Semisynthesis, Characterization, and Biological Evaluation of Fluorinated Analogues of the Spirobisnaphthalene, Diepoxin-η

Abstract

Diepoxin-η (1) is a cytotoxic fungal metabolite belonging to the spirobisnaphthalene structural class. In this study, four mono fluorinated analogues (2-5) of diepoxin-η (1) were semisynthesized in a single-step by selectively fluorinating the naphthalene moiety with Selectfluor. The structures of 2–5 were elucidated using a set of spectroscopic and spectrometric techniques and were further confirmed by means of TDDFT-ECD and isotropic shielding tensors calculations. Compounds 2–5 showed equipotent cytotoxic activity to 1 when tested against OVCAR3 (ovarian) and MDA-MB-435 (melanoma) cancer cell lines with IC₅₀ values that range from 5.7–8.2 μM.

Spirobisnaphthalenes are a structurally interesting class of fungal metabolites due to a 1,8-dihydroxynaphthalene-derived spiroketal unit linked to an oxidized naphthalene moiety.¹ Over two hundred members of this class have been identified since 1989, when the first spirobisnaphthalene metabolite, MK 3018, was disclosed (Fig. 1).^1–3 Their unique structures and diverse biological activities, including antitumor, anti-inflammatory, and enzyme inhibition activities,^{1, 4, 5} make them ripe for further optimization.^{1, 2}

Fig. 1.

Structures of MK 3018 and diepoxin-η (1).

It has recently become appealing to introduce fluorine atoms into secondary metabolites to modulate their acid/base properties, electronegativity, lipid solubility, and metabolic stability, potentially improving their pharmacokinetic and pharmacodynamic properties.^6–8 For instance, it has been estimated that over 20% of pharmaceutical agents available worldwide contain at least one fluorine atom.⁹ A current challenge with natural products is the inability to patent the isolated molecules,¹⁰ a fact that likely hampers the further development of promising leads. As such, we have been exploring ways to incorporate fluorine atoms into fungal metabolites, using both biosynthetic (i.e., precursor directed biosynthesis)^{11, 12} and semisynthetic approaches.^{13, 14}

The goal of this project was to use Selectfluor¹⁵ to explore the ability to fluorinate diepoxin-η (1; Fig. 1), which was first described in 1993.^{5, 16} Despite being in the literature for 30 years,^{5, 16} there are no reports of evaluating the cytotoxicity of 1, although diepoxin-η-related compounds have been reported recently to possess such activities.¹⁷

In the course of ongoing studies of fungi for anticancer drug leads,¹⁸ diepoxin-η (1) was isolated as a white powder with the molecular formula of C₂₀H₁₆O₇ (Fig. 1).^{17, 19} Its structure was determined by a combination of spectroscopic and spectrometric techniques, mainly HRESIMS and NMR. Our NMR spectroscopic data were in agreement with those reported previously, although we were able to improve the NMR assignments of all positions in the naphthalene moiety based on access to a higher field magnet and DFT chemical shifts calculations²⁰ (see Supporting Information, Table S1 and Figs. S3–S6 and S27). Fluorination was carried out by reacting 1 and Selectfluor at 0 °C (Scheme 1), as monitored by LC-MS.²¹ Selective fluorination of the naphthalene moiety in 1 resulted in four monofluorinated products (2–5) (Scheme 1); difluorinated products were not observed.

Scheme 1.

Fluorination of diepoxin-η (1) using Selectfluor, showing the relative yield of each monofluorinated analogue (2-5).

Compounds 2-5 were purified from the reaction mixture using preparative HPLC as white powders, and their purities were evaluated using UPLC (Fig. S1). An 18 amu difference in the HRESIMS data between compounds 2-5 vs 1 indicated the formation of four monofluorinated analogues, each of which had the molecular formula of C₂₀H₁₅FO₇ (Fig. S2). The ¹⁹F NMR of 2–5 revealed a fluorine peak resonating at −142.6, −142.2, −131.6, and −131.5 ppm, respectively (Figs. S8, S13, S18, and S23). The ¹H NMR data indicated the loss of one aromatic proton from compounds 2-5 compared to 1 (Fig. S28). The mechanism of fluorination of aromatic compounds via Selectfluor indicated that the ortho and the para positions of the naphthalene ring, relative to where the ether linkages are attached, are potential sites for fluorination.²² Hence, it was hypothesized that the four semisynthesized analogues (2-5) could consist of two ortho- and two para-fluorinated analogues.

