1. Introduction
Hibarimicin (HBM) B (
1a,
Fig. 1), an aromatic polyketide produced by a rare actinomyces
Microbispora rosea (
M. rosea) subsp.
hibaria TP-A0121, exhibits significant antitumor activity against mouse melanoma cell B16-F10 and the biological activity of inducing differentiation in human myeloid leukemia cell HL-60 [
1], [
2]. HBM B also shows attractive inhibitory activity against the protein tyrosine kinase. However, the concentration needed to inhibit proto-oncogene tyrosine-protein kinase (Src) activation is approximately 100 times higher than that required for growth inhibition and the induction of differentiation, indicating that a different mode of action may be associated with HBM B’s antitumor activity [
3]. Structurally, HBM B belongs to the largest and most complex class of Type-II polyketides, with a unique pseudo-dimeric structure, highly oxidized aglycon, and rich glycogen modifications [
4]. The compound’s unique structure and notable biological properties make it a crucial synthetic target for chemists [
5], [
6], [
7], [
8], [
9]. To date, the total syntheses of hibarimicinone (
2a,
Fig. 1), the core aglycon of HBM B, and its analogues HMP-P1 and HMP-Y1 have been achieved, although the whole molecule has not yet been chemically synthesized [
6], [
9]. Thus, it remains a challenging task to obtain enough HBM B and its analogues to further pursue related biological and pharmacological studies.
To harness the full potential of these compounds and expand the scope of potential analogues, a strategy utilizing a biosynthesis system was considered. Previously, traditional
13C-labeling precursor feeding experiments and intermediate analysis from blocked mutants suggested that a complex Type-II polyketide synthase (PKS) system is responsible for the biosynthetic pathway of HBMs [
10], [
11], [
12]. However, further deep studies using the rare actinomyces
M. rosea subsp.
hibaria TP-A0121 were unsuccessful. Due to the difficulty of genetic manipulation on the native host, we aimed to establish a hetero-biosynthetic platform to biosynthesize HBMs. Recently, the application of an optimized chassis strain to heterologously express the relevant biosynthetic gene cluster (BGC) has emerged as a promising method. This approach effectively activates silent BGCs and establishes a connection between natural product chemotypes and their corresponding genotypes. Consequently, it facilitates the investigation of biosynthetic mechanisms and enables the convenient acquisition of analogues for further study [
13], [
14], [
15], [
16], [
17]. In this article, we report on the identification, heterologous expression, and genetic characterization of the
hbm BGC in an appropriate chassis host, Streptomyces coelicolor (
S. coelicolor). Through rational engineering of the secondary metabolic pathways, we not only heterologously biosynthesized HBM B effectively but also obtained HBM derivates with improved antitumor activity. In addition, we propose a biosynthetic pathway based on the heterologous genetic investigation of several important genes and the chemical identification of the relevant key intermediates/shunt products.
2. Materials and methods
2.1. General
Table S1 in Appendix A provides a list of the bacterial strains, plasmids, and bacterial artificial chromosomes (BACs) utilized in this study. Polymerase chain reaction (PCR) primers are listed in Table S2 in Appendix A. General enzymes, kits, media, chemicals, and molecular biological reagents were purchased from standard commercial sources. DNA manipulations and isolation in
Escherichia coli (
E. coli) were carried out using standard protocols [
18].
E. coli GB05-red was used for gene deletion by means of Redα/Redβ/Redγ (Red)/RecE/RecT (ET) recombination [
19]. Triparental conjugation was used to transfer BACs from
E. coli to
S. coelicolor according to standard protocols [
20]. For
S. coelicolor, Mannitol soya flour (MS) medium was used for the sporulation and conjugation of
S. coelicolor at 30 °C. #22 liquid medium (containing 35.0 g∙L
−1 malt extract, 30.0 g∙L
−1 cornstarch, 15.0 g∙L
−1 corn steep liquor, 15.0 g∙L
−1 cotton seed flour, and 2.0 g∙L
−1 CaCO
3) supplemented with 5% HP-20 resin and R3 solid medium plates was used for metabolite production at 30 °C.
