Introduction

The Veratrum alkaloids constitute a family of steroidal alkaloids characterized by a unique C-nor-D-homo steroid skeleton (Fig. 1). These alkaloids are widely distributed across various species of Veratrum and Zigadenus, comprising three subtypes (jervanine, veratramine, cevanine) and encompassing over 200 compounds1,2,3. Notably, several members of this alkaloid family are recognized for their distinctive biological activities or toxic effects. For instance, cyclopamine and jervine exhibit pronounced teratogenicity by blocking the Hedgehog signaling pathway4,5,6; zygadenine, found in death camas species, induces poisoning in livestock7,8; germine and protoverine demonstrate antihypertensive properties9,10, and veratridine binds to voltage-gated sodium channels, acting as either an agonist or antagonist in a context- and subtype-dependent manner11,12,13. The intricate structures and significant biological activities of these compounds have sparked considerable interest within both the synthetic and biological research communities.

Fig. 1: Reported syntheses of Veratrum alkaloids.
figure 1

a Framework of the jervanine subgroup. b Framework of the veratramine subgroup. c Framework of the cevanine subgroup.

Full size image

Over the past 50 years, continuous synthetic breakthroughs have significantly advanced the synthesis of the jervanine and veratramine subtypes. In 1967, Masamune’s group and Johnson’s group independently achieved the synthesis of jervine and veratramine, respectively14,15,16. Kutney’s group reported the total synthesis of verarine in 197617, employing a strategy involving the hydrogenation of a pyridine unit to construct the F ring; they subsequently applied the same strategy in the synthesis of verticine18,19. In 2009, Giannis’ group introduced a semisynthetic route to cyclopamine20, while in 2023, both Baran’s group21 and Gao’s group22 separately disclosed the total synthesis of (–)-cyclopamine. Most recently, in 2024, Qin, Liu and coworkers reported divergent and gram-scale syntheses of (–)-veratramine and (–)-cyclopamine from dehydro-epi-androsterone in an efficient manner23. Shifting focus to the cevanine subtype, Rawal’s group achieved the synthesis of (+)-heilonine in 202124, and more recently in 2024. Dai’s group also reported its synthesis25. However, total syntheses of highly oxidized cevanine subtype alkaloids have remained limited. An important contribution in this context was made by Stork’s group26, where they successfully synthesized (±)−4-methylenegermine, which features only one additional methylene group compared to germine.

Zygadenine (1), germine (2) and protoverine (3) are highly oxidized members of the cevanine subgroup and serve as core alkamines for various natural products that exist as esters or polyesters of their hydroxyl groups, including germemine27, neogermitrine28, germitetrine29, and protoveratrines30. Several of these alkaloids were previously used clinically for hypertension management9. However, due to adverse reactions such as neurotoxicity, safer alternatives have since replaced them10. Nevertheless, compounds 1, 2 and 3, distinguished by increasing oxidation levels and multiple hydroxyl groups intricately embedded in the A-E rings, present opportunities for strategic innovation in accessing Veratrum alkaloids. In our prior work, we employed an intramolecular Diels−Alder reaction/radical cyclization strategy to construct the hexacyclic carbon skeleton, followed by a meticulously designed sequence of redox manipulations to complete the total synthesis of (–)-zygadenine (1)31. Herein, we unveil a late-stage neighboring group participation strategy, resulting in a divergent and enantioselective total synthesis of zygadenine (1), germine (2), protoverine (3), and the alkamine of veramadine A (4)30.

Results and discussion

Retrosynthetic design

By analyzing the differences in these target molecules, we envisioned a divergent synthetic strategy and logically traced them back to intermediate 5 (Fig. 2). Starting from compound 8 communicated in our previous publication31, hydrogenation of the Δ6,7-alkene followed by subsequent redox and protection/deprotection manipulations could transform it into zygadenine (1). In comparison, for germine (2), protoverine (3), and the alkamine of veramadine A (4), the installation of the C6 and/or C7 hydroxyl groups must be achieved in a regio- and stereoselective manner, alongside the introduction of the C3 and C4 oxidation states. Tactically, the functionalization of the Δ6,7-alkene could proceed before the C3 reduction, regioselective elimination, dihydroxylation, and subsequent stepwise protections that install the protected hydroxyl groups at C3 and C4 (approach A), and intermediate 6 was hence devised. Alternatively, the protected hydroxyl groups at C3 and C4 could be secured first, followed by the functionalization of the Δ6,7-alkene (via 7, approach B). Either way, the challenges associated with the chemo-, regio-, and stereochemical selectivity of late-stage transformations in a complicated polycyclic skeleton should not be underestimated, and significant experimentation was anticipated to achieve the desired results.

Fig. 2: Retrosynthetic analysis of complex Veratrum alkaloids.
figure 2

The structure of veramadine A was highlighted in light gray background.

