Recent Advances in the Heterologous Biosynthesis of Natural Products from Streptomyces

Abstract

Streptomyces is a significant source of natural products that are used as therapeutic antibiotics, anticancer and antitumor agents, pesticides, and dyes. Recently, with the advances in metabolite analysis, many new secondary metabolites have been characterized. Moreover, genome mining approaches demonstrate that many silent and cryptic biosynthetic gene clusters (BGCs) and many secondary metabolites are produced in very low amounts under laboratory conditions. One strain many compounds (OSMAC), overexpression/deletion of regulatory genes, ribosome engineering, and promoter replacement have been utilized to activate or enhance the production titer of target compounds. Hence, the heterologous expression of BGCs by transferring to a suitable production platform has been successfully employed for the detection, characterization, and yield quantity production of many secondary metabolites. In this review, we introduce the systematic approach for the heterologous production of secondary metabolites from Streptomyces in Streptomyces and other hosts, the genome analysis tools, the host selection, and the development of genetic control elements for heterologous expression and the production of secondary metabolites.

1. Introduction

Streptomyces is a group of filamentous, Gram-positive bacteria with a high GC (guanine-cytosine) content in their genomes. The secondary metabolites or natural products of Streptomyces, such as antibiotic, anticancer, antitumor, antiviral, and immune suppressive compounds, are some of the industrially important compounds that are used for biotechnological applications [1]. Streptomyces is the main source of novel secondary metabolites, including polyketides, peptides, terpenoids, alkaloids, and saccharides [2,3,4,5,6]. Some natural secondary metabolites from Streptomyces are shown in Figure 1.

Previous studies showed that natural products have greater chemical diversity and a more extensive chemical space than synthetic compounds [16]. The structure of natural products is also more complex than the structure of synthetic compounds. Therefore, natural products are useful templates for developing relevant bioactive compound classes [17]. Many chemical synthetic drugs cause several side effects when curing disease. In contrast, natural products are produced by living cells and they may avoid the side effects caused by synthetic drugs [18]. Furthermore, bacterial resistance emerges as a current problem for treatment; therefore, discovering novel drugs is required and the exploration of natural products is one solution [19].

Genome sequences and genome mining revealed that Streptomyces contains many secondary metabolite biosynthetic gene clusters (BGCs); however, some BGCs of native strains are silent or cryptic under laboratory conditions [20]. The heterologous expression is a powerful strategy for discovering uncharacterized BGCs or increasing the production of many secondary metabolites BGCs. For example, hybrubins were successfully explored via the heterologous expression of the hbn BGC from Streptomyces variabilis Snt24 in S. lividans SBT5 [21]. The yield of pikromycin from the heterologous expression in S. lividans TK21 showed a 2.1-fold increase compared to the production of pikromycin from the parental Streptomyces venezuelae ATCC 15439 [22].

The BGCs vary in size from a small size to large sizes, such as 12 kb for the kocurin BGC [23], 60 kb for the pikromycin BGC [22], 65 kb for the daptomycin BGC [24], 90 kb for the meridamycin BGC [25], and 141 kb for the vancoresmycin BGC [26]. Depending on the size of the BGC, several cloning methods and vector systems have been developed for the heterologous expression of BGCs. Many techniques are applied for cloning target BGCs, such as the Gibson assembly [27], site-specific recombination-based tandem assembly (SSRTA) [28], the bacterial artificial chromosomal (BAC) system [2], and the transformation-associated recombination (TAR) system [29]. The selection of hosts and other elements for heterologous expression depends on the targeted products.

In this review, we provide insight into the systematic approach for the heterologous production of secondary metabolites from Streptomyces in Streptomyces and other hosts, such as Escherichia coli, Pseudomonas putida, Saccharomyces cerevisiae, Bacillus subtilis, Corynebacterium glutamicum, and Rhodococcus erythropolis. The tools for genome analysis assist with precisely identifying and cloning BGCs. This review also presents the process for selecting hosts and developing genetic control elements for heterologous expression.

2. In Silico Prediction of BGCs

Sequence analysis provides opportunities to produce novel secondary metabolites based on their genetic potential. The BGCs are chosen for their heterologous expression of the natural products by using bioinformatics tools. Nowadays, various bioinformatics tools are developed for analyzing the BGCs, such as antiSMASH [30], NaPDoS [31,32], NRPSpredictor [33], ClusterFinder [34], PRISM [35], ClustScan [36], NP.searcher [37], EvoMining [38], ARTS [39], SBSPKS [40], BAGEL4 [41], RiPPER [42], NeuRiPP [43], RippMiner [44], and RODEO [45].

AntiSMASH, PRISM, and Clusterfinder are widely used to detect the secondary metabolite gene clusters. They are powerful tools for identifying the number of putative BGCs. These tools determine the BGCs of polyketide synthase (PKS), nonribosomal peptide synthase (NRPS), hybrid, ribosomally synthesized and post-translationally modified peptide (RiPP), terpene, melanin, and siderophore [30,34,35]. For example, antiSMASH was used to identify the genome of Streptomyces actuosus ATCC 25421, and the avermipeptin BGC achieved from that analysis was successfully expressed in S. lividans TK24 [46].

