Biotechnology Approaches for Natural Product Discovery, Engineering and Production based on Burkholderia Bacteria

Abstract

Bacterial natural products retain high value in discovery efforts for applications in medicine and agriculture. Burkholderia β-Proteobacteria are a promising source of natural products. In this review, we summarize the recent genetic manipulation techniques used to access silent/cryptic biosynthetic gene clusters from Burkholderia native producers. We also discuss the development of Burkholderia bacteria as heterologous hosts and the application of Burkholderia to industrial-scale production of natural products. Genetic engineering and fermentation media optimization have enabled the industrial-scale production of at least two Burkholderia natural products. The biotechnology approaches discussed here will continue to facilitate the discovery and development of natural products from Burkholderia.

Introduction

Natural products (NPs) of bacterial origin continue to provide an abundant source of novel chemical entities with applications in agriculture and medicine [1]. It has been recently estimated that only 3% of bacterial NPs have been discovered [2]. Gram-negative β-Proteobacteria belonging to the Burkholderia genus are an emerging source of NPs [3]. Typically found in diverse terrestrial and aquatic niches, Burkholderia exist as free-living species, or in symbiotic/pathogenic associations with eukaryotic hosts [3]. Bioactivity-guided assays and genome mining approaches have led to the discovery of nonribosomal peptides (NRP), polyketides (PK), hybrid polyketide-nonribosomal peptides, polyynes, terpenes, ribosomally synthesized and posttranslationally modified peptides (RiPP), and others. Interested readers are directed to a recently published review on NPs isolated from Burkholderia [3].

Here we offer an account of biotechnology approaches to NP discovery and production from Burkholderia bacteria. This review is not intended to be comprehensive. Rather, it emphasizes the genetic manipulation techniques applied to Burkholderia native producers, and the Burkholderia heterologous hosts tested to discover, study, or produce NPs. We describe the genetic manipulation techniques used to connect NPs to biosynthetic gene clusters (BGCs) or to activate silent BGCs from Burkholderia native producers in the last five years. We then follow with an account of Burkholderia strains that have been used as heterologous hosts. Lastly, we highlight studies on fermentation media optimization and applications of Burkholderia bacteria to industrial-scale production of select NPs (Figure 1).

Figure 1. Biotechnology approaches applied to Burkholderia bacteria discussed in this review.

Schematic illustration using an optimized culture medium to profile BGCs via (A) targeted mutagenesis of Burkholderia native producers involving either host-based homologous recombination, phage-based recombineering, or CRISPR-Cas9 based techniques; (B) untargeted mutagenesis of native producers involving transposon-based techniques; and (C) heterologous expression in Burkholderia hosts. BGC, biosynthetic gene cluster; sgRNA, single guide RNA.

For complementary reviews, interested readers are referred to recently published articles on accessing and refactoring BGCs for NP discovery and production [4–8].

Genetic engineering approaches applied to Burkholderia native producers

Advances in bioinformatics predictive tools and repositories for BGCs [9–15], combined with genetic engineering via targeted or untargeted mutagenesis have helped elucidate NP biosynthesis. Genetic engineering of native producers can aid with NP discovery by either comparative metabolite analysis of wild-type and knockout mutant strains or by activation of poorly expressed (silent) BGCs. Here, we focus on engineering techniques applied to native Burkholderia producers to discover new NPs (Figure 1, Table 1). Moreover, we highlight the application of these techniques to study the biosynthesis of known NPs and to connect NPs to BGCs. We attempted to focus on references from the last five years but also cite seminal and review papers to place the techniques into context.

Table 1.

Natural products discovered in the last 5 years using genetic manipulation of Burkholderia native producers.

