Structural Revision and Total Synthesis of Azaspiracid‐1, Part 2: Definition of the ABCD Domain and Total Synthesis

Abstract

The chase for the structure of azaspiracid-1 has ended with a revised structure (see picture). The total synthesis of this challenging target has closed another chapter in this intriguing chemical detective story and allowed the assignment of the relative and absolute configuration of the molecule. The preceding communication in this issue1 detailed studies that led to the absolute structural definition of the EFGHI segment (C21–C40 domain) of azaspiracid-1 (see one of the four originally proposed structures, 1-i, Figure 1), but left the final structural assignment for the ABCD framework (C1–C20 domain) as an open question, despite considerable effort to resolve this issue. Herein we present the remainder of the azaspiracid-1 saga that led to the complete definition of the ABCD ring framework. The total synthesis of azaspiracid-1 (1-c) has finally revealed the long-sought structure of this marine neurotoxin, rendering it readily available for biological investigations. Originally (1-i) and newly (1-c) proposed structures of azaspiracid-1, and unsuccessful attempts to match synthetic materials (2 and 3) with naturally derived ABCD fragment. In recapping the studies on the ABCD domain of azaspiracid-1 described in the preceding paper,1 we note that the degradatively derived fragment did not match either structure 2 (Δ8,9, Figure 1), which was designed on the basis of the originally proposed structure of azaspiracid-1 (1-i), or compound 3 (Δ7,8, Figure 1), whose adoption as possible candidate for the correct architecture was based on the structure of lissoketal. These findings still left us with a multiple choice as to which stereoisomer of structure 3 (which by now became our favorite representation of the coveted subtarget molecule) to pursue next. However, based on our knowledge at the time, we could decrease the reasonable isomer possibilities to a small number that included compound 4 (Figure 2). This isomer appeared to benefit from a double anomeric effect (based on manual molecular modeling) and to fulfil the crucial requirement for the NOE effect between 6-H and the methyl group at C14 (see Figure 2) observed for the natural product2 and its ABCD-domain degradation product.1 It was precisely because of this latter observation that compound 4 became our next priority target. Its attempted synthesis (Scheme 1) was met with unexpected, but nevertheless, path-pointing observations. Thus, substrate 9 was prepared from building blocks3 6, 7, and 8 according to our previously reported route,3 with the appropriate modification of building block 7 to accommodate the new stereochemical requirement at C14 in subtarget 4. The intermediate 9 was then subjected to deprotection–cyclization cascade conditions.3 Interestingly, however, the major product formed in this double spiroketalization process (9→10) was that with the opposite stereochemistry at C13 from the one proposed in the C14-epi target intermediate 4. This rather distressing signal became even more ominous four steps further along the sequence when an attempt was made to correct the stereochemistry