|Home | About | Journals | Submit | Contact Us | Français|
Polyoxygenated1 and polysulfated steroids with atypical, modified side chains,2 are prevalent among compounds isolated from marine organisms. Sponge-derived sulfated sterols have a wide array of reported biological activities in a variety of therapeutic areas. Of particular note is their activity against HIV-1.3–5 In addition, the spheciosterol sulfates A–C, isolated from a Spheciospongia sp. sponge, were recently shown to inhibit protein kinase C ζ (PKCζ), as well as downstream NF-κB activation.6 PKCζ has been implicated as an integral factor in several types of cancer,7–11 obesity,12 and osteoarthritis.8,13 Consequently, the detection and identification of PKCζ-specific inhibitors could have a potentially profound impact on the treatment of a number of diseases and disorders.
As part of an ongoing search for bioactive marine metabolites, crude extracts from our marine invertebrate library were screened for PKCζ inhibition. The methanol extract of a Lissodendoryx (Acanthodoryx) fibrosa sample, collected from Coron Island, Philippines, showed promising PKCζ inhibition in the initial screening. No natural products had been reported from this sponge, suggesting L. (A.) fibrosa would be an attractive source for chemical investigation. As a result, fibrosterol sulfates A–C (1–3), three new sulfated bis-steroids, were isolated from the sponge. Data from several experiments verified that fibrosterol sulfates A (1) and B (2) were, in fact, PKCζ inhibitors.
The L. (A.) fibrosa specimen (PC00-04-56) was exhaustively extracted with MeOH and the crude extract separated on an HP20SS resin using a step gradient of H2O to isopropanol (IPA) (25% steps, 5 fractions). Bioassay-guided fractionation of the second (75/25 H2O/IPA) and third fractions (50/50 H2O/IPA), utilizing reversed-phase column chromatography and reversed-phase HPLC, resulted in the isolation of fibrosterol sulfates A–C (1–3).
The molecular formula for fibrosterol sulfate A (1), C54H84O19S4Na4, was derived from NMR data and the HRESIMS ion at m/z 605.2147 ([M−2Na]−2; Δ +0.21 ppm ). Utilizing the ultra-high resolution capabilities of FTMS, the 34S peak could be resolved from the 13C2 peak, indicating the presence of four sulfurs in 1. The presence of multiply charged ions in the mass spectra coupled with characteristic sulfate losses in FT-MS/MS experiments indicated that sulfate esters were present in 1. The structure of fibrosterol sulfate A (1) was established on the basis of extensive 1D and 2D NMR studies. Initial interpretation of the NMR data (Table 1) suggested that 1 was an isoprenoid containing five methyl singlets (δH 0.71, 0.72, 1.05, 1.08, 1.15), two methyl doublets (δH 0.97, 0.99), seven oxygenated methines (δH 3.58, 3.98, 4.06, 4.19 (2), 4.46, 4.64), a trans olefin (δH 5.35, J = 15.7; 5.54, J = 15.7) and a terminal olefin (δH 4.85, 4.81). The data also indicated that fibrosterol sulfate A (1) contained six quaternary carbons, 22 methines, and 19 methylenes. Many of the signals, specifically the methyls, appeared in pairs, indicating that 1 was a pseudo-symmetrical bis-steroid. The pseudo-symmetrical nature of the molecule created an overlap of spectral resonances which complicated data interpretation at lower field strength (500 MHz). However, the increased sensitivity and resolution, afforded by a 600 MHz instrument equipped with a cryoprobe, allowed for clear interpretation of the overlapping signals in 1. Analysis of COSY and HMBC data led to the assignment of the ABCD and A′B′C′D′ steroid ring systems in fibrosterol sulfate A (1) (Table 1). Rings A and B were assembled based on COSY correlations among all adjacent protons between H-1 and H-9. HMBC correlations from Me-19 to C-1, C-5, C-9 and C-10 completed the structural assignment of the A and B rings. Rings C and D were assigned based on HMBC correlations from Me-18 to C-12, C-13, C-14 and C-17; and H-14 to C-8, C-13, and C-15. COSY correlations between H-11 and H-9, and H-11 and H-12; and HMBC correlations from H-11 to C-9 and C-12 supported the connectivity of C-9 through C-12. HMBC correlations from H-16 to C-17 and C-15 completed the structural assignment of the C and D rings. The A′B′C′D′ rings were assembled in the same fashion; all of the correlations used to assign the ABCD rings were identical to the correlations used to assign the A′B′C′D′ rings. HMBC correlations from Me-21, Me-26 and Me-21′ were particularly useful in constructing the side chain that fuses rings D and D′ in 1 (HMBC correlations from Me-21 to C-17, C-20 and C-22; Me-21′ to C17′, C-20′ and C-22′; Me-26 to C-24, C-25, C-27 and C-24′). COSY correlations between H-22 and H-23, and H-23 and H-24, in addition to the methyl correlations, supported all connectivities between C-20 and C-27. HMBC correlations from H-23′ to C-22′ and C-24′; and H-26′ to C-24′ and C-27′, supported all connectivities between C-20′ and C-27′. COSY correlations between H-27 and H-27′, as well as HMBC correlations from H-27′ to C-27, completed the assignment of the side chain that joins the two steroid units in fibrosterol sulfate A (1). Sulfates induce a downfield 13C shift of ~5–9 ppm on the δC of C-2, C-3 and C-6 when compared to the corresponding triols.14–16 As such, the sulfate groups were assigned to C-2, C-6, C-2′, and C-6′, and hydroxyls were placed at C-3, C-3′, and C-22 in fibrosterol sulfate A (1).