Burkholderia	Natural Product	Class	Yield (mg/L)	Technique used	Ref.
B. gladioli BCC0238	Bolagladins	NRP	0.75–0.85	Homologous recombination	[23]
B. glumae	Burrioglumins	NRP	N/A1	Homologous recombination	[29]
B. plantarii DSM9509	Haereoplantins	NRP	0.28–1.05	RecET-like recombineering from Burkholderia spp.	[33]
B. gladioli ATCC 10248	Burriogladiodins	NRP	0.3–0.85	Redy-BAS recombineering from Pseudomonas aeruginosa	[32]
	Haereogladiodins		0.25–0.4
B. gladioli pv. agaricicola	Burriogladins	NRP	0.1–1.75	RecET recombineering from E. coli	[29]
	Haereogladins		0.15–6.45
B. glumae	Haereoglumins	NRP	NR	RecET-like recombineering from Pseudomonas syringae
B. gladioli HKI0739	Gladiofungins	PK	0.09–2.3	CRISPR-Cas9 engineering	[36]
B. thailandensis	Thailandenes	PK	0.05–0.08	Transposon mutagenesis	[44]
B. gladioli ATCC1024	Gladiobactin A	NRP	0.75	Transposon mutagenesis	[46]
B. plantarii ATCC4373	Haereoplantins A		0.25 – 0.5
	Burrioplantin A

NRP, nonribosomal peptide. PK, polyketide. NR, not reported.

¹Burrioglumins were isolated from an E. coli heterologous expression host to afford 0.0625–0.075 mg/L

Host-based homologous recombination.

Gene deletion or replacement using the host’s innate, recA-dependent homologous recombination system has been widely used in bacteria [16]. The examples highlighted below illustrate the different vector systems and modes of mutagenesis that have been recently applied to Burkholderia. To link BGCs to target NPs and to study biosynthesis, mutagenesis to disrupt metabolism has been exploited by either replacing target genes with antibiotic resistance markers or by performing markerless gene deletions.

To cite a recent example of gene replacement with an antibiotic resistance marker, Dose et al. knocked out isonitrile synthase genes by substitution with a kanamycin resistance gene, enabling the characterization of the biosynthetic route to the rare antifungal sinapigladioside [17]. While replacements with resistance markers are convenient due to the relative ease of selecting mutants, polar effects – unintended effects on other genes in the same operon – may be an issue [18].

In addition to avoiding transcriptional polar effects due to readthrough from the marker’s promoter [18], markerless deletion has the advantage of enabling the use of the same antibiotic resistance marker for sequential genetic manipulations. Considering that two species of Burkholderia bacteria (B. mallei and B. pseudomallei) belong to the category B select bioterrorism agents in the United States, only a few antibiotic resistance markers are allowed for genetic manipulation with these bacteria. Besides, Burkholderia have been reported to possess high levels of resistance to commonly used antibiotic markers so that the number of markers available are limited [19–21]. We highlight three vector systems below to illustrate different modes of achieving markerless deletions.

Flannagan et al. adapted a system based on the yeast homing endonuclease I-SceI for use in B. cenocepacia. A suicide vector, pGPI-SceI, containing a trimethoprim resistance gene and an 18-bp recognition site for I-SceI in the vector backbone is used to construct a replacement allele. The gene replacement plasmid should contain ~1-kb homology arms to the flanking sequences of the target gene. After integration of the replacement plasmid into the genome via homologous recombination, a second plasmid, pDAI-SceI, that constitutively expresses I-SceI is introduced into the single crossover mutant. I-SceI activity leads to a double stranded break at the I-SceI recognition site which induces a second intramolecular recombination event yielding either the wild-type or the desired deletion of the target gene [20]. Double crossovers can then be identified by loss of trimethoprim resistance and confirmed by e.g., PCR. The I-SceI approach was more recently expanded by Abdu et al. to include other resistance markers and enable its use to more members of the B. cenocepacia complex [21].

Jenul et al. applied the I-SceI approach to demonstrate the biosynthetic origin of the antifungal metallophore fragin in B. cenocepacia H111 [22]. Dashti et al. applied it in constructing a non-producing gladiolin mutant of B. gladioli BCC1622, with an in-frame deletion in gbnD1 encoding a polyketide synthase. The gbnD1 mutant was then used as parent for subsequent deletions during biosynthetic studies of the non-ribosomal peptide and potential siderophore bolagladin [23].

