Extending the Structural Diversity of Labdane Diterpenoids from Marine-Derived Fungus Talaromyces sp. HDN151403 Using Heterologous Expression

Abstract

Heterologous biosynthesis has become an effective means to activate fungal silent biosynthetic gene clusters (BGCs) and efficiently utilize fungal genetic resources. Herein, thirteen labdane diterpene derivatives, including five undescribed ones named talarobicins A–E (3–7), were discovered via heterologous expression of a silent BGC (labd) in Aspergillus nidulans. Their structures with absolute configurations were elucidated using extensive MS and NMR spectroscopic methods, as well as electronic circular dichroism (ECD) calculations. These labdanes belong to four skeleton types, and talarobicin B (4) is the first 3,18-dinor-2,3:4,18-diseco-labdane diterpene with the cleavage of the C2–C3 bond in ring A and the decarboxylation at C-3 and C-18. Talarobicin B (4) represents the key intermediate in the biosynthesis of penioxalicin and compound 13. The combinatorial heterologous expression and feeding experiments revealed that the cytochrome P450 enzymes LabdC, LabdE, and LabdF were responsible for catalyzing various chemical reactions, such as oxidation, decarboxylation, and methylation. All of the compounds are noncytotoxic, and compounds 2 and 8 displayed inhibitory effects against methicillin-resistant coagulase-negative staphylococci (MRCNS) and Bacillus cereus.

1. Introduction

Labdane-related diterpenes are widely distributed secondary metabolites discovered from fungi, insects, higher plants, and marine organisms [1], possessing high structural diversity and a broad spectrum of biological activities [2,3,4,5,6,7,8,9,10]. The basic core structure of labdane diterpenes is composed of a fused decalin system (C1–10) and a branched six-carbon side chain (C11–16, with C16 attached to C13) at C9, generally with additional four carbons (C17–20) on C8, C4 (with C18 and C19 being attached), and C10 of the decalin system, respectively [6,11]. Originating from geranylgeranyl diphosphate (GGPP), two classes of specific diterpene synthases (diTPSs) are involved in the biosynthesis of labdane diterpenes [12,13,14,15,16]: (1) mostly the class II diTPS, copalyl pyrophosphate synthase (CPS), catalyzing the cyclization of GGPP to produce a bicyclic diphosphate intermediate (formed labdane- [17], clerodane- [18] and halimane-related [19] diterpenoids by removing the phosphate group, Figure S1), and (2) class I diTPS, converting into the final diterpenoids (including kaurane-, abietane-, pimarane-, beyerane-, atisane-, canssane-, stemodane-, and manoyl oxide-types) [20].

The reported class II labdane diTPSs are dual functional enzymes composed of terpene cyclase and prenyltransferase domain, for example, PvCPS (GenBank: LC316181.1) and PfCPS (JGI protein ID: 403578), which were discovered from Penicillium verruculosum TPU1311 and Penicillium fellutanum ATCC 48694, respectively [15]. The C-terminal prenyltransferase domain of PvCPS catalyzes the tandem condensation of C5 dimethylallyl diphosphate (DMAPP) with three successive isopentenyl diphosphates (IPPs) to form geranylgeranyl diphosphate (GGPP), which is subsequently cyclized to form copalyl diphosphate (CPP) catalyzed by the N-terminal terpene cyclase domain (Figure S1) [15]. Although the mechanism of diTPSs are insightfully understood, the secondary metabolites of the biosynthetic gene clusters (BGCs) harboring diTPS, such as PvCPS/PfCPS, especially those with various tailoring enzymes, are still poorly explored.

In this study, an attractive gene cluster (named labd) containing PvCPS homologous protein and various tailoring enzymes was found in the Antarctic sponge-derived fungus Talaromyces sp. HDN151403 based on bioinformatic analysis. Due to no labdane-related diterpene detected in the wild-type, reconstructing the entire biosynthetic pathway in genetically treatable fungal hosts may address this issue. Genetically tractable filamentous fungi such as Aspergillus nidulans have compatible transcription, translation, post-translational modification, and secretion machineries for the expression of foreign fungal genes [21], and therefore, the heterologous expression of the labd cluster in Aspergillus nidulans A1145 was attempted, which led to the production of 13 labdane diterpene derivatives, including five undescribed ones (namely talarobicins A–E). Structurally, these diterpenoids could be classified into four types, 3,18-dinor-2,3:4,18-diseco-, 18-nor-4,18-seco- and 3-nor-2,3-seco-labdane, and classical bicyclic labdane diterpenoids. Talarobicin B (4) is the first 3,18-dinor-2,3:4,18-diseco-labdane diterpene and served as the key intermediate in generation of labdane diterpenes with ring A cleaved and rearranged. Based on biosynthetic pathway reconstruction and feeding experiments in A. nidulans, the biosynthetic pathway of these compounds was proposed, which involved P450-catalyzed multiple oxidation.