SBSPKS, NaPDoS, ClustScan, and NP.searcher focus on nonribosomal peptide and polyketide biosynthetic pathways. SBSPKSv2 is a powerful web server. It facilitates the analysis and identification of not only NRPS/PKS domains but also other clusters with similar open reading frames (ORFs); three-dimensional modeling; chemical structure similarities; a simplified molecular-input line-entry system (SMILES) for starter, extender, intermediates, and final secondary metabolites; specificity prediction [40]. NaPDoS predicts and analyzes ketosynthase and the condensation domain from deoxyribonucleic acid (DNA) and protein sequences based on the domain phylogeny. This tool contributes to assessing the secondary metabolite gene diversity [31]. ClustScan and NP.searcher also predict NRPSs, PKSs, and hybrid NRPSs/PKSs. ClustScan can predict linear and cyclical structures of polyketides, while NP.searcher can predict the putative structure of both non-ribosomal peptides and polyketides. NRPSpredictor2 only works on NRPS to predict specific substrates of adenylation domains based on transductive support vector machines [33].

In contrast to PKS and NRPS BGCs, some RiPPs do not exhibit common genetic features for identification, where the gene encoding precursors are small in size and easily ignored [42]. Many bioinformatics tools were developed for predicting the precursor peptide sequences of RiPPs, such as BAGEL, RODEO, RiPPMiner, RiPPER, and NeuRiPP. BAGEL also enables the analysis of ORF predictions and facilitates discovering novel classes of peptides, and it is used to analyze small gene encoding for RiPPs [41]. RODEO mainly focuses on exploring certain RiPP classes [45]. RiPPMiner is an informatics tool for identifying the correct crosslink sequence in a core peptide of RiPPs [44]. RiPPER and NeuRiPP are new tools for identifying an RiPP precursor peptide family. The genes encoding for RiPP precursors are often missed in genome analysis. These tools are advantageous for predicting the novel RiPP classes because they identify precursor peptide events of unknown homology or novel RiPP structural classes [47].

Minimum Information about a BGC (MIBiG) is the global repository of microbial BGCs, which provides another useful resource for the accurate prediction and analysis of Streptomyces BGCs. MIBiG was established in 2015 with 1170 known BGCs; up to now, MIBiG contains 1923 known secondary metabolite BGCs [48,49]. These resources provide new opportunities to discover Streptomyces natural products using the bottom-up approach [1].

In addition to the exploration of unidentified or less studied resources, the metabolomics-based approach is based on reference compounds as potential biomarkers or the identification of molecular fingerprints that determine bioactivities has recently been used to identify new chemical diversities with versatile activities. The top-down approach focuses on the production of natural products without the knowledge of BGCs. In contrast, the bottom-up approach first focuses on BGCs and then uses genetic techniques to produce natural products from target BGCs [50]. Molecular networking was introduced in 2012 [51] and was built using simplified tandem mass spectrometry (MS/MS) data for molecular network analysis. The mass spectrometry data of putative compounds can reveal similarities in the fragmentation patterns of a single compound using the Global Natural Products Search (GNPS) website. The putative structure of molecules is closely linked to the structure of compounds found in the GNPS. The GNPS is a powerful tool for the discovery of new drugs and metabolites [52].

Molecular networking entails the computational analysis of metabolic fragments. It enables the rapid identification of compounds in a broad diversity of unknown natural products related to potential drugs, where even the compounds are in the low titer or difficult to separate. Based on the MS/MS data, bioactive natural products were explored by visualizing and organizing them using untargeted mass spectrometry. The advantage of molecular networking is the possibility of detecting analog compounds and identifying the biosynthetic pathways. Molecular networking permits finding unknown compounds and connecting “standard networks” for the annotation of new compounds [52]. For example, a saponin glycoside from the British Bluebell, a map of matlystatin congeners in various media, and the proposal of bottromycin pathway were explored via molecular networking [53,54,55].

3. Selecting a Suitable Host for Heterologous Streptomyces Products

3.1. Streptomyces

Heterologous expression is a promising tool for exploring compounds from silent BGCs or increasing the titers of natural products. Streptomyces produces numerous biologically active compounds with various effects, including agrobiological and pharmacologic agents. Many Streptomyces strains grow slowly, are difficult to genetically manipulate, and can produce many types of compounds. Thus, to select strains for the activation or enhancement of target products, host strains should contain some attractive features, such as fast growth, a well-studied secondary metabolism, and being easily genetically manipulated. Based on this, some Streptomyces hosts have been used for heterologous expression, such as S. albus, Streptomyces ambofaciens, S. avermitilis, Streptomyces coelicolor, Streptomyces fradiae, Streptomyces roseosporus, Streptomyces toyocaensis, S. venezuelae, and S. lividans [56]. Therefore, heterologous expression in Streptomyces enables the activation or enhancement of the production of secondary metabolic products. For example, albucidin was produced in S. albus Del14 from S. albus subsp. chlorinus NRRL B-24108 [57], kinamycin from Streptomyces galtieri Sgt26 was successfully biosynthesized in S. albus J1074 [58], and thaxtomins were successfully produced at a 10-fold higher titer in S. albus J1074 than in native Streptomyces scabiei [59].

Streptomyces host can also be used for the production of metabolites from rare Actinomycetes. Some rare Actinomycetes exhibit slow growth and carry a smaller population than Streptomyces, which is also considered a bioactive compound source. Some compounds were discovered from rare Actinomycetes via heterologous expression in a Streptomyces host, such as chuangxinmycin (from Actinoplanes) [60], tunicamycin (from Actinosynnema) [61], huimycin (from Kutzneria) [62], thiocoraline (from Micromonospora) [63], nargenicin (from Nocardia) [64], kocurin (from Nonomuraea) [65], GE2270 (from Planobispora) [66], shinorine (from Pseudonocardia) [67], taromycin B (from Saccharomonospora) [68], and A201A (from Saccharothrix) [69].