M. rosea subsp.
hibaria TP-A0121 was cultured in V-22 medium for 3 d at 28 °C as a seed culture; 1 mL of the seed culture was then transferred into 100 mL of production medium (MP) for metabolite production [
11].
2.2. Heterologous expression of BAC hbm gene clusters
BAC 3C16 and its derivative plasmids were transferred into
S. coelicolor through ET12567/pUB307-mediated triparental conjugation [
20]. In brief, ET12567/BACs and the helper strain ET12567/pUB307 were cultured in Luria-Bertani (LB) medium for 6-8 h at 37 °C to an optical density at 600 nm (OD
600) of 0.5-0.7. Then 2 mL of ET12567/BACs and 1 mL of ET12567/pUB307 were collected by centrifugation, washed twice with antibiotic-free LB medium, and resuspended in 200 μL of LB medium. Freshly collected
S. coelicolor spores were washed twice with LB medium, resuspended in 400 μL of 2× yeast extract tryptone (YT) medium heat-shocked at 50 °C for 10 min, mixed evenly with the washed ET12567/BACs and ET12567/pUB307, and spread on MS medium. After incubating at 30 °C for 13-16 h, the MS plates were flooded with apramycin (50 µg∙mL
−1) and trimethoprim (40 µg∙mL
−1). The conjugates were allowed to grow for an additional 6-7 d. The conjugants were then verified through PCR analysis.
2.3. Metabolite production and analysis
To produce the HBMs, 100 μL of spore suspension of S. coelicolor derivatives was inoculated into 50 mL of liquid tryptic soy broth (TSB) and incubated at 30 °C for 1-2 d to establish a seed culture. Then, 2 mL of the seed culture was transferred into 100 mL of fermentation medium (#22) and incubated for an additional 5 d at 30 °C. For solid fermentation, 30 μL of spore suspension was spread on an R3 solid plate and incubated in a 30 °C culture chamber for 4 d. For the analysis of metabolites, the solid plate was chopped and extracted three times with 35 mL of 99:1 (v/v) ethyl acetate/acetic acid. After evaporation using a rotary evaporator, the residue was dissolved in 1 mL of methanol. A 20 μL sample was taken for high-performance liquid chromatography (HPLC) detection. In the case of liquid cultures, the culture broth was adjusted to pH 3-4 and then extracted with an equal volume of ethyl acetate. The ethyl acetate extract was concentrated using rotary evaporation, and the final extract was dissolved in methanol. Sample analysis was performed using an Agilent 1200 HPLC system (USA) with an Acclaim120 C18 column (5 μm, 4.6 mm × 250 mm) using acetonitrile and water containing 0.1% formic acid at a flow rate of 1.0 mL∙min-1. Gradient elution of acetonitrile was changed from 25% to 90% for 20 min, followed by a 10 min hold. The detection wavelength of HBMs was 276 or 500 nm.
2.4. Construction of S. coelicolor M1156
To delete the remaining
act genes in
S. coelicolor M1154 [
21], PCR was used to amplify two homologous fragments from
S. coelicolor M1154 genomic DNA with the primers Δ
act-L-
HindIII-F, Δ
act-L-
XbaI-R, Δ
act-R-
XbaI-F, and Δ
act-R-
EcoRI-R (sequences are provided in Table S2). The PCR products obtained were purified and digested using the corresponding restriction enzymes. Subsequently, they were ligated into the predigested plasmid pKC1139, resulting in the creation of the gene knockout plasmid pKC1139-Δ
act. The constructed plasmid was then introduced into
S. coelicolor M1154 through conjugation with
E. coli ET12567/pUZ8002. The conjugants were cultured for one generation at 37 °C and then further cultured for 3-4 successive generations at 30 °C to allow for double-crossover events. The double-crossover mutants were verified by PCR using the flanking primers Δ
act-flank-F/R, and the results were further confirmed via PCR product sequencing.