Because of a significant number of overlapping 1H signals, few coupling constants could be obtained for fibrosterol sulfate A (1), complicating the relative configuration analysis. However, biosynthetic precedents and ROESY data clearly indicated that 1 contained two standard 5α10β steroid nuclei (Figure 1). ROESY data, w-coupling, and 13C chemical shift analysis were all necessary to determine the configuration at C-2, C-3, C-6, C-2′, C-3′, and C-6′. H-2 was given an equatorial assignment based on the observed w-coupling between H-2 and H-4α in the COSY spectrum, and the narrow multiplicities for H-2 in the 1H spectrum (absence of large vicinal coupling constants). H-3 was designated axial based on the presence of a large vicinal coupling constant (J = 11.8 Hz) and the ROE observed between H-3 and H-5. H-6 was given an axial assignment based on the ROE observed between H-6 and Me-19. The 13C chemical shifts in the AB rings were compared with AB steroid rings containing different configurations at C-2, C-3, and C-6.16–21 The configuration of the AB rings in fibrosterol sulfate A (1) corresponded best to that observed in amaranzole A,19 as 2S, 3R, 6S. Chemical shift variances observed between the AB and A′B′ rings suggested configuration differences, which were confirmed after careful observation of several parameters. The narrow multiplicities for H-2′ and H-3′ in the 1H spectrum and the lack of a ROE between H-2′ and Me-19′, and H-3′ and H-5′ implied that H-2′ and H-3′ were equatorial. It is probable that w-coupling between H-2′ and H4 α′ and H-3′ and H1β′ occurs; however, the signal overlap of H4α′ and H1β′ (δH 2.04 and 2.06, respectively) precludes discrete observation in the COSY spectrum. H-6′ was given an axial assignment based on the ROE observed between H-6′ and Me-19′. 13C chemical shifts in the A′B′ rings were compared with AB steroid rings containing different configurations at C-2, C-3, and C-6.16–21 The configuration of the A′B′ rings in fibrosterol sulfate A (1) was consistent with the configurations of halistanol sulfate16 and a semi-synthetic sterol,22 as 2′S, 3′S, 6′S. J-based analysis of 1 was used in an attempt to assign the relative configuration of the C-22 alcohol,23 but the results were inconclusive (see Supporting Information). Molecular modeling studies were conducted on four diastereomers of 1 (25R,24′S; 25S,24′R; 25R,24′R; and 25S,24′S), and interatomic distances were calculated (Supplemental Table 1) in an effort to determine if ROEs could be used to distinguish whether the ring in 1 had a cis or trans relative configuration. However, no conclusive results could be obtained (see Supporting Information). Ongoing studies using RDC analysis should aid in the assignment of the C-22 configuration as well as the configuration at C-25 and C-24′. Based on the aforementioned data, fibrosterol sulfate A (1) was assigned as 2S, 3R, 5S, 6S, 8S, 9S, 10R, 13S, 14S, 17R, 20R, 23E, 2′S, 3′S, 5′S, 6′S, 8′S, 9′S, 10′R, 13′S, 14′S, 17′R, 20′R.
