Total Synthesis of (±)-N-Methyldibromoisophakellin and N-Methylugibohlin via Net [3+2] Cycloguanidinylations Employing 2-Amido-1,3-Diamino-Allyl Cations

Abstract

A concise total synthesis of (±)-N-methyldibromoisophakellin, a member of the monomeric pyrrole–aminoimidazole alkaloid family isolated from the marine sponge Stylissa carbica, was achieved via a net [3+2] cycloaddition to install the cyclic guanidine. This ring annulation employs a 2-amido-1,3-aminoallyl cation obtained under oxidative conditions from variously N-substituted guanidines which in one instance led to isolation of a tetracycle bearing a carbinolamine center through subsequent benzylic oxidation. Finally, the serendipitous formation of a unique, related alkenyl guanidine, N-methylugibohlin, achieved via ring opening of cyclic guanidine under acidic conditions suggests that ugibohlin may be an artifact of isolation.

Graphical Abstract

Introduction

Pyrrole-aminoimidazole alkaloids (PAIs) are one of the largest natural product families isolated from marine sponges derived from oroidin as a building block.¹ This family is categorized based on the number of oroidin monomers incorporated and whether cyclization has occurred including linear monomers (e.g. hymenidin),² cyclic monomers (e.g. dibromophakellin),³ cyclic dimers (e.g. palau’amine),⁴ cyclic tetramers (e.g. stylissadine A),⁵ and acyclic dimers (e.g. nagelamide A)⁶ PAIs. Biosynthetic studies have revealed that linear monomeric PAIs such as oroidin are the progenitors of more complex PAIs⁷ and oroidin is likely derived from the amino acids, ornithine proline and histidine.⁸ Moreover, PAIs have unique structures with broad bioactivities^1a–c,9 and these features have attracted the attention of synthetic chemists leading to several synthetic studies and completed total syntheses^{1b, c, e, f,10} including those reported by our group.¹¹

Dibromophakellin (1)³ is the first cyclic monomeric PAI isolated and subsequently (+)-dibromocantharelline (2, also known as (+)-dibromoisophakellin),¹² (−)-dibromoisophakellin (5a),¹³ and cylindradine A (3)¹⁴ were reported differing in the connectivity of the pyrrole ring (N1, C4, or C5; Figure 1). A number of other unique, cyclized monomeric PAIs have been isolated including the further oxidized (−)-dibromohydroxyphakellin (4),¹⁵ the apparently rearranged dibromoagelaspongin (7),¹⁶ and the alkenyl guanidine-containing ugibohlin (6a).¹⁷ Furthermore, variations in the degree of pyrrole bromination and differences in N-alkylation are found in (−)-monobromoisophakellin (5a),¹⁸ monobromougibohlins (6b, c),¹⁹ (−)-N-methyldibromoisophakellin (8),²⁰ and N-methylugibohlin (9).²¹ The first total synthesis of a monomeric PAI, (±)-dibromophakellin, was reported by Foley and Büchi in 1982.²² Subsequently, several total syntheses of cyclic monomeric PAIs have been reported,²³ including an elegant biomimetic synthesis of (±)-dibromophakellstatin and (±)-dibromoisophakellin by Horne²⁴ and synthesis of ugibohlin and (±)-N-methylisophakellin by Lindel²⁵ building on Horne’s work. We previously described total syntheses of (+)-phakellin and (+)-monobromophakellin,²⁶ (+)-dibromophakellstatin,²⁷ and a bioinspired strategy to (±)-agelastatin A.¹¹

Figure 1.

Tri- and tetracyclic pyrrole-aminoimidazole alkaloids (PAIs) with isomeric pyrrole connectivity and cyclic or acyclic guanidines.

In 1968, Sheehan described the thermal generation of aza-oxyallyl cations from α-lactams which engaged alkenes in [3+2] cycloadditions leading to γ-lactams.²⁸ Building on this early precedent, the Jeffrey²⁹ and Wu³⁰ groups reported [3+2] cycloadditions with alkenes employing the aza-oxyallyl cations generated under oxidative conditions for the synthesis of γ-lactams. In addition, Jeffrey extended this chemistry to the oxidative generation of diaza-oxyallyl cations for the synthesis of cyclic ureas from indole derivatives.³¹ Building on these precedents, we recently described the development of a net [3+2] cycloaddition for direct cyclic guanidine annulation onto alkenes employing 2-amido-1,3-diaminoallyl cations as 1,3-dipoles generated under oxidative conditions leading to a concise synthesis of (±)-phakellin.³² Herein, we describe the first total syntheses of (±)-N-methyldibromoisophakellin (10) employing the 2-amido-1,3-diaminoallyl cation and identify an over-oxidized isophakellin structure bearing a carbinolamine. Further, our studies reveal the ease at which dibromoisophakellin isomerizes to N-methylugibohlin (9) under acidic conditions raising the possibility that the ughibohlins could be an artifact of isolation. This type of elimination was previously proposed by Mourabit³³ and observed by Lindel.²⁵

Our synthetic strategy toward (±)-N-methyldibromoisophakellin (10) would utilize the 2-amido-1,3-diaminoallyl cation 12 for direct guanidine annulation onto the tricyclic dibromoenamide 13 in turn derived from prolinol and a dibromo pyrrole trichloroketone 14 (Scheme 1).