at C13. Thus, when intermediate 11 (obtained from 10 by 1) pivaloate formation; 2) dithiane cleavage; and 3) reduction with sodium borohydride) was exposed to TFA, an equilibrium was reached in which the isomer 12 with the isomerized C13 center predominated (≈55:45). Although NOE studies revealed a strong effect between 6-H and the methyl group at C14 in 12, the chemical shift of the latter signal differed significantly from the corresponding signal in the natural product, suggesting the axial orientation for this methyl substituent (C14). Indeed, the observed thermodynamic stability of azaspiracid-1 under acidic conditions2 did not bode well for isomer 4 as the correct representation of the ABCD domain of the natural product. Its construction, therefore, was abandoned at this stage in favor of another synthesis, this time directed toward the next suspected structure, namely 5 (Figure 2). The latter structure, in which the C6 stereocenter is inverted, also enjoys the thermodynamic stability (manual molecular modeling) derived from two anomeric effects, provided that the two spirocenters are inverted into the opposite configurations as shown in Figure 2 (structure 5). Third and fourth proposed structures (4 and 5) of the ABCD domain of azaspiracid-1 with expected anomeric effects and NOE interactions. On the way to proposed intermediate 4 (C14-epi): a) TMSOTf (4.0 equiv), CH2Cl2, −90 °C, 30 min, 89 %; b) PivCl (3.0 equiv), pyridine (10.0 equiv), 4-DMAP (0.1 equiv), CH2Cl2, 0→25 °C, 12 h; 95 %; c) NBS (8.0 equiv), 2,6-lutidine (16 equiv), MeCN/H2O (4:1), 25 °C, 1 h, 80 %; d) NaBH4 (1.1 equiv), MeOH, −78→−60 °C, 3 h, 92 %; e) TFA (2.0 equiv), CH2Cl2, 0 °C, 1.5 h, 98 % (12/11≈55:45). TMS=trimethylsilyl, Piv=trimethylacetyl, 4-DMAP=4-(dimethylamino)pyridine, NBS=N-bromosuccinimide, TFA=trifluoroacetic acid. Compound 5 was synthesized stereoselectively as shown in Scheme 2. Thus, starting with the ketophosphonate 6 derived from L-malic acid (natural), and following our previously described route,3 intermediate 13 was prepared in 12 steps in good overall yield. To accommodate the requirements of the new subtarget 5, a substantially modified sequence from 13 to 5 had to be developed. To facilitate the pending cyclization to the ABCD tetracycle (see below), the TBS group of 13 was exchanged with a TES group, followed by cleavage of the double bond to give aldehyde 14 in excellent overall yield. Coupling of 14 with dithiane 15 (also derived from L-malic acid)3 under the influence of nBuLi afforded the corresponding secondary alcohol as a mixture of diastereoisomers (≈1:1, 98 % yield) whose oxidation with DMP transformed it into the required cyclization precursor 16 in 90 % yield. The anticipated deprotection–double spirocyclization of 16 proceeded smoothly under the influence of TMSOTf in CH2Cl2 at −90 °C to furnish the desired polycycle 17 in 89 % yield as a single stereoisomer. The observed NOE interaction between 6-H and C14-Me confirmed the anticipated structure 17 and provided support for our proposition of its thermodynamic stability. To decrease the number of steps required for the side-chain attachment, an olefin metathesis protocol was applied instead of the previously