Identification and evaluation of soft coral diterpenes as inhibitors of HIF-2α induced gene expression

Kidney cancer was the cause of almost 13,000 deaths in the United States in 2009, with clear cell carcinoma, the most common histological form of kidney cancer, responsible for a majority of deaths.1 Loss of function of the VHL tumor suppressor gene (von Hippel-Lindau disease) dramatically increases the risk of developing clear cell kidney carcinoma, as well as other tumors.2 Additionally, loss of VHL protein function is prevalent in clear cell renal carcinoma. The VHL protein has a number of functions, many of which appear to contribute to its role as a tumor suppressor, in particular its regulation of hypoxia inducible factor (HIF). HIF responds to changes in O2 levels in the cell and is responsible for mediating the transcriptional response to hypoxia. Under hypoxic conditions, or in the absence of functional VHL, HIF enhances the transcription of a number of downstream genes thought to be important in cancer.3 Of the three known HIFα gene products, HIF-2α appears to play a fundamental role in renal carcinoma.4,5 Therefore, a high throughput screen (HTS) was developed to identify small molecule inhibitors of HIF-2 gene expression that could potentially modulate downstream effectors of tumorigenesis.3 In the assay, HIF-2α transcription activity was monitored in the renal cell carcinoma cell line 786-O engineered with 5 copies of the minimal HIF-2α hypoxia responsive element (HRE) of the vascular endothelial growth factor (VEGF) linked to a luciferase reporter gene.3 The screen was performed on 146,814 natural product extracts sourced from a diverse collection of marine invertebrates, plants and fungi from the Natural Products Repository of the National Cancer Institute and yielded 153 confirmed active extracts. Three of the active extracts were from marine soft corals of the order Alcyonacea: Sarcophyton sp.,6 Lobophytum sarcophytoides7 and Asterospicularia laurae.8 Bioassay-guided fractionation led to the isolation of two new cembrane diterpenes, (4Z,8S*,9R*,12E,14E)-9-hydroxy-1-(prop-1-en-2-yl)-8,12-dimethyl-oxabicyclo[9.3.2]-hexadeca-4,12,14-trien-18-one (1), and (1E,3E,7R*,8R*,11E)-1-(2-methoxypropan-2-yl)-4,8,12-trimethyl-oxabicyclo[12.1.0]-pentadeca-1,3,11-triene (7), as well as eight known compounds, 2-6 and 8-10. Soft coral specimens were repeatedly extracted with CH2Cl2 – MeOH (1:1) and 100% MeOH according to the methodology outlined in McCloud9 to give the organic solvent crude extracts. A portion of the crude organic Sarcophyton sp. extract (165 mg) was subjected to a solvent-solvent partition, with the activity concentrated in the hexane and EtOAc fractions. The EtOAc fraction was subjected to size-exclusion LH-20 chromatography and semi-preparative C18 HPLC eluting with a gradient from MeCN-H2O (60:40) to 100% MeCN to yield the new natural product 1 (7.2 mg, 4.4 % crude extract weight) and a known cembrane 210,11 (2.3 mg, 1.4 % crude extract weight). The hexane fraction was subjected to reversed-phase C8 flash chromatography followed by normal-phase SiO2 flash chromatography to yield known compounds 3 (3.7 mg, 2.2% crude extract weight) and 4 (0.8 mg, 0.5 % crude extract weight).12 HRESIMS data for compound 1 revealed a molecular formula of C20H28O3, accounting for seven double bond equivalents. A comparison of the 1H and 13C NMR spectroscopic data13 with those observed for the known natural product 2,11 suggested a common cembrane core containing an α-s-unsaturated seven-membered lactone ring system. The major difference between compounds 1 and 2 centered on the C-1 isopropyl substituent, where the two doublet methyl resonances in 2 were replaced by a pair of broad singlets of an exo-methylene group (δH 4.91, δH 4.99; δC 112.1) and an allylic singlet methyl (δH 1.90; δC 21.2) in 1. Interpretation of the 2D HMBC NMR data associated with the two resonances and the extra degree of unsaturation in 1 were all consistent with an isopropenyl substituent at C-1. The structure of compound 1 