3.2. Other Hosts

Naturally, Streptomyces exhibits a slow growth rate (about 0.16–0.25 h⁻¹) [70], and their species are many times more difficult to genetically engineer. In most cases, the BGCs corresponding to potent compounds are silent/cryptic or the production titers are only produced in a detectable range. The BGCs derived from the native organism can be introduced into surrogate hosts to overcome the undesirable biological features of native producers. Streptomyces compounds were expressed heterologously in other hosts, such as E. coli, P. putida, S. cerevisiae, B. subtilis, C. glutamicum, and R. erythropolis.

S. cerevisiae is a generally recognized as safe (GRAS) organism with DNA recombinant stability, easy genome engineering, well-known genomics and proteomics, and a cell factory for producing natural products. Engineered S. cerevisiae was successfully used to produce a high level of natural products from other sources, for example, artemisinic acid up to 100 mg/L [71], tetrahydrocannabinolic acid up to 8.0 mg/L, cannabidiolic acid, and tetrahydrocannabivarinic acid up to 4.8 mg/L [72]. Engineered S. cerevisiae also produced some natural products from Streptomyces with high titers, such as gamma aminobutyric acid up to 62.6 g/L [73] and cannabigerolic acid up to 299.8 mg/L [74].

B. subtilis exhibits a rapid growth rate of 0.5 h⁻¹ [75], which facilitates its use for heterologous expression. In previous work, B. subtilis was successfully used for the heterologous expression of secondary metabolites from Streptomyces, such as 6-deoxyerythronolide B from Saccharopolyspora erythraea [76] and enniatin from Fusarium oxysporum [77]. Therefore, B. subtilis is a promising host for the heterologous expression of BGCs from Streptomyces.

E. coli also exhibits fast growth with a growth rate of 2–3 h⁻¹ [78] and it is an easy target for genetic manipulation. Its native biochemistry, physiology, and metabolomics are extensively understood [79]. Natural products, such as isoprenoids, non-ribosomal peptides, alkaloids, and polyketides, were successfully biosynthesized in E. coli [79]. However, the disadvantage of using E. coli for the heterologous expression of natural products from Streptomyces is the lack of building blocks and phosphopantetheinyl moiety in E. coli. Many complex enzymes were expressed in insoluble form in E. coli. E. coli lacks some specific building blocks, such as propionyl-CoA, methylmalonyl-CoA, and benzoyl-CoA, which can be supplied to the culture [80] or an engineered strain can be used that could provide the necessary precursors [81]. Many factors should be considered when doing a heterologous expression in E. coli, such as cofactors, the functional generation of the holoenzymes, the copy number of BGCs, promoters, native regulatory elements, and transcription factors [82,83]. Therefore, for the heterologous expression of BGCs, the specific genes must be introduced or re-engineered to produce attractive products. On the other hand, the BGCs were re-engineered for expression in E. coli. For example, type II PKS systems from Streptomyces contains an insoluble ketosynthase chain length factor in E. coli. Cummings and co-workers developed a plug-and-play production system to express type II PKSs for generating aromatic polyketides. An iterative monomodular type I was used for minimal PKS (mPKS) components expressing type II PKS clusters in E. coli using mPKS. They successfully used Photorhabdus luminescens mPKS combined with tailored enzymes from other organisms to produce anthraquinones, dianthrones, and benzoisochromanequinones [84]. Liu et al. also successfully used mPKS to generate TW95C and dehydrorabelomycin in E. coli [85].

C. glutamicum is a GRAS strain, generates amino acids at a large scale, and highly resists aromatic compounds. C. glutamicum belongs to Actinomycetes and is closely related to Streptomyces. Several investigators reported that engineered Corynebacterium could biosynthesize target compounds, such as alcohols, aromatic compounds, and other secondary metabolites. Corynebacterium possesses endogenous 4’-phosphopantetheinyl transferase (PPTase), PptA_Cg, which can be utilized in heterologous expression of type I PKS, and NPRS from Streptomyces [86]. For example, roseoflavin from Streptomyces davaonensis was produced in C. glutamicum via the heterologous expression of its BGCs [87].

P. putida is also a GRAS strain with a rapid growth rate of 0.3–0.6 h⁻¹ [88]. It has a high GC content in its genome and a wide range of PPTases, similar to Streptomyces [89]. P. putida is a useful host for natural products because many natural products have been produced in this strain using heterologous expression [90]. For instance, flaviolin, mono-rhamnolipid, zeaxanthin, and prodigiosin were produced via the heterologous expression of their BGCs from Streptomyces griseus, Pseudomonas aeruginosa, Pantoea ananatis, and Serratia marcescens in P. putida, respectively [90,91]. However, P. putida requires introducing intracellular substrates and other features for the expression of target products, such as methylmalonyl-CoA [92] and coumaroyl-CoA [93].

R. erythropolis is an aerobic, non-motile, and Gram-positive bacteria with a high GC content in its genome, similar to the Streptomyces genome. Kasuga et al. demonstrated that Rhodococcus is a candidate for the expression of BGCs from Streptomyces [94]. For example, kasugamycin from Streptomyces kasugaensis was produced in R. erythropolis [94]. The processing of heterologous in different hosts is shown in Figure 2.