2. Results and Discussion

2.1. Prediction of the Labdane-Related Diterpenes BGC in Talaromyces sp. HDN151403

By using the PvCPS gene (GenBank: LC316181.1) as a query to LocalBlast on our sequenced fungal genomes, a series of conservative homologous gene clusters were found in several fungi (Figure 1). The orthologous gene cluster was also conserved in Talaromyces and Penicillium fungi when using the remote search tool cblaster [22] to search the online database (Figures S2 and S3). These clusters contain two conserved genes encoding a phosphatase and a bifunctional terpene synthase, adjacent to various post-modification genes (Figure S3). Among them, a gene cluster (labd) from Talaromyces sp. HDN151403 is attractive for the abundant oxidoreductase. The labd gene cluster contains a bifunctional diterpene synthase LabdA, a phosphatase LabdB, a short-chain dehydrogenase/reductase (SDR) LabdG, an aldehyde dehydrogenase LabdD, two P450 monooxygenases LabdE and LabdF, and a dual domain P450-methyltransferase gene labdC (Figure 1, Table S3). Detailed bioinformatics analysis indicate that the diterpene synthase LabdA possesses both type II terpene cyclase and prenyltransferase domains with 89% identity to PvCPS and 59% identity to PfCPS. The sequence similarity search results indicate that the P450s LabdC, LabdE, and LabdF have high similarity with many sequences in the public database, but the functions are unknown (Figures S6–S8). Further phylogenetic analysis indicated that LabdE and LabdF may have the ability to catalyze multiple oxidation reactions (Figure S4), while the domain analysis of LabdC suggested that it may have both P450 and methyltransferase catalytic activities (Figure S5).

2.2. Heterologous Expression the labd BGC in A. nidulans A1145

The reverse transcriptase PCR (RT-PCR) results show that the labd gene cluster was not expressed under laboratory culture conditions (Figure S9). To activate this cryptic gene cluster, we expressed the genes in the engineered A. nidulans A1145. The transformants of A. nidulans were grown on solid CD-ST medium for 5 days and extracted with ethyl acetate (EtOAc). Two compounds (1 and 2) were obtained from the core genes transformant AN-labdAB, and eleven compounds (3–13), including five undescribed ones, talarobicins A–E (3–7), were obtained from the whole genes transformant AN-labdABCDEFG (Figure 2 and Figure 3, Figure S10).

Compound 3 was obtained as a colorless powder. The molecular formula was determined to be C₂₀H₃₂O₄ (five degrees of unsaturation) on the basis of the cationic molecular ion peak at m/z 354.2648 [M + NH₄]⁺ in the high-resolution electrospray ionization mass spectrometry (HRESIMS) (Figure S18). The ¹H NMR spectrum of 3 (

Table 1. ¹H and ¹³C NMR spectroscopic data for compounds 3 and 4 in CD₃OD (125 and 500 MHz).

No.	3		4
	δC, Mult	δH (J in Hz)	δC, Mult	δH (J in Hz)
1a	49.0, CH2	2.08, m	42.9, CH2	2.49, d (15.7)
1b		1.00, m		2.14, d (15.7)
2	65.5, CH	3.81, m	174.3, C	–
3a	45.0, CH2	2.17, m	–	–
3b		0.89, m
4	41.5, C	–	215.1, C	–
5	56.9, CH	1.26, m	63.6, CH	3.39, d (11.0)
6a	25.2, CH2	1.85, m	71.5, CH	3.85, td (11.0, 5.0)
6b		1.34, m
7a	39.5, CH2	2.42, m	47.6, CH2	2.65, dd (12.0, 5.0)
7b		1.96, m		2.02, overlapped
8	148.7, C	–	145.3, C	–
9	57.4, CH	1.70, m	49.5, CH	2.27, m
10	41.8, C	–	42.0, C	–
11a	23.0, CH2	1.75, m	23.8, CH2	1.82, m
11b		1.57, m		1.58, m
12a	40.7, CH2	2.31, m	40.7, CH2	2.29, m
12b		2.02, m		2.02, overlapped
13	161.9, C	–	161.8, C	–
14	116.8, CH	5.62, d (1.2)	116.8, CH	5.64, d (1.5)
15	170.2, C	–	170.3, C	–
16	18.9, CH3	2.14, d (1.2)	19.0, CH3	2.15 d (1.5)
17a	107.8, CH2	4.90, s	111.0, CH2	5.08, s
17b		4.57, s		4.75, s
18a	65.5, CH2	3.61, d (11.5)	–	–
18b		3.29, d (11.5)
19	28.1, CH3	1.01, s	35.2, CH3	2.32, s
20	16.7, CH3	0.74, s	17.9, CH3	0.81, s

, Figure S12) displayed three olefinic protons (δ_H 5.62, d, J = 1.2 Hz; δ_H 4.90, s; δ_H 4.57, s), two oxygenated methylene protons (δ_H 3.61, d, J = 11.5 Hz; δ_H 3.29, d, J = 11.5 Hz), one oxygenated methine proton (δ_H 3.81, m), and three methyl groups (δ_H 2.14, d, J = 1.2 Hz; δ_H 1.01, s; δ_H 0.74, s). The ¹³C and heteronuclear single quantum coherence (HSQC) NMR experiments showed 20 carbon resonances, including five non-protonated carbons (one carbonyl at δ_C 170.2, two olefinic carbons at δ_C 148.7 and 161.9, and two sp3 carbons at δ_C 41.5 and 41.8), eight methylenes (one olefinic carbon at δ_C 107.8, one oxygenated carbon at δ_C 65.5, and six sp3 methylene carbons at δ_C 23.0, 25.2, 39.5, 40.7, 45.0, and 49.0), four methines (one olefinic carbon at δ_C 116.8, one oxygenated carbon at δ_C 65.5, and two sp3 carbons at δ_C 56.9 and 57.4), and three methyls (δ_C 16.7, 18.9, 28.1) (Table 1, Figures S13 and S14).