Full size image

Implementation of a late-stage neighboring-group participation to functionalize the Δ6,7-alkene

We commenced by adjusting the oxidation states of C6 and C7 through transformations of the Δ6,7-alkene moiety within the cis-decahydronaphthalene framework of 8. Initially, a direct oxidation strategy was employed (Fig. 3a). However, dihydroxylation of 8 with OsO4 resulted in diol 9 with incorrect stereochemistry at C6 and C7, likely due to significant steric hindrance on the α surface in the AB-ring system. Evaluating Woodward cis-hydroxylation conditions (AgOAc, I2) on 8 revealed a side reaction involving the nucleophilic attack of the C9-OH on the iodonium intermediate32, yielding the undesired product 10.

Fig. 3: Evolution of strategies to functionalize the Δ6,7-alkene.
figure 3

a Attempted dihydroxylation of the Δ6,7-alkene. b Initial exploration of the neighboring-group participation strategy. c Analysis of the successful elimination reaction in 18. d The failed attempt to access the substrate with free C9-OH via a carbonate intermediate. Reagents and conditions: (a) OsO4, pyridine, 0 °C to rt, overnight, 85%. b AgOAc, I2, AcOH, overnight, 70%. c NaBH4, DCM/MeOH=1:4, 0 °C, 30 min, 82%. d NaHMDS, TBSCl, then Bz2O, THF, 0 °C, 1 h, 80%. e IBr, MeCN, buffer (pH = 7), 28 °C, 2 h, 85%. f AIBN, nBu3SnH, benzene, reflux, 1 h then TBAF, THF, rt, 2 h, 81%. g TsCl, pyridine, rt, 5 h, 90%. h NaHMDS, TBSCl, then Boc2O, THF, 0 °C, 1 h, 80%. i IBr, MeCN, buffer (pH = 7), 28 °C, 2 h, 80%.

Full size image

These initial endeavors sparked the investigation of a neighboring-group participation strategy by leveraging the C9-OH group (Fig. 3b). Reduction of 1131 by NaBH4, which approached the C3 carbonyl group from the convex face, gave 12 in 82% yield as a single isomer. Protection of the C3-OH in 12 using TBS, followed by benzoylation of the C9-OH in a one-pot procedure, provided 13 in 80% yield. Notably, this method (NaHMDS/Bz2O) exhibited remarkable efficiency for the acylation of the C9-OH, representing a rare example of benzoylation on a tertiary hydroxyl group despite the considerable steric hinderance. Treatment of 13 with IBr afforded orthoester 14 in 85% yield. This cascade event proceeded via oxocarbenium intermediate A, which engaged the neighboring oxygen at the C14 position to form the orthoester, leading to the selective deprotection of the C14, C15-benzylidene acetal. With the correct oxidation level and configuration at C7 successfully attained, we employed radical dehalogenation of the C-6 iodide and deprotected the TBS group to yield 15. We then examined the elimination/dihydroxylation strategy for introducing the C4 oxidation state. However, after tosylation of the C3-OH, the elimination did not occur upon treatment with tBuOK, resulting in the recovery of tosylate 16.

This experimental observation prompted a careful assessment of the mechanism underlying the elimination of C3-OTs. The elimination in substrate 18 to form the Δ3,4-alkene could be realized, as communicated in our former publication31. A plausible explanation is that the elimination reaction to form the Δ3,4-alkene requires C3–OTs and C4-β-H to be trans-coplanar, necessitating the A-ring to adopt a twist-boat conformation (B, Fig. 3c). This conformation can occur in substrate 18. However, in substrate 16, the B-ring conformation was locked by the orthoester structure, adding extra conformational constraints in the AB ring system, and making it difficult to achieve the necessary conformation in the A-ring for the trans-elimination. Therefore, we opted to protect the C9–OH with a Boc group, yielding intermediate 20, which was then treated with IBr to afford 21 in 80% yield (Fig. 3d). We then attempted to deprotect the C9–OH to release the conformational constraint in the AB system to ensure the elimination reaction. However, removing the carbonate group via alkaline hydrolysis proved challenging. Instead of obtaining 22 or the corresponding epoxide, we observed the recovery of starting material 21, indicating significant steric hinderance near the carbonate, likely due to the A-ring and the benzylidene acetal on C14,C15-diol.