reported procedure.1 Thus, Swern oxidation of alcohol 17 followed by Wittig olefination of the resulting aldehyde furnished the desired terminal olefin 19 in 81 % overall yield. Cross-metathesis of 19 with olefin 21 was effected by the action of Grubbs catalyst 20,4 yielding the corresponding E olefin as the major product, from which the minor Z olefin was easily separated chromatographically (95 % yield total after three cycles, ≈60 % conversion in each cycle, E/Z≈10:1). The dithiane moiety was then removed from the resulting olefin by exposure to PhI(OCOCF3)2, leading to ketone 22 in 90 % yield. The introduction of the required α,β-unsaturation within 22 was successfully carried out by the Mukaiyama method (LDA–tBuNSPhCl)5 to afford 23 in 81 % yield. Regio- and stereoselective reduction of the enone functionality of 23 with NaBH4–CeCl3, followed by exposure to ClCO2Me in the presence of pyridine, led to allylic carbonate 24 through the intermediacy of the corresponding alcohol in 70 % overall yield. Compound 24 served well as a precursor to olefin 25 under a palladium-catalyzed reductive protocol ([Pd2dba3]⋅CHCl3, nBu3P, LiBH4, DME),6 furnishing the desired Δ7,8 olefinic compound 25 as the predominant product (82 % yield, Δ7,8/Δ8,9≈7:1; the undesired Δ8,9 isomer was chromatographically removed at this stage). The key intermediate 5 was then generated from 25 in 71 % overall yield by a five-step sequence: 1) DIBAL-induced cleavage of the pivaloate group; 2) Swern oxidation of the resulting alcohol to the aldehyde; 3) further oxidation to the corresponding carboxylic acid; 4) methylation with diazomethane; and 5) desilylation with TBAF (Scheme 2). Synthetic 5 matched the degradatively derived material by 1H NMR spectroscopy.7 This was indeed a pleasing moment in the azaspiracid-1 campaign, although the minute supplies of the degradation product did not permit optical rotation comparisons, and, therefore, the absolute stereochemistry of 5 still remained elusive. This gap in our intelligence gathering dictated the synthesis of the other enantiomer of 25 as well (ent-25, see Scheme 3), a task that was easily accomplished starting from the enantiomer of 6 (prepared from D-malic acid, see Scheme 2) and proceeding in the manner described above for 25. Synthesis of key intermediate 5 and confirmation of the structure of the ABCD domain. Reagents and conditions: a) TBAF (1.0 M in THF, 3.0 equiv), THF, 25 °C, 48 h, 99 %; b) TBDPSCl (1.2 equiv), Et3N (3.0 equiv), 4-DMAP (0.1 equiv), CH2Cl2, 0 °C, 12 h, 90 %; c) TESCl (1.5 equiv), imidazole (3.0 equiv), 4-DMAP (0.1 equiv), CH2Cl2, 0 °C, 30 min, 95 %; d) OsO4 (0.03 equiv), NMO (2.0 equiv), tBuOH/THF/H2O (10:2:1), 25 °C, 12 h; then NaIO4 (5.0 equiv), pH 7 buffer, 25 °C, 5 h, 95 %; e) nBuLi (1.6 M in hexanes, 3.5 equiv), 15 (3.5 equiv), THF, −30 °C, 30 min, 98 %; f) DMP (1.2 equiv), py (10.0 equiv), CH2Cl2, 0 °C, 1 h, 90 %; g) TMSOTf (4.0 equiv), CH2Cl2, −90 °C, 30 min, 89 %; h) (COCl)2 (5.0 equiv), DMSO (11 equiv), CH2Cl2, −78→−60 °C, 1.5 h; then Et3N (22 equiv), −25 °C; i) Ph3PCH3+Br− (5.0 equiv), nBuLi (4.0 equiv), THF, −78→0 °C, 1 h; then −78 °C; then 18, −78→0 °C, 30 min, 81 % from 17; j) 21 (3.0 equiv), 20 (0.1 equiv), CH2Cl2, 40 °C, 12 h, 60 %, E/Z≈10:1; k) PhI(OCOCF3)2 (3.0 equiv, slow addition), MeCN/pH 7 buffer (4:1), 0 °C, 30 min, 90 %; l) LDA (1.5 equiv), tBuNSPhCl (1.5 equiv), THF, −78 °C, 10 min, 81 %; m) NaBH4 (3.1 equiv), CeCl3⋅7 H2O (1.0 equiv), MeOH, −65 °C, 40 min, 80 %; n) ClCO2Me (25 equiv), 4-DMAP (1.0 equiv) CH2Cl2:py (2:1), −10 °C, 3 h, 87 %; o) [Pd2dba3]⋅CHCl3 (0.125 equiv), nBu3P (0.48 equiv), LiBH4 (10.0 equiv), DME, 0 °C, 1 h, 82 %, Δ7,8/Δ8,9=7:1; p) DIBAL-H (1.0 M in CH2Cl2, 2.0 equiv), CH2Cl2, −78 °C, 15 min; q) (COCl)2 (5.0 equiv), DMSO (11 equiv), CH2Cl2, −78→−60 °C, 1.5 h; then Et3N (22 equiv), −25 °C; r) NaClO2 (10.0 equiv), NaH2PO4 (10.0 equiv), 2-methyl-2-butene (excess), tBuOH/H2O (4:1), 25 °C, 1 h; s) CH2N2 (excess), Et2O, 25 °C; t) TBAF (1.0 M in THF, 2.0 equiv), THF, 25 °C, 1 h, 71 % from 25. TBAF=tetra-n-butylammonium fluoride, TBDPS=tert-butyldiphenylsilyl, TES=triethylsilyl, NMO=N-methylmorpholine N-oxide, DMP=Dess–Martin periodinane, Tf=trifluoromethanesulfonyl, DMSO=dimethyl sulfoxide, LDA=lithium diisopropylamide, dba=dibenzylideneacetone, DME=1,2-dimethoxyethane, DIBAL-H=diisobutylaluminum hydride. Construction of key intermediate 32. Reagents and conditions: a) TBAF (1.0 M in THF, 2.0 equiv), THF, 25 °C , 2 h, 98 %; b) (COCl)2 (5.0 equiv), DMSO (11 equiv), CH2Cl2, −78 °C, 30 min; then Et3N (22 equiv), −78→0 °C; c) NaClO2 (6.0 equiv), NaH2PO4 (6.0 equiv), 2-methyl-2-butene (excess), tBuOH/H2O (4:1), 25 °C, 1 h, 90 % over two steps; d) PFPOH (1.5 equiv), DCC (2.0 equiv), CH2Cl2, 25 °C, 2.0 h, 84 %; e) 28 (7.0 equiv), nBuLi–nBu2Mg (1.1 M in hexanes, 4.7 equiv), THF, 0→25 °C, 1.5 h; then −90 °C; then 27, 30 min, 50 %; f) DIBAL-H (1.0 M in CH2Cl2, 10.0 equiv), CH2Cl2, −90 °C, 1 h; g) TBAF (1.0 M in THF, 2.0 equiv), THF, 25 °C, 8 h, 44 % over two steps; h) AcCl (20 equiv), 2,4,6-collidine (40 equiv), CH2Cl2, −78 °C, 6 h, 90 %; i) TBSOTf (10.0 equiv), 2,6-lutidine (20 equiv), CH2Cl2, −78→0 °C, 30 min, 86 %; j) PhI(OCOCF3)2 (2.0 equiv), MeCN/pH 7 buffer (4:1), 0 °C, 10 min, 68 %. PFP=pentafluorophenyl, DCC=dicyclohexylcarbodiimide, TBS=tert-butyldimethylsilyl. Having secured adequate supplies of both enantiomers of 25 and being assured that one of them corresponded to the ABCD domain of our target natural product, we set out to reach the final target. The first enantiomer to reach the end was that derived from non-natural D-malic acid (i.e. ent-25), only to find that it did not correspond to azaspiracid-1. The final drive toward azaspiracid-1 from enantiomer 25 (derived from L-malic acid) was then undertaken in earnest, following the convergent strategy previously employed in these laboratories for the synthesis of the originally proposed structure (1-i). Owing to the problems that arose in the final deprotection step of our previous route8 to azaspiracid-1, we opted to adopt a TBS protecting group for the hydroxy group of C25, as opposed to the originally employed acetate group. The new sequence toward advanced intermediate 32 (see Scheme 3), which was required for the planned palladium-catalyzed coupling reaction with the FHI segment, began with intermediate 25 and proceeded along the path delineated in Scheme 3. Thus, the TBDPS protecting group was removed from 