The planar structure of 3 was determined via detailed analysis of its two-dimensional (2D) NMR data (Figures S14–S17). Firstly, the ¹H-¹H correlation spectroscopy (COSY) spectrum revealed three partial structures, as shown in Figure 4, on the basis of the correlations between H₂-1 and H-2, H-2 and H₂-3, H₂-6 and H₂-7, H-9 and H₂-11, and H₂-11 and H₂-12. Further analysis of the heteronuclear multiple-bond connectivity (HMBC) spectrum key correlations from H-5 to C-4/C-6/C-10, from H₂-18 and Me-19 to C-3/C-4/C-5, from H-9 to C-8/C-10/C-11, from Me-20 to C-1/C-5/C-9/C-10, from H₂-17 to C-7/C-8/C-9, and from Me-16 to C-12/C-13/C-14/C-15, along with H-14 to C-12/C-13/C-15, revealed the formation of a labdane-related diterpene skeleton deriving from (+)-CPP [15]. The chemical shifts of the carbonyl carbon at C-15 (δ_C 170.2) and the two oxygenated carbons, including C-2 (δ_C 65.5) and C-18 (δ_C 65.5), located the corresponding positions of the carboxyl group and the remaining two hydroxy groups, respectively.

The relative configuration of 3 was deduced via the rotating frame Overhauser effect spectroscopy (ROESY) experiment (Figure 4 and Figure S17). The ROESY correlations of H₂-18/Me-20/H₂-11 and the ROESY correlation between Me-20 and H-2 indicated that H₂-18, Me-20, H₂-11, and H-2 were at the same side (β-orientation). The ROESY correlations of Me-19/H-5/H-9 suggested that Me-19, H-5, and H-9 were on the same side (α-orientation). The double bond between C-13 and C-14 was assigned E-geometry as a result of the ROESY correlation between H-14 and H₂-12.

As for the absolute configuration of 3, ECD calculations of the enantiomers (2R*,4S*,5R*,9S*,10R*)-3 and (2S*,4R*,5S*,9R*,10S*)-3 were carried out using B3LYP/6-31G(d)-GD3BJ optimized geometries at the B3LYP/6-311+g(d,p) level in methanol. The experimental ECD spectrum of 3 showed an ECD curve with a positive Cotton effect at 223 nm (Figure S11). The calculated ECD spectrum for (2R*,4S*,5R*,9S*,10R*)-3 showed a similar ECD curve with Cotton effects at 225 (+) nm and 259 (−) nm, allowing the assignment of the 2R,4S,5R,9S,10R configuration for 3 (Figure S11).

Compound 4 was isolated as a colorless powder. HRESIMS analysis established the molecular formula of 4 to be C₁₈H₂₆O₆, indicating six degrees of unsaturation (Figure S26). The ¹H NMR spectrum of 4 displayed three olefinic protons (δ_H 5.64, d, J = 1.5 Hz; δ_H 5.08, s; δ_H 4.75, s), one oxygenated methine proton (δ_H 3.85, td, J = 11.0, 5.0 Hz), and three methyl groups (δ_H 2.32, s; δ_H 2.15, d, J = 1.5 Hz; δ_H 0.81, s) (Table 1, Figure S20). Based on the analysis of ¹³C NMR and HSQC spectra (Table 1, Figures S21 and S22), 18 carbons could be designated as three methyls, five methylenes (one olefinic carbon at δ_C 111.0 and four sp3 methylene carbons at δ_C 23.8, 40.7, 42.9, and 47.6), four methines (one olefinic carbon at δ_C 116.8, one oxygenated carbon at δ_C 71.5, and two sp3 carbons at δ_C 49.5 and 63.6), and six non-protonated carbons (three carbonyls at δ_C 215.1, 174.3, and 170.3, two olefinic carbons at δ_C 145.3 and 161.8, and one sp3 carbons at δ_C 42.0), respectively.

The planar structure of 4 was determined via careful analysis of its 2D NMR data (Figures S22–S25). The ¹H-¹H COSY spectrum revealed two partial structures, as shown in Figure 4. The key HMBC correlations from H-5 to C-4/C-6/C-10, from Me-19 to C-4/C-5, from Me-20 to C-1/C-2/C-5/C-9/C-10, and from Me-16 to C-12/C-13/C-14/C-15, along with from H-14 to C-12/C-13/C-15, revealed the formation of a ring-opening labdane-related diterpene skeleton. The relative configuration of 4 was deduced via the ROESY experiment (Figure 4 and Figure S25). Specifically, the ROESY correlations between Me-20 and H-6, Me-20 and H₂-11, and H-5 and H-9 indicated that Me-20, H₂-11, and H-6 were at the same side (β-orientation) while H-5 and H-9 were on the other side (α-orientation). The ROESY correlation between H-14 and H₂-12 indicated that the double bond between C-13 and C-14 was E-geometry. Finally, the calculated ECD spectrum for (5S*,6S*,9S*,10R*)-4 showed a similar ECD curve with that of 4. Thus, the absolute configuration of 4 was determined as 5S,6S,9S,10R (Figure S11).

