First Divergent Total Synthesis of (±)-Nobilin D and Structurally Related Bibenzyl Natural Products

Abstract

We report the first concise and efficient racemic total synthesis of (±)-Nobilin D, a bioactive bibenzyl natural product isolated from Dendrobium nobile. Our unified synthetic approach also provides access to (±)-Combretastatin, Moscatilin, and Erianin, all derived from a readily available bromoethyl-substituted norbornyl α-diketone precursor. The key steps involve a Grob fragmentation–aromatization sequence to yield a crucial aromatic ester intermediate, which is then subjected to Baeyer–Villiger oxidation, O-methylation, dehydrohalogenation, and epoxidation to generate highly substituted styrene epoxides. Subsequent regioselective aryl-organolithium epoxide opening, followed by Pd–C/H₂ hydrogenolysis, enables the efficient assembly of all four target natural products. This work not only provides the first synthetic access to Nobilin D but also allows for unambiguous structure confirmation by X-ray crystallography. Furthermore, our strategy establishes a scalable platform for the preparation of pharmacologically relevant bibenzyl scaffolds, facilitating future structure–activity relationship (SAR) studies.

Introduction

Bibenzyl natural products (BNPs) constitute a structurally privileged class defined by an aryl-C­(sp³)–C­(sp³)-aryl core that supports a wide spectrum of biological activities. Representative members such as (−)-ε-viniferin, (±)-Nobilin D, (±)-Nobilin E, (±)-Combretastatin, Moscatilin, and Erianin illustrate the remarkable functional diversity accessible from this simple framework (Figure ). (−)-ε-Viniferin, a resveratrol-derived dimer, displays prominent antioxidant, neuroprotective, and anti-inflammatory properties. Bibenzyls isolated from Dendrobium nobile further broaden this profile: Nobilin D suppresses LPS-induced nitric oxide production and exhibits antioxidant activity, whereas Nobilin E shows complementary cytotoxic and anti-inflammatory effects. , In contrast, Moscatilin and Erianin act as potent pyruvate carboxylase (PC) inhibitors, disrupting cancer cell metabolism and exhibiting strong anticancer activity against hepatocellular carcinoma via metabolic reprogramming. , Distinct from these metabolic and immunomodulatory mechanisms, (±)-Combretastatin, a reduced member of the combretastatin family, retains high antimitotic potency through tubulin polymerization inhibition, paralleling the activity of its parent stilbenoid Combretastatin A-4 (CA-4). , Collectively, these natural products underscore the functional plasticity of the bibenzyl scaffold, wherein subtle variations in oxidation state and substitution redirect biological activity across cytoskeletal, metabolic, and inflammatory pathways, motivating unified synthetic approaches to access and systematically interrogate this underexplored natural product family.

1.

Potential polyphenolic bibenzyl natural products.

Despite their therapeutic potential, Nobilin D and structurally related Dendrobium bibenzyls remain uncharacterized with respect to their synthetic and stereochemical properties, except for the recent synthetic methods for Nobilin D related Dendrobium-derived bibenzyls. ,,, The original isolation report established its molecular framework spectroscopically but provided no X-ray crystallographic data, leaving its absolute configuration unresolved. Furthermore, no total or racemic synthesis of Nobilin D has been documented, which limits both its pharmacological validation and structural understanding.

Therefore, developing concise, protecting-group-efficient racemic routes that tolerate multifunctional aromatic substrates is crucial to access Nobilin D and related analogs for chemical and biological research. Achieving synthetic and stereochemical clarity will create new opportunities for structure–activity relationship (SAR) studies, enabling comparisons among bibenzyl-derived metabolic inhibitors, vascular-disrupting agents, and redox-modulating anti-inflammatory compounds within the broader spectrum of bibenzyl class natural products.

In this work, we describe the first racemic total synthesis of (±)-Nobilin D (1), together with the previously reported (±)-Combretastatin (2) and their corresponding 1-deoxy bibenzyl congeners, Moscatilin (3) and Erianin (4). The strategy hinges on a previously unexplored disconnection beginning from the readily accessible and structurally distinctive bromoethyl-substituted norbornyl α-diketone (5), as outlined in the retrosynthetic analysis (Scheme ). A pivotal intermediate, methyl 2-(benzyloxy)-5-(2-bromoethyl)-4-chloro-3-methoxybenzoate (10), can be obtained from 5 via a Grob fragmentation–aromatization sequence followed by a reduction–oxidation transformation, which efficiently converts the norbornyl framework into the corresponding aromatic aldehyde 15. From this point, a sequence comprising Baeyer–Villiger oxidation, O-methylation, dehydrohalogenation, and epoxidation furnishes the highly functionalized styrene epoxide 19 or 24, a versatile electrophilic intermediate suitable for divergent natural product assembly.

