Steroidal alkaloids, a class of nitrogen-containing derivatives of natural steroids, constitute a significant group of alkaloids and secondary metabolites present in plants, amphibians, and marine organisms [
1,
2]. Notably, several steroidal alkaloids have been successfully developed into therapeutic agents, with abiraterone acetate, a steroidal antiandrogen approved by the FDA in 2011 for treating metastatic castration-resistant prostate cancer, being a prominent example [
3,
4]. This has spurred extensive interest in the synthesis of bioactive steroidal alkaloids with unique structural features, including rearranged skeletons (
e.g., cyclopamine [
5–
8]) and unusual bond connectivities (
e.g., batrachotoxin [
9–
11]) [
12].
In light of the global opioid crisis, the development of non-opioid analgesics has become an important topic [
13,
14]. (-)-Veragranines A (
1) and B (
2), two non-opioid steroidal alkaloids isolated from
Veratrum grandiflorum by Luo and co-workers, have attracted significant attention owing to their potent analgesic activity (
Fig. 1a) [
15]. These compounds act as Ca
V2.2 voltage-gated calcium channel (VGCC) inhibitors, with IC
50 values of 45.76 ± 1.14 and 7.82 ± 1.10 µmol/L, respectively, making them promising candidates for further biological investigation. However, the extremely low natural abundance of
1 and
2 (approximately 0.00004 wt%) poses a significant challenge for conducting comprehensive biological evaluations. To overcome this limitation, efficient chemical syntheses are not only crucial but also indispensable. Structurally, both
1 and
2 feature a complex 6/6/6/5/6/6 hexacyclic ring system, a unique C12-C23 linkage, and a trisubstituted pyridine ring. Driven by our ongoing interest in the synthesis of steroidal alkaloids [
16–
19], veragranines have attracted our attention due to their natural scarcity, unique structures, and promising biological activities. Herein, we report our divergent synthesis of
1 and
2 along with their dihydrogenated analog 5
α,6-dihydroveragranines A (
3) through two distinct strategies. Notably, in 2022, concurrent with our study, Shi reported the synthesis of
3 and
4 [
20]. During the preparation of this manuscript, Dai disclosed a semisynthesis of
1 and its analogs [
21]. It is important to note that our work remains the first approach enabling the synthesis of both
1 and
2.
Two similar biosynthetic pathways were proposed by Luo and Dai [
15,
21], where they believed
1 and
2 could be biosynthesized from veramiline (
5), a major steroidal alkaloid of
V. grandiflorum. According to their hypothesis, veramiline could be converted to vermitaline (
6)
via an enzymatic C12-H hydroxylation and dehydrogenation of the piperidine ring. Subsequently, Luo postulated a stepwise process: Initially, oxidative aromatization of the F ring to a pyridine, followed by an S
N2-type substitution to form the C23-C12
σ-bond. Whereas Dai proposed an alternative sequence, first, construction of the C23-C12
σ-bond through a similar S
N2 reaction, then oxidative aromatization and epimerization of the C20 stereochemistry to give
1. We conducted a comprehensive evaluation of the biomimetic approach as a potential synthetic pathway. Based on a detailed analysis, we concluded that the S
N2-type substitution could be challenging because the hydroxyl group is a poor leaving group. Additionally, there could be difficulties in the subsequent steps, including the oxidation of an imine to a pyridine and the desaturation to form a C11-C12 double bond. We then proposed an alternative biosynthetic pathway involving two key transformations: First, the selective oxidation of the C12-OH in
6 to yield ketone
8; second, an aldol condensation to form the C23-C12 bond, generating the
α,β-unsaturated imine
9, which subsequently undergoes oxidative aromatization followed by double bond transposition to afford
1 (
Fig. 1b).