While the I-SceI endonuclease approach is effective, it is applicable only if the host’s genome does not already contain a I-SceI recognition site. An alternative approach we are using in our lab (unpublished) is based on the suicide vector pMo130 developed by Hamad et al. [19]. Vector pMo130 contains a xylE reporter gene encoding 2,3-catechol dioxygenase, providing a visual detection of plasmid integration in the genome because clones expressing xylE turn yellow when sprayed with catechol. Having xylE in addition to the kanamycin-resistance marker is helpful because many Burkholderia are naturally resistant to antibiotics and background colonies are common even with high antibiotic concentrations. The pMo130 vector also contains a sacB gene encoding levansucrase that confers lethal sensitivity on sucrose-containing media and serves as a counter-selection marker to force the resolution and loss of vector backbone. Thus, single crossover clones are identified guided by the xylE reporter and then cultivated on media containing sucrose where only double crossover mutants should be able to grow.

The last markerless deletion approach we want to highlight is based on Flp-FRT recombination. The flippase (Flp) protein is a site-specific recombinase that recognizes the flippase recognition target (FRT). In this approach, two FRT sequences are integrated into the genome via homologous recombination to flank the target region. Placing an antibiotic resistance marker between the FRT sites facilitates mutant selection. After flp is expressed from a plasmid, Flp-mediated recombination deletes the sequence between the FRT sites and leaves behind one FRT “scar” [24]. Flp-FRT recombination was used in the deletion of hamC, hamD, hamE, and afcA during elucidation of fragin biosynthesis in the study by Jenul et al. [22]. The Flp plasmid contained sacB for counterselection and plasmid curing after deletion mutants were obtained.

Phage-based RecET/Redαβγ recombineering.

Recombinase systems with broad functionalities and high efficiency for homologous recombination in target strains can allow scalable, efficient, and precise genomic manipulation. The two main advantages of bacteriophage-based recombineering are the ability to use short, ~50-bp homology arms and linear DNA for transformation so that the replacement allele can be conveniently generated by PCR rather than requiring cloning into a circular plasmid vector. The third advantage is improved recombination efficiency. The use of RecET from the Rac prophage of E. coli K12 and Redαβγ from the λ phage was reported for cloning and for genome manipulation in E. coli two decades ago [25,26]. RecET cloning or genome editing using linear DNA is possible in strains that have the exonuclease activity of RecBCD inhibited by either a sbcA suppressor mutation or via expression of λ-redγ [25]. RecET and λ-Redαβ analogously act as 5’→3’-exonuclease and single-stranded DNA binding proteins, respectively, to promote recombination between short homology sequences [25]. A distinction is that while λ-Redαβ promote linear - circular homologous recombination, RecET is more efficient at catalyzing linear-linear homologous recombination [25,27].

A method to apply λ-Red recombineering for naturally transformable B. thailandensis and B. pseudomallei strains was described about a decade ago and can be used to capture or delete genes [28]. In 2018, Thongkongkaew et al. used RecET recombineering to associate BGCs with haereogladins and burriogladins from B. gladioli pv. Agaricicola [29]. However, due to low efficiency with most Burkholderia strains [28] the need to establish recombineering systems that are broadly applicable in Burkholderiales remained.

In 2018, Wang et al. [30] identified seven Redαβ and ten RecET homologs in ~1,000 Burkholderiales genomes they searched. One Redαβ-like operon from the Burkholderiales strain DSM 7029 (now assigned as Schlegelella brevitalea DSM 7029) and one RecET-like operon from Burkholderia sp. BDU8 were tested, showing that the Redαβ-7029 system functions best for recombineering both in E. coli and in DSM 7029. The Redαβ-7029 operon was placed under a l-rhamnose inducible promoter and, together with λ-Redγ which enhanced recombination efficiency, was used for genome editing. Two examples of genome editing in DSM 7029 were provided, i.e., gene and BGC inactivation of up to 200-kb in size via replacement with a selectable marker using as low as 50-bp homology arms, and the activation of silent BGCs via promoter replacement. Promoter replacement led to the activation and discovery of NPs from DSM 7029 itself and from Mycetohabitans rhizoxinica HKI 454. Zheng et al. later applied the λRedγ-Redαβ7029 system in Paraburkholderia megapolitana DSM 23488 [31].

