Exciton Coupling Circular Dichroism of an Allylic N-imidazolyl Group in Amaranzole A, a Marine Natural Product from Phorbas amaranthus

Abstract

A new steroidal alkaloid amaranzole A (10) with a C24-imidazolyl group displays an unusually large split-CD spectrum at short wavelengths that we assign to exciton coupled circular dichroism (ECCD) between the polarized π-π* transitions of the C25 C=C double bond and the imidazolyl group. A model 4,5-disubstituted imidazole 11, prepared from optically pure (R)-(-)-2-aminobutanol, exhibited similar ECCD and solvent and pH-dependence consistent with changes in the the protonation state of the imidazole ring. Calculations and CD measurement of 12 (the dihydro-derivative of 11) suggest that the 4-hydroxyphenyl group is not strongly conjugated to the imidazole group in 10, and the observed ECCD is entirely accounted for by coupling between the C=C double bond and isolated imidazole π-π* transitions.

Introduction

Electronic circular dichroism has a long association with the study of natural products and the assignment of their absolute configurations. Outstanding examples include the interpretation of the Cotton effects associated with the forbidden n-π* transitions of cyclic ketones and applications to conformational analysis in steroids.ⁱ Exciton coupled circular dichroism (ECCD) is a powerful tool for assignment of configuration in natural products, and has been used to predict the configurations of cyclic diols, amino alcohols through their corresponding dibenzoates or other chromophoric diester derivatives.^ii,iii Weak Cotton effects (CE) in acyclic systems are also useful in assignment of configuration in acyclic molecules provided there is an understanding of the origin of the CE in context of the conformation and configuration of the molecular structure. For example, we assigned the configurations of the macrolide glycosides, phorbasides A (1) and B (2) (Figure 1a), by exploiting the CD of the remote hyperconjugated ene-yne chlorocyclopropane chromophore (λ 232 nm, Δε +9.1; 241, +8.1) which relays stereochemical information within the first sphere of asymmetry of the cyclopropyl group to the allylic acyloxy-substituted carbon C13.^iv The sponge Xestospongia muta, common throughout the Caribbean, typically produces long-chain brominated 1,3-diene and 1,3-ene-yne unsaturated fatty acids. We recently characterized seven minor components — mutafurans A-G (3-9, Figure 1b) which are the first chiral brominated lipids in this class^v — and assigned the absolute configuration of the tetrahydrofuran ring based on interpretation of the weak ene-yne associated Cotton effects (λ 227 nm, Δε —0.46; 237, —0.5; 252, +0.26) that are asymmetrically perturbed by the propargylic center C8. In these two special cases of naturally occurring molecules with ene-yne chromophores, it is important to realize that calculations based on force field models predict the sp-sp² bond is essentially a free rotor with a barrier to rotation of approximately 0 kcal/mol^-1. The important consequence of the latter property is that accurate time-dependent DFT calculations of the CD spectra will be compromised by small inaccuracies in Boltzman-weighted distributions. In fact, as the number of degrees of freedom in an acylic system increases, the confidence level of geometry calculations may even preclude accurate CD calculations. Under these conditions, empirical approaches appear more attractive.

Figure 1.

Phorbasides A (1), B (2), mutafurans A-G (3-9), amaranzole A (10) and model compounds (+)-(R)-11 and (-)-(R)-12.

Recently, we reported an unusual N-imidazolyl steroidal alkaloid, amaranzole A (10) from the tropical sponge Phorbas amaranthus.^vi While many imidazole and 2-aminoimidazole natural products and some steroidal alkaloids are known from marine invertebrates and microbes,^vii this is the first example of a steroidal imidazole.^viii The unusual structure of 10 implies a convergence of isoprenoid and amino acid secondary metabolism and presented a challenge for configurational assignment. Amaranzole A exhibited a negative Cotton effect at λ∼ 205 nm which we presumed to arise from ECCD between the imidazole and the 1,1-disubstituted olefin which absorbs λ< 200 nm. The C24 configuration was solved by matching the CD spectrum of 10 with that of a stereodefined model imidazole prepared from (R)-(-)-2-amino-1-butanol.⁶ In this report we characterize the CD properties of allyl imidazole and propose an assignment of the origin of the ECCD between the isolated imidazole and the 1,1-disubstituted alkene.

EXPERIMENTAL

General

All CD spectra were acquired with a JASCO-810 spectrophotometer at 23 °C using a 2 mm cell at concentrations of ∼1× 10^-4 M in spectroscopic or HPLC grade solvents (Aldrich, MeOH, CH₃CN, H₂O, or CF₃CH₂OH). Each CD spectrum was acquired by averaging 20 scans with correction for background (solvent). Smoothing was applied to the CD spectra in Figures 2a and 2b. No smoothing was applied to the data in Figure 4. pH measurements were carried out using a Denver Instrument Model 220 pH-conductivity meter which was calibrated against standard buffers. The isolation and purification of amaranzole A (10) and synthesis of compounds (+)-(R)-11 and (—)-(R)-12 are described elsewhere.⁶

Fig 2.