Scheme 1.

Retrosynthetic analysis of (±)-N-methyldibromoisophakellin (10) employing a [3+2] cycloguanidinylation.

Results and Discussion

The dibromopyrrole trichloroketone 14 was prepared from the commercially available 2-(trichloroacetyl)pyrrole in 2 steps according to the reported procedure³⁴ and was converted to alcohol 15 by treatment with L-prolinol under reflux conditions in CH₃CN (Scheme 2). The resulting alcohol was oxidized to the corresponding aldehyde by Dess–Martin oxidation³⁵ and converted to the tricyclic dibromoacylenamide 13 via an acid-catalyzed Friedel–Crafts alkylation and subsequent elimination.^{33, 36} With the desired substrate in hand, we studied the [3+2] annulation with the 2-amido-1,3-diaminoallyl cation generated from N-Ts guanidine 16 upon treatment with PhI(OAc)₂ (1.5 equiv) under our previously optimized conditions.³² However, the only isolable product (13%) was not the desired cyclic guanidine (cf. 18) but rather a compound that was missing one benzyloxy moiety and did not possess the expected downfield C6-proton based on NMR and MS. Ultimately, single crystal X-ray analysis of this adduct (see inset, Scheme 2).³⁷ revealed this product to be the carbinolamine 17, presumably arising from a benzylic oxidation followed by an elimination. While a sterically-driven elimination of benzyl alcohol and capture of water is possible, we hypothesize that under the mildly basic conditions, the OBn group at N-7 is likely oxidized since the corresponding methoxy substituted-guanidine did not eliminate under identical conditions (vide infra), the N-9 OBn group remains intact, and the potential for a charge-transfer complex³⁸ with the pyrrole ring. The pyrrole ring’s two bromine atoms through mesomeric effects and despite the weakly electron-withdrawing vinylogous acyl urea, may provide a sufficiently electron-rich aromatic to assist in this oxidation leading to selective loss of benzaldehyde at N-7 following hydrolysis of the resulting oxocarbenium. The exposed hydroxylamine in guanidine 19 can then undergo an elimination, perhaps facilitated by the hypervalent iodine reagent as shown, to generate an iminium species 20 that can capture water to provide carbinolamine 17 (Scheme 2). Interestingly, carbinolamine 17 is reminiscent of the previously isolated natural product, (−)-dibromohydroxyphakellin (4) (see Figure 1).

Scheme 2.

Attempted cycloguanidinylation with tricyclic dibromoenamide 13 and N-Ts guanidine 16. Proposed benzylic oxidation and elimination sequence leading to the observed carbinolamine 17 (inset: single crystal X-ray structure of carbinolamine 17•MeOH)

Given the observed benzylic oxidation with guanidine 16 leading to unexpected product 17 (Scheme 2), we sought to circumvent this issue by studying alternative 1,3-dipole precursors devoid of benzyl ethers such as bis-N-methoxy guanidine 22 prepared as previously described from thioimidate 21 (Scheme 3).^32,39 The chemical shifts observed by ¹H-¹⁵N HMBC correlations (compared to typical chemical shifts of hydroxylamines⁴⁰ and oximes⁴¹) suggested that guanidine reagent 22 exists primarily as the oxime isomer as shown versus the imine-containing isomer (iso-22).

Scheme 3.

Synthesis of a bis-OMe, Cbz-guanidinylating reagent 22.

With guanidine 22 in hand, we again attempted the cycloguanidinylation however this time with alkene 23^36b since the brominated pyrrole substrate alkene 13 displayed reduced reactivity likely due to the inductive effects of the bromine atoms (Scheme 4). We were pleased to find that this cycloguanidinylation proceeded smoothly under typical oxidative conditions to efficiently provide a mixture of the separable (MPLC), isomeric cyclic guanidine compounds 26 and 27 in 69% and 30% yield, respectively, without production of the previously observed carbinolamine. The major cyclic guanidine 26 was isolated as an ~1:1 mixture of E/Z oxime isomers (26a/b) which could be separated by RP-HPLC (for tentative assignment of each isomer, see SI). The structure of the isomeric guanidine 27 was supported by the observed nOe correlations indicated (N7-OCH₃ → C3-pyrrole NH and N9-OCH₃ → H11) with the latter nOe absent from the oxime isomers 26a/b in conjunction with ¹⁵N-NMR chemical shifts and ¹H-¹⁵N HMBC (see SI for details). Both cyclic guanidines 26 and 27 showed expected ¹H-¹⁵N HMBC correlations with N7 (δ 170 ppm; consistent with previously reported chemical shifts for hydroxylamine nitrogens).⁴⁰ Oximes 26 exhibited a ¹H-¹⁵N HMBC correlation between a ¹⁵N resonance (δ 310.0 and 318.0) and the N16-OCH₃ group which is consistent with literature ¹⁵N values for oxime-type nitrogens.⁴¹ In the case of isomer 27, the second methoxy group was assigned as the N9-OCH₃ based on a NOESY correlation with the C11 proton and a ¹H-¹⁵N HMBC correlation with a nitrogen resonance at δ 183.0. Based on this NMR analysis, it was determined that compound 26 has the oxime-type cyclic guanidine structure and compound 27 has the Cbz-imine-type cyclic guanidine structure. It should be noted that imine 27 appears as a single geometrical isomer by NMR indicating that it exists as a single isomer or rapidly interconverts relative to the NMR time scale. These two regioisomeric adducts arise from either hindered rotation about the N7-C8 bond (blue arrows, Scheme 4) or differential electron density at N9 and N16 (cf. 24 and 25) leading to the adducts 26 and 27 in this presumed stepwise cycloguanidinylation.