From previous experiments, Streptomyces spp., S. cerevisiae, B. subtilis, E. coli, C. glutamicum, P. putida, and R. erythropolis were shown to be hosts for the heterologous expression of BGCs from Streptomyces. Some natural products, which were produced by heterologous expression in Streptomyces and other hosts, are listed in Table 1. The successful heterologous expression of BGCs depends on many elements, such as regulators, promoters, terminators, ribosome binding sites (RBSs), riboswitches, and transfer ribonucleic acids (tRNAs). However, to achieve this, the host should be engineered by deleting unwanted BGCs, inducing transcriptional terminators, using riboswitches, engineering tRNAs, and deleting or inserting regulators.

4. Host Engineering Approaches

4.1. Host Cleaning

Analysis of the Streptomyces genome has demonstrated that its genomes contain many BGCs; therefore, it is required that the genome be minimized by deleting unnecessary BGCs. The advantages of a minimal genome include faster growth of the host strains, a low background of native secondary metabolites, absence of biological activity, and a high chance of successful heterologous expression of BGCs [102]. Clean host strains are used for the characterization and modification of the BGCs, activation of the silent BGCs, and increasing the titer of target compounds.

Streptomyces host strains were engineered by deleting or repressing the expression of unwanted BGCs, and the resulting engineered strains were employed as hosts for the integration of heterologous BGCs, leading to the production of the target products [103]. Several methods are available to develop clean hosts, such as a PCR-targeted system [104], site-specific recombination using Cre-loxP [105,106], meganuclease I-SceI [107], and clustered regularly interspaced short palindromic repeat (CRISPR)/Cas genome editing tools [108]. A PCR-targeted system is a useful tool for gene knockout. This system contains a selectable gene, two FRT or loxP sites with 40–50 bp of homologous extension flank at both ends, and a λ recombinant system. Xu et al. successfully deleted 10 regulator genes in S. coelicolor by using a PCR-targeted method [109]. Site-specific recombination using Cre/loxP is applied in the knockout and/or insertion of huge BGC regions in many Streptomyces spp. The Cre-loxP system needs a double crossover to introduce the loxP site into the genome and the expression of Cre recombinase to delete target genes. The Cre-loxP system was successfully used to delete 1.4 Mb of S. avermitilis’s genome [110]. Meganuclease I-SceI is an endonuclease from S. cerevisiae that recognizes an 18 bp unique sequence and cleaves DNA for engineering a genome with high efficiency and multiple deletions [111]. This method was successfully applied to delete the red-pigmented undecylprodiginine BGC in S. coelicolor M1141 [111]. Recently, the CRISPR/Cas system of Streptococcus pyogenes (SpCas9) is a vital tool in genome engineering that is widely applied in different strains [112]. It has been developed and used to clean many Streptomyces hosts, such as S. coelicolor, S. lividans, and S. albus [113,114]. However, SpCas9 does not properly work in some Streptomyces strains [115,116]; therefore, the FnCas12a system (from Francisella novicida U112), fnCpf1 (from Francisella tularensis subsp. novicida U112 Cpf1), sth1Cas9 (from Streptococcus thermophilus CRISPR1 Cas9), and saCas9 (from Staphylococcus aureus Cas9) were developed for genome editing Streptomyces strains [116,117]. For example, the FnCas12a system was used to disrupt 128 kb of chromosome of S. hygroscopicus [117].

Additionally, the specific integration ΦC31 attB loci sites were introduced into a clean host to facilitate the integration of exogenous biosynthetic clusters into the chromosomes [11,102]. For example, in the S. albus Del14 strain, fifteen BGCs were deleted, and one, two, and four ΦC31 attB sites were introduced to make S. albus B2P1, B3P1, and B4, respectively [102]. S. lividans ΔYA9, ΔYA10, and ΔYA11 were achieved by deleting 9, 10, and 11 BGCs and one, two, and three additional introduced attB sites, respectively [11].

4.2. Transcription Terminators

Besides the regulator and promoter, the transcriptional terminator is another essential element that affects the gene expression level. The transcription terminator includes protein-dependent factors, such as rho, nusA, and tau, and protein-independent factors structured by a stem-loop form in messenger ribonucleic acid (mRNA) [118]. Based on the heterologous expression level of glucuronidase, Horbal et al. found that the tt_sbiB terminator is the strongest compared with U, V, T4 lang, and T4 kurz terminators [119]. Curran et al. reported that terminators impact the mRNA half-life and can be used to modulate gene expression [120]. The terminators carry larger stem-loop structures to increase the efficiency of the mRNA released. Inverted repeat sequences play a significant role in the stability of the transcription by a stem-loop structure. The Streptomyces genome, with its high GC, requires a long terminator and a powerful stem-loop structure to increase the transcription termination efficiency [118]. For example, the TD1 terminator from B. subtilis was successfully introduced into S. lividans [121]. Many transcription terminator sequences were found in several genes, such as aph, vio, tsr, and hyg. The terminator of the aph gene was successfully utilized for the heterologous expression of the interferon α2 gene from humans in S. lividans [118].

4.3. Riboswitches

A riboswitch is a non-coding RNA structure, which includes an aptamer domain and an expression platform. It can regulate the expression of the genes at both the transcription and translation levels by interacting with mRNA [122]. Rudolph et al. found several theophylline-dependent riboswitches to induce gene expression in S. coelicolor. Riboswitch E* was the best-activated regulator for the expression of agarase gene dagA [123]. The heterologous expression of btm BGC from Streptomyces scabies, together with a gene from yeast in S. coelicolor, lead to a 120-fold increase in bottromycin production [124]. The cumate and theophylline dual-riboswitch control system were expressed in S. albus. They could control the production of pamamycin [125].