The molecular formulas of compounds 5–7 were determined to be C₂₀H₃₀O₄, C₂₀H₃₀O₄, and C₂₀H₃₀O₅ on the basis of HRESIMS data (Figures S34, S42, and S50), respectively. The similarity of the NMR data of 5–7 and callilongisin D [25] suggested that these compounds were close analogues. The ¹H-¹H COSY correlations of H-1/H-2/H-3, the HMBC correlations from Me-18 and Me-19 to C-3/C-4/C-5, and the HMBC correlations from Me-20 to C-1/C-5/C-9/C-10, along with the C-2 chemical shift of 5 (δ 64.6) (

Table 2. ¹H and ¹³C NMR spectroscopic data for compounds 5–7 in CD₃OD (125 and 500 MHz).

No.	5		6		7
	δC, Mult	δH (J in Hz)	δC, Mult	δH (J in Hz)	δC, Mult	δH (J in Hz)
1a	48.4, CH2	2.16, m	39.6, CH2	1.91, m	48.3, CH2	2.17, m
1b		1.22, m		1.35, m		1.25, m
2a	64.6, CH	3.83 m	18.8, CH2	1.57, m	64.1, CH	3.82, m
2b				1.50, m
3a	52.6, CH2	1.65, m	37.2, CH2	1.94, m	45.6, CH2	2.25, m
3b		1.14, m		0.96, m		0.91, m
4	35.1, C	–	39.0, C	–	41.0, C	–
5	63.6, CH	2.21, s	65.5, CH	2.43, s	64.6, CH	2.44, s
6	202.0, C	–	203.0, C	–	202.1, C	–
7	129.2, CH	5.77, s	128.8, CH	5.76, s	129.0, CH	5.78, s
8	161.9, C	–	162.6, C	–	162.4, C	–
9	57.0, CH	2.32, m	57.2, CH	2.27, m	57.2, CH	2.35, m
10	45.8, C	–	44.5, C	–	45.7, C	–
11a	26.5, CH2	1.69, m	26.5, CH2	1.66, m	26.6, CH2	1.70, m
11b				1.61, m
12a	43.9, CH2	2.49, m	43.9, CH2	2.48, m	43.9, CH2	2.49, m
12b		2.27, m		2.25, m		2.29, m
13	160.5, C	–	160.6, C	–	160.5, C	–
14	117.5, CH	5.74, d (1.2)	117.5, CH	5.73 d (1.2)	117.5, CH	5.74, d (1.2)
15	170.1, C	–	170.1, C	–	170.0, C	–
16	19.0, CH3	2.20, d (1.2)	19.0, CH3	2.18, d (1.2)	19.0, CH3	2.20, d (1.2)
17	22.3, CH3	2.02 s	22.4, CH3	2.01, s	22.3, CH3	2.02, s
18a	22.9, CH3	1.20, s	63.7, CH2	4.10, d (11.0)	64.3, CH2	4.11, d (11.0)
18b				3.63, d (11.0)		3.57, d (11.0)
19	33.8, CH3	1.17, s	27.3, CH3	1.15, s	27.5, CH3	1.20, s
20	15.9, CH3	0.91, s	15.8, CH3	0.84, s	16.7, CH3	0.88, s

), which was downfield compared to that of callilongisin D (δ 18.3), indicated the presence of a -OH group substituted at C-2 in 5 (Figures S31 and S32). The shift value of C-18 (δ 63.7) in 6 was downfield compared to that of the same carbon in callilongisin D (δ 33.6), suggesting that a -OH group was substituted at C-18 in 6, as evidenced by the HMBC correlation from H₂-18 to C-3/C-4/C-5/C-19 (Figure S39). The C-2 chemical shift of 7 (δ 64.1) was downfield compared to that of callilongisin D (δ 18.3), and the shift value of C-18 (δ 64.3) was downfield compared to that of the same carbon in callilongisin D (δ 33.6), indicating that a -OH group was substituted at C-2 and C-18 of 7, respectively. The ROESY correlations of Me-20/H-2, Me-20/Me-18, Me-18/H-2, Me-20/H₂-11, Me-19/H-5, and H-5/H-9 in 5, of Me-20/H₂-18, Me-20/H₂-11, Me-19/H-5, and H-5/H-9 in 6, and of Me-20/H-2, Me-20/H₂-18, H₂-18/H-2, Me-20/H₂-11, Me-19/H-5, and H-5/H-9 in 7 indicated that the stereostructure of 5–7 is highly similar (Figure 4, Figures S33, S41, and S49). The ROESY correlation between H-14 and H₂-12 indicated that the double bond between C-13 and C-14 of 5–7 was E-geometry. Finally, the good agreement between the experimental and calculated ECD curves evidenced the absolute configuration of 5–7 as 2S,5S,9S,10R, 4S,5R,9S,10R, and 2R,4S,5R,9S,10R, respectively (Figure S11).

According to the comparison of NMR and MS data with those reported in the literature, the eight known compounds obtained in this study were identified as 3β-hydroxy-8(17),13(E)-labdadien-15-oic acid (1) [26], 3β-hydroxy-labd-8(17)-en-15-oic acid (2) [27], labda-8(17),13(E)-diene-15,19-dioic acid (8) [28], 2α-hydroxy-labda-8(17),13(E)-diene-15,19-dioic acid (9) [29], paecilomycine B (10) [29], penicichrysogene B (11) [30], penitholabene (12) [31], and 5-(4-carboxy-3-methyl-3-buten-1-yl)octahydro-4-(methoxycarbonyl)-3-methyl-6-methylene-2-oxo-4-benzofuranacetic acid (13) [32].