1. Retro Synthetic Analysis of BNP’s: (±)-Nobilin D (1); (±)-Combretastatin (2); Moscatilin (3); Erianin (4).

The target natural products arise from a streamlined series of transformations involving tributyltin hydride-mediated dechlorination, organolithium-promoted epoxide opening with the appropriate aryl bromides, and benzyl deprotection under Pd–C hydrogenation conditions, providing racemic (±)-Nobilin D (1) and (±)-Combretastatin (2). The 1-deoxy analogs, Moscatilin (3) and Erianin (4), can be accessed directly from their hydroxylated counterparts through Pd-catalyzed hydrogenolysis of the benzylic hydroxyl group under extended hydrogenation conditions. This overall strategy delivers a unified and efficient entry to four structurally and biologically significant bibenzyl natural products, as shown in Scheme .

Results and Discussion

For the synthesis of the targeted natural products 1–4, α-diketone 5 was first treated with trifluoroacetic acid (TFA) in refluxing 1,2-dichloroethane, followed by selective methylation using diazomethane as the methylating agent to obtain a regio-isomeric mixture of products 6 and 7 in a 3:1 ratio, favoring 6 as shown in Scheme . This unique structural transformation was achieved using the Grob fragmentation sequence. The mixture of fragmentation products 6 and 7 was then benzylated under K₂CO₃–BnBr conditions in acetone to produce aromatized regioisomeric analogs 8 and 9, which could be separated on a silica gel column, yielding 87% overall over three steps.

2. Synthesis of 8 from Grob Fragmentation of α-Diketone 5: (a) TFA, 1,2-DCE, Reflux, 1 h; (b) CH₂N₂–Et₂O, MeOH, 0 °C, 0.5 h; (c) K₂CO₃, BnBr, Acetone, 0 °C-rt, 36 h, 87% (3 Steps).

Comparison of the ¹H NMR spectra of these two regioisomers revealed that the aromatic proton next to the ester in product 8 was more deshielded (δ 7.49) than the proton next to chlorine in product 9 (δ 7.11). Further 2D NMR analysis, including HSQC and HMBC experiments, conclusively established the regioisomeric identities of 8 and 9, thereby confirming the structural assignments of the Grob fragmentation products 6 and 7. The unambiguous characterization of 8 provided the foundation for the unified synthetic route to the bibenzyl natural products 1–4, as outlined in the retrosynthetic analysis (Scheme ).

DBU-mediated dehydrohalogenation of 8 afforded styrene 10 in excellent yield (92%). The ester group in compound 10 was subsequently converted into aldehyde 11 through a two-step reduction–oxidation process, involving DIBAL-H reduction followed by PCC oxidation, both of which produced excellent results (Scheme ). At this stage, it was anticipated that Dakin oxidation and epoxidation of styrene-aldehyde 11, followed by O-methylation and reductive dehalogenation, would yield the desired substituted epoxide 19 as mentioned in the retrosynthetic analysis (Scheme ). However, treating 11 with 30% alkaline hydrogen peroxide (H₂O₂) in methanol afforded a mixture of polar compounds, which, upon methylation, yielded three separable products (Scheme ). Among these, two minor products (13 and 14) were identified as epoxide-opened aromatic esters, whereas the major component (12) corresponded to an undesired epoxide-containing methyl ester. The identity of minor compounds 13 and 14 was established by HMBC analysis (SI). These outcomes suggest that the styryl double bond undergoes premature epoxidation, followed by reoxidation of aldehyde 11 to the corresponding acid under the strongly oxidative Dakin conditions.

3. Unsuccessful Attempt to Synthesize the Key Styrene-Epoxide, 19: (a) DBU, DCM, rt, 36 h, 92%; (b) DIBAL-H, DCM, −78 to 0 °C; (c) PCC, DCM, 0 to 32 °C, 90% (2 Steps); (d) 30% H₂O₂, 1N NaOH, MeOH, 50 °C, 1 h; (e) MeI, K₂CO₃, Acetone, 32 °C, 2 h (12:40%; 13 15%; 14:23%).