4.4. tRNA Engineering

Studies on tRNA have demonstrated the presence of 30–40 different tRNAs in bacteria. The tRNAs carry amino acids to build proteins. A few specific aminoacyl-tRNAs also participate in the biosynthesis of natural products, such as non-ribosomal peptides, tRNA-dependent cyclodipeptides, diketopiperazines, and cyclodipeptides, by using aminoacyl-tRNA to form peptide bonds. Gondry et al. reported that the active form of AlbC from Streptomyces noursei required aminoacyl-tRNA as a substrate to catalyze the cyclodipeptide formation [126]. Some BGCs required specific tRNAs to build the diketopiperazine precursors for the biosynthesis of natural compounds. For example, the BGC of bicyclomycin from Streptomyces sapporonensis ATCC 21532 was expressed in E. coli, suggesting that Leu-tRNA^Leu and Ile-tRNA^Ile are precursors to initiate the biosynthesis of bicyclomycin [127].

The biosynthesis of some compounds occurs in different ways, such as the C4 (Shemin pathway) or the C5 pathway. For example, the biosynthesis of tetrapyrroles required 5-aminolevulinic acid as the precursor, which is generated via the condensation of glycine and succinyl-CoA in the C4 pathway in animal mitochondria and alpha-proteobacteria, whereas glutamyl-tRNA is needed in the C5 pathway in all other cells (chloroplasts) [128]. The heterologous expression of the gene encoding aminoacyl-tRNA led to increasing the titer of secondary metabolites in expression hosts. For example, the overexpression of hemA1 encoding glutamyl-tRNA in S. roseosporus HCCB10043 and Streptomyces gilvosporeus HCCB 13086 led to a 1.4-fold increase in daptomycin and a 1.64-fold increase in natamycin, respectively [129]. However, specific tRNAs also indirectly affect the morphology or the yield of secondary metabolites depending on the need for a specific tRNA as an adaptor molecule for the biosynthesis of polypeptide chains.

The bldA gene from S. coelicolor encodes specific ${\mathrm{tRNA}}_{UUA}^{Leu}$, which recognizes the rare UUA codon in Streptomyces. It can control the expression of phenotypic carB gene in S. coelicolor [130] and the heterologous expression of pur BGC from Streptomyces alboniger in S. lividans [131]. The interaction between bldA-tRNA and DnrO led to increasing the daunorubicin production in S. peucetius [8]. Some natural products were produced by the heterologous expression of BGCs involving tRNA synthases genes; for instance, the valanimycin BGC consists of the gene vlmL encoding Ser-tRNA^Ser that produces seryl-tRNA, which may take part in the biosynthetic pathway of valanimycin [132].

4.5. Regulators

Regulators positively or negatively affect the expression of target BGCs by directly or indirectly regulating the transcription of target genes. Streptomyces contains many transcriptional regulator families, such as TetR, LuxR, MarR, and ArfR. Each family has different functions. The TetR family regulator is the most popular in Streptomyces and controls the expression of multiple genes in secondary metabolite BGCs. For example, the expression of GdmRIII led to increasing the production of geldanamycin while decreasing the production of elaiophylin in Streptomyces autolyticus CGMCC0516 [133]. AccR bound to a 16-nucleotide palindromic binding motif in the promoter of target genes resulted in the decreasing biosynthesis of malonyl-CoA and methylmalonyl-CoA in S. avermitilis. The production of avermectin was enhanced by 14.5% when accR gene was deleted [134]. The LuxR family regulator is a pathway-specific transcriptional regulator of BGCs, including AniF-activated anisomycin production [135]. The MarR family regulators contain winged helix-turn-helix DNA binding domains, which can work like transcriptional repressors or activators. Gou et al. indicated that SAV4189, a MarR-family regulator, acts as an activator of avermectin production in S. avermitilis. They also produced the heterologous expression of sav_4189 in S. coelicolor, resulting in enhancing the antibiotic production of this strain [136]. Peng et al. engineered S. lividans STB5 by integrating global regulator genes, such as mdfA_co, lmrA_co, nusG_sc, and afsRS_cla, and deleting a transcription regulator gene, namely, wblA_sl, to achieve S. lividans LJ1018. The production of murayaquinone, hybrubins, actinorhodin, dehydrorabelomycin, and actinomycin D achieved via heterologous expression in S. lividans LJ1088 were increased 96-, 29-, 23.3-, 12.7-, and 3.5-fold compared with the production achieved via heterologous expression in S. lividans STB5, respectively [137]. Therefore, for improving the heterologous expression level of BGCs, host strains should be engineered by deleting repressor genes and integrating activator genes.

5. Strategies for the Construction of a Biosynthetic Gene Cluster

5.1. Vectors and Cloning Methods

BGCs contain many genes, which encode one or some secondary metabolites. Some cloning methods were used to clone large BGCs. Traditionally based on a library, the large genomic DNA has been digested and ligated into high-capacity vectors, such as cosmids, fosmids, or BACs. Cosmids can insert DNA fragments ranging in size between 32 kb and 45 kb based on the bacteriophage vector [138]. Fosmids, which are similar to cosmids but contain an F-factor to control the copy number of vectors, can insert DNA fragments up to 100 kb in size [139]. For example, the mat gene cluster used a fosmid vector to produce deshydroxymatlystatins via heterologous expression in S. coelicolor and S. albus [53]. The BAC vector contains the F-factor, multiple cloning sites, and is capable of inserting DNA that is approximately 490 kb in size [140]. PAC vector could insert DNA that was between 100 and 300 kb in size [141].

