Marine-derived actinomycetes are an important source of a variety of novel natural products with comprehensive biological activities. In recent years, the rediscovery rate for known compounds from marine-derived actinomycetes is increasing [1]. Thus, novel methods have been developed for the rapid dereplication of known natural products and targeted identification of novel compounds, such as the imaging mass spectrometry method [2], the strain prioritization by real-time polymerase chain reaction [3], the resistance gene-guided genome mining [4], the bioinformatics-guided search strategy [5], and the CRISPR/Cas9-based editing-directed activation of silent clusters [6]. Nevertheless, heterologous expression is still thought as a promising tool for natural products studies and has been widely applied to enable the discovery of new compounds from silent biosynthetic gene clusters, and to improve the yield of natural products [7].

During our previous studies, the intact fluostatin gene cluster (fls) from South China Sea-derived Micromonospora rosaria SCSIO N160 has been successfully expressed in different heterologous hosts, such as Streptomyces coelicolor YF11 [8], and Streptomyces albus J1074 [9], to afford a variety of C–C/C–N-coupled dimeric or trimeric fluostatins, and several aromatic polyketides derived from different chain lengths and diverse cyclization patterns [9, 10]. A further careful inspection of the metabolite extracts of the recombinant strain S. albus J1074 harboring the fluostatin biosynthetic gene cluster revealed the presence of two minor untapped peaks, the UV spectra of which were strikingly distinct from the identified aromatic polyketides. Subsequent-targeted isolation afforded to two products, a new isoindolequinone albumycin (1) and a known isoquinolinequinone mansouramycin A (2) [11,12,13]. Herein we reported the detailed isolation, structural elucidation, and antibacterial activities of 1 and 2.

The 20 l of fermentation cultures of the recombinant strain S. albus J1074/pCSG5033 have been previously described, which led to the discovery of a number of dimeric or trimeric fluostatin analogues, and several aromatic polyketides differed in chain lengths and cyclization patterns [9, 10]. Further careful and detailed investigation on the crude extracts enabled the isolation and characterization of two trace compounds, including a new isoindolequinone albumycin (1) and the known isoquinolinequinone mansouramycin A (2) (Fig. 1).

Fig. 1
figure 1

Chemical structures of compounds 1 and 2, and selected key COSY and HMBC correlations of 1

Albumycin (1) was obtained as a red powder. Its molecular formula was established as C10H10N2O2 (m/z 191.0817 [M + H]+, calcd. for 191.0815) with seven degrees of unsaturation by high-resolution electrospray ionization mass spectroscopy (HRESIMS). The UV maximum absorption at 375 nm indicated a highly delocalized conjugated system. Analysis of the 1H NMR spectrum of 1 (Table 1) revealed the presence of a singlet methyl at δH 2.40, a doublet methyl at δH 2.68 (d, J = 5.1 Hz), and two olefinic protons at δH 5.12 (s) and 7.43 (d, J = 2.5 Hz). In addition, the presence of two amino proton signals were implied by a broad quartet at δH 7.01 (q, J = 5.1 Hz) and a broad singlet at δH 11.90. The DEPT 135 spectrum classified the 10 carbons in 1 as two methyls, two olefinic methines, and six quaternary carbons (four olefinic ones, and two keto signals at δC 177.0 and 181.7). The presence of a member of isoindole-4,7-diones [14] in 1 was supported by HMBC correlations from H-3 (δH 7.43) to C-1/C-3a/C-7a and H-6 (δH 5.12) to C-4/C-5/C-7/C-7a, and the COSY correlation between NH-2 and H-3 (Fig. 1). Location of the singlet methyl (Me-8, δH 2.40, s) at C-1 was strongly supported by HMBC correlation from H3-8 to C-1/C-7a (Fig. 1). Furthermore, the presence of the methylamino group (CH3NH−) was confirmed by the COSY correlation between H3-9 (δH 2.68, d, J = 5.1 Hz) and 5-NH (δH 7.01, q, J = 5.1 Hz) (Fig. 1). Location of the methylamino group at C-5 was based on HMBC correlations from H3-9 to C-5 and from 5-NH to C-4/C-6 (Fig. 1). Taken together, the structure of 1 was unambiguously determined to be 1-methyl-5-methylamideisoindole-4,7-dione, designated albumycin (1).