All of the isolated compounds were evaluated for their antimicrobial activity against methicillin-resistant coagulase-negative staphylococci (MRCNS), methicillin-resistant Staphylococcus aureus (MRSA), Acinetobacter baumannii, Bacillus cereus, Pseudomonas aeruginosa, S. aureus, and Canidia albicans. Compounds 2 and 8 displayed weak inhibitory activity against B. cereus (MIC 128 µg/mL) and MRCNS (MIC 128 µg/mL), respectively. Additionally, these compounds were not cytotoxic against the K562, ASPC-1, MDA-MB-231, H69AR, and L-02 cell lines, with inhibition lower than 50% at a concentration of 30 μM.

2.3. Heterologous Reconstitution of the Labdane Diterpenes Biosynthetic Pathway

To investigate the biosynthetic pathway of isolated compounds, we sought to reconstitute its biosynthesis in A. nidulans. Firstly, compounds 1 and 2 were produced from the transformant AN-labdAB (Figure 5A, trace iii). Structurally, 2 is the product of the reduction of the C13/C14 double bond of 1, which is inferred to be catalyzed by the endogenous oxidoreductase of A. nidulans. Subsequently, 5 and 8 were produced from the two three-gene-expressing transformants AN-labdABE and AN-labdABF, respectively (Figure 5A, traces iv and v). Compounds 3 and 6–12 were produced by the four-gene-expressing transformants labdABEF (Figure 5B, trace iii). In addition, trace amounts of 4 and 13 were produced by strains AN-labdABDEF and AN-labdABCDEF, respectively (Figure 5B, traces iv and v). Only 5 was produced from the transformant strain AN-labdABCDEG (Figure 5B, trace vi), while 8 was extensively produced in the strain AN-labdABCDFG (Figure 5A, trace xi), indicating that the two genes labdE and labdF catalyze the most critical reaction in the biosynthesis pathway of other compounds.

Next, we aimed to reconstitute the complete biosynthetic pathway using A. nidulans feeding experiments. As shown in Figure 6A,B, LabdE can catalyze the conversion from 6 to 7 and 8 to 9–12, respectively, indicating that LabdE is a multifunctional cytochrome P450 enzyme that catalyzes multiple oxidization. The strain AN-labdG fed 11 and produced 9, indicating that LabdG can catalyze the reduction of the ketone group at C-2 in 11 (Figure 6A, traces iii and iv). To identify whether 4 and 13 are products of the labd gene cluster, strain AN-labdCDEFG was constructed and fed 8 and 9 separately. As a result, only 8 was successfully converted to 4 and 9–13 (Figure 6B, traces i and ii). Aside from that, 4 and 13 were also successfully produced from strain AN-labdCDEFG fed 10 (Figure 6A, traces v and vi), confirming that 10 was the direct precursor for the synthesis of 4 and 13.

2.4. The Proposed Biosynthetic Pathways

On the basis of the heterologous reconstitution of 1–13, a plausible biogenetic pathway is proposed (Scheme 1). Firstly, the PT domain of LabdA catalyzes the condensation of DMAPP with three successively added IPPs to form GGPP, which is then cyclized to form copalyl diphosphate by the type II TC domain of LabdA [15]. After that, an endogenous phosphatase in A. nidulans or the phosphatase LabdB could convert copalyl diphosphate into (+)-CPP. The (+)-CPP is converted into compound 1 through intermediate A under the catalysis of endogenous oxidase in A. nidulans. Moreover, 1 can be converted into 2 under the catalysis of endogenous reductase in A. nidulans. Compound 5 is derived from the multi-step oxidation of intermediate A catalyzed by the P450 LabdE. The P450 LabdF can catalyze the hydroxylation of Me-18 in A, followed by oxidation to aldehydes and finally, dehydrogenates to 8. During this cascade reaction, LabdE can catalyze the continued oxidation reaction of hydroxylation intermediate B to generate 3, 6, and 7.

On the one hand, compound 8 can be converted to 9–12 under the catalysis of LabdE. Compound 11 can also be reduced to 9 under the catalysis of the SDR LabdG. In the process of converting 8 to 12, it is speculated that LabdE catalyzes multi-step reactions, including oxidation and decarboxylation. On the other hand, 10 undergoes a ring-opening reaction under the catalysis of LabdE to form aldehyde intermediate F, which is catalyzed by aldehyde dehydrogenase LabdD to form intermediate G. Since G is unstable, it will oxidize and decarboxylate spontaneously or under the catalysis of LabdE to produce 4. The Intermediate H is catalyzed by LabdE to form I, which can further spontaneously or enzymatically form the compound penioxalicin. Bioinformatic analysis showed that the two P450s LabdE and LabdF share 39% and 20% identity to LabE, which is responsible for oxidizing the C-20 methyl group of raimonol in Streptomyces sp. KIB 015 to an aldehyde or carboxyl group, respectively [33]. Therefore, it is speculated that penioxalicin undergoes multi-step oxidation and methylation reactions under the catalysis of LabdE and LabdC, respectively, to generate 13.

