Synthesis of Anhydro-dc-Tetracyclic Saxitoxin Via a Photoinduced Cycloaddition

Abstract

(+)-Saxitoxin, a potent and reversible blocker of voltage-gated sodium channels, has attracted considerable interests as a scaffold for the development of novel analogs. Here, we report the design and synthesis of a tetracyclic analogue featuring an additional cis-fused five-membered ring (C5–C6), constructed via a novel photoinduced radical cycloaddition reaction. This transformation efficiently established the quaternary carbon center at C5, which is difficult to access by conventional methods. Although the IC₅₀ values of two analogues against hNa_V1.4 showed a significant decrease in potency, this work introduces a new chemotype of saxitoxin, offering a foundation for future optimization efforts.

Voltage-gated sodium channels (Na_Vs) belong to the cation channel family − and are widely distributed across neuronal, skeletal muscle, and cardiac muscle cell membranes. In humans, ten Na_V isoforms (Na_V1.1 to Na_V1.9 and Na_X) have been identified, each exhibiting distinct tissue distributions. Dysfunction of specific Na_V subtypes is associated with severe diseases, such as epilepsy (Na_V 1.1 and 1.2), periodic paralysis (Na_V 1.4), and pain (Na_V 1.7)conditions that affect millions of patients and impose substantial economic burdens annually. To address these disorders, the discovery of small-molecule inhibitors with high subtype selectivity has attracted considerable interest. However, currently available small-molecule Na_V inhibitors, including clinical local analgesics (e.g., lidocaine, carbamazepine) and aryl/acyl sulfonamide derivatives (e.g., PF-05089771, GX-936), typically act as state- and frequency-dependent blockers, often accompanied by undesirable pharmaceutical properties such as high plasma protein binding and off-target activity. In contrast, naturally occurring Na_V inhibitors like tetrodotoxin (TTX) exhibit state-independent channel blockade with remarkable selectivity. Structural modification of these natural toxins thus represents a promising strategy for developing potent and selective Na_V inhibitors.

Saxitoxin (STX, 1a, Figure ), a paralytic shellfish toxin, is a potent inhibitor Na_Vs. Unlike TTX, STX demonstrates the ability to distinguish Na_V1.7 from other isoforms, a selectivity largely attributable to the Na_V1.7-specific two amino acid sequence motif in domain III (T1398/I1399), which reduces the IC₅₀ of STX by 250-fold compared to isoforms such as Na_V1.4. This unique feature has motivated considerable interest in modifying STX to elucidate its structure–activity relationships (SAR) and binding modes across Na_V subtypes. Pioneering studies by the Du Bois and Nagasawa groups revealed that modifications at the C13 carbamate significantly influence Na_V1.7 binding affinity, , with the ability to form a hydrogen bond being essential. Additionally, Du Bois and co-workers also shown that substitutions on the guanidine moiety substantially impair binding affinity. The efforts to modify the C11 and C13 positions of STX by both groups demonstrated that alterations at these sites modulate both binding affinity and isoform selectivity (e.g., 2 and 3). , In contrast, modifications at the C5 position remain unexplored, likely due to the inherent chemical inertness at this site. This sparks our curiosity to investigate how C5 modification would affect STX activity toward Na_Vs. Given that dicarbamoyl saxitoxin (dc-STX, 1b) retains potency comparable to STX (1a), we designed a tetracyclic analog, dc-tetracyclic STX 4, featuring a fused five-membered ring spanning C5–C6. This deep-seated structural modification was envisioned to conformationally constrain the C13 hydroxyl group into its bioactive orientation, potentially enhancing binding affinity.

1.

Saxitoxin, analogues, and the design of dc-tetracyclic STX.

Inspired by our prior total synthesis of STX (1a), we devised a retrosynthetic route to 4 via urea 5, involving guanidine installation and redox adjustments (Figure ). The additional five-membered ring fused across the C5–C6 bond was envisaged to arise from radical cyclization of hemiaminal 6, with an alkyne moiety serving as the radical acceptor. Conventional approaches for forging the C5 quaternary centersuch as alkylation of 7were deemed unfeasible due to prohibitive ring strain upon enolization (e.g., enolate 8). Instead, we proposed accessing 6 from 9, which could be derived from ozonolytic cleavage of the cycloheptene ring in compound 10 Z. The synthesis of 10 Z was planned via an intramolecular [5 + 2]-type cycloaddition between an 8-oxoxanthine and a vinylcyclopropane (VCP), leveraging the excited state of intermediate 11. In this strategic disconnection, the VCP unit served as a key synthon, enabling the formation of two adjacent quaternary centers at C4 and C5 in a single step. While metal-catalyzed [5 + 2] cycloadditions of VCPs with π-systems are well-established, intramolecular [5 + 2]-type cycloadditions initiated from the triplet-state of heterocycles remain rare, with only one precedent reported by You, Zheng, and co-workers.

2.

Retrosynthetic analysis of dc-tetracyclic STX (4).