Table 1 1H and 13C NMR data for 1 and 2 (TMS, δ in ppm)

Detailed analysis of the 1H and 13C NMR data of compound 2 (Table 1) revealed that it was identical to isoquinolinequinone mansouramycin A (2), previously reported from a marine-derived Streptomyces sp. Mei37 [12]. 1H NMR (700 MHz, CD3OD) δ 8.93 (1H, s, H-1), 7.59 (1H, br s, 7-NH), 5.71 (1H, s, H-6), 2.92 (3H, br s, 7-NCH3), 2.76 (3H, s, 4-CH3), 2.69 (3H, s, 3-CH3). 13C NMR (176 MHz, CD3OD) δ 186.2 (C-5), 182.4 (C-8), 167.9 (C-3), 149.8 (C-7), 145.6 (C-1), 137.1 (C-4a), 133.5 (C-4), 125.2 (C-8a), 103.3 (C-6), 29.5 (NCH3), 24.7 (3-CH3), 16.7 (4-CH3).

Compounds 1 and 2 were evaluated for their antibacterial activities against five indicator strains: Staphylococcus aureus ATCC 29213, Acinetobacter baumannii ATCC 19606, Bacillus subtilis SCSIO BS01, Micrococcus Luteus SCSIO ML01, and methicillin-resistant S. aureus ATCC 43300 by measuring minimal inhibition concentrations (MIC). Compound 2 exhibited more potent antibacterial activities than compound 1 toward the tested strains (Table 2).

Table 2 Antibacterial activities of 1 and 2

In conclusion, a novel compound albumycin (1) and the known mansouramycin A (2) were isolated and identified from S. albus J1074 harboring the heterologous fluostatin biosynthetic gene cluster (fls). A variety of mansouramycin analogues have been previously identified from several actinomycetal strains [11,12,13]. Mansouramycins (such as 2) feature an isoquinolinequinone skeleton, in which a p-benzoquinone is fused to a six-membered pyridine ring. Albumycin (1) is different from mansouramycins by possessing an isoindolequinone scaffold, in which a five-membered pyrrole ring is fused to a p-benzoquinone. Obviously, no enzymes encoded in the fls-gene cluster could be found suitable to govern the biosynthesis of albumycin (1) and mansouramycin A (2), given the recent understanding of the fluostatin biosynthesis [8,9,10, 15, 16]. We proposed that the production of both 1 and 2 was probably associated with other biosynthetic gene clusters encoded in S. albus J1074, although 1 and 2 were not dectected in the native host S. albus J1074 under the same cultivation conditions. Consistent with this proposal, mansouramycin analogues have been reported from S. albus J1074 by activating silent gene clusters using chemical elicitors [13]. Also, there have been examples of producing secondary metabolites that are non-relevant to the heterologous biosynthetic gene clusters. For example, introduction of the thienamycin biosynthetic gene cluster from Streptomyces cattleya into S. albus J1074 led to the production of paulomycins and paulomenols but not thienamycins [17]. Similarly, production of actinorhodins were strongly activated in Streptomyces lividans TK21 through the transformation of the same thienamycin biosynthetic gene cluster [17]. Heterologous expression of the positive regulatory gene pimM of the pimaricin cluster from Streptomycesnatalensis activated the simultaneous production of candicidins and antimycins in S. albus J1074 [18]. The activation of a native biosynthetic pathway in the host through expressing a heterologous gene cluster reflects an extensive ‘cross-talk’ between pathway-specific regulators in different biosynthetic pathways that are previously reported in streptomycetes [19]. S. albus J1074 has been widely known as a heterologous host for expressing biosynthetic gene clusters from other actinomycetes [18]. Recent studies of genome mining and activation of ‘silent’ gene cluster have revealed that S. albus J1074 is also a producer for a variety diverse natural products [13, 18]. Herein, this study provides evidence that S. albus J1074 has the ability to produce isoindolequinone (albumycin, 1). However, the biosynthetic mechanism of 1 and 2 has not been well understood. The production of 1 and 2 might be an interaction between the host genes (such as the antimycin biosynthetic genes) in S. albus J1074 and the heterologous fluostatin biosynthetic genes (probably the fls regulators). Plausibly, 1 and 2 were proposed to be derived from peptide carrier protein (PCP)-tethered 3-formamidosalicyclic acid, a well established precursor en route to the biosynthesis of antimycins (Fig. 2) [20, 21]. Alternatively, after condensation of 3-formamidosalicyclic acid with l-threonine (or glycine), the resulting product would be released from the nonribosomal peptide synthetase (NRPS) assembly of antimycins, to form antimycic acid-like precursors, which would undergo further oxidoreduction and cyclization to generate 1 and 2 (Fig. 2). Nonetheless, experimental data are required to understand the intriguing biosynthetic machinery for 1 and 2. This study highlights again that heterologous expression plays an increasingly substantial role in the discovery of novel natural products.

Fig. 2
figure 2

Proposed antimycin-associated biosynthetic pathway of 1 and 2