Total syntheses of (–)-zygadenine, (–)-germine, (–)-protoverine, and the alkamine of veramadine A

Subsequently, the reaction sequence was modified so that the neighboring-group participation reaction occurred after the elimination/dihydroxylation of the C3-C4 position (Fig. 4). Initially, compound 12 underwent the elimination of the C3-OH through tosylation and treatment with tBuOK, leading to the formation of the Δ3,4-alkene product in 20% yield, which was contaminated by Δ2,3-alkene and Δ4,5-alkene byproduct. After screening conditions, we were able to regio-selectively obtain the desired Δ3,4-alkene product (23) in 80% isolated yield by controlling the reaction temperature to –25 °C and adding tBuOK solid slowly in batches. Next, we attempted the regio- and facial-selective dihydroxylation of the Δ3,4-alkene of 23. Under OsO4/pyridine treatment, the desired product 21 was obtained in only 13% yield, along with byproducts from dihydroxylation of the Δ6,7-alkene and both the Δ3,4-alkene and Δ6,7-alkene. After careful optimization, the regioselective product 24 was obtained in 57% yield (80% brsm) by dropwisely adding the OsO4 solution at low temperature. Mono-protection with TBS provided separable regioisomers 25a and 25b in 31% and 51% yields, respectively, with the undesired regioisomer (25a) being recyclable. Intermediate 25b was further converted to 7, the late-stage common intermediate for our total synthesis endeavor, by installing a BOM protecting group on the C3–OH. For the synthesis of (–)-zygadenine (1), the C6‒C7 alkene was hydrogenated, and the C4–OH was deprotected via TBAF treatment in a one-pot fashion, resulting in the intermediate 26, which could be directed towards (–)-1 using identical reaction sequences in our previous publication31.

Fig. 4: Formal synthesis of (–)-zygadenine (1) and synthesis of the alkamine of veramadine A (–)-4.
figure 4

Reagents and conditions: a TsCl, pyridine, 50 °C, 2 h then tBuOK, toluene, DMSO, –78 °C to –25 °C, 30 min, 80%. b OsO4, pyridine, THF, –40 °C, 2 h, 57% (80% brsm). c TBSCl, imidazole, DMAP, DCM, rt, overnight, 82% (68a/68b = 3:5) d BOMCl, DIPEA, TBAI, DCE, 80 °C, 1 h, 97%. e Pd/C, EtOAc, Et3N, H2 (7 MPa), rt, 20 h then TBAF, THF, 70 °C, 5 h, 65%. f NaHMDS, Bz2O, THF, 0 °C, 1 h, 97%. g IBr, MeCN, buffer (pH=7), 28 °C, 2 h, 56%. h NaHMDS, CS2, THF, –78 °C, 10 min then MeI, –78 °C, 30 min, 72%. i AIBN, nBu3SnH, benzene, reflux, 1.5 h then TBAF, THF, reflux, 3 h, quant. j DMP, NaHCO3, DCM, rt, 15 h, 84%. k Pd(OH)2/C, MeOH, H2 (1 atm), rt, 5 h. l Et2SiH2, [Ir(COE)2Cl]2, toluene, reflux, 3 h then MeCN, HF (40% aq.), rt, overnight, 62% (2 steps). brsm, yield based on the recovered starting material; BOMCl, Benzyl chloromethyl ether; DMP, Dess-Martin periodinane.

Full size image

We then proceeded to diverge from 7 to obtain 2, 3 and 4. The alkamine of veramadine A (4)30, a congener with the same oxidation level as zygadenine (1), was targeted first. A benzoyl group was introduced to the C9-OH to give 27, and subsequent IBr treatment afforded orthoester 28 in 56% yield. Conversion of the C15–OH to the corresponding xanthate 29 was followed by radical defunctionalization to adjust the oxidation levels at C6 and C15, and one-pot TBAF treatment afforded alcohol 30. The C4-OH group was then oxidized to a ketone to give 31 in 84% yield, and global deprotection via palladium-catalyzed hydrogenation released the C9-OH to form the hemiketal (32), serving as protection of the C4-ketone during iridium-catalyzed silane-reduction33. Notably, silylation of the hydroxyl groups occurred during this final transformation; therefore, hydrofluoric acid treatment was performed in the same reaction flask after the reduction, leading to the isolation of the alkamine of veramadine A (4) in 62% yield over two steps.

To access germine (2), radical dehalogenation of the iodide in 28 was followed by protecting the C15-OH group with a BOM group, and TBAF treatment was performed in a one-pot fashion to yield 33 (Fig. 5). This was followed by DMP oxidation of the C4-OH group to deliver ketone 34. Sequential global deprotection and final amide reduction then afforded (–)-germine (2) in 45% yield over two steps. The elaboration of 28 to protoverine (3) commenced with the transformation of C6-I to a TEMPO group, resulting in 35 using Boger’s protocol34. Subsequent peroxy acid oxidation produced the C6 ketone 36, which was then reduced by NaBH4 to introduce the β-C6-OH group, yielding 37. Intermediate 37 underwent BOM protection of both the C6 and C15 hydroxyl groups, followed by desilylation of the TBS-protected C4–OH group, leading to the formation of 38. A subsequent DMP oxidation step resulted in ketone 39 in 78% yield. The final steps involved the same global deprotection and amide reduction sequence as previously utilized for 4, culminating in the total synthesis of (–)-protoverine (3). The NMR spectra of the synthesized samples of 2 were consistent with those reported in the literature (Supplementary Table 2)30,35, while the structures of 2 and 4 were unambiguously verified by X-ray diffraction analysis; the stereochemistry of 3 was determined through 2D NMR experiments due to the unavailability of NMR spectra data of the isolated sample.