25 by exposure to TBAF to reveal the primary alcohol (98 % yield), which was oxidized in a two-step procedure to carboxylic acid 26 in 90 % overall yield. Compound 26 was converted into its pentafluorophenyl ester 27 by DCC-facilitated coupling with pentafluorophenol in 84 % yield. The activated form of dithiane 28 (nBuLi–nBu2Mg) was then treated with 27 at −90 °C to generate the coupled product 29 in 50 % yield. A two-step sequence involving excess DIBAL-H followed by TBAF was then used to convert 29 into 30 in 44 % overall yield. The indicated stereochemistry at C20 was assigned to tetraol 30 by analogy to our previous results8, 9 in the synthesis of the originally proposed azaspiracid-1 structure. The activation–protection required for 30 prior to coupling with the FHI segment was introduced by selective acetylation of its primary hydroxy groups (90 % yield) and silylation of the C25-OH group (86 % yield) to afford intermediate 31. Final unmasking of the keto group of 31 by treatment with PhI(OCOCF3)2 furnished the advanced C1–C27 fragment 32 (68 % yield) ready for further elaboration. The second crucial coupling in the synthesis, that between the ABCD (C1–C27) and FHI (C28–C40) fragments 32 and 33, was brought about by the action of [Pd2dba3] catalyst in the presence of LiCl, AsPh3, and EtNiPr2 and led to intermediate 34 (55 % yield), which contains all the carbon atoms required for azaspiracid-1, but lacks rings G and E (Scheme 4). Ring G was cast first through the intervention of TBAF to remove the TES group selectively followed by ring closure as induced by NIS and reductive removal of the residual iodine atom (nBu3SnH–Et3B) to afford octacyclic compound 36 through intermediate iodide 35 in 43 % overall yield. The construction of ring E was left until last, and not before all the necessary functionality for the final target was in place. Thus, a TES group was called upon to temporarily protect the C20-OH group of compound 36 (91 % yield); removal of the acetate group from C1 with K2CO3 in MeOH furnished the corresponding primary alcohol (81 % yield), which was sequentially oxidized, first to the aldehyde and then to the carboxylic acid, to afford the penultimate precursor, compound 37 (70 % over two steps). Finally, all three remaining protecting groups on 37 (TES, TBS, and Teoc) were removed concurrently by exposure to excess TBAF in THF at ambient temperature to reveal the desired structure 1-c, which was spontaneously formed upon collapse of the liberated C25-OH group onto the carbonyl functionality at C21 (Scheme 4). This time, the TLC profile of the synthetic material matched that of natural azaspiracid-1 obtained from the Yasumoto–Satake group.7 More convincingly, and delightfully, all data (1H NMR, 13C NMR, MS, TLC, optical rotation) for the synthetic and natural samples of azaspiracid-1 were identical.7 Total synthesis of azaspiracid-1 (1-c). Reagents and conditions: a) 32, [Pd2dba3] (0.3 equiv), AsPh3 (0.3 equiv), LiCl (6.0 equiv), EtNiPr2 (12 equiv); then 33 (3.0 equiv, 0.03 M in THF, syringe pump addition), NMP, 40 °C, 1 h, 55 %; b) TBAF (1.0 M in THF, 1.2 equiv), THF, 0 °C, 20 min, 80 %; c) NIS (2.0 equiv), NaHCO3 (10.0 equiv), THF, 0 °C, 12 h, 62 %; d) Et3B (1.0 M in hexanes, 0.2 equiv), nBu3SnH/toluene (1:2), 0 °C, 5 min, 86 %; e) TESOTf (10.0 equiv), 2,6-lutidine (20 equiv), CH2Cl2, −78→0 °C, 10 min, 91 %; f) K2CO3 (2.0 equiv), MeOH, 25 °C, 1 h, 81 %; g) (COCl)2 (10.0 equiv), DMSO (20 equiv), CH2Cl2, −78 °C, 1 h; then Et3N (50 equiv), −78→−20 °C, 30 min; h) NaClO2 (10.0 equiv), NaH2PO4 (10.0 equiv), 2-methyl-2-butene (excess), tBuOH/H2O (4:1), 25 °C, 30 min, 70 % over two steps; i) TBAF (1.0 M in THF, 5.0 equiv), THF, 25 °C, 15 h, 60 %. NIS=N-iodosuccinimide, Teoc=2-(trimethylsilyl)ethoxycarbonyl, NMP=N-methylpyrrolidinone. Thus, the long-sought structural elucidation of azaspiracid-1, the notorious marine neurotoxin originally isolated from mussels (Mytilus edulis) collected in European waters, ended with its total synthesis, an accomplishment that also renders this valuable, but scarce, substance available for much-needed biological investigations. Among such investigations, those directed towards the development of assays for its detection and measurement in seafood and seawater and the elucidation of its mechanism of action must rank high on the list. 5: Rf=0.24 (silica gel, EtOAc/hexanes 2:1); [α]=−2.7 (CHCl3, c=0.45); IR (film): max=3413, 2929, 2871, 1739, 1433 cm−1; 1H NMR (500 MHz, CDCl3): δ=5.77–5.69 (m, 2 H), 5.63 (dt, J=10.5, 1.0 Hz, 1 H), 5.49 (dd, J=15.5, 7.5 Hz, 1 H), 4.80 (br s, 1 H), 4.35 (ddt, J=6.0, 5.5, 3.0 Hz, 1 H), 4.22 (m, 1 H), 3.90 (m, 1 H), 3.73 (dd, J=7.0, 3.0 Hz, 1 H), 3.67 3 H), (dd, J=7.0, Hz, 1 H), Hz, 1 H), (m, 3 H), (dt, 7.5 Hz, 1 H), (dt, Hz, 1 H), (m, 2 H), (dt, 7.5 Hz, 1 H), (dd, Hz, 1 H), (m, 4 H), (dd, Hz, 1 H), Hz, 3 13C NMR MHz, CDCl3): for (silica gel, EtOAc/hexanes IR (film): cm−1; 1H NMR (500 MHz, (m, 1 H), (dt, J=10.5, 1.0 Hz, 1 H), 5.63 (dt, J=10.5, 1.0 Hz, 1 H), (ddt, J=15.5, 1.0 Hz, 1 H), Hz, 2 H), (br s, 1 H), Hz, 1 H), Hz, 1 H), Hz, 1 H), (dt, Hz, 1 H), (br s, 1 H), Hz, 2 H), (br s, 1 H), Hz, 1 H), Hz, 1 H), Hz, 1 H), Hz, 1 H), (dt, 7.5 Hz, 1 H), (m, 5 H), 3 H), 3 H), (m, 3 H), (m, 1 H), (dd, Hz, 1 H), (m, 3 H), (dd, Hz, 2 H), (br s, 1 H), Hz, 3 H), 9 H), Hz, 3 H), Hz, 3 H), 3 H), 3 13C NMR MHz, for (silica gel, EtOAc/hexanes IR (film): cm−1; 1H NMR MHz, (m, 1 H), (m, 2 H), (dd, Hz, 1 H), (br s, 1 H), (br s, 1 H), (br s, 1 H), (m, 1 H), (m, 2 H), Hz, 1 H), (br s, 1 H), (m, 2 H), Hz, 2 H), (m, 1 H), (br s, 1 H), (br s, 1 H), 3.73 Hz, 1 H), Hz, 1 H), Hz, 1 H), (m, 1 H), (m, 1 H), (m, 1 H), (dt, 1 H), (m, 7 H), 3 H), (m, 2 H), (m, 1 H), (dd, Hz, 1 H), (m, 2 H), (m, 4 H), (m, 6 H), Hz, 2 H), (dt, Hz, 1 H), Hz, 2 H), Hz, 3 H), (m, 2 H), Hz, 3 H), 9 H), Hz, 3 H), Hz, 3 H), Hz, 3 H), Hz, 3 H), 9 H), 0.03 3 H), 3 13C NMR MHz, for (silica gel, 2:1); IR (film): cm−1; 1H NMR (500 MHz, (m, 2 H), Hz, 1 H), (dd, J=15.5, 7.5 Hz, 1 H), (br s, 1 H), (br s, 1 H), (m, 2 H), Hz, 1 H), (m, 1 H), (br s, 1 H), Hz, 1 H), (dd, J=10.5, Hz, 1 H), (m, 1 H), (m, 1 H), (br s, 1 H), Hz, 1 H), (m, 1 H), Hz, 1 H), Hz, 1 H), Hz, 1 H), (m, 1 H), (m, 3 H), (m, 4 H), (m, 3 H), (m, 1 H), (m, 2 H), (dd, Hz, 1 H), (m, 1 H), (m, 1 H), (m, 3 H), (m, 2 H), (m, 3 H), (m, 2 H), (m, 4 H), Hz, 3 H), 3.0 Hz, 2 H), Hz, 3 H), Hz, 9 H), 9 H), Hz, 3 H), Hz, 3 H), Hz, 3 H), Hz, 3 H), Hz, 6 H), 9 H), 3 H), 3 13C NMR MHz, for