Diketopiperazines, inhibitors of sterol O-acyltransferase, produced by a marine-derived Nocardiopsis sp. KM2-16

Sterol O-acyltransferase (SOAT, also known as acyl-CoA:cholesterol acyltransferase (ACAT), EC 2.4.1.26), an endoplasmic reticulum (ER) membrane protein, catalyzes the synthesis of cholesteryl ester (CE) from free cholesterol and long-chain fatty acyl-CoA. SOAT has been postulated as a target for modulation by a new type of antiatherosclerotic agent. Recent molecular biological studies revealed the existence in mammals of two different SOAT isozymes, SOAT1 and SOAT2.^{1, 2, 3, 4} SOAT1-selective inhibition may cause detrimental effects,^{5, 6, 7} whereas SOAT2-selective inhibition has consistently shown antiatherosclerotic activity.^{8, 9} Therefore, it is important to determine the selectivity of inhibitors toward the two SOAT isozymes for development as new antiatherosclerosis agents. Our group has focused on the discovery of SOAT2-selective inhibitors of microbial origin in cell-based assay or an enzyme assay using microsomes prepared from SOAT2-expressing Chinese hamster ovary (CHO) cells. During the course of our screening program, two diketopiperazines, 1 and amauromine¹⁰ (2) (Figure 1), were isolated as SOAT2 inhibitors from the culture broth of actinomycete strain Nocardiopsis sp. KM2-16. Amauromine was originally isolated as a vasodilator from the culture broth of the fungus Amauroascus sp. Yin et al.¹¹ reported that 1 was produced by bioconversion from cyclo-L-tryptophan-L-tryptophan using two recombinant enzymes involved in acetylaszonalenin biosynthesis.¹² Thus, 2 and structurally related acetylaszonalenin were fungal secondary metabolites. In this study, we showed that 1 and 2 were isolated as actinomycete secondary metabolites and that 2 selectively inhibited SOAT2 activity.

Figure 1

Structures of 1, amauromine (2) and gypsetin (3).

The strain KM2-16 was isolated from sea sediments collected off Iriomote Island in Okinawa, Japan in 2012. In a BLAST search, the 16S rRNA sequence of the strain KM2-16 indicated that it could be considered to belong to the actinomycete genus Nocardiopsis. The strain was inoculated into a 500-ml Erlenmeyer flask containing 100 ml of production medium (1.0% starch, 0.40% yeast extract, 0.20% peptone, 0.10% CaCO₃, 0.010% KBr, 0.0040% Fe₂(SO₄)₃•nH₂O, 100 ml of natural seawater). Fermentation was carried out at 27 °C for 11 days under shaking conditions (180 r.p.m.). The culture broth (100 ml × 3) was extracted with acetone (300 ml). This extract was concentrated and extracted with EtOAc to yield the crude extracts (42 mg), which was purified by HPLC using a PEGASIL ODS column (10 × 250 mm², Senshu Scientific Co., Tokyo, Japan) under the following conditions: solvent, a 30-min linear gradient from 50 to 85% CH₃CN; flow rate, 3.0 ml min⁻¹; detection, UV at 285 nm. Compounds 1 and 2 were eluted as peaks with retention times of 14 and 29 min, respectively. Each peak was collected and concentrated to yield pure 1 (8.6 mg) and 2 (4.6 mg) as colorless solids. Interestingly, 1 and 2 were produced in the seawater-supplemented medium, whereas almost no production was observed in an analogous medium made with distilled water.