was therefore concluded to be the 16,17-dehydro derivative of 2. The EE geometry of the Δ12,13 and Δ1,14 double bonds in 1 was confirmed upon the observation of a strong ROESY correlation between H-14 (δH 6.47, d, J = 11.6 Hz) and Me-20 (δH 1.83, s) as well as a 11.6 Hz coupling constant between H-13 (δH 5.72) and H-14 (δH 6.47), consistent with other related EE diene cembranes.11,12,14 The relative stereochemistry at C-9 was established by comparison of the H-9 and Me-19 1H NMR chemical shifts with that of the known natural product (4Z,8S,9R,12E,14E)-2 and the semi-synthetic C-9 epimer (4Z,8S,9S,12E,14E)-2. For the naturally-occurring (8S,9R)-2, the H-9 and Me-19 chemical shifts have been reported as δH 4.29 and δH 1.35, respectively,11 whereas the C-9 epimer (8S,9S)-2 had significantly upfield-shifted resonances for H-9 and Me-19 at δH 4.02 and δH 1.10, respectively.14 With the observed H-9 and Me-19 chemical shifts of δH 4.25 and δH 1.38, in agreement with those reported for (4Z,8S,9R,12E,14E)-2, the stereochemistry of C-9 in 1 was concluded to be 9R. The H-9 resonance did not show a ROESY correlation to Me-19, which suggested that the two were on the opposite face of the molecule and that the configuration at C-8 was S. Therefore, the configuration of the new natural product was assigned as (4Z,8S*,9R*,12E,14E)-1. Once purified, compound 1 was found to be unstable and it decomposed in solution as well as within a month at −20° C dry storage. A portion of the Lobophytum sarcophytoides organic extract (1.02 g) was subjected to a solvent-solvent partitioning scheme, concentrating the HIF-2α activity into the MeOtBu fraction. The MeOtBu fraction was subjected to two rounds of size exclusion chromatography on Sephadex LH-20 (2:5:1 hexanes/CH2Cl2/MeOH; 1:1 CH2Cl2/MeOH) followed by reversed-phase C18 flash chromatography to yield 5 (32.6 mg, 3.2 % crude extract weight) and 7 (1.2 mg, 0.1% crude extract weight). The molecular formula for 7, C21H34O2, was derived from NMR and HRESIMS data. Analysis of the spectroscopic data for 7,15 and comparison with the reported data for 6,16 indicated they were closely related. The major chemical shift differences between 6 and 7 occurred around the tertiary alcohol. The presence of a methoxyl signal in 7, the downfield shift of the quaternary oxygenated carbon (δC 78.0 in 7; δC 74.2 in 6), and the molecular formula for 7 all suggested that 7 was the methoxyl derivative of 6. HMBC correlations confirmed the location and presence of the methoxyl group in 7. Methanol and acetic acid were utilized in the isolation of 7. Attempts to re-isolate 7 without using MeOH were unsuccessful; compound 6 was isolated when MeOH was not used (2.1 mg, 0.6% crude extract weight). Therefore, compound 7 appears to be an artifact of isolation. A portion of the Asterospicularia laurae organic extract (229 mg) was separated by two Diol SPE cartridges (2 g resin each), and the equivalent fractions were combined to give five total fractions; Fraction 1 = 9:1 hexanes/CH2Cl2, Fraction 2 = 20:1 CH2Cl2/EtOAc, Fraction 3 = EtOAc, Fraction 4 = 5:1 EtOAc/MeOH, Fraction 5 = MeOH. Size exclusion chromatography of fraction 2 on Sephadex LH-20 using hexanes/CH2Cl2/MeOH (2:5:1) yielded 9 (56.3 mg, 24.6% crude extract weight). Size exclusion chromatography of fraction 1 on Sephadex LH-20 using hexanes/CH2Cl2/MeOH (2:5:1) followed by reversed-phase C4 flash chromatography yielded 8 (7.1 mg, 3.1% of crude extract weight) and 10 (7.6 mg, 3.3% of crude extract weight). The known compounds 2-6 and 8-10 were identified upon comparison of the [α]D, 1H NMR, 13C NMR and HRESIMS data with published values.10-12,16-23 Data on the in vitro HIF-2α activity24 of compounds 1-10 is outlined in Table 1. Cembranes 1-7 showed poor HIF-2α inhibition, and were not evaluated further. The xenicin-type diterpenes 8-10 had the lowest IC50 values with low cytotoxicity. They are currently undergoing further in vitro evaluation, the results of which will be published at a later date. Table 1 HIF-2α inhibitory activity of compounds 1-10