2.5. Construction of gene disruption, complementation, and overexpression mutants
Genetic manipulations of 3C16 were performed using Red/ET recombination-mediated gene deletion. About 0.5 μg of one linear PCR product of the kanamycin resistance gene cassette flanked by flippase recognition target (FRT) sites from the pJTU4659 DNA fragment with a 39-nucleotide homologous arm for the target gene was electroporated into 50 μL of GB05-red/3C16 competent cells. Correct colonies were verified using PCR. Subsequently, these constructs were transferred to
E. coli BT340 for the excision of the kanamycin resistance gene, mediated by the flippase (FLP) recombinase. Before being transferred into
S. coelicolor, the final knockout strains containing the FLP scar were confirmed by PCR. For overexpression of
hbmRg2 in situ, an ampicillin-resistant gene cassette with flanking FRT sites next to a
kasOp* promoter [
22] was synthesized, and the procedures were repeated using the new synthesized ampicillin resistance cassette.
The integrative plasmid pMS82 or pMS4F (a pMS82 derivative with the
stnYp promoter [
23]) was used for gene complementation. The
hbmD gene was cloned into the downstream of the
stnYp promoter in the pMS4F plasmid at the
BcuI and
EcoRV restriction sites to generate pMS4F-
hbmD plasmid. The complementation strain Δ
hbmD/M1156::pMS4F-
hbmD was obtained by introducing the pMS4F-
hbmD plasmid into the Δ
hbmD/M1156 mutant
. Similarly,
actVI-ORF1 was cloned to pMS82 and then introduced into 3C16/M1156 to obtain the complementation strain 3C16/M1156::
actVI-ORF1.
2.6. Isolation and structure elucidation of metabolites from related strains
For the preparation of compounds 3, 4, 5, and 6, liquid fermentation broth (6 L) of the mutant 3C16/M1154 was extracted three times with an equal volume of ethyl acetate after adjusting the pH to 3-4. After rotary evaporation of the extract, the extract was dissolved in an appropriate amount of methanol. The solution was then loaded onto a reverse-phase C18 column and eluted with a gradient of methanol:H2O mixture. The desired fractions were further purified using a semi-preparative column to obtain the target compound. For the remaining strains, a 5 L solid fermentation was performed. The isolation procedures were similar to those of 3, 4, 5, and 6; ultimately, the pure compounds 1b/c, VB1/2 (from the fermentation of the 3C16/M1156 mutant), 1a (from the fermentation of the ΔhbmD mutant), 12 (from the fermentation of the ΔhbmO6 mutant), and 15b (from the fermentation of the ΔhbmO8 mutant) were obtained through semi-preparative HPLC.
2.7. Cell viability assay
Cell Counting Kit-8 (CCK-8) assays were adopted to evaluated the viability of cells after incubation with gradient compounds. In brief, cells were seeded into 96-well plates overnight at a seeding density of 2500 cells per well. On the following day, the medium was discarded and replaced with fresh medium containing the gradient compound. After 72 h of incubation with the compound, 10 μL of CCK-8 (K1018; APExBIO, USA) was added into each well for 4 h at 37 °C. The absorbance of each well at 450 nm was recorded using an EnVision Multimode Plate Reader (PerkinElmer, USA).
2.8. RNA extraction and reverse transcription-polymerase chain reaction (RT-PCR) analysis
The total messenger RNA (mRNA) was isolated from M. rosea subsp. hibaria TP-A0121 after 48 h of growth in MP at 30 °C using a MiniBEST Universal RNA Extraction Kit (Takara, Japan). PCR was used to verify that there was no DNA contamination. Isolated RNA was then reverse transcribed into complementary DNA (cDNA) with a cDNA Synthesis Kit (Takara). To analyze gene expression, the cDNA sample was denaturalized at 95 °C for 5 min before being amplified by 30 PCR cycles, as follows: 95 °C denaturalization (30 s), hybridization at the optimal temperature (30 s), extension at 72 °C (15-30 s), and a final extension at 72 °C (10 min).
2.9. Hydroxycobalamin (OHCbl) feeding experiments
For the OHCbl feeding experiments, spore suspensions of the M. rosea subsp. hibaria TP-A0121 were inoculated in V-22 medium and incubated for 48 h as a seed culture. Then 1 mL of seed culture was inoculated into 50 mL of MP. The cultivation temperature was set at 30 °C with agitation at a speed of 220 r∙min−1. After 24 h, OHCbl was introduced into the MP at a final concentration of 60 µmol∙L−1. Following a cultivation period of 3 d, the culture broths were collected and subjected to metabolite analysis. Prior to the analysis, the pH of the broths was adjusted to 3-4; then, an equal volume of ethyl acetate was used for extraction.