Based on our biosynthetic proposal, the first-generation retrosynthetic analysis is presented in
Fig. 1c. Since the C11 ketone of
2 could be constructed from a C11-C12 epoxide
via Meinwald rearrangement [
20], we prioritized the synthesis of
1. The Δ
5(6) double bond of
1 could be introduced from
10, through a challenging HAT-based late-stage C6-H halogenation/elimination strategy. We envisaged
10 could be obtained from
11 through our proposed biomimetic oxidative aromatization, double bond transposition. Then
11 could be synthesized from
12 through C26 azidination followed by reductive imine formation and C16 debromination. Moreover,
12 could be easily prepared from hecogenin acetate (
13) through BF
3·Et
2O-promoted bromination/ring-opening of steroidal sapogenins, the methodology we previously developed [
16].
As shown in
Fig. 2a, our synthesis started with the preparation of dibromide compound
12 from
13. Then regioselective azidination of the less hindered C26-Br gave the desired product
14 in 92% yield. We hypothesized that the C26-azide reduction, C22 imine formation, and C16-Br hydrodebromination transformations could be performed in the same flask. After systematic optimization (see Supporting Information for details), we discovered that the Raney Ni/H
2 reduction conditions facilitated this cascade transformation efficiently, affording the desired product
11 in 83% yield. Remarkably, extending the reaction time enabled the formation of the complete hydrogenation product
15 in high yield. Subsequent hydrolysis afforded product
16, featuring a piperidine as the F ring, which was also evaluated for bioactivity. With
11 in hand, we proceeded to investigate the key biomimetic oxidative aromatization of the F ring with transposition of the Δ
12(23) double bond to Δ
11(12). After unsuccessful attempts with reagents such as DDQ, IBX, and others, we found that using Pd/C as the catalyst and oxygen as the oxidant provided the desired product
10 in 24% yield. After hydrolysis, the desired product 5
α,6-dihydroveragranine A (
3) was obtained in good yield.
In order to obtain the natural product veragranine A (
1), we decided to implement a strategy featuring HAT-based late-stage C(sp
3)-H halogenation/elimination [
22–
28]. Inspired by the pioneering work reported by Wu [
29], we designed a decatungstate photoredox-catalyzed C—H chlorination (
Fig. 2b). Prior to applying the desired substrate
10, we selected an easily accessible compound
17 as the model substrate. After optimization (see Supporting Information for details), the desired C—H chlorination product could be obtained in 36% yield, with 1.7:1 (C6:C7) regioselectivity (
Fig. 2c). A further elimination step was carried out to transfer the C6-Cl product
18 into Δ
5(6) compound
19. Although this strategy could be successfully applied to introduce a Δ
5(6) double bond in model substrate
17, unfortunately, it did not work on real substrate
10. After spending half a year studying HAT-based C—H functionalization conditions to introduce the Δ
5(6) double bond, we could not find an efficient solution. We therefore started our second-generation synthetic route.
Given the challenge of late-stage introduction of a Δ
5(6) double bond in the A/B
trans-fused steroid system, and considering that it is easier to introduce a Δ
4(5) double bond in A/B
cis-fused steroids (which can further undergo transposition to Δ
5(6)) [
30,
31], we chose the abundantly available deoxycholic acid (
24) as the starting material. Our second-generation retrosynthetic analysis is presented in
Fig. 3, where both
1 and
2 could be synthesized from
20 through late-stage oxidation to introduce a Δ
4(5) double bond. According to Shi’s result [
20], late-stage introduction of a C11 ketone group in
2 from the Δ
11(12) olefin in
20 is also feasible. The E ring of
20 could be constructed through a photoredox-catalyzed decarboxylative Minisci reaction from
21 [
32–
34]. The pyridine fragment of
21 could be introduced through the addition of organolithium reagent
22 to the C12-ketone of
23. Subsequent elimination of the newly formed C12-OH afforded the Δ
11(12) double bond. Also, the C22 carboxylic acid of
21 could be easily obtained
via oxidative cleavage of the terminal Δ
22(23) olefin of
23. In turn,
23 could be constructed from
24 through photoredox-catalyzed oxidative degradation of the side chain [
35,
36].