In contrast, Chen et al. showed that recombineering with the λRedγ-Redαβ7029 system was ineffective in Burkholderia gladioli ATCC 10248 whereas a Redγ-BAS recombination system, built from the λ Red-like operon BAS from Pseudomonas aeruginosa phage Ab31, successfully activated silent BGCs to discover lipopeptides burriogladiodins and haereogladiodins [32]. Thongkongkaew et al. used a RecET-like recombineering system from Pseudomonas syringae in the study of lipopeptides haereoglumins A and B from Burkholderia glumae with roles in phytopathogenesis [29]. In another study, Li et al. tested three RecET-like recombination systems from different Burkholderia spp. for genome editing, facilitating the discovery of lipopeptides haereoplantins F – H via promoter replacement in B. plantarii DSM 9509 [33].

CRISPR-Cas9 engineering.

The clustered regularly interspaced short palindromic repeats (CRISPR) and the CRISPR-associated protein 9 (Cas9) that endow Streptococcus pyogenes with RNA-guided phage immunity have been widely applied for genome engineering from bacteria to humans [34]. Here we highlight two examples of CRISPR-Cas methods applied to Burkholderia to either effect gene silencing or genome editing.

In 2019, Hogan et al. reported the adaptation of the CRISPR interference (CRISPRi) technology to Burkholderia [35]. CRISPRi is based on the formation of a RNA-protein complex between a nuclease-inactive version of Cas9 (dCas9, ‘dead’ Cas9) and a single guide RNA to sterically inhibit transcription, and observe phenotypic changes. A codon-optimized dcas9 gene from S. pyogenes was placed under control of a l-rhamnose inducible promoter and the CRISPRi approach was tested with B. cenocepacia, B. thailandensis, and B. multivorans. The authors targeted the conditionally essential gene paaA necessary for phenylacetate degradation to detect reduced growth on phenylacetate as the sole carbon source. Although CRISPRi has yet to be applied to natural products, it may offer a valuable tool to link BGCs to NPs in the future by rapidly silencing BGCs.

In 2020, Niehs et al. reported the application of CRISPR-Cas9 for excision or point mutation of an A-factor-synthase-like domain within the polyketide synthase encoding gladiofungin biosynthesis in the insect-associated B. gladioli HKI0739 [36]. A codon optimized cas9 gene from S. pyogenes was placed under a l-rhamnose-inducible promoter and used to catalyze double strand breaks. The l-rhamnose-inducible λRedγ-Redαβ7029 system [30] was cloned in the same pRO1600-based, temperature sensitive vector and used to improve homologous recombination between the target domain and the replacement allele provided in the same vector as well [36].

Moreover, endogenous CRISPR-Cas systems have been identified in Burkholderia such as in plant pathogenic B. gladioli, B. plantarii and B. glumae strains [37–39], and in soil Burkholderia sp. SMB0852 and lichen-associated Burkholderia sp. strains [40] that may facilitate further CRISPR-Cas-based genome editing in the Burkholderia genus.

Transposon mutagenesis.

Transposases catalyze insertional mutagenesis by integrating exogenous sequences (transposon) randomly in the genome. To afford the transposition event, a suicide vector is introduced into the host. The vector typically carries a resistance marker, a transposon flanked by specific inverted sequences and a conditional origin of replication. The transposase recognizes and forms a complex with the inverted sequences resulting in the excision of the transposon. The host DNA is also excised by the transposase which is followed by integration of the transposon, gap filling, and gap repairing by the host’s DNA polymerase and ligase, respectively [41]. Transposon mutagenesis kits are commercially available. For example, EZ-Tn5 transposome kits (Lucigen) were used in some of the studies highlighted below.

Untargeted or random genomic manipulation as afforded by transposon mutagenesis is useful in the biosynthetic profiling of Burkholderia genomes particularly when NP biosynthesis cannot be easily predicted. However, this approach results in large libraries (>100 mutants) and a need for a convenient read-out method for screening of obtained clones. To select mutants for further investigations, various techniques have been employed.

For instance, loss of activity in transposon mutants help link biosynthetic genes to the corresponding NP. Klaus et al. observed a loss of antibacterial activity to be caused by transposon insertion in the hmqA BGC that was shown to encode for 4-hydroxy-3-methyl-2-alkenylquinolines (HMAQ) in B. thailandensis [42]. Similarly, following a loss in antibacterial activity, the biosynthesis of the antibacterial ditropolonyl sulfide was linked to phenylacetic acid catabolism and glutathione metabolism in B. cepacia strain R-12632 by Depoorter et al. [43].