3. Results and discussion
Bioinformatic analysis of the genome data of
M. rosea subsp.
hibaria TP-A0121 using antiSMASH 6.0 [
24] revealed a Type-II PKS-related (
hbm) BGC with 40% of genes showing similarity to those of Type-II polyketide mithramycin biosynthesis [
25]. The potential
hbm BGC is about 61 kb and contains 48 open reading frames (ORFs) that may encode the biosynthesis, regulation, and resistance of HBMs (
Fig. 2(a)). Their putative functions were predicted by conducting a basic local alignment search tool (BLAST) analysis with the corresponding amino acid sequences (
Table 1), and the boundaries of the BGC were determined based on sequence analysis and biosynthetic logic. Considering the large size of the cluster, we selected BAC cloning to acquire the entire
hbm BGC. An
M. rosea subsp.
hibaria TP-A0121 genomic BAC library was prepared, and the targeted BACs were screened via PCR. Several BACs containing the expected full BGC region were identified (Fig. S1 in Appendix A).
A positive BAC (3C16) was subsequently introduced into the widely used chassis host
S. coelicolor M1154 [
21] by conjugation. Next, extracts of cultures from the
S. coelicolor M1154 containing the empty vector pSET152 (as a control) and 3C16 were analyzed via HPLC and electrospray ionization-mass spectrometry (ESI-MS). However, only trace amounts of HBM-related compounds (
m/z 1739.12 [M-H]
-,
m/z 1753.13 [M-H]
-) were detected, and plenty of lower molecular weight metabolites were accumulated in the recombinant strain (
Fig. 2(b)). After a large-scale fermentation of this strain (3C16/M1154), compounds
3-
6 were isolated and structurally identified by nuclear magnetic resonance (NMR;
Fig. 2(c), Tables S3-S7 and Figs. S2-S24 in Appendix A). The yield of HBMs was too low to be isolated, however. Compound
3 was previously found by the heterologous expression of the
skt cluster in
S. coelicolor M1154 or M1152 [
26]. The structures of compounds
4 and
5 are similar to those of streptoketide A and UWM5, except for an extra carbon extension at the starter unit [
26]. Compound
6 was identified as HMP-M1, which was first isolated in the culture broth of a mutant strain BN-Y185 [
12] and was suggested to be related to HBM B. This result demonstrates that the
hbm BGC does indeed govern the biosynthesis of HBMs. In addition, the low production of end-compounds and precursor
6 and the large accumulation of compounds
3-
5 strongly hint that some off-pathway enzymes originating from the heterologous host
S. coelicolor M1154 may drive the biosynthetic intermediates into branch roads during the early stage of the HBM pathway.
In order to obtain the target products, the biosynthetic pathways had to be reformed to significantly improve the yield of HBMs. We noticed that ① compounds
3 and
4 share a common pyran ring group, which also exists in actinorhodin (ACT;
Fig. 3(a)); ② S2502 and S2507, which are highly related to compound
4, were previously obtained via the heterologous expression of the nogalamycin anthraquinone aglycone biosynthetic genes in
S. lividans TK24 [
27]; and ③
S. lividans TK24 has the capability to produce the blue pigment actinorhodin due to the activation of the
act BGC in its genome. This capability is attributed to the presence of the
rpsL K88E mutation, in comparison with the parent strain
S. lividans TK66, which lacks the mutation and cannot produce actinorhodin [
28]. Therefore, we speculated that compounds
3 and
4 are hybrid compounds derived from a combination of the HBM and ACT biosynthetic pathways. In ACT biosynthesis, it has been demonstrated that ACTIV acts as a bifunctional cyclase-thioesterase, generating an acyl carrier protein (ACP)-unbound bicyclic intermediate. This intermediate is subsequently reduced by a dedicated reductase (RED1) to establish the (S)-configuration at C3 (
Fig. 3(a)) [
29]. After reduction, the product undergoes hemiketal formation and subsequent dehydration to produce the monomer precursor of ACT [
30]. Considering the common pyran ring group in compounds
3,
4, and ACT, we speculated that ACTIV and RED1 might “hijack” the nascent precursor synthesized by the HBM biosynthetic pathway as a substrate to produce a great deal of the hybrid products
3 and
4 (
Figs. 3(b) and
(c)), leading to the low yield of the final products. Although most of the
act BGC (a 16 985 bp deletion from 5 515 934 to 5 532 918 bp of the genome sequence) was deleted in the