Scheme 4.

Guanidine [3+2] annulation toward (±)-N-methyldibromoisophakellin (10) leading to isomeric cyclic guanidines 26a,b and 27.

We next attempted N–O bond cleavage of the resulting cyclic guanidines 26 and 27 via treatment with SmI₂ (Scheme 5).⁴² Interestingly, the isomeric cyclic guanidines 26 and 27 showed divergent reactivity as they each gave rise to different products during attempted deprotections. Surprisingly, cyclic guanidine 26 returned the cycloaddition substrate 23 as the major product through an overall deguanidinylation with SmI₂. In addition, alkenyl guanidine 28 was obtained through a presumed ring-opening/elimination of the cyclic guanidine as a minor product possibly initiated by Lewis acidic Sm(III) species. On the other hand, cyclic guanidine 27 was readily converted to the desired N–O bond cleaved cyclic guanidine 29 (74%). The synthesis of (±)-N-methyldibromoisophakellin (10) was then achieved in 2 steps by dibromination of the pyrrole and deprotection of the Cbz group under basic hydrolytic conditions rather than hydrogenolysis.⁴³ Characterization data obtained for synthetic N-methyldibromoisophakellin matched that previously reported.²⁰

Scheme 5.

Divergent reactivity of the isomeric guanidine annulation adducts 26 and 27: synthesis of (±)-N-methydibromoisophakellin (10).

We also studied an alternative route to (±)-N-methyldibromoisophakellin (10) from cyclic guanidine 26 which ultimately led to a synthesis of an ugibohlin (Scheme 6). Protected guanidine 26 was converted to Boc-protected N-methyldibromoisophakellin 30 in 3 steps involving hydrogenolysis leading to N–O bond cleavage and cleavage of the Cbz group,⁴⁴ transient Boc protection of the guanidine and finally dibromination of the pyrrole ring. Attempted Boc group deprotection under acidic conditions (TFA) led to a single compound which was identified as N-methylugibohlin (9) with spectral data matching that previously reported.²¹ The synthesis of ugibohlin from ring cleavage of dibromoisophakellin was previously proposed by Mourabit³³ and subsequently described by the Lindel group²⁵ under strongly acidic conditions (6M aqueous HCl). We propose that Boc-protected N-methyldibromoisophakellin 30 is readily converted to N-methylugibohlin (9) via guanidium ion 31 leading to the N-acyl iminium species 32 and elimination. Based on these results, we hypothesize that ugibohlins may be artifacts of isolation of the corresponding isophakellins, as originally posited by Mourabit,³³ since silica gel chromatography and/or HPLC purification with TFA-containing solvents may lead to ring-cleavage to ugibohlin framework which were co-isolated with the corresponding isophakellins.^{17, 19b, 21}

Scheme 6.

Synthesis of N-methylugibohlin (9) through ring cleavage of dibromoisophakellin core.

Conclusions

In summary, we report a total synthesis of (±)-N-methyldibromoisophakellin (10) in 7 steps from commercially available 2-trichloroacetyl-1-methylpyrrole and prolinol employing a net [3+2] cycloguanidinylation with a 2-amido-1,3-diaminoallyl cation generated under oxidative conditions. A novel carbinolamine-containing tetracycle 17 was also produced during an attempted cycloaddition with the 1,3-dipole derived from a Ts-protected bis-N-benzyloxy guanidine. This adduct was confirmed by X-ray analysis and presumably arises from a selective, oxidative benzyl cleavage of one N-OBn, an elimination, and water capture under the conditions of the cycloaddition. We found that a Cbz-protected bis-methoxy guanidinylating agent is useful for generation of the corresponding 2-amino-1,3-diaminoallyl cations and this was employed for the synthesis of (±)-N-methyldibromoisophakellin (10). Finally, the first total synthesis of N-methylugibohlin (9) is described and occurs by cyclic guanidine ring cleavage and elimination under mild acidic conditions suggesting that the ughibohlins may in fact be artifacts of isolation formed during purification of isophakellins when using trifluoroacetic acid as a modifier in RP-HPLC.