Our synthesis commenced with the N ³-alkylation of 7-benzyl-8-bromoxanthine (12) using iodide 13, followed by sequential N ¹-benzylation and hydrolytic cleavage of the C8–Br bond in a one-pot fashion (Scheme ). This afforded the photocycloaddition substrate 11 as a 1:1 mixture of inseparable Z/E olefin configurational isomers in 45% yield. Irradiation of 11 at 300 nm in toluene at room temperature promoted an intramolecular [2 + 2] cycloaddition, yielding product 14 as a 3.5:1 mixture of inseparable diastereomers in 70% yield (Table , entry 1). While DCM as solvent slowed the reaction but maintained a comparable yield of 14 (entry 2), acetone completely suppressed the transformation (entry 3). Notably, when xanthone was employed as a photosensitizer, we isolated 15 as the only identifiable product, which was unambiguously characterized by X-ray crystallography; full conversion was achieved using 0.5 equiv of xanthone, affording 15 in 48% isolated yield (entry 4). In contrast, substitution with benzophenone as the photosensitizer resulted in only a 10% yield of 15 (entry 5). Mechanistically, we propose that 11 underwent energy transfer to generate triplet-state A (Figure ), which then underwent a 5-exo radical cyclization followed by cyclopropane ring opening to form biradical intermediate B. Following intersystem crossing (ISC) to the singlet state, ring-closing radical–radical recombination would furnish either 10 E or 10 Z, though the formation of 10 E might be disfavored due to the strained E-olefin geometry in a seven-membered ring. Nonetheless, subsequent hydrogen atom abstraction of 10 by exited-state xanthone could generate allylic radical C, enabling alkene migration and ultimately producing 15. Further mechanistic studies are warranted to validate this hypothesis.

1. Total Synthesis of Anhydro-dc-Tetracyclic Saxitoxin (26) .

^a Reagents and conditions: (a) 13, K₂CO₃, DMF, rt, 15 h then BnBr, K₂CO₃, DMF, rt, 8 h then KOH, EtOH/H₂O, 95 °C, 18 h, 45%. (b) Xanthone, DCM, 302 nm, rt, 24 h, 48%. (c) O₃, MeOH, −78 °C, 20 min, then FeSO₄·7H₂O, TEMPO, rt, 2 h, 35%. (d) DIBAL, DCM, −78 °C, 30 min, 52%. (e) TBDPSCl, DCM, imdazole, 0 °C, 30 min, 72%. (f) mCPBA, DCM, 0 °C, 30 min, 81%. (g) Bestmann-Ohira reagent, K₂CO₃, MeOH, 0 °C to rt, 5 h, 85%. (h) LiBH₄, THF/MeOH, 0 °C, 30 min, 88%. (i) Cp*TiCl₃, TESCl, 4 Å MS, Zn, THF, 55 °C, 15 h, 72%. (j) O₃, DCM, −95 °C then NaBH₄, MeOH, 0 °C, 71% (21a/21b = 1:3). (k) MOMBr, TBAI, DCE, 60 °C, 2 h then TBAF, THF, rt, 30 min, 83%. (l) MsCl, NEt₃, DCM, 0 °C, 2 h, quant. (m) DBU, NaI, DME, 100 °C, 18 h, 88%. (n) DTBBP, Li, THF, 0 °C, 2 h, 73%. (o) MeOTf, TTBP, DCM, rt, 24 h then EtCOONH₄, NH₃, MeOH, 65 °C, 36 h. (p) TFA, DCM, rt, 12 h, 54% (2 steps). (q) O₃, MeOH, −95 °C, 10 min then Me₂S, rt, 2 h, 87%. (r) O₃, MeOH, −95 °C, 10 min then Me₂S, rt, 2 h then TFA, DCM, rt, 12 h, 43% (2 steps). DIBAL, Diisobutylaluminum; TBAI, Tetrabutylammonium iodide; DTBBP, 4,4-di-tert-butylbiphenyl; TTBP, 2,4,6-tri-tert-butylpyrimidine.

1. Optimization of the Photocycloaddition.

Entry	Conditions	Yield
1	toluene, 302 nm, rt, 12 h	70% 14
2	DCM, 302 nm, rt, 24 h	68% 14
3	acetone, 302 nm, rt, 12 h	NR
4	0.5 equiv xanthone, 302 nm, DCM, rt, 10 h	48% 15
5	0.5 equiv Ph2CO, 302 nm, DCM, rt, 10 h	10% 15

^a[11] = 0.01 M (10 mg).

^bIsolated yield after flash chromatography. NR = no reaction.

3.

Proposed mechanism for the formation of 15.