S. coelicolor M1154 genome, the
actVI-ORF1 and
actIV genes were retained by chance. In order to eliminate the shunt products
3 and
4, the retained region of
act BGC (6 246 bp) in
S. coelicolor M1154 containing intact
actVI-ORFB, A, 1,
actIV,
actVB, and partial
actVI-ORF2 and
actVII genes was totally deleted, resulting in the
act BGC being completely removed from the strain
S. coelicolor M1156 (23 231 bp deletion from 5 511 901 to 5 535 131 bp of the
S. coelicolor A3(2) genome sequence) (Fig. S25 in Appendix A).
BAC 3C16 was then introduced into
S. coelicolor M1156 via intergeneric conjugation, and its production profiles were analyzed by means of HPLC/ESI-MS. As expected, compounds
3 and
4 were eliminated, and the yields of precursor
6 and new members of HBMs (
1b/
c) were substantially improved (
Fig. 4(a)). When this mutant strain was gene-complemented with the
actVI-ORF1 gene under the control of its own prompter, the recombinant strain 3C16/M1156::
actVI-ORF1 restored the production of compounds
3 and
4. These pieces of evidence support our hypothesis and suggest that
actVI-ORF1 alone is sufficient to divert the HBM biosynthetic route to the shunt product pathway (
Fig. 3(b)). After a large-scale fermentation of the strain
S. coelicolor 3C16/M1156, the new compounds
1b/
c were isolated and structurally identified by NMR (
Fig. 4(b), Tables S8 and S9 and Figs. S26-S39 in Appendix A). Compounds
1b/
c are structurally similar to HBM B (
1a), except that the 3′-OMe or 3′,3-OMe are respectively replaced by the -OEt group. Compounds
VB1/
2, which were also isolated from the culture broth of
S. coelicolor 3C16/M1156, respectively arose from
1b/
c through the formation of the ether bridge via an addition-elimination reaction between the D and E aromatic groups with the release of an ethanol molecule (
Fig. 4(b), Tables S10 and S11 and Figs. S40-S53 in Appendix A).
Crosstalk between different secondary metabolic pathways in bacteria is a common occurrence. For example, the synthesis of hybrubins involves combining a truncated undecylprodigiosin biosynthetic pathway with a heterologous bipyrrole tetramic acids biosynthetic pathway [
31]. Combined biosynthesis from different pathways often occurs due to the substrate promiscuity of key enzymes that bridge the two pathways. However, the promiscuity of key synthetic enzymes also presents new opportunities for combinatorial biosynthesis. By leveraging the substrate promiscuity of glycosyltransferases (GTs), it is possible to combine the synthesis of substrate molecules with different sugar modifications, thereby altering the molecules’ biological activity or water solubility. For example, by heterologously overexpressing deoxysugar biosynthesis pathways in the mithramycin producer
S. argillaceus, it is possible to obtain mithramycin analogs with improved activity [
32]. The GT-methyltransferase (MT) biosynthetic module derived from Beauveria bassiana, which exhibits high substrate promiscuity, can be used to modify a broad range of drug-like substrates, including polyketides, anthraquinones, flavonoids, and naphthalenes [
33]. The reductase RED1 mentioned in this study also demonstrates certain substrate promiscuity, as the heterologous expression of multiple Type-II polyketide BGCs such as streptoketide, HBM, and nogalamycin resulted in the production of structurally similar byproducts. Moreover, compound
4 is similar to streptoketide A except for an extra carbon extension at the starter unit, suggesting that RED1 has some tolerance toward the structure of the starter unit. Therefore, it is possible to obtain streptoketide A analogs by heterologously expressing the scaffold genes of anthracycline-type Type-II polyketides with different starter units.