The molecular formula for fibrosterol sulfate B (2), C54H83O22S5Na5, was derived from NMR data and the HRESIMS ion at m/z 656.1843 ([M-2Na]−2, Δ−0.35 ppm). FT-MS analysis indicated five sulfur atoms were present in the molecular formula for 2. Characteristic sulfate losses in FT-MS/MS experiments combined with multiply charged ions in the mass spectra indicated that 2 also contained sulfate esters. Comparison of the molecular formulas for 1 and 2 revealed that fibrosterol sulfate B (2) contained an additional sulfate group. The NMR spectra of fibrosterol sulfates A (1) and B (2) are very similar, with the major 1H and 13C chemical shift differentials between the two molecules occurring in the A′ ring (Table 1 and Table 2). All of the data suggested that the only difference between the two molecules was an additional sulfate in 2 at C-3′.
The molecular formula for the minor compound fibrosterol sulfate C (3), C54H84O19S4Na4, was derived from NMR data and the HRESIMS ion at m/z 1233.4167 ([M-Na]−, Δ−1.3 ppm). Compound 3 showed a similar 1H NMR spectrum and identical molecular formula when compared with 1, which suggested that 3 was also a sulfated bis-steroid. Comparison of the NMR data for fibrosterol sulfates A (1) and C (3) indicated that the ABCD and A′B′C′D′ rings are virtually indistinguishable, while the side chains are quite different (Table 1 and Table 3). The side chain of fibrosterol sulfate C (3) lacked the terminal olefin seen in 1, but contained a cis olefin (δH 5.29, J = 10.8; 5.02, J = 10.8), three singlet methyls, an oxygenated methine (δC 82.4; δH 4.16) and a deshielded quaternary carbon (δC 91.0). HMBC correlations from Me-21, Me-26, Me-27, Me-21′, and Me-26′ were essential for constructing the side chain that fuses rings D and D′ in 3. HMBC correlations from Me-21 to C-17, C-20 and C-22, Me-26 to C-24, C-25, C-27 and C-27′, Me-27 to C-24, C-25, C-26 and C-27′, and COSY correlations between H-22 and H-23, and H-23 and H-24, supported all connectivities between C-20 and C-27. HMBC correlations from Me-21′ to C-17′, C-20′ and C-22′, Me-26′ to C-24′, C-25′ and C-27′, and COSY correlations between H-22′ and H-23a/b′, and H-23a/b′and H-24′, supported all connectivities between C-20′ and C-27′. HMBC correlations from Me-26 and Me-27 suggested that C-25 and C-27′ were connected, which was further supported by HMBC correlations between H-27a/b′ and C-25, C-26, C-27, C-25′, C-26′. COSY correlations between H-24 and H-24′ and HMBC correlations from H-24 to C-24′, and H-24′ to C-24 supported a connection between C-24 and C-24′ to form the cyclopentane ring in 3. The molecular formula for fibrosterol sulfate C (3) required 11 degrees of unsaturation, suggesting that the final degree of unsaturation was an ether linkage between C-22′ and C-25′to give the oxabicyl[3,3,0]octane seen in 3. The unusual quaternary ether carbon 13C chemical shift (δC 91.0, C-25′) was compared to similar ring systems seen in ibhayinol (δC 91.4),24 and kuhistaferone (δC 100.4),25 and was consistent with the oxabicyl[3,3,0]octane for 3 as drawn.
Fibrosterol sulfate C (3) exhibited identical ROESY data to 1 for the ABCD and A′B′C′D′ rings, supporting a 2S, 3R, 6S, 2′S, 3′S, 6′S configuration for 3. ROESY data and molecular modeling were utilized to determine the relative configuration of the oxabicyl[3,3,0]octane in 3 (Figure 2 and Supporting Information). ROEs between Me-26′and H-24′supported a cis ring juncture, with both Me-26′ and H-24′ being in the α configuration. ROEs between Me-26′ and H-22′ indicated that H-22′ was also in the α configuration. ROEs between H-24′ and Me-27 established that Me-27 was in the α configuration, while ROEs between Me-26 and H-24 suggested that H-24 was in the β configuration. The large coupling constant observed between H-24 and H-24′ (J = 10.1 Hz) also supported a trans relationship between H-24 and H-24′. Based on these data, fibrosterol sulfate C (3) was assigned as 2S, 3R, 5S, 6S, 8S, 9S, 10R, 13S, 14S, 17R, 20R, 22Z, 24S*, 2′S, 3′S, 5′S, 6′S, 8′S, 9′S, 10′R, 13′S, 14′S, 17′R, 20′R, 22′R*, 24′S*, 25′R*. The relative configuration between C-24 and C-22′ could not be relayed through C-20′ due to overlapping signals; H-20′ and H-23β′ chemical shifts are identical (δH1.84), as well as H-21′ and H-17′(δH 1.00).
