Histone deacetylases (HDACs) play an important role in the epigenetic regulation of gene expression by catalyzing the removal of acetyl groups from lysine residue of histone protein, stimulating chromatin condensation and promoting transcriptional repression.1, 2 HDACs are divided into four classes on the basis of their homology to yeast HDACs: class I (HDAC1, 2, 3 and 8), class IIa (HDAC4, 5, 7 and 9), class IIb (HDAC6 and 10), class III (SIRT1, 2, 3, 4, 5, 6 and 7) and class IV (HDAC11). As aberrant epigenetic changes are a hallmark of cancer, HDACs are a promising target for an anticancer drug. The inhibitors of HDACs can induce cell-cycle arrest, promote differentiation and stimulate tumor cell death. In fact, several HDAC inhibitors are currently in clinical trials both for solid and hematological malignancies.1, 2 Therefore, we attempted to search new HDAC inhibitors. As a result, we isolated a novel compound designated as JBIR-17 (1) from Streptomyces sp. 26634 (Figure 1). We report herein the isolation, structure elucidation and biological activity of 1.

Figure 1
figure 1

Structures of JBIR-17 (1) and trichostatin A (2).

Streptomyces sp. 26634 was isolated from a leaf of Kerria japonica collected in Iwata, Shizuoka Prefecture, Japan, and cultured on a rotary shaker (220 r.p.m.) at 28°C for 4 days in a 500-ml Erlenmeyer flask containing 60 ml of a medium consisting of 4% β-cyclodextrin, 0.5% glycerol, 2% Pharmamedia (Traders Protein, Lubbock, TX, USA), 0.0005% CuSO4·5H2O, 0.0005% MnCl2·4H2O and 0.0005% ZnSO4·7H2O.

n-BuOH (37.5 ml) was added to the fermentation broth (60 ml) and shaken for 15 min. After centrifugation, the organic layer was evaporated in vacuo. The dried residue (107 mg) was subjected to reversed-phase medium-pressure liquid chromatography (Purif-Pack ODS 100, Moritex, Tokyo, Japan) and eluted with a MeOH–H2O (5-100% MeOH) linear gradient system. The 70–90% MeOH eluate (3.7 mg) was further purified by reversed-phase HPLC using an XBridge Prep C18 column (5 μm optimum bed density (OBD), 4.6 i.d. × 250 mm, Waters, Milford, MA, USA) with 35% aqueous CH3CN containing 0.2% formic acid (flow rate, 1 ml min–1) to yield JBIR-17 (1, 0.9 mg; retention time (Rt) 9.7 min) and trichostatin A (2, 0.7 mg; Rt 10.3 min).3

The physicochemical properties of 1 are summarized in Table 1. Compound 1 was obtained as a colorless amorphous solid, and its molecular formula was determined to be C20H26N2O5 by HR-electrospray ionization (ESI)-MS. The IR spectrum revealed the characteristic absorptions of the aroyl and/or amide carbonyl (νmax 1652 cm–1) and amide N-H (νmax 1597 cm−1) groups. The structure of 1 was mainly determined by NMR spectral analyses as follows.

Table 1 Physicochemical properties of 1

The direct connectivity between each proton and carbon was established by the heteronuclear single quantum coherence spectrum, and the 13C and 1H NMR spectral data for 1 are shown in Table 2. A total of 20 signals were observed in the 13C NMR spectrum, consistent with the HR-MS data. These signals included three carbonyl (C-1, C-7, C-1′), four olefinic (C-2 to C-5) and six aromatic (C-8 to C-13) carbons. The proton spin couplings between two olefinic protons 2-H (δH 5.91) and 3-H (δH 7.28), between an olefinic proton 5-H (δH 6.04) and a methyl proton 6-Me (δH 1.28) through a methine proton 6-H (δH 4.36), and between two equivalent aromatic protons 9,13-H (δH 7.84) and 10,12-H (δH 6.69) on a p-disubstituted benzene ring were observed in a double-quantum filtered (DQF)-COSY spectrum as shown in Figure 2 (bold line). The constant time heteronuclear multibond correlation (CT-HMBC) experiment revealed the presence of 1H-13C long-range couplings from an N,N-dimethyl proton (δH 3.05) to an aromatic carbon C-11 (δC 155.2), from 9,13-H to C-11 and a carbonyl carbon C-7 (δC 198.7), from 10,12-H to an aromatic carbon C-8 (δC 127.0), from 6-Me, 6-H and 5-H to C-7, from the vinyl methyl proton 4-Me (δH 1.89) to three olefinic carbons C-3 (δC 147.5), C-4 (δC 132.5) and C-5 (δC 141.9), and from 2-H and 3-H to an amide carbonyl carbon C-1 (δC 168.6). The stereochemistries of two olefins were determined as 2E and 4E according to the coupling constant (J2,3=15.4 Hz) and the high-field-shifted 13C chemical shift at 4-Me (δC 12.6). Thus, the partial structure was elucidated as a trichostatic acid (3) moiety, and their 13C and 1H NMR signals are superimposable with those of 23 and 34.