In the
hbm BGC,
hbmD encodes a radical
S-adenosyl-
L-methionine (SAM) protein belonging to a family of methylcobalamin-dependent radical SAM C-MTs exemplified by KedN5, which has been proposed to catalyze tandem C-methylation to furnish the isopropoxy group in kedarcidin biosynthesis [
34]. Therefore, we speculated that the formation of the 3′ and 3 ethyl groups were catalyzed by HbmD. To prove this hypothesis,
hbmD was in-frame deleted to explore the function of this radical SAM C-MT
in vivo (Fig. S54 in Appendix A). The mutant,
S. coelicolor Δ
hbmD/3C16/M1156, exhibited an inability to produce
1b/
c. Instead, it accumulated a red compound (
1a) with an ultraviolet (UV) spectrum similar to that of the desired product (
Fig. 5(a)). Liquid chromatography-mass spectrometry (LC-MS) analysis revealed it to be HBM B (
1a;
m/z 1723 [M-H]
-; Table S12 and Figs. S55-S59 in Appendix A), which is a known target compound previously isolated from
M. rosea subsp.
hibaria TP-A0121. Furthermore, HBM B was detected in minimal amounts in the
S. coelicolor 3C16/M1156 strain. The production of compounds
1b/
c was restored via complementation with the
hbmD gene in the Δ
hbmD mutant strain by means of intergeneric conjugation (
Fig. 5(a)). These
in vivo results support HbmD as the C-MT in the biosynthesis of the 3′ and 3 ethyl groups. To our puzzlement, compounds
1b/
c were not detected in the fermentation broth of the original strain
M. rosea subsp.
hibaria TP-A0121. Thus, it is unclear why HbmD does not function in the native producer. We detected the transcription of the
hbmD gene in the strain
M. rosea subsp.
hibaria TP-A0121, and the results showed that there was no problem in the transcription of
hbmD (
Fig. 5(b)). Therefore, we speculated that the HbmD protein may lack the cobalamin cofactor, which is necessary to the function of the class B family radical SAM protein.
To verify this hypothesis, we first fed hydroxocobalamin into
M. rosea subsp.
hibaria TP-A0121 fermentation broth. After feeding, we successfully detected the production of compound
1b and the 3′ ethyl substituted derivatives of HBM C and HBM D-namely, HBM C′ and D′ (
Figs. 5(c) and
(d)). However, we did not detect the production of the 3′,3-OEt substitution product,
1c. Therefore, we propose that the
M. rosea subsp.
hibaria TP-A0121 strain may obtain cobalamin from the microbial community through cofactor cross-feeding [
35]. That is, in its natural habitat, the native strain TP-A0121 may acquire cobalamin through microbial community interactions. Since cobalamin is only synthesized by certain bacteria and archaea [
36], most plant-based culture media contain negligible amounts of cobalamin. The fermentation medium for the native strain consists of 2% cottonseed powder and 4% mannose. However, cottonseed powder medium lacks cobalamin. Conducting a supplementation test by adding a yeast extract medium, which is known for its high content of B-group vitamins, including cobalamin, to the original fermentation culture medium holds promising potential in facilitating the production of ethyl derivatives of HBM. This approach warrants further investigation and experimentation.
With several new HBM congeners in hand, we were interested to test their inhibitory potential against tumor cells. To evaluate the impact of glycosyl groups on antitumor activity, the aglycons
2a/
b (Tables S12-S15 and Figs. S60-S69 in Appendix A) were obtained via acid-catalyzed methanolysis in 1.5 MHCl-MeOH (100 μg∙mL
-1) [
11], which involved hydrolyzing the
1a/
b glycosylation group. The antitumor activities against various cancer cell lines are summarized in
Table 2. Among these compounds,
1c showed the best antitumor activities in most cases, except for SKMel-28 cells.
VB2 displayed a slightly higher inhibitory activity against SKMel-28 than
1c, which can be quickly converted into
VB2 in pH 7.5 or heated conditions. Interestingly, the sugar appendages seem to be important for the bioactivities against A375, while the 1-OH and 3′-OMe group seem to be crucial for the bioactivities against B16-F10 cells.