3. Materials and Methods

3.1. General Experimental Procedures

NMR spectra were recorded on a JEOLJN M-ECP 600 spectrometer (JEOL, Tokyo, Japan), an Agilent DD2-500 spectrometer, or a Bruker Avance NEO 400 MHz spectrometer (Bruker, Karlsruhe, Germany) using tetramethylsilane (TMS) as an internal standard. Specific rotations were obtained on a JASCO P-1020 digital polarimeter. UV-vis spectra were recorded on UFLC system (Shimadzu, Tokyo, Japan). HRESIMS spectra were obtained on a Thermo Scientific LTQ Orbitrap XL mass spectrometer (Thermo Fisher Scientific, Bremen, Germany) or Micromass Q-TOF ULTIMA GLOBAL GAA076 LC mass spectrometer (Waters Aisa Ltd., Singapore). A JASCO J-715 spectropolarimeter (JASCO, Tokyo, Japan) was used to obtain ECD spectra. LC−MS analyses were performed using a UFLC system (Shimadzu, Tokyo, Japan) with a C18 column (Shimadzu, 4.6 mm × 150 mm, 5 μm, 1 mL/min) coupled to a LCMS-2020 mass spectrometer (Shimadzu, Tokyo, Japan). Column chromatography (CC) was performed with silica gel (200–300 mesh, Qingdao Marine Chemical Ltd., Qingdao, China), SiliaSphere C18 (Octadecylsilyl, ODS) monomeric (SiliCycle Inc., Quebec, Canada, 50 μm, 120 A), and Sephadex LH-20 (GE Healthcare, Uppsala, Sweden). MPLC was performed using a C18 column (Welch Materials Inc., Ultimate^® XB-C18, 21.2 mm × 250 mm, 5 μm, 20 mL/min). Fractions were monitored via thin layer chromatography (TLC) (GF 254, Qingdao Haiyang Chemical Co., Ltd., Qingdao, China). The compounds were purified using a HPLC Hitachi 1110 system equipped with a 1430 PDA detector or HPLC system (PuriMaster-2000, Shanghai Kezhe Biochemical Technology Co., Ltd., Shanghai, China) and a C18 column (SilGreen, 10 mm × 250 mm, 5 μm, 3 mL/min).

3.2. Materials and Culture Conditions

The fungal strains used in this study are listed in Table S1. The fungus, Talaromyces sp. HDN151403, was isolated from an unidentified sponge sample collected from Prydz Bay, Antarctica (−624 m, 75.5° E, 69.7° S) and identified based on internal transcribed spacer DNA sequencing (GenBank: MW888514) [34]. Talaromyces sp. HDN151403 was cultured on potato dextrose agar (PDA, BD) at 28 °C for 5–7 days, and then the mycelia were transferred into potato dextrose broth (PDB, BD) for the isolation of genomic DNA, the isolation of total RNA, or the OSMAC analysis of secondary metabolites. Escherichia coli XL1-Blue was used for DNA manipulation. Heterologous expression was carried out with Aspergillus nidulans A1145 as the host. Saccharomyces cerevisiae BJ5464-NpgA was used for plasmid construction.

3.3. Gene Cloning, Plasmid Construction, and Genetic Manipulation

The plasmids utilized in this work are listed in Table S1. The oligonucleotide sequences for PCR primers are given in Table S2. The Q5^® High-fidelity DNA Polymerase and Restriction endonucleases used for all DNA manipulation were from New England Biolabs (NEB). The Frozen-EZ Yeast Transformation II Kit (Zymo Research) and the Zymoprep Yeast Plasmid Miniprep I Kit (Zymo Research) were used for transformation and plasmid recombination in yeast.

Mycelia were collected from the liquid medium PDB via centrifugation. After grinding the mycelia in liquid nitrogen with a mortar, genomic DNA was extracted using cetyltrimethylammonium bromide (CTAB) buffer and phenol–chloroform extraction methods. All seven genes and their native terminators (300–500 bp downstream from the stop codon) were amplified via PCR using the gDNA of the original host as the template. Then, the recombinant plasmids were constructed according to the description from Yee and Tang [35]. The glaA, gpdA, and amyB promoters were amplified from vectors pANU, pANR, and pANP by using primer pairs glaA-F/glaA-R, gpdA-F/gpdA-R, and amyB-F/amyB-R, respectively. The labd genes were amplified via PCR using primer pairs as listed in Table S1 and then combined with the PacI-linearized vector pANU, pANP, or pANR in vivo using yeast homologous recombination to yield several plasmids. The obtained yeast colonies were characterized using PCR. Yeast plasmids were isolated by using a Zymoprep Yeast Plasmid Miniprep I Kit and transformed into E. coli XL1 Blue. All the plasmids were confirmed via restriction enzyme digestion and DNA sequencing.

3.4. RNA Extraction and Reverse Transcriptase PCR (RT-PCR)

The RNA manipulations were carried out using the Direct-zol™ RNA Microprep Kit (Zymo Research) and the HiScript^® II 1st Strand cDNA Synthesis Kit (Vazyme Biotech Co., Ltd., Nanjing, China). The spores of Talaromyces sp. HDN151403 were inoculated into 100 mL of several media and cultivated at 28 °C for 14 days under dark conditions with two replicates each. Then, the mycelia were harvested and the total RNA was extracted using the Direct-zol™ RNA Microprep Kit according to the instructions. To evaluate the transcription of the labd gene cluster under different fermentation conditions, single-stranded cDNAs were synthesized using the HiScript^® II 1st Strand cDNA Synthesis Kit. The synthetic cDNA was used as a template to amplify gene fragments via PCR using primer pairs labdA-F/labdA-R, labdB-F/labdB-R, labdC-F/labdC-R, labdD-F/labdD-R, labdE-F/labdE-R, labdF-F/labdF-R, and labdG-F/labdG-R. The target gene fragment synthesized via PCR using gDNA as a template was used as a control. Then, we checked whether the target gene was expressed via DNA sequencing. The PCR primer pairs are listed in Table S2.