Transposon mutagenesis has also been used to activate silent BGCs. For example, thailandene A – C were discovered from B. thailandensis by Park et al. following the detection of orange pigmented mutants from a transposon library [44]. Four of the five orange mutants identified had transposon insertions upstream of the silent thailandene BGC (termed org for orange) that led to its activation due to a promoter read through from the gentamicin resistance gene [44]. The remaining orange mutant had a transposon insertion in an operon that included a putative σ54-dependent transcriptional regulator, activation of which was hypothesized to indirectly lead to org BGC activation.

So as not to be restricted to NPs that absorb in the visible range and can be visualized with the naked eye, a reporter-guided transposon mutagenesis approach was developed by Mao et al. in 2020 to screen for mutations that induce the expression of any target BGC [45]. In this approach, the gene encoding Green Fluorescent Protein (GFP) was first inserted in the target BGC as a reporter to provide a readout that can be assayed in high throughput. A transposon library was generated using the GFP strain as parent. Mutants in which GFP is expressed were then studied to understand the mechanism of activation. The authors applied the approach to B. thailandensis, activating the BGCs mal, cap and tomm that encoded malleilactone, capistruin and an uncharacterized RiPP, respectively [45].

To circumvent the need to do two rounds of genetic engineering with the reporter approach above, a follow up study by Yoshimura et al. utilized untargeted metabolomics to screen transposon mutants. The authors developed a HPLC-MS-coupled self-organizing map analytics platform combined with rapid imaging mass spectrometry to identify mutants with activated BGCs leading to the discovery of haereoplantin A–E and burrioplantin A from B. plantarii, and gladiobactin A from B. gladioli [46].

Another example of improved NP production was provided by Martinez et al. who screened for an increase in rhamnolipid surfactant activity using an oil spray method. An observed increase in production of rhamnolipid surfactants revealed the suppressive effects of scmR, a LysR-type transcriptional regulator, on the transcription of the rhl operon [47].

Burkholderia heterologous hosts

Heterologous host strains provide the capability to streamline the exploration, discovery, and production of NPs [48]. We have identified only a few reported examples of Burkholderia strains used as heterologous hosts. Herein we summarize the reported Burkholderia hosts in chronological order, the modifications made to improve the host (if any), the NPs they were used to produce, and the NP yields that were obtained.

Burkholderia graminis C4D1M.

The first reported use of a Burkholderia host we identified was in 2010, when Craig et al. screened environmental DNA libraries using six different proteobacterial hosts, Burkholderia graminis C4D1M being one of them [49]. However, no small molecules of interest were produced by the B. graminis host strain. Nevertheless, this study provided evidence that the use of a diversity of bacterial hosts increases the chance of detecting NPs from environmental DNA clones.

Burkholderia gladioli pv agaricicola HKI0676.

In 2019, Bratovanov et al. reported the heterologous production of lasso peptide burhizin-23 from the endosymbiont Mycetohabitans (basonym Burkholderia) rhizoxinica in Burkholderia gladioli pv agaricicola HKI0676 [50]. The burhizin BGC had been previously expressed in E. coli by Hegemann et al. However, expression in E. coli had led to truncated products [51]. B. gladioli pv agaricicola HKI0676 was selected by Bratovanov et al. for use as a heterologous host given the closer phylogenetic relationship to the source DNA. Expression of the burhizin BGC in B. gladioli was accomplished by using an l-arabinose inducible araC/P_BAD promoter. Production in M20 defined medium containing l-arabinose resulted in the full-length product burhizin-23 in an isolated yield of 1 mg/L.

Burkholderia sp. FERM BP-3421.