To address challenges encountered during the Dakin oxidation sequence, the synthetic route was reoptimized to include a Baeyer–Villiger oxidation prior to the DBU-mediated elimination. Aromatic ester 8 was reduced with DIBAL-H, then oxidized with PCC to produce aromatic aldehyde 15 in nearly quantitative yield (Scheme ). The Baeyer–Villiger (BV) oxidation of 15 with m-CPBA, followed by hydrolysis of the resulting formate intermediate, efficiently yielded phenol 16. Methylation of 16, followed by DBU-mediated elimination, generated the substituted chlorostyrene derivative 17 (Scheme ).

4. Synthesis of (±)-Nobilin D (1) and Moscatilin (3): (a) DIBAL-H, DCM, −78 °C to rt, 30 min; (b) PCC, DCM, rt, 5 h, 98% (2 Steps); (c) m-CPBA, DCM, 0 °C to rt, 40 h, Et₃N, MeOH/DCM (1:1), 90%; (d) MeI, K₂CO₃, Acetone, rt, 2 h; (e) DBU, DCM, 0 °C to rt, 11 h (97%, 2 Steps); (f) TBTH, AIBN, C₆H₆, Reflux, 4 h, 88%; (g) m-CPBA, Na₂CO₃, DCM, 0 °C to rt, 36 h, 88%; (h) 1-(Benzyloxy)-4-bromo-2-methoxybenzene, t-BuLi, THF, −40 °C to rt, 0.5 h, 85%; (i) Pd–C/H₂, EtOAc, rt (1: 97%; 3: 95%); (j) Ac₂O or PhCOCl, Et₃N, DMAP, DCM, rt, 5 h, 98%.

In a complementary sequence toward Nobilin D, dechlorination of 17 with tributyltin hydride (TBTH) gave des-chloro-styrene 18, which, upon epoxidation with m-CPBA, yielded the corresponding styrene epoxide 19 in 88% overall yield. Initial attempts to open the epoxide of 19 using n-BuLi and 1-(benzyloxy)-4-bromo-2-methoxybenzene were unsuccessful. However, replacing n-BuLi with t-BuLi resulted in efficient and regioselective aryl-organolithium addition, delivering the penultimate bis-O-benzyl intermediate 20 in excellent yield (Scheme ). The distinct reactivity observed between n-BuLi and t-BuLi in the epoxide-opening step can be attributed to differences in their nucleophilicity, steric profile, and efficiency of halogen–lithium exchange. When n-BuLi was employed in the presence of substituted aryl bromide, direct nucleophilic attack on the styrene epoxide 19 predominated, leading to the butylated product and suppressing the formation of the desired aryl lithium intermediate. This behavior arises from the relatively high nucleophilicity and lower steric hindrance of n-BuLi, combined with slower and incomplete halogen–lithium exchange. In contrast, t-BuLi undergoes rapid halogen–lithium exchange to generate the aryl lithium species efficiently, while its steric bulk minimizes direct nucleophilic attack on the epoxide. As a result, the aryl lithium intermediate selectively attacks the styrene epoxide 19 at less substituted nonbenzylic position, affording the desired bibenzyl product 20 in high regioselectivity.

The subsequent Pd–C catalyzed hydrogenolysis of 20 was monitored by TLC, which showed the formation of two products within 45 min: a predominant low-polarity compound and a high-polarity compound. Spectroscopic analysis ¹H and ¹³C NMR indicated that the low-polarity product corresponds to the fully 1-deoxygenated natural product Moscatilin (3), while the high-polarity product matches (±)-Nobilin D (1), as shown in Scheme .

Notably, shortening the hydrogenation time to <10 min produced (±)-Nobilin D (1) in 97% yield, whereas prolonging the process to 1 h resulted exclusively in Moscatilin (3) in 95% yield (Scheme ). Multiple attempts to obtain single crystals of 1 or 20 suitable for X-ray diffraction were unsuccessful. However, single-crystal X-ray crystallographic analysis of the acetyl derivative 21 prepared from intermediate 20 enabled the definitive structural confirmation, thereby unambiguously establishing the molecular framework of Nobilin D, as depicted in Scheme .