BGCs can be cloned in a directed manner by employing in vivo cloning methods, such as TAR, LLHR, ExoCET, and PISR. TAR constructs target BGC from a digested genome to an expression vector using yeast. This technique successfully cloned a DNA fragment with a size up to 250 kb [142]. As the recombinant vector was constructed in yeast, maintained in E. coli, and expressed in a host, it has to carry elements from each of the yeast, E. coli, and the host. For example, the 67 kb BGC of taromycin A was successfully expressed from Saccharomonospora sp. CNQ-490 in S. coelicolor using the TAR cloning system [143]. The LLHR used E. coli to clone a large gene cluster up to 100 kb to an expression vector based on the activity of RecE/RecT. Fu et al. successfully used this method to clone and express PKS-NRPS BGCs from P. luminescens in E. coli [144]. Recently, ExoCET was used to directly clone large selected regions (up to 106 kb) into a BAC vector in a single step. ExoCET combines an in vitro exonuclease of T4 polymerase with a full-length RecE/RecT to clone homology regions in E. coli. The efficiency of vector construction depends on the position of the homologous sequence in the digested genome. For increasing the efficiency of this method, Cas9 was applied to digest the genome. For instance, the salinomycin gene cluster was cloned to a BAC vector by using the ExoCET method [145]. PISR was used to clone and delete large BGCs directly from the Streptomyces genome based on the activity of the ϕBT1 integrase. Du et al. applied this technique to delete and clone actinorhodin, nikkomycin, and gougerotin from S. coelicolor, Streptomyces ansochromogene, and Streptomyces graminearus, respectively [146].

Besides in vivo cloning, some BGCs were also successfully cloned using PCR gene and DNA assembly in different ways. In vitro cloning includes DNA assembly methods, such as Gibson assembly, DiPAC, PfAgo-based AREs, SSRTA, SIRA, and CATCH. Gibson assembly was applied to clone DNA fragments with a size up to 14 kb, such as bicyclomycin [147], kocurin [23], and lasso peptides [148]. Gibson assembly needs DNA polymerases, exonucleases, and DNA ligases to assemble BGCs with vectors. DiPAC was applied to clone DNA fragments amplified from PCR based on HiFi DNA assembly. DiPAC is suitable for cloning short BGCs, with a capacity of up to 23 kb, such as apt [98], sodorifen [149], and hapalosin [150] BGCs. SIRA used Φ31 integrase and a pair of different orthogonal attP/attB recombination sites to form attL/attR for a recombinational assembly and cloning strategy [151]. For example, Gao et al. successfully cloned and expressed erythromycin BGC using SIRA [152]. The SSRTA method is based on the operation of Streptomyces φBT1 integrase and attB/attP recombination sites for the tandem assembly of multiple DNA fragments. Some BGCs were cloned using this method, such as epothilone [28] and lycopene BGCs [153]. PfAgo-based AREs utilize PfAgo to create artificial restriction enzyme sites, which possess the ability to recognize and cleave a specific DNA sequence. Therefore, PfAgo-based AREs can become a robust tool for cloning BGCs because PfAgo can induce the unique restriction sites, which are often limited in the target DNA [154]. The AGOS is a plug-and-play method based on a SuperCos1 and Red/ET-mediated assembly that introduces unique restriction sites to target BGCs. From this target, BGCs can be cloned by using the restriction enzymes. AGOS is a power tool for refactoring and cloning target BGCs [155]. CATCH assembles DNA target sequences that are digested using Cas9 from the bacterial chromosomes with vectors via Gibson assembly [156]. In vivo cloning methods are powerful tools for cloning large-size BGCs (up to 250 kb when using the TAR cloning method). However, the efficiency of these tools is not so high, from 8.3 to 58% [143,145]. In vitro cloning methods are useful for cloning multiple DNA fragments with very high cloning efficiencies from 87% up to almost 100%; however, these methods are usually used for cloning the small and medium DNA sizes (up to 100 kb) [98,147,152,153]. The cloning and expression protocol for the BGCs is shown in Figure 3.

5.2. Promoter Engineering

The promoter is one of the most important factors for the heterologous expression of BGCs at the transcription level. Each host has different promoter systems. Therefore, after the engineering host, powerful and suitable promoters should be selected for developing heterologous expression systems. When the BGCs are cloned to vectors, the powerful and suitable promoter can replace the native promoter. This helps to enhance or activate the heterologous expression of BGCs.

The most suitable part of the promoter activity relies on the −10 and −35 regions. These regions determine the strength of the RNA polymerase binding. There are three types of promoters in Streptomyces. A type I promoter includes the −10 element (5′-TANNNT-3′) and −35 element (5′-TTGNC-3′) with an 18–19 nucleotide spacer length [157,158], such as the hrdB promoter from S. coelicolor [159]. In contrast, a type II promoter contains the −10 element, similar to type I, and other elements (5′-TGTC-3′) with a different spacer length [160], and a type III does not display the typical −10 and −35 sequences that are present in different types [157,161].