Fibrosterol sulfates A (1) and B (2) inhibited PKCζ with IC50 values of 16.4, and 5.6 µM, respectively. Fibrosterol sulfate C (3) was not tested for biological activity due to the limited amount of material isolated. The risk of a false-positive was eliminated by employing a counter screen that ensured 1 and 2 were not interfering with the signal detection. Compounds 1 and 2 were incubated with the phosphorylated ULight™-PKC peptide and the antibody, and the TR-FRET signals were measured at 665 nm. The TR-FRET signals remained constant when 1 and 2 were incubated with the phosphorylated ULight™-PKC peptide and the antibody, thereby eliminating the possibility of false-positive inhibition by this mechanism. Light scattering measurements also indicated that fibrosterol sulfates A (1) and B (2) were soluble at PKCζ assay concentrations, ruling out the possibility that 1 and 2 were false positives due to aggregate formation.
The spheciosterol sulfates A-C, isolated from a Spheciospongia sp., were recently shown to inhibit PKCζ and NF-κB activation.6 It was shown that the sterol side chain factors into their PKCζ activity; the longer side chain seen in spheciosterol sulfate C is 5-fold more active than spheciosterol sulfate B, and 10-fold more active than the shorter side chains seen in spheciosterol sulfate A and topsentiasterol sulfate E.6 The number of sulfates also appears to be important for PKCζ activity, as fibrosterol sulfate B (2) is 3-fold more active than fibrosterol sulfate A (1). Interestingly, the spheciosterol sulfates and the fibrosterol sulfates share a similar oxygenation pattern in the steroid rings, suggesting that the steroid oxygenation pattern may also be important for PKCζ inhibition.
Fibrosterol sulfates A–C (1–3) appear to be composed of two cholestene monomers, with differing configuration at C-3, and oxygenation at C-22 in only one monomer. These molecules would be excellent candidates for future biosynthetic studies.
Previous investigations of sponges from the genus Lissodendoryx have yielded steroids,26,27 pyrrololactams,28 cembranes,29 and polyether macrolides.30,31 Only a few steroid dimers have been isolated from marine organisms such as cephalostatins,32–39 crellastatins,40–42 ritterazines,43–46 hamigerols,47 bistheonellasterone,48 and amaroxocanes.49 The cephalostatins,32–39 ritterazines43–46 and bistheonellasterone48 are all fused between the A and A′ rings, while the crellastains,40–42 hamigerols,48 and amaroxocane B49 are fused through the side chains to form a dioxabicyclononane ring system. Amaroxocane A49 and crellastatin M42 contain a carbacylic ring fusion similar to the cyclopentane seen in fibrosterol sulfates A and B. Fibrosterol sulfate C (3) is unique in that there has never been a report of the oxabicyclo[3,3,0] octane in a dimeric sterol.
L. (A.) fibrosa (Lévi, 1961) (Coelosphaeridae), sample PC00-04-56, was collected by SCUBA from Coron Island (11° 57.833′N, 120° 06.311′E), northern Palawan, Philippines; a voucher specimen is maintained at the University of Utah.
The Lissodendoryx (Acanthodoryx) fibrosa specimen (PC00-04-56) was exhaustively extracted with MeOH to yield 2.60 g of crude extract. The crude extract was separated on HP20SS resin using a step gradient of H2O to IPA in 25% steps, and a final wash of 100% MeOH, to yield 5 fractions. The second fraction (0.46 g; 75/25 H2O/IPA) was chromatographed on C18 (32 × 10 cm) using a step gradient of 100% 0.2 M aq. NaCl to CH3CN in 10% steps, and a final wash of 100% MeOH, to yield 12 fractions (175A – 175L). The third HP20SS fraction (50/50 H2O/IPA) was also chromatographed on C18 (32 × 10 cm) using the same gradient, to yield 12 fractions (185A – 185L). Fractions 175D (20.4 mg; 30% CH3CN/70% 0.2 M NaCl in H2O), 175E (7.2 mg; 40% CH3CN/60% 0.2 M NaCl in H2O), 185E (27.4 mg; 40% CH3CN/60% 0.2 M NaCl in H2O) and 185F (15.4 mg; 50% CH3CN/50% 0.2 M NaCl in H2O) were combined. The combined fractions were purified by HPLC using a C18 column (250 × 10 mm) employing a gradient of 2% CH3CN/98% 0.2 M NaCl in H2O to 30% CH3CN/70% 0.2 M NaCl in H2O over 5 min, followed by a gradient of 30% CH3CN/70% 0.2 M NaCl in H2O to 50% CH3CN/50% 0.2 M NaCl in H2O at 4.5 mL/min over 32 minutes to yield fibrosterol sulfate A (1, 6.4 mg) eluting at 18.6 min and fibrosterol sulfate B (2, 8.0 mg) eluting at 16.1 min.