3.5. Transformation of A. nidulans

A. nidulans A1145 was used as the recipient host. Fungal protoplast preparation and transformation were carried out as described by Yee and Tang [35]. PEG-mediated protoplast transformation was employed to construct the 12 A. nidulans transformant strains AN-labdA, AN-labdE, AN-labdG, AN-labdAB, AN-labdABE, AN-labdABF, AN-labdABEF, AN-labdABDEF, AN-labdABEFG, AN-labdABCDEF, AN-labdABCDEG, and AN-labdABCDEFG. Transformation with three empty pANU, pANR, and pANP vectors was used as the control (AN-empty vectors). Transformants were verified using PCR.

3.6. Fermentation and LC/LC−MS Analysis

A. nidulans transformants were cultivated on solid CD-ST medium (20 g/L starch, 10 g/L casein hydrolysate (acid), 50 mL/L nitrate salts, 1 mL/L trace elements, 20 g/L agar) at 28 °C for 5 days. Then, the cultures were extracted three times with ethyl acetate (EtOAc). The organic phase was evaporated to dryness and re-dissolved in 200 μL of MeOH/H₂O (90:10). Then, 50 μL of dissolved extract was injected for high-performance liquid chromatography–photodiode array detection-mass spectrometry (HPLC-DAD-MS) analysis (C18 column, Shimadzu, 4.6 mm × 150 mm, 5 μm, 1 mL/min); the samples were separated on a linear gradient of 15–50% acetonitrile (CH₃CN) in water (0.1% trifluoroacetic acid) for 25 min at a flow rate of 1 mL/min, followed by a linear gradient of 50–100% acetonitrile (CH₃CN) in water (0.1% trifluoroacetic acid) for 10 min, and then followed by isocratic 100% CH₃CN for 5 min. For LC analysis (C18 column, Shimadzu, 4.6 mm × 150 mm, 5 μm, 1 mL/min), samples were separated on a linear gradient of 5−50% acetonitrile (CH₃CN) in water (0.1% trifluoroacetic acid) for 40 min at a flow rate of 1 mL/min, followed by a linear gradient of 50–100% acetonitrile (CH₃CN) in water (0.1% trifluoroacetic acid) for 10 min, and then followed by isocratic 100% CH₃CN for 5 min.

3.7. Extraction, Isolation, and Purification

The large-scale solid fermentation culture (CD-ST medium, 5 L) of the transformant strain AN-labdAB was extracted four times with EtOAc. The crude extract (9 g) was subjected to a C18 column using a stepped gradient elution of MeOH/H₂O, yielding five subfractions (Fr.1–Fr.5, 30% to 70%). Fr.3 was purified on a preparative C18 HPLC column with a gradient of MeOH/H₂O (72:28) to yield 2 (10.5 mg, retention time (t_R) = 22 min). Fr.4 was purified on a preparative C18 HPLC column with a gradient of MeOH/H₂O (75:25) to yield 1 (12.0 mg, t_R = 16.5 min).

The large-scale solid fermentation culture (CD-ST medium, 20 L) of the transformant strain AN-labdABCDEFG was extracted four times with EtOAc. The crude extract (20 g) was subjected to a C18 column using a stepped gradient elution of MeOH/H₂O, yielding six subfractions (Fr.1–Fr.6, 30% to 80%). Fractions were monitored via TLC and LC–MS. Fr.2 was purified on a preparative C18 HPLC column with a gradient of MeOH/H₂O (48:52) to yield 13 (4.0 mg, t_R = 19.0 min). Fr.3 was purified on a preparative C18 HPLC column with a linear gradient of MeOH/H₂O (40:60-90:10) to yield two subfractions (Fr.3.1 and Fr.3.2). Then, Fr.3.1 was purified on a preparative C18 HPLC column with a gradient of MeOH/H₂O (58:42) to yield 7 (22.5 mg, t_R = 11.9 min) and 4 (1.7 mg, t_R = 14.4 min). Fr.4 was purified on a preparative C18 HPLC column with a gradient of MeCN/H₂O (30:70) to yield 12 (5.5 mg, t_R = 23.5 min), 10 (6.3 mg, t_R = 25 min), 5 (6.7 mg, t_R = 38 min), 6 (11.0 mg, t_R = 44.5 min), and 11 (45.4 mg, t_R = 47.5 min). Fr.5 was purified on a preparative C18 HPLC column with a gradient of MeCN/H₂O (35:65) to yield 9 (64.8 mg, t_R = 25.0 min) and 3 (18.5 mg, t_R = 27.1 min). Fr.6 was purified on a preparative C18 HPLC column with a gradient of MeCN/H₂O (50:50) to yield 8 (51.7 mg, t_R = 31.2 min).

Talarobicin A (3): colorless powder; ${\left[\alpha \right]}_{\mathrm{D}}^{25}$ +24.24, (c 0.21, CH₃OH); UV (CH₃OH) λ_max (log ε) 223 (3.13) nm; ECD (1.25 × 10⁻³ M, MeOH) λ_max (∆ε) 223 (+22.24) nm; ¹H NMR (500 MHz) and ¹³C NMR (125 MHz) data (CD₃OD), see Table 1; positive ion HRESIMS m/z 354.2648 [M + NH₄]⁺ (calcd. for C₂₀H₃₆O₄N⁺, 354.2639).