Native promoters were used in studies for a long time, but they do not work at a wide range of transcription levels. Based on the properties of native promoters, powerful synthetic promoters were developed and selected for the expression of the target gene at a high level. Promoter engineering is performed by changing the sequence between the −10 and −35 regions of a native promoter. Several native promoters have been studied and used to drive the expression of genes in Streptomyces, such as inducible promoters tipAp from S. lividans 66 [162] and nitAp from Rhodococcus rhodochrous J1 [163], and the constitutive promoters ermEp from Streptomyces erythraeus [164], SF14p from Streptomyces ghanaensis phage I19 [165], and kasOp from S. coelicolor [166]. The activity of promoter thlM4p from Streptomyces chattanoogensis L10, which was selected via a transcriptome-based screening method, was seven times stronger than the promoter ermE*p [167]. Many synthetic promoters were developed; among them, ermE*p and kasO*p were widely used in the gene expression in Streptomyces. Wang et al. demonstrated that the kasO*p promoter is stronger than the SF14p promoter, which is stronger than the erm*p promoter, as seen from the production level of actinorhodin when used in S. coelicolor [166].

5.3. RBS Tuning

Engineering the expression of genes is focused not only on the transcriptional level but also on the translation level. The protein expression strongly depends on the RBS, which controls the gene expression at the translation level. The position and sequence of the RBS affect the translational efficiency and protein expression. The experiment showed that an unsuitable RBS could reduce the gene expression to zero [119], which then affects the biosynthesis of secondary metabolite products. For identifying a suitable RBS, Farasat et al. developed a Library Calculator [168] and Na et al. introduced the RBSDesigner [169] for designing a powerful synthetic RBS sequence [168] to enhance the protein expression.

6. Conclusions

The development of genome sequencing and bioinformatics tools has greatly facilitated the discovery of BGCs. Advances in heterologous expression have resulted in the development of novel secondary metabolites. The heterologous expression of natural products depends on two crucial factors: host engineering and BGC construction. Host engineering involves choosing hosts, deleting unwanted BGCs, inducing transcriptional terminators, riboswitches, engineering tRNAs, and deleting or inserting regulators. In comparison, BGC construction involves vectors, cloning methods, promoters, and RBSs.

Many hosts were used for the heterologous expression of natural products from Streptomyces, such as Streptomyces spp., S. cerevisiae, B. subtilis, E. coli, C. glutamicum, P. putida, and R. erythropolis. For easily detecting and purifying heterologous products, unwanted secondary metabolite BGCs from host strains should be deleted. Many genetic systems were developed for genome editing, such as CRISPR/Cas, meganuclease I-SceI, site-specific recombination using Cre-loxP, and PCR-targeted systems. The transcriptional regulator and terminator are two factors that affect the transcription level. Negative regulator genes of heterologous BGCs should be deleted. Moreover, positive regulator genes and transcriptional terminator genes should be introduced in hosts to improve the heterologous expression level of natural products. Riboswitches could regulate the heterologous expression of BGCs in both transcription and translation levels. They are also crucial factors that can be used to control the heterologous expression. Furthermore, some active tRNAs are precursors for the biosynthesis of secondary metabolites. Therefore, inducing specific tRNA genes into hosts could also enhance the production achieved from the heterologous expression of BGCs.

Two cloning method classes have been applied to clone the target BGCs, namely, in vitro and in vivo cloning methods. The vectors and the cloning methods were chosen depending on the heterologous hosts and the size of the target BGCs. In vitro cloning methods are often used for cloning the small- or medium-sized attractive BGCs, while in vivo cloning methods are usually applied for cloning the large-sized target BGCs. For the activation or enhancement of the expression products, recombinant vectors are also needed for the reconstruction. Robust and suitable promoters and RBSs were used to replace the native promoters and RBSs.

In conclusion, the advances of synthetic biology and powerful genome mining techniques led to rapidly discovering and enhancing the production of natural products. Besides OSMAC, the overexpression/deletion of regulatory genes, ribosome engineering, and promoter replacement strategies that are utilized in heterologous expression provide robust strategies.

Author Contributions

Conceptualization, V.T.T.P., C.T.N. and J.K.S.; writing-original draft preparation, V.T.T.P. and C.T.N.; writing-review and editing, V.T.T.P., C.T.N., D.D., H.T.N., and T.-S.K.; supervision, J.K.S.; funding acquisition, J.K.S. All authors have read and agreed to the published version of the manuscript.

Funding

This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MEST) (NRF-2017R1A2A2A05000939.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

Not applicable.

Conflicts of Interest

The authors declare no conflict of interest.

Abbreviations

AGOS, artificial gene operon assembly system; ARE, artificial restriction enzyme; BAC, bacterial artificial chromosomal; BGC, biosynthetic gene cluster; CATCH, Cas9_Assisted Targeting of Chromosome segments; CLF, chain length factor; CRISPR, clustered regularly interspaced short palindromic repeat; DiPAC, direct pathway cloning; DNA, deoxyribonucleic acid; ExoCET, exonuclease combined with RecET recombination; GC, guanine-cytosine; GNPS, Global Natural Products Search; GRAS, generally recognized as safe; LLHR, linear-plus-linear molecule homologous recombination; MIBiG, Minimum Information about a Biosynthetic Gene Cluster; mPKS, minimal PKS; mRNA, messenger ribonucleic acid; MS/MS, tandem mass spectrometry; NRPS, nonribosomal peptide synthase; ORF, open reading flame; OSMAC, One Strain Many Compounds; PCR, polymerase chain reaction; PfAgo-based ARE, Pyrococcus furiosus Argonaute protein-based artificial restriction enzyme; PISR, phage ϕBT1 integrase-mediated site-specific recombination; PKS, polyketides synthase; PPTase, 4′-phosphopantetheinyl transferase; RBS(s), ribosome binding site(s); RiPP, ribosomally synthesized and post-translationally modified peptide; SIRA, serine integrase recombinational assembly; SMILES, simplified molecular-input line-entry system; SSRTA, site-specific recombination-based tandem assembly; TAR, transformation associated recombination; tRNA(s), transfer ribonucleic acid(s).