The physico-chemical properties of 1 and 2 are summarized in Table 1. The molecular formulas of 1 and 2 were determined to be C₂₇H₂₈N₄O₂ and C₃₂H₃₆N₄O₂ on the basis of HR-ESI-MS measurement, respectively. From ¹H and ¹³C NMR and specific rotation, 2 was identified as amauromine previously reported as a fungal vasodilator.¹⁰ Compound 1 had absorption maxima at 210, 219, 244, 282 and 290 nm in the UV spectrum. The IR absorption maxima of 1 at 3276 and 1664 cm⁻¹ suggested the presence of amino and carbonyl moieties, respectively. Although 1 appeared to be a known compound reported by Yin et al.,¹¹ the structural determination has not been described in detail. The structure of 1 was mainly elucidated by analysis of NMR spectra, including 2D NMR. The ¹³C NMR spectrum (in CDCl₃) showed 27 resolved signals, which were classified into two methyl carbons, two sp³ methylene carbons, three sp³ methine carbons, one sp² methylene carbon, 10 sp² methine carbons and 9 quaternary carbons, including two carbonyl carbons. The connectivity of proton and carbon atoms was established by the HMQC spectrum, as shown in Table 2. Analysis of the ¹H-¹H COSY spectrum gave the 7 partial structures I (NH-1 to C-2), II (C-4 to C-7), III (C-10 to C-11), IV (C-14 to C-17), V (C-19 to NH-20), VI (C-22 to C-25) and VII (C-1′ to C-2′) drawn with the bold lines in Figure 2. The ¹H-¹³C long-range couplings of ²J and ³J observed in the HMBC experiments gave the following information. (1) The cross-peaks were observed from 1-NH (δ 5.06) to C-3 (δ 61.6) and C-9 (δ 128.9), from 2-H (δ 5.55) to C-8 (δ 149.9), C-9, C-11 (δ 59.0) and C-13 (δ 166.0), from 4-H (δ 7.14) to C-3 and C-8, from 5-H (δ 6.76) to C-9, from 6-H (δ 7.11) to C-8, from 7-H (δ 6.61) to C-9, from 10-H₂ (δ 2.42, 2.51) to C-3, C-9 and C-16 (δ 168.9), from 11-H (δ 3.91) to C-10 and C-16, from 14-H (δ 4.30) to C-13, and from 15-NH (δ 5.69) to C-11, C-13, C-14, C-16 and C-17 (27.0), suggesting the presence of 6-, 5-, 5- and 6-membered ring systems (Part A), including indoline and diketopiperazine rings, which contained the partial structures I, II, III and IV. (2) The cross-peaks were observed from 19-H (δ 7.08) to C-18 (δ 109.7), C-21 (δ 136.6) and C-26 (δ 126.6), from NH-20 (δ 8.19) to C-18, C-21 and C-26, from 22-H (δ 7.38) to C-26, from H-23 (δ 7.22) to C-21, from 24-H (δ 7.12) to C-26 and from 25-H (δ 7.55) to C-18, C-21 and C-26, suggesting the presence of the indole ring (Part B) containing the partial structures IV and V. (3) The cross-peaks were observed from 1′-H₂ (δ 5.08, 5.12) to C-3′ (δ 40.8), from 2′-H (δ 5.97) to C-3, C-4′ (δ 22.8) and C-5′ (δ 22.4), from 4′-H₃ (δ 1.01) to C-2′ (δ 143.5), C-3′ and C-5′ and from 5′-H₃ (δ 1.11) to C-2′, C-3′ and C-4′, suggesting the presence of the prenyl moiety (Part C) containing the partial structure VII. Finally, (4) the cross-peaks from 17-H₂ (δ 2.97, 3.74) to C-19 (δ 123.2) and C-26, from 19-H to C-17, from 2-H and 10-H₂ to C-3′ and from 2′-H, 4′-H₃ and 5′-H₃ to C-3 indicated that Part A is attached to Parts B and C, as shown in Figure 2. The structure satisfied the degrees of unsaturation and the molecular formula.

Table 1 Physico-chemical properties of 1 and 2

Table 2 ¹H and ¹³C NMR chemical shifts of 1 (600 MHz for ¹H, 150 MHz for ¹³C)

Figure 2

Key cross-peaks observed in ¹H-¹H COSY and ¹H-¹³C HMBC experiments of 1.

The relative configurations of C-2, C-3, C-11 and C-14 were elucidated by NOE experiments. As shown in Figure 3, the cross-peaks between 2-H and 4′-H₃/5′-H₃ proved the relative configurations, 2S* and 3R*. Furthermore, the correlation between H-11 and H-14 proved the relative configurations, 11S* and 14R*. As shown in Table 1, the CD spectra of 1 showed positive Cotton effects at 272 and 220 nm and negative Cotton effects at 300 and 248 nm. These data suggested that 1 has the same absolute configurations, 2S, 3R, 11S and 14R, as 2.¹³

Figure 3

Key NOE experiment of 1.

SOAT inhibitory activity of 1 and 2 was investigated in the enzyme assay using microsomes prepared from SOAT1- and SOAT2-expressing CHO (hereafter referred to as SOAT1-CHO and SOAT2-CHO, respectively) cells.¹⁴ As summarized in Table 3, 1 and 2 are SOAT inhibitors rather selective toward SOAT2 isozyme in the enzyme assay. Furthermore, the SOAT inhibition was evaluated in a cell-based assay using SOAT1- and SOAT2-CHO cells.¹⁴ As shown in Table 3, 2 inhibited CE synthesis with an IC₅₀ value of 0.45 μM in SOAT2-CHO cells, and it became clear that 2 is a SOAT2-selective inhibitor with a selective index (SI) value of 62 in the cell-based assay. However, 1 showed no inhibition of SOAT1 and SOAT2 at 22 μM. It might be that 1 cannot penetrate CHO cells. Shinohara et al.¹⁵ reported that structurally related gypsetin (3), produced by the fungus Nannizzia gypsea var. incurvata IFO9228, inhibited rat liver microsomal SOAT activity (IC₅₀, 18 μM) and cholesteryl ester synthesis in macrophage J774 (IC₅₀, 0.65 μM). Now it is known that J774 cells exclusively express SOAT1 and rat liver mainly expresses SOAT2. Therefore, it will be worth testing 3 in our assay system to make the SOAT selectivity clear.

Table 3 Effects of 1 and 2 on SOAT isozymes in cell-based and enzyme-based assays

A number of indoline alkaloid-containing diketopiperazines carry a prenyl moiety at C3. For example, amauromine (2), gypsetin (3), epiamauromine,¹⁶ roquefortine C¹⁷ and fructigenines A and B¹⁸ were discovered as fungal metabolites, mainly produced by Penicillium and Aspergillus. In this study, we discovered 1 and 2 from a marine-derived actinomycete, Nocardiopsis sp. KM2-16. It will be worth comparing the biosynthetic genes from fungal and actinomycete strains.