Considering that the unique diethyl derivative
1c from the heterologous biosynthetic system exhibits better activity, we wanted to improve the yield of the final products,
1b/
c. The
hbm BGC contains a transcriptional regulator gene
hbmRg2 encoding an AfsR/Streptomyces antibiotic regulatory protein (SARP) family transcriptional activator, which usually acts as a pathway-specific activator directly affecting the transcription of the specific BGC for antibiotics synthesis [
37]. Therefore, a strong promoter
kasOp* [
22] was placed directly upstream of
hbmRg2 in BAC 3C16 by means of Red/ET recombineering to create the construct OE
Rg2-3C16, which was then introduced into
S. coelicolor M1156 to yield the OE
Rg2-3C16/M1156 mutant. Recently, deletion of a single
tacP gene encoding a malonyl-coenzyme A (CoA) decarboxylase homologue resulted in higher titers of the final product in comparison with the wild-type strain in thioangucycline biosynthesis [
38]. In the
hbm BGC, the gene
hbmRg2 is located upstream of the minimal PKS genes
hbmA1-A3 and downstream of
hbmP1, which encodes a propionyl-CoA carboxylase beta subunit. However, deletion of
hbmP1 failed to improve the yield of HBMs (Fig. S70 in Appendix A). Next, we replaced the
hbmP1 with
kasOp* promoter located about 973 bp upstream of
hbmRg2 by means of Red/ET recombineering to create another
hbmRg2 overexpressing construct named OE
Rg2*-3C16. OE
Rg2*-3C16 was then introduced into
S. coelicolor M1156 to yield the OE
Rg2*-3C16/M1156 mutant. Both of the
hbmRg2 overexpressing mutants showed higher titers of
1b/
c than
S. coelicolor 3C16/M1156 (
Fig. 6); the yields were improved by about two times and up to about 30 mg∙L
−1 for
1b in shake bottles.
After the establishment of the optimized heterologous expression system, the two genes
hbmO6 and
hbmO8, which encode a flavin adenine dinucleotide (FAD)-dependent monooxygenase without homologous protein in mithramycin biosynthesis, were in-frame deleted in order to preliminarily explore the late oxidative modification steps of HBM. The Δ
hbmO6 mutant accumulated a new compound
12 (
Fig. 7(a)), which is a homodimer structure featuring the same 2,2′-coupling position as HBM B (
Fig. 7(b), Table S16 and Figs. S71-S77 in Appendix A). This finding indicated that the phenol-coupling conversion occurs before the generation of the previously known intermediate, HMP-Y1 (
14a,
Fig. 3(c)). The Δ
hbmO8 mutant accumulated compound
15b, which is similar to HMP-Y6 (
15a) except for an additional ethyl at C3′-OH (
Fig. 7, Table S17 and Figs. S78-S84 in Appendix A). The FAD-dependent monooxygenase HBM O8 may catalyze the first oxidation reaction from HMP-Y1 (
14a) to hibarimicinone (
2,
Fig. 3(c)). Compounds
1b/
c feature a unique 3′-ethyl or 3′,3-diethyl group, which grants them better antitumor activity than HBM B (
Table 2). However, it is unclear what the authentic substrate of HBM D is, because abundant compound
15b—which also features a C3′ ethyl group—was accumulated in the culture broth of the Δ
hbmO8 mutant (
Fig. 7(a)). It seems that HBM D is able to catalyze HMP-Y1 to form
14b, which is then glycosylated to yield compound
15b. It is also possible that HBM D catalyzes ethyl formation immediately after the formation of the C3′-OMe group (
Fig. 3(c)). Further
in vitro enzymatic reaction or biotransformation research on HBM D is needed to address this issue.