Fig. 5: Syntheses of (–)-germine (2) and (–)-protoverine (3).
figure 5

Reagents and conditions: a nBu3SnH, AIBN, benzene, reflux, 1.5 h then BOMCl, TBAI, DIPEA, DCE, 80 °C, 2 h then TBAF, THF, reflux, 3 h, 89%. b DMP, NaHCO3, DCM, 30 °C, 13 h, 72%. c Pd(OH)2/C, MeOH, H2 (1 atm), rt, 5 h. d Et2SiH2, [Ir(COE)2Cl]2, toluene, reflux, 3 h then MeCN, HF (40% aq.), rt, overnight, 45% (2 steps). e TEMPO, nBu3SnH, toluene, reflux, 3 h, 87%. f UHP, Na2CO3, TFAA, DCM, 0 °C, 6 h, 81%. g NaBH4, MeOH, 0 °C, 30 min, 85%. h BOMCl, TBAI, DIPEA, DCE, 80 °C, 2 h, then TBAF, THF, reflux, 3 h, 81%. i DMP, NaHCO3, DCM, 30 °C, 14 h, 78%. j Pd(OH)2/C, MeOH, H2 (1 atm), rt, 5 h. k Et2SiH2, [Ir(COE)2Cl]2, toluene, reflux, 3 h then MeCN, HF (40% aq.), rt, overnight, 53% (2 steps). UHP, urea hydrogen peroxide; TFAA, trifluoroacetic anhydride.

Full size image

In summary, we have developed a successful divergent approach for the syntheses of (–)-zygadenine (1), (–)-germine (2), (–)-protoverine (3) and the alkamine of veramadine A (4). The primary challenge lay in achieving chemo-, regio- and stereoselectivity during the late-stage transformations, especially redox sequence, within this complex and highly oxidized framework. Ultimately, the functionalization of the C6-C7 alkene using neighboring-group participation (C9-OBz) after the introduction of the protected C4-OH proved to be an essential tactic. This synthetic strategy not only provides a pathway to access other important members of the Veratrum family of alkaloids but also offers opportunities for further exploration of their pharmacological properties.

Methods

Unless otherwise mentioned, all reactions were carried out under a nitrogen atmosphere with dry solvents under anhydrous conditions. All the chemicals were purchased commercially, and used without further purification. Anhydrous THF was distilled from sodium-benzophenone, toluene was distilled from sodium, dichloroethane and dichloromethane were distilled from calcium hydride according to Purification of Laboratory Chemicals (Peerrin, D. D.; Armarego, W. L. and Perrins, D. R., Pergamon Press: Oxford, 1980). Reactions were monitored by thin layer chromatography (TLC) carried out on 0.25 mm E. Merck silica plates (60F-254), using UV light as the visualizing agent and an ethanolic solution of ammonium molybdate and heat, or KMnO4 and heat as developing agents. If not specially mentioned, flash column chromatography uses silica gel (200–300 mesh) supplied by Tsingtao Haiyang Chemicals (China). NMR spectra were recorded on Brüker Advance 400 (1H 400 MHz, 13C 100 MHz), Brüker Advance 500 (1H 500 MHz, 13C 125 MHz), Brüker Advance 600 (1H 600 MHz, 13C 150 MHz) or Brüker Advance 800 (1H 800 MHz, 13C 200 MHz). TMS was used as internal standard for 1H NMR (0.00 ppm), and solvent signal was used as reference for 1H NMR (CDCl3, 7.26 ppm, CD2Cl2, 5.32 ppm, CD3OD, 3.31 ppm, CD3COCD3, 2.05 ppm, C6D6, 7.16 ppm), 13C NMR (CDCl3, 77.16 ppm, CD2Cl2, 54.0 ppm, CD3OD, 49.00 ppm, CD3COCD3, 206.26, 29.84 ppm, C6D6, 128.06 ppm). The following abbreviations were used to explain the multiplicities: s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet, br = broad. Mass spectrometric data were obtained using Brüker Apex IV FTMS using ESI (electrospray ionization) and Waters GCT (GC-MS) using EI (electron impact ionization). Infrared spectra were recorded on a Thermo Nicolet iS5 spectrometer. Optical rotations were measured on an InsMark IP-digi300 digital polarimeter with a LED light source at ambient temperature and are reported as follows: [α]λ (c g/100 mL).