Analyses of the 1D and 2D NMR data of 2 indicated the replacement of the H-7´ aromatic proton (δ_H 6.94, d, J = 7.6 Hz) in 1 with a fluorine atom in 2, which was supported by both the deshielding of C-7´ (δ_C 146.2, d, ¹J_C-7´/F-7´ = 244.3 Hz) in 2 relative to 1 (δ_C 109.8) and the splitting of C-7´ by the fluorine atom into a doublet with a ¹J coupling value of 244.3 Hz (Tables S1 and S2; Figs. S3 and S7). Likewise, the chemical shift values and the splitting patterns of H-6´/C-6´ (δ_H 7.40, dd, J_H-6´/F-7´ = 10.8 Hz; J_H-5´/H-6´ = 9.1 Hz/δ_C 118.9, d, ²J_C-6´/F-7´ = 21.4 Hz) in 2 were different from those of 1 (δ_H 7.45, t, J_H-6´/H-7´ = 7.6 Hz/δ_C 128.5) (Tables S1 and S2; Figs. S3 and S7). Furthermore, signals for most of the aromatic carbon atoms in 2 relative to 1 were split by the fluorine atom into doublets with varying coupling constants (Table S2). COSY data identified two spin systems within the naphthalene ring (i.e., H-2´/H-3´/H-4´ and H-5´/H-6´). The position of fluorination (i.e., C-7´) was further confirmed by HMBC correlations of H-5´ and H-6´ to C-7´ and H-4´ to C-5´. These data established the structure of 2 as 7´-fluorodiepoxin-η.²³

Analyses of the NMR data of 3 revealed an ortho-fluorinated analogue of 1 that was similar to 2, except for the position of fluorination (C-2´ in 3 versus C-7´ in 2). As seen in 2, most of the aromatic carbon signals in the ¹³C NMR spectrum of 3 were split by the fluorine atom into doublets (Table S3). The replacement of the H-2´ aromatic proton in 1 with a fluorine atom in 3 was supported by both the deshielding of C-2´ and its splitting pattern (δ_C 146.4, d, ¹J_C-2´/F-2´ = 244.9 Hz) (Tables S1 and S3, Fig. S12). Likewise, the chemical shift values and the splitting patterns of H-3´/C-3´ (δ_H 7.43, dd, J_H-3´/F-2´ = 10.6 Hz; J_H-3´/H-4´ = 9.2 Hz/δ_C 119.1, d, ²J_C-3´/F-2´ = 21.4 Hz) in 3 were different from those of 1 (Tables S1 and S3, Figs. S3 and S12). COSY data identified two spin systems within the naphthalene ring (i.e., H-3´/H-4´ and H-5´/H-6´/H-7´). The position of fluorination (i.e., C-2´) was further confirmed by HMBC correlations of H-3´ and H-4´ to C-2´ and H-5´ to C-4´. These data established the structure of 3 as 2´-fluorodiepoxin-η.²⁴

Analyses of the 1D and 2D NMR data of 4 indicated a para-fluorinated analogue of 1 at C-4´. As in 2 and 3, most of the ¹³C NMR signals for the aromatic carbon atoms in 4 were split by the fluorine atom into doublets (Table S4). The H-4´ aromatic proton in 1 (δ_H 7.55, d, J = 7.5Hz) was replaced with a fluorine atom in 4. Such replacement was supported by both the deshielding of C-4´ (δ_C 154.6, d, ¹J_C-4´/F-4´ = 244.5 Hz) relatives to 1 (δ_C 121.8) and the splitting of C-4´ into a doublet with a ¹J coupling value of 244.5 Hz (Tables S1 and S4, Figs. S3 and S17). Moreover, the chemical shift values and the splitting patterns of H-3´/C-3´ (δ_H 7.22, dd, J_H-3´/F-4´ = 10.8 Hz; J_H-2´/H-3´ = 8.3 Hz/δ_C 112.1, d, ²J_C-3´/F-4´ = 22.2 Hz) in 4 were different from those of 1 (δ_H 7.49, t, J_H-3´/H-4´ = 7.5 Hz/δ_C 128.7) (Tables S1 and S4, Figs. S3 and S17). Analysis of the COSY data identified two spin systems within the naphthalene ring (i.e., H-2´/H-3´ and H-5´/H-6´/H-7´). The position of fluorination (i.e., C-4´) was further confirmed by HMBC correlations of H-2´, H-3´ and H-5´ to C-4´. These data established the structure of 4 as 4´-fluorodiepoxin-η.²⁵

Analyses of the NMR data of 5 revealed a para-fluorinated analogue of 1 that was similar to 4, except for the position of fluorination (C-5´ in 5 versus C-4´ in 4). As in compounds 2-4, the ¹³C NMR spectrum of 5 showed splitting of most of the aromatic carbon atoms by the fluorine atom into doublets (Table S5). The H-5´ aromatic proton in 1 (δ_H 7.54, d, J = 7.6 Hz) was replaced with a fluorine atom in 5, as supported by both the deshielding of C-5´ and its splitting pattern (δ_C 154.6, d, ¹J_C-5´/F-5´ = 245.9 Hz) relative to 1 (δ_C 121.8) (Tables S1 and S5, Figs. S3 and S22). Moreover, the chemical shift values and the splitting patterns of H-6´/C-6´ (δ_H 7.19, dd, J_H-6´/F-5´ = 10.8 Hz; J_H-6´/H-7´ = 8.3 Hz/δ_C 111.9, d, ²J_C-6´/F-5´ = 22.2 Hz) in 5 were different from those of 1 (δ_H 7.45, t, J_H-5´/H-6´ = 7.6 Hz/δ_C 128.5) (Tables S1 and S2, Fig. S3 and S22). Analysis of the COSY data identified two spin systems within the naphthalene ring (i.e, H-2´/H-3´/H-4´ and H6´/H-7´). The position of fluorination (i.e., C-5´) was further confirmed by HMBC correlations of H-4´, H-6´ and H-7´ to C-5´. These data established the structure of 5 as 5´-fluorodiepoxin-η.²⁶