An aliquot of the third HP20SS fraction (186 mg; 50/50 H2O/IPA) was chromatographed on LH20 (24 × 2.5 cm) using MeOH as eluant, to yield 67 fractions (138.1–138.67). Fraction 138.4 was purified by HPLC using a C8 column employing a gradient of 10% CH3CN/H2O to 35% CH3CN/H2O at 4 mL/min over 33 minutes to yield 9 fractions (159A–159I). Fraction 159I was further purified by HPLC using a C18 column (250 × 10 mm) employing a gradient of 2% CH3CN/98% 0.2 M NaCl in H2O to 30% CH3CN/70% 0.2 M NaCl in H2O over 5 min, followed by a gradient of 30% CH3CN/70% 0.2 M NaCl in H2O to 50% CH3CN/50% 0.2 M NaCl in H2O at 4.5 mL/min over 32 minutes to yield fibrosterol sulfate C (3, 0.3 mg) eluting at 28.4 minutes.
Desalting of the flash column and HPLC fractions was achieved by filtering the samples through C18 Sep-Pak cartridges; salts were removed by washing with 100% H2O, and individual fractions or compounds were eluted with 100% MeOH.
Amorphous white solid, [α]22 D +16.2 (c 0.08, MeOH); UV (MeOH) λmax (log ε) 206 (3.70) nm; IR (film, NaCl) νmax 3377 (br), 1660, 1641, 1444, 1221, 1063, 966, 708 cm−1; 1H NMR and 13C NMR data, see Table 1; HRESIMS m/z 605.1459 [M−2Na]−2 (calcd for C54H84O19S4Na2, 605.1480).
Amorphous white solid, [α]22 D +19.1 (c 0.10, MeOH); UV (MeOH) λmax (log ε) 206 (3.91) nm; IR (film, NaCl) νmax 3224 (br), 1662, 1639, 1444, 1221, 968, 714 cm−1; 1H NMR and 13C NMR data, see Table 2; HRESIMS m/z 656.18432 [M−2Na]−2 (calcd for C54H83O22S5Na3, 656.18418).
Amorphous white solid,[α]20 D +29.2 (c 0.13, MeOH); UV (MeOH) λmax (log ε) 204 (4.04) nm; IR (film, NaCl) νmax 3346 (br), 2953, 1662, 1641, 1446, 1219, 1063, 962, 708 cm−1; 1H NMR and 13C NMR data, see Table 3; HRESIMS m/z 1233.4167 [M−Na]−(calcd for C54H84O19S4Na3, 1233.4183).
IC50 values were determined in a homogenous TR-FRET based PKCζ kinase activity assay (LANCE-Ultra, Perkin Elmer). Compounds 1 and 2, at various concentrations, were incubated with the ULight™-PKC peptide substrate (50 nM), ATP (2 µM), and PKCζ (25 pM). The reaction was stopped with EDTA (15 mM) after 60 min, Eu-labeled anti-phospho-PKC peptide antibody was added, and the extent of phospho-peptide product formation was determined through the measurement of the TR-FRET signals at 615 nm and 665 nm wavelengths upon excitation at 340 nm. A decrease in the TR-FRET signal ratio was observed as a function of increasing compound concentration and the IC50 for each compound was determined accordingly.
The authors would like to thank Dr. M. He, Dr. F. Ritacco, Y. Zhang, Dr. G. Schlingmann, Dr. F. E. Koehn, and Dr. G.T. Carter, Wyeth Research, for their kind support. We acknowledge the NIH, Grants RR13030, RR06262, and NSF Grant DBI-0002806 for funding NMR instrumentation in the University of Utah Health Sciences NMR Facility. This research was supported by NCI grant CA67786 and NIH grant CA36622 (C.M.I.).
Supporting Information Available: 1H spectra of 1–3, 13C spectra of 1 and 2, HSQC spectrum of 3, methods attempted to determine the configuration of C-22 and the cyclopentane ring in 1, a table of interatomic distances for 3 from molecular modeling, and a simplified 3D representation of 3. This material is available free of charge via the Internet at http://pubs.acs.org.