(a) CD spectra of (A) amaranzole A (10), (B) compound 11, and compound 12 (3:7 CH₃CN/H₂O, T= 23 °C). (b) CD spectra of compound 11 and 12 (CH₃CN, T= 23°C).

Fig 4.

pH dependent CD spectra for compound 11 (30% CF₃CH₂OH + H₂O, 20 mM Na₂HPO₄ /NaH₂PO₄), T= 23°C.

pH Dependence of the CD spectrum of 11

The CD spectrum of a solution of 11 (0.1 mM, 500 μL) in 30% CF₃CH₂OH/ 20mM aqueous Na₂HPO₄, pH 8.2) was measured (pH = 8.2). The solution was titrated with aqueous HCl (0.375 M, 5 μl, resultant pH = 7.2), the CD spectrum measured again, and the procedure repeated (pH= 6.4). The solution was titrated sequentially three more times with 0.125 M HCl (5 μl) and CD measurements made in between additions at resultant pH’s = 5.9, 4.9, and 3.3.

Calculations

Calculations were performed at semi-empirical level (PM3) or DFT (B3LYP-/6-31G(D), Spartan 2004) on a Macintosh (G4 Motorola processor, 1.44 GHz, or iMac (Intel Core2 Duo, 2.4 GHz processor).

RESULTS AND DISCUSSION

The marine sponge Phorbas amaranthus occurs in waters of the Florida Keys and, less commonly, in the Bahamas. Amaranzole A (10) was isolated from the polar extracts P. amaranthus collected at Key Largo by extensive gel filtration chromatography and final purification by HPLC. The substitution pattern and stereochemistry of the sterol core was assigned using conventional NMR techniques, in particular analysis of the vicinal coupling constants and the NOESY spectra. Compound 10 is the first example of a steroidal imidazole natural product and presented a unique challenge for stereochemical analysis. The configuration of the remote C24 —imidazolyl group in 10 could not be assigned by standard methods, however, the CD spectrum of 10 (MeOH, 22 °C) showed rich information that allowed configurational assignment by comparison with suitable model compounds 11 and 12.

Compound 11 was prepared in several steps from (R)-(-)-2-amino-1-butanol.⁶ The CD spectra of 10 and 11 were originally reported in MeOH, however, the low-wavelength Cotton effects (CE’s) are less obscured in 3:7 CH₃CN/H₂O as reported here (Figure 2a). Compound 11 displayed a weak positive CE at ∼256 nm (<0.2), a negative CE at 218 nm (Δε -0.9) and stronger positive effect (λ 201 nm, Δε +4.5) followed by a strong negative CE (189 nm, Δε -6.4). Almost identical CE’s were seen in amaranzole A (10), except the intensity of the CE at λ 201 nm was increased (Δε +8.4) therefore we may assign a (24R)-configuration to 10. The weak band at 256 nm was assigned to a weakly perturbed phenol chromophore, however the pH dependent UV spectra suggested this chromophore was, at best, poorly conjugated to the imidazole group. Vicinal substitution of the imidazole ring by the N1-alkenyl group and C4-para-hydroxyphenyl group gives rise to non-bonded interactions that force the two rings out of planarity; a phenomenon that is supported by molecular mechanics calculations which reveal that the imidazole and benzene rings 11 are skewed at an angle of ∼77 ° (Figure 3c).⁵

Figure 3.

Dominant conformation of (3R)-N-allylic heterocycles (a) (R)-phthalimide 14 showing directions of π-π* electronic transition dipole moments and helicity of chromophores (from Gawronski et al. Ref 12) (b) (R)-N-allyl imidazole 11 and nOe (c) Force field minimized conformation of 11 (MMFF94, Spartan 2004).

Due to the limited solubility of 11 in solvents other than MeOH, the analysis of the ECCD of allylic N-imidazoles were interpreted measurements of 11 and 12 in CH₃CN and comparisons with 1. The electronic spectrum of the parent heterocycle, imidazole 13, has been studied both experimentally and by quantitative MO methods.^ix Two shortwavelength transitions of moderate intensity are seen in neutral aqueous solutions of 13 at λ =187 nm (ε 4200) and 207 nm (ε 4500); the former band is blue shifted at pH = 2 (λ =178 nm).^9b Solvent and pH dependence of the CD spectrum of substituted imidazole chromophores may be expected if these transitions participate in ECCD, however, it was not clear from the outset how this would influence the CD spectrum of 10 and 11. Despite these uncertainties, the expected directionality of the transition dipole moments in 10 and 11 based upon examination of HOMO and LUMO should roughly bisect C2 (imidazole numbering) and the mid-point of C4-C5. The 1,1-disubstituted olefin π-π* transition of 10-12 lies at the edge of our observation window, however its participation in ECCD is readily evident in the CD spectrum of 11 (Figure 2b) where the strong negative shortwavelength component of the split Cotton effect is observed (λ 187 nm, Δε -11). The latter effect is absent in the CD spectrum of dihydro-derivative 12^x where only shallow Cotton effects remain from weak asymmetric perturbations of the imidazole and phenol chromophores (λ_max 200 nm, Δε < -0.2; ∼250 Δε +1.5, note reversal of sign at ∼250 nm, Δε -2, compared to 11).