Electronic circular dichroism (ECD) can be used to explore the absolute configuration of compounds with a UV chromophore in the vicinity of the stereogenic element. In particular, this technique allows one to evaluate the calculated vs measured spectra.^27–30 Interestingly, compounds 2-5 showed distinct ECD spectra. For instance, the ECD spectrum of 2 displayed a positive Cotton effect at λ = 321 nm, while compound 3 exhibited two distinct Cotton effects, one positive at λ = 288 nm and another negative at λ = 324 nm. The ECD spectra of compounds 4 and 5 demonstrated a positive Cotton effect at λ = 294 nm. However, unlike compounds (1–3), compounds 4 and 5 did not exhibit the same positive Cotton effect in the range of 211–225 nm. Hence, the structure elucidation of compounds 1–5 were further confirmed by means of TDDFT-ECD calculations, where good agreement was observed between experimental and calculated spectra (Fig. S29).

We have been exploring the use of isotropic shielding tensors calculations using the GIAO (gauge-including atomic orbitals)^31–33 method as an aid for structural elucidation, as they represent a valuable tool for calculating theoretical chemical shifts.^34–36 In the current study, despite the detailed characterization of the fluorinated products, the NMR data for the two ortho-substituted analogues were similar to each other, making it hard to unequivocally differentiate them; the same was true for the two para-substituted analogues. Thus, to further confirm the proposed assignments, the isotropic shielding tensors of compounds 2-5 were computed and analyzed. Using the DP4+ probability methodology described by Sarotti et al.,³⁷ the calculated ¹H and ¹³C NMR chemical shifts for the proposed structures of 2-5 were in agreement with the experimental data with a global probability >98% (Figs. S30–S33); these results agreed with those observed from the TDDFT-ECD calculations described above.

The parent compound [diepoxin-η (1)] and its fluorinated analogues (2-5) were evaluated for cytotoxic activity against both the OVCAR3 (ovarian) and MDA-MB-435 (melanoma) cancer cell lines. Interestingly, fluorination did not negatively affect the cytotoxicity, as 2-5 all had IC₅₀ values that were comparable to the cytotoxicity of 1 (Table 1). Compounds 1-5 were also tested against three other cancer cells lines (MDA-T32, DU-145, and HPAC) but were inactive (see Supporting Information). Additionally, 1-5 were tested for antimicrobial activity against a panel of bacteria and fungi and were found to be inactive when tested up to 500 μM.

Table 1.

Cytotoxicity of compounds 1-5 against two human tumor cell lines.

Compound	IC50 values in μM
	OVCAR3a	MDA-MB-435b
diepoxin-η (1)	5.7 ± 0.1	8.9 ± 0.2
7´-fluorodiepoxin-η (2)	6.5 ± 0.1	8.6 ± 0.2
2´-fluorodiepoxin-η (3)	5.8 ± 0.2	8.2 ± 0.2
4´-fluorodiepoxin-η (4)	5.9 ± 0.3	8.6 ± 0.1
5´-fluorodiepoxin-η (5)	7.4 ± 0.1	8.6 ± 0.2
taxolc (paclitaxel)	0.0026 ± 0.0001	0.0010 ± 0.0001

Human ^aovarian and ^bmelanoma cell lines.

^cPositive control.

All samples were analyzed as three biological replicates, each in triplicate.

This is the first report involving the incorporation of a fluorine atom into spirobisnaphthalenes by means of semisynthesis. However, fluorinated analogues of palmarumycin B6 were reported previously by total synthesis using fluorinated starting materials, resulting in modifications in positions 6 and 8 and leaving the naphthalene moiety intact.³⁸ Consistent with our findings, the incorporation of fluorine did not negatively affect the cytotoxic activity of those analogues relative to the parent compound.³⁸ It is challenging to patent isolated natural products,¹⁰ and a semisynthetic strategy that incorporates a fluorine atom may present a reasonable avenue toward new intellectual property. Specific to the spirobisnaphthalenes, which have been reported to have a wide range of biological activities,^{1, 2, 4, 5, 16} these molecules seem to be amenable to such modifications. For example, based on some earlier work with resorcylic acid lactones,¹³ we believe it would be possible to add other halogens (i.e., Br) to the naphthalene moiety of these compounds by the use of other electrophilic halogenating reagents.