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Figures and Table

Figure 1. Natural secondary metabolites synthesized in Streptomyces: avermectin, pentalenene from Streptomyces avermitilis [2,7], doxorubicin and geosmin from Streptomyces peucetius ATCC 27952 [8,9], oxazolomycin from Streptomyces albus JA3453 [10], germicidin from Streptomyces lividans TK24 [11], isoafricanol from Streptomyces malaysiensis [4], echinomycin from Streptomyces lasaliensis ATCC 35851 [12], bosamycin from Streptomyces sp. 120454 [3], cyclizidine from Streptomyces sp. HNA39 [13], legonimide from Streptomyces sp. CT37 [6], avilamycin from Streptomyces viridochromogenes Tü57 [5,14], and hygromycin from Streptomyces hygroscopicus [15].

Figure 2. Processing of heterologous biosynthetic gene clusters. RBS, ribosome binding site; CDS, coding sequence; TAR, transformation-associated recombination; LLHR, linear-plus-linear homologous recombination-mediated recombineering; PISR, phage ϕBT1 integrase-mediated site-specific recombination; ExoCET, exonuclease combined with RecET recombination; DiPAC, direct pathway cloning; SSRTA, site-specific recombination-based tandem assembly; SIRA, serine integrase recombinational assembly; CATCH, Cas9-assisted targeting of chromosome segments; PfAgo-based AREs, Pyrococcus furiosus Argonaute-protein based artificial restriction enzymes; AGOS, artificial gene operon assembly system; PCR, polymerase chain reaction.

Figure 3. Schematic diagram of the BGC cloning methods. The in vivo cloning methods include TAR, PIRS, LLHR, and ExoCET systems, while the in vitro cloning methods include CATCH, SIRA, and SSRTA assembly. RE, restriction enzymes; kanR, kanamycin restriction gene gene; ampR, ampicillin restriction gene; tetR, tetracycline restriction enzyme; gRNA/Cas9, guide RNA/Cas9; T4 pol, T4 polymerase.

Table 1

List of compounds synthesized heterologously.

Native Strain	Heterologous Host	Method Clone	Products
Streptomyces leeuwenhoekii C34T	S. coelicolor M1152 and M1154	PCR, cloning to pIJ10257	Leepeptin [95]
S. variabilis Snt24	S. lividans SBT5	BAC clone	Tetramic acid [21]
Streptomyces koyangensis SCSIO 5802	S. coelicolor M1152	Phage-P1-derived artificial chromosome (PAC) library	Neoabyssomicin, abyssomicin [96]
Streptomyces sp. Acta1362	S. albus J1074	TAR system	Grecocycline [97]
S. venezuelae	S. lividans and S. coelicolor	BAC cloning	Pikromycin [22]
S. galtieri Sgt26	S. albus J1074	BAC cloning	Kinamycin [58]
S. albus subsp. chlorinus NRRL B-24108	S. albus Del14, S. lividans TK24	BAC clone	Albucidin [57]
S. erythraea DSM 40517	S. coelicolor M1152 and M1154	Direct pathway cloning (DiPaC)	Erythromycin [98]
Streptomyces tsusimaensis ATCC 15141	E. coli	Co-expression genes	Valinomycin [99]
S. erythraea	B. subtilis	PCR, cloning	6-deoxyerythronolide B [76]
S. davaonensis	C. glutamicum	PCR, cloning	Roseoflavin [87]
S. griseus	P. putida	PCR, cloning	Flaviolin [91]
S. kasugaensis	R. erythropolis	PCR, cloning	Kasugamycin [94]
Actinosynnema mirum DSM 43827	S. avermitilis SUK22	PCR, cloning	Mycosporine-glycine-alanine [67]
Kocuria rosea s17	S. coelicolor M1146, S. sp. s120	Gibson assembly, integration vector pSET152	Kocurin [23]
Amycolatopsis sp. DEM30355	S. coelicolor M1152	PAC library	Vancoresmycin [26]
Myxococcus xanthus DK1622	S. albus Del14	PCR and subcloned	Cittilins [100]
Actinoplanes tsinanensis CPCC 200056	S. coelicolor M1146	DNA assembler by yeast	Chuangxinmycin [60]
Kutzneria albida DSM 43870	S. albus Del14	BAC vector	Huimycin [62]
Planobisporarosea ATCC 53733	S.coelicolor M1146	Cosmid vector	GE2270 [66]
A. mirum DSM 43827	S. avermitilis SUKA22	In vivo by phage λ-Red recombining system	Shinorine, porphyra-334, mycosporine-glycine, and mycosporine-glycine-alanine [67]
Kitasatospora setae KM-6054	Streptomyces lohii JCM 14114, S. griseus DSM 2608	BAC cloning	Bafilomycins A1, C1, and B1 (setamycin) [101]