For the synthesis of (±)-Combretastatin (2) and Erianin (4), intermediate 18 was first debenzylated under standard BCl₃ conditions, followed by methylation, yielding the 1,2,3-trimethoxy-5-vinylbenzene (23) as shown in Scheme . Next, epoxidation of 23 produced the corresponding styrene epoxide 24 with high overall yield. Opening the epoxide 24 with 2-(benzyloxy)-4-bromo-1-methoxybenzene under conditions similar to those used for 19, gave the O-benzyl intermediate 25 in excellent yield. Hydrogenolysis of 25 under Pd–C/H₂ cleanly produced (±)-Combretastatin (2) in 1.5 h (96%). Conversely, producing Erianin (4) required more than 10 h of hydrogenation to reach a similar yield (95%), as shown in Scheme . The ¹H and ¹³C NMR data for the synthesized (±)-2 have been well established by comparison with those of the isolated natural product. Further, a single-crystal X-ray crystallographic analysis of the acetyl derivative 26 prepared from intermediate 25 enabled the definitive structural confirmation, thereby unambiguously establishing the molecular framework of (±)-Combretastatin (2).

5. Synthesis of (±)-Combretastatin (2) and Erianin (4): (a) BCl₃–C₆H₅Me, DCM, −78 °C, 0.5 h; (b) MeI, K₂CO₃, Acetone, rt, 2 h, 96% (2 Steps); (c) m-CPBA, Na₂CO₃, DCM, 0 °C to rt, 36 h, 84%; (d) 2-(Benzyloxy)-4-bromo-1-methoxybenzene, t-BuLi, THF, −40 °C to rt, 0.5 h, 85%; (e) Pd–C/H₂, EtOAc, rt (for 2: 96%, for 4: 95%); (j) Ac₂O, Et₃N, DMAP, DCM, rt, 5 h, 98%.

However, the hydrogenolysis experiments summarized in Table demonstrate a pronounced time- and catalyst-loading dependence in the Pd–C/H₂ reduction of the benzyl-protected intermediates 20 and 25, enabling selective access to either hydroxylated or 1-deoxy bibenzyl natural products. When benzyl-protected precursor 20 was subjected to hydrogenation using 10 mol % Pd–C, extended reaction times (1 h) led predominantly to the formation of the fully deoxygenated product Moscatilin (3) (entry 1), consistent with complete hydrogenolysis of the benzylic C–O bond. At an intermediate reaction time (45 min), a mixture of (±)-Nobilin D (1) and Moscatilin (3) was obtained (entry 2), whereas short hydrogenation times (<10 min) afforded exclusive formation of (±)-Nobilin D (1) (entry 3). Prolonged hydrogenation (16 h) resulted in a complex mixture, indicating over-reduction and diminished chemoselectivity (entry 4). Notably, decreasing the Pd–C loading to 5 or 2 mol % restored complete selectivity for (±)-Nobilin D (1) even at longer reaction times (entries 5 and 6), highlighting catalyst loading as a critical parameter for suppressing undesired benzylic deoxygenation.

1. Hydrogenolysis Summary Experiments of BNP’s 1–4: Pd–C (5–10 Mol %), H₂ Balloon (∼1–2 mmHg), EtOAc, rt (25 °C).

S. No.	SM	Pd–C loading	Time	% Products (1–4)
1	20	10 mol %	1 h	100 (3)
2	20	10 mol %	45 min.	20:80 (1:3)
3	20	10 mol %	<10 min.	100 (1)
4	20	10 mol %	16 h	not determined
5	20	5 mol %	1 h	100 (1)
6	20	2 mol %	3 h	100 (1)
7	25	10 mol %	45 min.	100 (2, conv. 70%)
8	25	10 mol %	1.5 h	100 (2)
9	25	10 mol %	3 h	70:30 (2:4)
10	25	10 mol %	10 h	100 (4)

^aIsolated yields 1 (97%), 2 (96%), 3 (95%), 4 (95%).

^bProduct ratios determined from ¹H NMR analysis of the crude product mixture.