Analysis of the origin of the ECCD in 10 and 11 is made difficult by conformational mobility of the N1-side chain on the imidazole ring, uncertainties of the conformer populations about several rotors, including the C3-N, C2-C3, and C3-C4 bonds, and direction of the transition dipole polarization. However, early studies of allylic alcohol and amine derivatives provide some guidance. Nakanishi et al demonstrated that the major conformation of benzoates esters of secondary allylic alcohols is well-defined and consistent for all members; the lowest energy conformation is that which eclipses the allylic methine H_a by the C=C bond.^xi The transition dipole moment of the benzoate chromophore is oriented roughly along the C-O bond and subtends a negative helicity in (R)- benzoates of allylic secondary alcohols that gives rise to a negative ECCD. In (S)-enantiomers, the sign of the ECCD is reversed. Gawronski et al. showed that the phthalimides derived from simple (R)-allylic amines (e.g. 14) also give rise to a negative ECCD that arises from essentially the similar conformation (Figure 3).^xii Similar trends were observed for 2-naphthamides of allylic amines.^xiii In contrast, the sign of the ECCD in 10 and (+)-(R)-11 is reversed; (R)-Allylic imidazoles 10 and 11 show a positive ECCD effect. The sign reversal implies either the transition dipole moment vector of the imidazole has changed direction with respect to the π-π* transition of the C=C double bond, or the dominant conformation of the molecules are different, or both.

The conformation of 11 was defined by nOe measurements and calculations. Irradiation of the vinyl methyl group in 11 gave rise to a significant nOe at H3, but no nOe was observed between the vinyl proton signals and H3, suggesting the major conformation is that depicted in Figure 3b. NOe was also observed from H3 to the ortho-protons (H1′) on the phenyl ring of 11 consistent with a conformation in which the H3 methine is syn to the phenyl ring (Figure 3b). The lowest energy conformation of 11 was also calculated using molecular mechanics (Figure 3, MMFF94, Spartan 2004) and found to be entirely consistent with the nOe results, although, other low-lying rotamers may make minor contributions to the global conformer distribution in 2 and 11.

In order to further characterize the ECCD of allylic imidazoles, we carried out pH-dependent measurements of the CD spectrum of 11 (Figure 4). At neutral or alkaline pH, the CD spectrum of 11 (30% CF₃CH₂OH: aqueous phosphate buffer) were essentially the same, however, lowering the pH lead to a progressive diminution of the positive Cotton effect at λ ∼200 nm until, at pH = 3.3, the sign of the long-wavelength Cotton effect was inverted (Δε = -8). Within the limits of experimental error, it appeared that all CD spectra passed through an isobestic point at λ∼ 210 nm suggesting the pH-dependency of the CD spectrum was associated with reversible protonation of the imidazole ring. Grebow and Hooker showed that the orientation of the transition dipole moment of the neutral form of imidazole and protonated imidazole are significantly different.^9b A corresponding change in transition dipole orientation in 10 and 11 upon protonation of the imidazole nitrogen (Figure 5) may be responsible for the significant pH-dependent Cotton effects we observe. We cannot exclude other factors such as small changes in the Boltzman-weighted populations of conformations upon protonation of the imidazole N, although this seems unlikely, and the former explanation would seem to prevail.

Figure 5.

Calculated frontier orbitals of imidazole (13) (a) HOMO (b) LUMO (PM3 level, Spartan 2004) and approximate direction of electronic transition dipole.

CONCLUSION

In conclusion, the structure elucidation and configurational assignment of amaranzole A (10) has lead to discovery of a useful ECCD in allylic imidazoles. Both components of the split Cotton effect are observable when the CD spectra were recorded in acetonitrile. From the pH dependency and solvent effects of the CD spectra of allylic imidazoles, it appears the ECCD is fully accounted by coupling between the terminal C=C double bond and the imidazole, with negligible electronic contributions from the p-hydroxyphenyl group. The present study should find utility in assignment of minor congeners of amaranzole A and possible extension to other allylic azoles (e.g. 1,2-pyrazole, 1,2,4-triazole and benzimidazole).