As the key step in our synthetic plan, the decarboxylative Minisci reaction faced two major challenges. First, radical cyclization to the pyridine ring could lead to ortho-addition (C22) and para-addition (C24), thus regioselectivity needs to be controlled. We believed that the Me- group on the pyridine ring (C25) would make the para-addition to the C24 position more sterically hindered, therefore our desired ortho-addition product would be the major product. Second, the C21-Me configuration needs to be inverted through this step. Since the decarboxylation step would generate a radical at the C20, we hypothesized that during the cyclization, the C21-Me would prefer to be in the pseudo-equatorial position to avoid strong 1,3-diaxial interaction with the C18-Me, thus providing the desired inverted stereochemistry at C20.
As illustrated in
Fig. 4, utilizing the method developed by the Ritter and Larionov group [
35,
36], we initiated our synthesis with photoredox-catalyzed decarboxyolefination of
24, yielding the desired product
25 in 83% yield. After selective TBS-protection of the less hindered C3-OH and subsequent Dess-Martin oxidation of the C12-OH, intermediate
23 was obtained in 89% yield on a 2-gram scale. Following successful preparation of organolithium reagent
22 through
in-situ lithium-bromine exchange, we investigated the addition of the
22 to the C12-ketone of
23.
To our delight, this reaction proceeded smoothly, affording the desired product
26 in 97% yield, with a 10:1 diastereomeric ratio. We next explored the oxidative cleavage of the terminal Δ
22(23) olefin to a carboxylic acid. Several conventional methods, including the RuCl
3/NaIO
4 system or ozonolysis followed by Pinnick oxidation, gave low yields (see Supporting information for details). Fortunately, we found that with a basic KMnO
4/NaIO
4 system [
37], the desired product could be obtained in moderate yield. Since the desired carboxylic acid was highly polar and difficult to isolate, the crude product was directly subjected to the EDCI condensation. Delightfully, the desired redox-active-ester
27 could be obtained in 53% yield. We then proceeded to perform a C12-OH elimination reaction to introduce the Δ
11(12) double bond. Delightfully, with Burgess reagent, the desired product
21 was obtained in 52% yield. Other elimination conditions, such as SOCl
2/pyridine and POCl
3/pyridine, were found to be unsuccessful (see Supporting information for details).
We then proceeded to investigate the key decarboxylative Minisci reaction. Initially, the photoredox-catalyzed direct oxidative decarboxylative Minisci reaction from the carboxylic acid was also tested [
38–
40], but it failed to yield the desired product. Consequently, we shifted to redox-active-ester
21. As shown in
Fig. 4, after screening a series of conditions, the desired product
20 was successfully synthesized in 68% yield, accompanied by 18% of the regioisomer
20′, using Ir[dF(CF
3)ppy]
2(dtbbpy)PF
6 as the photocatalyst, TFA (1.5 equiv.) as the acid, and DMA as the solvent under blue LED irradiation (Entry 1). Using Ir(ppy)
3 as the photocatalyst resulted in a lower yield of
20 and reduced regioselectivity, whereas photocatalysts Ru(bpy)
3Cl
2 and Eosin Y showed no reactivity (entries 2–4). Both reducing or increasing the amount of TFA resulted in lower yields (entries 5 and 6). Replacement of TFA with TsOH or H
2SO
4 also decreased the yield (entries 7 and 8). Switching the solvent to DMF resulted in a decrease in the yield and regioselectivity, whereas MeCN provided a comparable yield and improved regioselectivity (entries 9 and 10).
With the advanced intermediate
20 in hand, we proceeded to finish the divergent synthesis of
1 and
2 (
Fig. 5). After IBX/NMO oxidation [
41],
20 was converted to Δ
4(5)-3-one product
28 in 54% yield. Then, after enolization with
t-BuOK/
t-BuOH and subsequent quenching in acidic conditions, the Δ
4(5) double bond underwent transposition to Δ
5(6) [
30]. Finally, reduction of the C3-ketone with NaBH
4 afforded the desired natural product (-)-veragranine A (
1) in 83% yield. Next, we focused on the synthesis of
2. Treatment of
20 with NBS/HOAc in aqueous THF solution, followed by hydrolysis of the crude bromohydrin acetate afforded
β-epoxide
29 in 68% yield. Subsequent IBX/NMO oxidation yielded Δ
4(5)-3
-one product
30 in 54% yield. The Et
2AlCl-promoted Meinwald rearrangement afforded the C11-ketone product
31 in 82% yield [
20]. Subsequently, after a similar Δ
4(5) double bond transposition to Δ
5(6) and selective reduction of the less hindered C3-ketone with NaBH
4, the desired natural product (-)-veragranine B (
2) was obtained in 64% yield.