A complementary trend was observed for benzyl-protected precursor 25, which leads to (±)-Combretastatin (2) and Erianin (4). Under 10 mol % Pd–C, short hydrogenation times (45–90 min) resulted in selective formation of (±)-Combretastatin (2) (entries 7 and 8). With increasing reaction time, progressive benzylic C–O hydrogenolysis was observed, leading to mixtures of (±)-Combretastatin (2) and Erianin (4) (entry 9), and ultimately to exclusive formation of Erianin (4) upon prolonged hydrogenation (entry 10). These results indicate that benzylic deoxygenation proceeds more slowly in the combretastatin series than in the Nobilin series, reflecting substrate-dependent reactivity. Collectively, these data establish that benzyl-protected intermediates 20 and 25 serve as bifurcation points from which either hydroxylated or 1-deoxy bibenzyl natural products can be accessed through kinetic control of Pd–C/H₂ hydrogenolysis. Precise modulation of reaction time and catalyst loading enables chemoselective differentiation between benzyl deprotection and benzylic C–O bond cleavage, underscoring the versatility of this divergent synthetic platform. The structural assignments of the racemic products (±)-Nobilin D (1), (±)-Combretastatin (2) were further confirmed unambiguously by directly comparing their spectroscopic data with the corresponding isolated natural product, as shown in Tables S1 and S2.

Palladium-catalyzed debenzylation typically proceeds via a surface-mediated mechanism involving substrate adsorption and dissociative activation of hydrogen to generate palladium-hydride species. In general, aryl benzyl ethers undergo benzylic C–O bond cleavage through a surface-stabilized benzyl intermediate, followed by hydride transfer to furnish toluene and the corresponding phenol upon protonation of the phenoxide in the first step (Scheme ). Subsequent deoxygenation of the benzylic alcohol occurs through a similar surface-mediated C–O bond activation (Int-1, Scheme ), leading to the formation of a stabilized benzylic carbocation-like intermediate (Int-2, Scheme ), which is further stabilized by π-conjugation with the adjacent aromatic ring. This conjugated intermediate undergoes rapid hydrogenation on the Pd surface to afford the bibenzyl framework (Scheme ).

6. A Plausible Hydrogenation Mechanism for the Formation of Products 1 and 3 from 20 .

The combined effects of benzylic stabilization, surface coordination, and conjugative activation significantly lower the barrier for hydrogenolysis, enabling efficient stepwise transformation from benzyl-protected intermediates to fully deoxygenated products under prolonged hydrogenation conditions.

In the present system, Pd–C-mediated hydrogenation initially promotes efficient debenzylation to yield (±)-Nobilin D (1). The newly generated para-phenolic hydroxyl group is likely to facilitate the formation of a p-quinone methide-type intermediate (Int-3), which undergoes hydrogenation further to furnish Moscatilin (3), as illustrated in Scheme . Accordingly, precise control of reaction time and catalyst loading is essential to achieve selective debenzylation of intermediate 20, allowing isolation of (±)-Nobilin D (1) while minimizing over-reduction to Moscatilin (3). These mechanistic observations are consistent with the Pd–C hydrogenation behavior observed for (±)-Combretastatin (2) and Erianin (4), where the corresponding para-phenolic functionality in intermediate 25 is protected as a methyl ether (Scheme ), thereby relatively preventing analogous quinone methide formation and subsequent quick deoxygenation.

Conclusions

In this work, we present an efficient, divergent racemic synthesis of four biologically significant natural products: (±)-Nobilin D, Moscatilin, (±)-Combretastatin, and Erianin. Starting from a unique norbornyl α-diketone feedstock, the route employs Grob fragmentation–aromatization to rapidly generate functionalized aromatic intermediates. Key oxygenation patterns are introduced via Dakin or Baeyer–Villiger oxidations, and regioselective aryl-organolithium epoxide openings provide high-yield access to the final products. Notably, this study reports the first total synthesis of (±)-Nobilin D, enabling future structure–activity relationship studies and the development of new analogs within the bibenzyl natural product family.

The data underlying this research are available in the published article and its Supporting Information.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.6c01624.

• Experimental procedures and ¹H and ¹³C magnetic resonance (NMR) spectra for all compounds (PDF)

• CCDC 2525583 (for 21) crystallography data can be obtained free of charge from the Cambridge Crystallographic Data Center via https://www.ccdc.cam.ac.uk/structures (CIF)

• CCDC 2525584 (for 26) crystallography data can be obtained free of charge from the Cambridge Crystallographic Data Center via https://www.ccdc.cam.ac.uk/structures (CIF)

The authors declare no competing financial interest.