at 261 nm (ε = 1,200 M⁻¹ cm⁻¹).

Time-dependent density-functional theory (TD-DFT) analysis of 2 at the B3LYP-D4/def2-TZVP-SMD(1,4-dioxane) level^34–42 corroborated the experimental light absorption characteristics, displaying a vertical excitation from the ground state (S₀) to the first excited singlet state (S₁) at 3.35 eV (370 nm) with a weak oscillator strength (ƒ = 0.0009), corresponding to an n → π* transition (Table S2 and Figure S4). The analogous S₀ → S₁ transition for 22 was calculated at 3.91 eV (317 nm, ƒ = 0.0003), also corresponding to an n → π* transition (Table S3 and Figure S5). The major experimental peaks for 2 (300 nm) and 22 (261 nm) were also captured by the calculations with π → π* transitions at 4.15 eV (298 nm, ƒ = 0.748) and 4.70 eV (264 nm, ƒ = 0.036), respectively.

Based on the absorption characteristics of 2 and 22, the TD-DFT calculations, and similarity to previously-reported examples of photometathesis,^{43, 44} the wavelength-dependent formation of 22 and 23 is explained as follows. Irradiation of 2 at either 370 nm or 300 nm causes excitation to the S₁ (n,π*) state or higher π,π* singlet state, respectively, followed by intersystem crossing from S₁ to triplet diradical (π,π*)-2 (Figure 2d).⁴⁵ The [2+2] cycloaddition, as described by the Li/She group,¹⁵ proceeds to give cyclobutane intermediate 22. Further irradiation at 370 nm causes no transformation of 22 due to insufficient energy to cause an S₀→S₁ transition at 3.91 eV (317 nm), as revealed in the TD-DFT calculations and light absorption spectrum. However, irradiation at 300 nm causes the n → π* transition, again, followed by intersystem crossing to triplet diradical (π,π*)-22. The proximity of the β-radical to the cyclobutane ring results in homolysis of the C1-C10 bond to generate 24, followed by fragmentation of the six-membered ring to form 23. It is worth noting that the fragmentation of 22 was not reported by the Nay or Li/She groups and may explain the limited success the Stoltz group reported with the [2+2] cycloaddition of 2 to cyclobutane 22.^13–15

With a clear path to 22, introduction of the C6 hydroxyl group was then investigated. An indirect approach of C6 bromination followed by nucleophilic displacement with an oxygen nucleophile was explored to install the C6 hydroxyl group, a strategy employed by the Baran group in their synthesis of decinnamoyltaxinine E and taxabaccatin III.⁴⁶ Allylic bromination of enone 22 with NBS and (BzO)₂ generated 25, which was used immediately in the next step without purification (Scheme 4). Attempted nucleophilic displacement of the bromide with NaOTMS resulted in intramolecular displacement instead, generating the cyclopropane-containing compound 26 in 80% yield. Interestingly, the Stoltz group also reported the formation of 26 by treating the mesylate of 6-epi-daphnepapytone A (27) with NaOH.¹³ Likewise, the formation of 26 is proposed to occur from backside enolate attack on bromide 21, suggesting that the initial bromination occurs from the less hindered convex face of 22. While approach from the concave face was disfavored for the initial bromination, we hoped the presence of the bromide would facilitate an S_N2 reaction from the concave face with inversion of the configuration at C6. To prevent deprotonation of C8 with NaOTMS, TMSOH was used instead, followed by deprotection of the TMS ether to reveal the hydroxyl group. Unfortunately, the C6-epimer of daphnepapytone A (27) was produced in 47% yield. To determine if 26 was an intermediate in the formation of 6-epi-daphnepapytone A (27), via acid-promoted opening of the cyclopropane ring with concomitant addition of water to C6, cyclopropane intermediate 26 was subjected to similar conditions used to convert 25 to 27 (AgOTf and aqueous HCl). However, 27 was not observed, suggesting it is formed through an S_N1-type reaction of bromide 25 and TMSOH facilitated by AgOTf.⁴⁷

Scheme 4.

Unsuccessful attempts at C6 hydroxylation to 5.

To leverage the concavity of the molecule in installing the C6 hydroxyl group, an alternative strategy of C6 ketone reduction from the convex face was attempted, similar to the recently reported approaches towards daphnepapytone A (5).^13–15 As reported by the Stoltz group, allylic oxidation of enone 22 with Mn(OAc)₃•2H₂O and TBHP generated 6-oxo-daphnepapytone A (28) in 27% yield (Scheme 5).^13,48 The low yield was found to be a result of incomplete conversion of the intermediate 6-(tert-butylperoxy)daphnepapytone A (S11) and 27 to 28.

Scheme 5.

Completion of the synthesis of daphnepapytone A.

For the final step of the synthesis, we anticipated that reduction of triketone 28 would proceed regioselectivity at C6 due to the kinetic favorability of reducing a six-membered cyclic ketone (C6) over five-membered cyclic ketones (C3 and C9).⁴⁹ Indeed, treatment of triketone 28 with NaBH₄ at 0 °C selectively reduced the ketone at C6 to deliver daphnepapytone A (5) in 64% yield, consistent with the yields reported by the Li/She, Nay, and Stoltz groups (43–75%).^13–15 Similar to the stereoselectivity observed in the bromination of diketone 22, hydride approach was rationalized to occur from the less hindered convex face of 28 to generate the C6 alcohol with the correct configuration. X-ray crystallographic analysis confirmed the absolute configuration of 5 (CCDC 2453138).

The photoconversion of 22 to 23 raised the question of whether daphnepapytone A (5) might undergo a similar degradation upon exposure to ultraviolet light. However, irradiation of 5 with 300 nm and 370 nm light in 1,4-dioxane resulted in complete decomposition with no discernable products. The UV-Vis spectrum of 5 displayed peak absorption around 248 nm (Figure S3), consistent with the excitation calculated at 5.08 eV (244 nm) using TD-DFT (Table S4). Like 22, the calculations suggest that 300 nm light should supply enough energy to cause an S₀ → S₁ excitation (3.84 eV, 323 nm, ƒ = 0.0006). However, the presence of the C6 hydroxyl group, absent in 22, likely leads to alternative photodegradation pathways after the initial excitation rather than cyclobutane opening.

Conclusions

In conclusion, daphnepapytone A (5) was synthesized from (R)-carvone using an allenic Pauson-Khand reaction to construct the guaiane skeleton and a [2+2] photocycloaddition to generate the tetracyclic core of the natural product. Optimization of the PKR conditions and substrate, as well as a pot-economical desilylative oxidation reaction, were critical to generating ample amounts of oleodaphnone (2). Additionally, a highly stereo- and regioselective reduction, as featured in the elegant approaches by the Li/She, Nay, and Stoltz groups, was accomplished by taking advantage of the natural concavity of 28 and its inherent kinetic differences of ketone reduction. Lastly, the unexpected transformations of 22 highlight the synthetic potential of the unusual tetracyclic core of daphnepapytone A. In particular, it was shown that the linearly conjugated diene-diketone 2 underwent an intramolecular photometathesis reaction to generate 23, proceeding through 22 in a wavelength-dependent manner. The pentacyclic scaffold of 26, the C6 epimer of daphnepapytone A (27), and 23 could all be accessed in one step from 22, making it a valuable intermediate for further exploration into its chemistry and biology.

Supplementary Material

Supplementary Material

Experimental procedures, computational details, UV-vis spectra, NMR spectra, X-ray crystallographic data of 5. (PDF)

Cartesian coordinates of optimized geometries (XYZ)