Prior to initiating our biological assays, we noted that Dai
et al. had concurrently reported their synthesis of veragranine A and analogs [
21]. To secure the novelty of our research and facilitate early-stage scholarly communication, we therefore made our synthetic routes to veragranines A and B publicly available as a preprint on ChemRxiv [
42]. Subsequently, with (-)-veragranines A/B, 5
α,6-dihydroveragranine A and their analogs in hand (
Fig. 6, top), we performed a comprehensive evaluation of their bioactivity to explore their therapeutic potential. The detailed synthetic procedures for analogs
32–35 are provided in Supporting information. Building on prior reports demonstrating that veragranines A and B potently inhibit Ca
V2.2 channels [
15], key mediators of nociceptive signaling, we employed whole-cell patch-clamp recordings to evaluate the inhibitory effects of a panel of synthetic derivatives. Initial screening revealed compound
16 as the most potent inhibitor, exhibiting activity comparable to the natural compound veragranine B (
Figs. 6A and B). Concentration-response analysis revealed that
16 suppressed Ca
V2.2 currents in a concentration-dependent manner (
Fig. 6C), with a half-maximal inhibitory concentration (IC
50) of 15.96 ± 1.71 µmol/L at a holding potential of −80 mV (
Fig. 6D). Notably, the IC
50 of
16 was intermediate between those of veragranines A and B, positioning it as a synthetic analog with balanced efficacy (
Figs. 6D and E).
Steady-state current-voltage (
I-V) relationship studies further demonstrated that
16 did not alter the voltage sensitivity of Ca
V2.2 channels, as indicated by the unperturbed activation curve (
Figs. 6F-H). This behavior mirrored the effects of
16 and veragranines, indicating a shared mechanistic basis involving direct interaction with the channel pore region to block ion permeation, rather than modulation of voltage-gated domains. Combining high synthetic efficiency, a streamlined preparation route, and selective Ca
V2.2 antagonism,
16 emerges as a promising lead compound for the development of targeted therapeutics against Ca
V2.2-dependent pathologies.
In conclusion, we have explored two distinct strategies for the synthesis of veragranines. The first approach enabled the concise synthesis of 5α,6-dihydroveragranine A in 5 steps (14.5% overall yield) from hecogenin acetate, featuring: (1) Two well-designed cascade processes that significantly improve the synthetic efficiency; (2) Differing from the proposed SN2-type biogenetic pathway, we proposed an aldol-type condensation to form the α,β-unsaturated imine, which was successfully converted into veragranines through oxidative aromatization and double bond transposition, thereby supporting this alternative biosynthetic route. The second strategy provided the first divergent synthesis of veragranines A and B in 10–12 steps (6%–2.6% overall yield) from deoxycholic acid, featuring: (1) Photoredox-catalyzed decarboxylative Minisci reaction for E ring closure with C20 stereoinversion; (2) Late-stage introduction of the Δ5(6) double bond via a strategic A/B cis-fused intermediate, circumventing regioselectivity challenges between the Δ11(12) and Δ5(6) double bonds and enabling the first synthesis of (–)-veragranine B. After further bioactivity studies of veragranines and their 10 analogs, we are delighted to identify the easily accessible compound 16 as a promising lead compound. Its dual advantages of mechanistic clarity and synthetic feasibility establish it as a versatile scaffold for probing CaV2.2 biology and accelerating analgesic drug discovery.
PDF
Full
89
47
关闭全屏
No. 86 Xueyuan South Road, Haidian District, Beijing
010-62662699(Operation and Technical Support)
qkjq@cast.org.cn