Table 2 13C and 1H NMR data for 1
Figure 2
figure 2

Key correlations in DQF-COSY (bold line) and CT-HMBC (arrow) spectra of 1.

Additional substructure was elucidated as follows. A proton spin coupling between an α-methine proton 2′-H (δH 4.60; δC 54.8) and oxymethylene protons 3′-H (δH 4.18, 3.85; δC 62.1) was observed. A long-range coupling from 3′-H to a carboxylic carbonyl carbon C-1′ deduced that the remaining structure was a serine moiety, and the serine was assumed to attach to C-1 of trichostatin through an amide bond.

The linkage position and the absolute configuration of the serine moiety of 1 were confirmed as follows. To determine the absolute configuration of the serine moiety, Marfey's method was adopted. Compound 1 (0.8 mg) was hydrolyzed with 6 N HCl (0.2 ml) at 120°C overnight to obtain the serine residue. After acid hydrolysis, the reaction solution was adjusted to neutral pH and evaporated in vacuo. The residue was dissolved in an aqueous solution of 0.1 M NaHCO3 (0.6 ml), and 10 mM Nα-(5-fluoro-2,4-dinitrophenyl)-L-alaninamide (FDAA) in Me2CO (0.6 ml) was successively added. The mixture was kept at 70°C for 10 min with frequent shaking. After work-up with the addition of 0.2 N HCl, the filtered reaction mixture was subjected to ultra performance liquid chromatography (UPLC) analysis (Acquity UPLC BEH C18 1.7 μm, 2.1 × 50 mm, Waters; 10% aqueous CH3CN containing 0.1% formic acid; flow rate, 0.3 ml min–1). The authentic D- and L-serine were reacted with FDAA in the same manner as described above. The serine residue obtained from the hydrolysate was determined to be L-serine (Rt 8.6 min; L-Ser, 8.5 min; D-Ser, 9.8 min). To confirm the linkage position of serine moiety, 1 was semi-synthesized from 2 as shown in Figure 3. Briefly, 2 (7.5 mg) was converted to 3 (3.4 mg) by HClO4. Compound 3 was coupled with an O-t-butyl-L-serine t-butyl ester in the presence of PyBOP and N,N-diisopropylethylamine followed by deprotection in acidic condition to yield an L-serine adduct of 3 (0.9 mg). This synthetic compound showed an identical 1H NMR spectrum to that of naturally isolated 1 from Streptomyces sp. 26634.

Figure 3
figure 3

Scheme of chemical conversion from 2 to 1. (a) 1.5 N HClO4 aq, 50°C, overnight. (b) O-t-butyl-L-serine t-butyl ester, PyBOP, N,N-diisopropylethylamine in CH2Cl2/DMF (N,N-dimethylformamide), room temperature, 2 h. (c) 90% aqueous trifluoroacetic acid, room temperature, 1.5 h.

To evaluate inhibitory activity for HDACs of 1, we used the reporter gene assay system using a luciferase gene as described earlier.5, 6 The human embryonic kidney 293T cells, transformed with the luciferase reporter gene driven by the cytomegalovirus promoter, produced 2.5 times more luciferase compared with the untreated control, when they were treated with 1 at a concentration of 30 μM. Furthermore, to clarify the selectivity against HDAC subtypes, 1 was tested in the HDAC inhibitory activity using HDAC1 (class I), 4 (class IIa) and 6 (class IIb) enzymes of 293T cell origin, which are usually used as the representative HDACs among each HDAC subtype.7 Compound 1 showed inhibitory activity against HDAC4 and 6 with IC50 values of 69 and 4.7 μM, respectively, but no activity against HDAC1 at a concentration of 100 μM. In contrast, 2 showed strong, but not selective, inhibitory effects against these HDACs (IC50 values, 18, 30 and 92 nM against HDAC1, 4 and 6, respectively) as reported earlier.8, 9 These results indicated that 1 selectively inhibited HDAC6 compared with HDAC1 and 4. HDAC6 is a cytoplasmic enzyme that regulates many important biological processes, including cell migration, immune synapse formation, viral infection and the degradation of misfolded proteins. Furthermore, HDAC6 deacetylates tubulin, Hsp90 and cortactin.10, 11, 12 The diverse functions of HDAC6 suggest that it is a potential therapeutic target for a wide range of diseases. Thus, JBIR-17 could be a valuable tool for the studies of HDAC6 and enzymatic property among HDAC subtypes.