The intramolecular [5 + 2]-type photocycloaddition reaction of 11 proceeded smoothly on a 500 mg scale to afford 15. We next investigated the oxodealkenylative cleavage of the seven-membered carbocycle in 15 via ozonolysis, followed by treatment with ferrous sulfate and TEMPO. Interestingly, the only isolable product was compound 16, which was obtained in a moderate 35% yield. The outcome suggested that, contrary to the typical preference for formation of the more-stable radical species during C–C bond cleavage, a less-stable primary radical was generated and subsequently trapped by TEMPO. This unusual selectivity may be attributed to the electron-deficient nature of C12, which disfavored the formation of the secondary carbon radical to be captured by TEMPO (Figure S1; see DFT calculations in the Supporting Information). The methyl ester of 16 was then reduced with DIBAL and protected as the TBDPS ether to give 17. Oxidation of the TEMPO moiety to the corresponding aldehyde, followed by a Bestmann–Ohira reaction, furnished alkyne 18 in 69% yield over two steps. Subsequent reduction of the C6-carbonyl group with LiBH₄ gave hemiaminal 19 in 88% yield. At this stage, the competing formation of a carbocation could potentially interfere with the desired homolytic C–O bond cleavage at C6. To address this, we employed Ti­(III) as a single-electron reductant, which both coordinated to and weakened the C–O bond while simultaneously promoting radical formation. This Ti­(III)-mediated radical cyclization proceeded smoothly to deliver tetracyclic intermediate 20 in 72% yield. Ozonolysis of 20 under −95 °C follow by NaBH₄ reduction produced separable epimers 21a and 21b in 17% and 54% yield, respectively, whereas the major epimer 21b bore the desired stereogenic center at C13. Subsequent MOM protection of the C13-OH, in situ TBDPS deprotection, and elimination of the resulting primary alcohol via mesylation/DBU-NaI treatment afforded 22 in 73% yield over three steps; the structure of 22 was confirmed by X-ray crystallography analysis. The removal of benzyl groups was achieved using Li/DTBBP, furnishing 23 in 73% yield. Final-stage transformations included O-alkylation of both ureas and aminolysis to form diguanidine 24, which was subjected to the MOM deprotection to give 25 in 54% yield over two steps. Ozonolysis of the C12 alkene in 25 afforded 26 in 87% yield. Notably, the crude product of 24 could alternatively undergo one-pot ozonolysis/MOM-deprotection to directly provide 26.

However, the ¹³C NMR analysis of compound 26 indicated that the C12 carbonyl group remained unhydrated. A comparable observation was recently been reported by Looper and co-workers, who conducted a detailed computational investigation on the hydration propensity of the C12 carbonyl group. Their results suggested that the conformation of the C4–C12 bond plays a critical role: compounds adopting a staggered arrangement of the C4 and C12 substituents readily undergo hydration, whereas those with an eclipsed conformation are less susceptible. To evaluate the conformation of 26, we performed density functional theory (DFT) calculations to refine its geometry: interestingly, in the absence of macrocyclic constrains, 26 adopted a staggered C4–C12 alignment (Figure S2; see Supporting Information for details). We thus hypothesize that the presence of the additional cis-fused five-membered ring hinders the approach of water molecules needed to stabilize the hydrated form. As a result, compound 26 preferentially remains in its carbonyl form, whereas dc-STX (1b) more readily exists in the hydrated state.

The Na_V-inhibitory activity of analogs 25 and 26 were evaluated using whole-cell patch-clamp recording in CHO cells expressing human Na_V1.4 (Figures S3 and S4), as STX is known to exhibit approximately 250-fold greater potency against Na_V1.4 compared to Na_V1.7. , The IC₂₅ values of analogue 25 were found to be >30 μM and 19 μM in the resting and half-inactivated states, respectively (Figure ). Analog 26 exhibited IC₂₅ values of 7.9 μM for the resting state and 1.3 μM for the half-inactivated state. For comparison, synthetic (±)-dcSTX (1b) was used as a control, displaying significantly higher potency, with IC₂₅ values of 0.049 μM in the resting state and 0.014 μM in the half-inactivated state.

4.

Bioactivity of racemic dc-STX (1b), 25, and 26.

In summary, we have designed a novel analogue of dc-STX and completed its racemic synthesis in 16 steps. A key feature of this route was the development of a photoinduced cycloaddition that simultaneously constructed the fused core and adjacent quaternary carbon centers. Subsequent transformations, including ozonolysis and Ti­(III)-mediated radical cyclization, enabled completion of the synthesis. Although the final compound (26) exhibited significantly lower potency compared to racemic dc-saxitoxin (1b), this study expands the structure–activity relationship understanding of saxitoxin and establishes a new synthetic platform for its future modification.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacsau.5c01028.

• Experimental procedures, characterization data, and NMR spectra of all newly synthesized compounds (PDF)

∥.

R. Fang and Y. Jiao contributed equally. CRediT: Runting Fang conceptualization, data curation, formal analysis, investigation, writing - original draft, writing - review & editing; Yang Jiao conceptualization, data curation, formal analysis, investigation, writing - original draft, writing - review & editing; Tianrun Xia conceptualization, data curation, investigation; Jiaqi Liu data curation, investigation; Tuoping Luo conceptualization, formal analysis, funding acquisition, project administration, supervision, writing - original draft, writing - review & editing.

The authors declare no competing financial interest.