Talarobicin B (4): colorless oil; ${\left[\alpha \right]}_{\mathrm{D}}^{25}$ +9.90, (c 0.03, CH₃OH); UV (CH₃OH) λ_max (log ε) 225 (2.71) nm; ECD (1.14 × 10⁻³ M, MeOH) λ_max (∆ε) 219 (+8.28), 288 (−2.57) nm; ¹H NMR (500 MHz) and ¹³C NMR (125 MHz) data (CD₃OD), see Table 1; positive ion HRESIMS m/z 339.1793 [M + H]⁺ (calcd. for C₁₈H₂₇O₆⁺, 339.1802).

Talarobicin C (5): colorless powder; ${\left[\alpha \right]}_{\mathrm{D}}^{25}$ +0.90, (c 0.03, CH₃OH); UV (CH₃OH) λ_max (log ε) 232 (2.92) nm; ECD (1.32 × 10⁻³ M, MeOH) λ_max (∆ε) 221 (−1.99), 240 (−3.94), 329 (+0.91) nm; ¹H NMR (500 MHz) and ¹³C NMR (125 MHz) data (CD₃OD), see Table 2; positive ion HRESIMS m/z 335.2210 [M + H]⁺ (calcd. for C₂₀H₃₁O₄⁺, 335.2217).

Talarobicin D (6): colorless powder; ${\left[\alpha \right]}_{\mathrm{D}}^{25}$ +6.00, (c 0.07, CH₃OH); UV (CH₃OH) λ_max (log ε) 227 (3.48) nm; ECD (1.16 × 10⁻³ M, MeOH) λ_max (∆ε) 224 (−2.48), 243 (−6.67), 320 (+1.51) nm; ¹H NMR (500 MHz) and ¹³C NMR (125 MHz) data (CD₃OD), see Table 2; positive ion HRESIMS m/z 335.2209 [M + H]⁺ (calcd. for C₂₀H₃₁O₄⁺, 335.2217).

Talarobicin E (7): yellowish oil; ${\left[\alpha \right]}_{\mathrm{D}}^{25}$ +0.34 (c 0.29, CH₃OH); UV (CH₃OH) λ_max (log ε) 230 (3.80) nm; ECD (1.28 × 10⁻³ M, MeOH) λ_max (∆ε) 218 (−2.08), 242 (−8.42), 329 (+2.04) nm; ¹H NMR (500 MHz) and ¹³C NMR (125 MHz) data (CD₃OD), see Table 2; positive ion HRESIMS m/z 351.2156 [M + H]⁺ (calcd. for C₂₀H₃₁O₅⁺, 351.2166).

3.8. A. nidulans Feeding Experiments

The A. nidulans transformant strain was grown on CD-ST liquid culture (50 mL) at 28 °C 180 rpm. After growth for 2 days, substrate (1–2 mg) dissolved in DMSO (50 μL) was added to the culture. After further incubation for 3 days, the mycelium was collected via filtration and extracted with acetone, while the media was extracted with ethyl acetate. The crude extracts were reduced to dryness in vacuo and re-dissolved in methanol for LC–MS analysis.

3.9. ECD Calculations

Conformation searches based on molecular mechanics with MMFF force fields were performed for stereoisomers to obtain stable conformers. All the stable conformers were further optimized using the density functional theory (DFT) method at the B3LYP/6-31G(d)-GD3BJ level via Gaussian 16 program package [36]. The ECD were calculated using time-dependent density functional theory (TDDFT) at a B3LYP/6-311+G(d,p) level in methanol with IEFPCM model. The calculated ECD curves were all generated using the SpecDis 1.71 program package, and the calculated ECD data of all conformers were Boltzmann averaged using Gibbs free energy [37].

3.10. Antimicrobial Activity Assay

The isolated compounds were assayed against the pathogenic bacteria methicillin-resistant coagulase-negative staphylococci (MRCNS), methicillin-resistant Staphylococcus aureus (MRSA), Acinetobacter baumannii, Bacillus cereus, Pseudomonas aeruginosa, and S. aureus, as well as the pathogenic fungus Canidia albicans. The detailed methodologies for biological testing were performed as previously described [38,39]. Ciprofloxacin or Nystatin was used as the positive control.

3.11. Cytotoxicity Assay

Five human cancer cell lines, K562 (using the MTT method), ASPC-1, MDA-MB-231, H69AR, and L-02 (using the SRB method) cells, were used in the cytotoxic assay. Adriamycin (Sigma-Aldrich Inc., Washington, USA) was used as the positive control. The detailed methodologies for biological testing were performed as previously described [40,41].

4. Conclusions

Fungi-derived secondary metabolites have always been an important source of natural medicines [42]. However, the success rate for discovering new compounds from fungi is hindered by the gene being silent or poorly expressed under laboratory conditions [43]. In this study, a total of 13 labdane diterpenes belonging to four skeleton types, including five undescribed analogs, were obtained by heterologous expression in A. nidulans. Compounds 3–13 were all produced under the catalysis of tailoring enzymes in the BGC labd. Moreover, in the in vivo conversion process of isolated compounds, the three P450s LabdC, LabdE, and LabdF were disclosed for the first time to be responsible for catalyzing multi-step chemical reactions, such as hydroxylation, methylation, decarboxylation, etc. It is remarkable that 4 possessed a unique 3,18-dinor-2,3:4,18-diseco-labdane core structure which could serve as a key intermediate in the biosynthetic route of a group of ring A-cleaved and rearranged labdane diterpene derivatives, such as penioxalicin and 13.