Based on the bioinformatic analysis of the
hbm BGC and the structure elucidation of the intermediates accumulated in the mutants, a general biosynthetic pathway for Type-II polyketides in a heterologous chassis host is proposed (
Fig. 3). The biosynthesis of compounds
1a/
b/
c begins with the condensation of nine malonates and the incorporation of a C4 starter unit to form a 22-carbon poly-β-ketone chain. Next, HBM C5 regioselectively reduces the backbone at C9, followed by sequential cyclizations. The latter are likely directed by HBM C4—which shares a high similarity to SsfY1, which has been proposed to catalyze the first cyclization in the SF2575 biosynthesis [
39]—to generate
7. This intermediate can undergo spontaneous nonenzymatic cyclization or is catalyzed by an unknown protein to form the ACP free compound
8, which can be reduced by RED1 at C3 to form
9. Hemiacetal formation in
9 delivers the intermediate
10, which-upon oxidative C,C-bond cleavage-yields streptoketide C (
3). Alternatively, dehydration and cyclization/dehydration yield
4. HBM C2, which is 71% identical to the cyclase MtmY, is predicted to be the second and third ring cyclase. HBM P, a member of the ATP-dependent acyl-CoA ligase family, together with the cyclase HBM C6, are proposed to be responsible for the fourth ring formation. The linear, tetracyclic aromatic compound
11 undergoes multiple redox steps to form HMP-M1 (
6). Based on the proven structure of compound
12, which was accumulated by the Δ
hbmO6 mutant, HMP-M1 (
6) needs to undergo complicated rearrangement and phenol-coupling steps. The gene
hbmO6 encodes a FAD-dependent monooxygenase and HBM O6, probably together with another MT, and catalyzes the hydroxylation and O-methylation of the aromatic ring at C3 and C3′ to form
13. In addition, HBM O7 and another O-MT catalyze the hydroxylation and O-methylation of the D and E ring at C4 and C4′ to form the proven intermediate HMP-Y1 (
14a). Three MTs are encoded by the
hbm BGC. HBM M1 has 66% similarity to CmmMII, which is more similar to C-MT MtmMII. HbmM2 and HbmM3 bear a strong resemblance to the O-MTs and may be responsible for the methylation at the C3 and C3′ and C4 and C4′ hydroxyl groups. HBM D may catalyze the C-methylation of HMP-Y1 or compound
13 at the C3′ O-Me group. Controlled oxidization chemistry catalyzed by oxygenase HBM O8 and other unidentified proteins paves the way to compounds
2a/
b/
c. Finally, compounds
2a/
b/
c are glycosylated to yield the final products,
1a/
b/
c.
4. Conclusions
In summary, the entire BGC of the HBMs was identified through the heterologous expression of BAC clones derived from the genomic library of M. rosea subsp. hibaria TP-A0121. Furthermore, we constructed a background optimized host S. coelicolor M1156, which can be used as a host for Type-II polyketide heterologous biosynthesis via deletion of the remnant ACT biosynthetic genes. In addition, we found that the radical SAM protein HBM D was responsible for the ethyl derivatization at the 3′ and 3 sites to form compounds 1b/c, which have enhanced anti-tumor activities. In aggregation, this work paves the way for further investigation of the biosynthesis of HBMs, including the enzymatic reactions of pseudo-dimerization and aglycon oxidizations, as well as combinatorial biosynthesis toward more potent analogues for drug discovery and development.
Acknowledgments
This work was supported in part by grants from the National Key Research and Development Program of China (2018YFA0901900), the National Natural Science Foundation of China (22137009), and the China Postdoctoral Science Foundation (2020M671271). We also thank Prof. Yongsheng Che (Institute of Medicinal Biotechnology, Chinese Academy of Medical Sciences & Peking Union Medical College) for helpful discussions.
Authors’ contribution
Xiangyang Liu: conducted the majority of the experiments, analyzed the majority of the data, and wrote the manuscript; Fei-Peng Zhao: constructed some of the mutants; Tian Tian: conducted the biochemical experiment; Wei-Chen Wang and Xian-Feng Hou: isolated some of the compounds; Zaizhou Liu and Jing Wang: performed the biological activity assay; Qiang Zhou: conducted the genetic assay; Wenli Guo and Shuangjun Lin: provided the promoter and plasmid; Yasuhiro Igarashi: provided the strain and sequenced the genome; Gong-Li Tang: designed the project, analyzed the data, and wrote the manuscript.
Compliance with ethics guidelines
Xiangyang Liu, Fei-Peng Zhao, Tian Tian, Wei-Chen Wang, Zaizhou Liu, Qiang Zhou, Xian-Feng Hou, Jing Wang, Wenli Guo, Shuangjun Lin, Yasuhiro Igarashi, and Gong-Li Tang declare that they have no conflict of interest or financial conflicts to disclose.
Appendix A. Supplementary data
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.eng.2024.01.012.
PDF
(2773KB)