In 2020, Kunakom & Eustáquio reported Burkholderia sp. FERM BP-3421 as a heterologous host to produce lasso peptide capistruin from Burkholderia thailandensis [52]. As it will be discussed later, FERM BP-3421 had been previously used to produce autologous spliceostatins at industrial, gram-per-liter scale [53]. For capistruin heterologous expression, both the wild-type FERM BP-3421 strain as well as a spliceostatin polyketide synthase mutant (fr9DEF⁻) were tested using the l-arabinose-inducible araC/P_BAD promoter. The FERM BP-3421 wild-type strain produced capistruin at an average titer of 1 mg/L in M20 medium and 3.2 mg/L in the complex 2S4G medium described below. Use of the fr9DEF⁻ strain led to 13 mg/L capistruin on average in the 2S4G medium, a four-fold improvement compared to the wild type. Finally, an overproducer clone was identified from the transformation of the capistruin expression vector into the in FERM BP-3421 wild-type strain. This outlier clone was capable of consistently producing capistruin at 116 mg/L on average, which is 580-times the production levels previously reported with E. coli as host (0.2 mg/L yield from M20 medium) [54].

Bioprocess development

The transition of drug discovery to industrial scale production is a crucial step for facilitating downstream processes and preclinical development of NPs of interest. Here, we highlight the development and application of Burkholderia bacteria in bioreactor fermentation processes (Table 3).

Table 3.

Application of Burkholderia bacteria in bioreactor fermentation processes.

Burkholderia	Natural Product	Class	Yield (mg/L)	Bioreactor scale (L)	Fermentation medium	Ref.
B. sp. FERM BP- 3421 fr9P− / fr9R++	Thailanstatin A	PK-NRP	2,400	10	2S4G with 100 mM L-ara	[53]
B. thailandensis MSMB43 tstP	Thailanstatin A	PK-NRP	714	90	2S4G	[59]
B. thailandensis MSMB43 tdpR+	Romidepsin	PK-NRP	168	20	M8	[56]
B. rinojensis A396	Romidepsin	PK-NRP	400	5*	E-LB-Sabouraud	[60]

PK-NRP, polyketide-nonribosomal peptide; L-ara, L-arabinose

Low titers are a bottleneck in NP discovery and development. Therefore, coupling genetic engineering with media optimization can fast-track the process (Figure 1). The optimal media component should be cost-effective and sustain high titer production of the NP. By systematically evaluating sources for carbon and nitrogen, sustainable NP production can be achieved in high yields.

To give a few examples, a medium containing soy peptone and glycerol (2S4G medium) as nitrogen and carbon sources, respectively, was found to be optimal to produce spliceostatin congeners in Burkholderia sp. FERM BP-3421 [53]. In another study, mannitol and urea were observed to be the preferred sources of carbon and nitrogen for optimal production of rhamnolipids in B. glumae BGR1 [55]. Likewise, the M8 medium containing glucose and peptone was selected for the production of FK228 in B. thailandensis MSMB43 [56].

Considering that target NPs usually direct media optimization approaches, the integration of biosynthetic engineering extends the utility of media optimization studies. Take for instance spliceostatin/thailanstatin/FR901464 produced by some Burkholderia spp. (note different names for the same compound class). To streamline the metabolite profile of Burkholderia sp. FERM BP-3421 and afford only the desired thailanstatin A congener, a two-step engineering strategy was implemented by Eustáquio et al. [53]. The deletion of an iron/α-ketoglutarate-dependent dioxygenase gene, fr9P, using host-based homologous recombination blocked the production of the hemiketal FR901464 from the thailanstatin A precursor. However, the fr9P⁻ mutants generated the 4-deoxy congener of thailanstatin A (spliceostatin C / thailanstatin D) as the main metabolite. To relieve the hydroxylation bottleneck, an extra copy of the cytochrome P450 gene fr9R under an l-arabinose inducible promoter was introduced, yielding thailanstatin A as a nearly single component. Combined with media optimization studies, titers of thailanstatin A reached 2.4 g/L in a 10-L bioreactor [53], facilitating preclinical testing of thailanstatin A as antibody drug conjugates [57,58].

In a subsequent study by Liu et al., the knockout of tstP, which encodes a Fr9P dioxygenase homolog in B. thailandensis MSMB43, resulted in a 58% increase of thailanstatin A in the 2S4G medium [59]. A 90-L fed-batch fermentation of the ΔtstP mutant produced 714 mg of thailanstatin A with > 98.5% purity [59].

Another NP produced by Burkholderia that has clinical relevance is the FDA-approved anticancer agent romidepsin (Istodax^®) originally discovered from Chromobacterium violaceum No. 968, which is the current commercial source of the drug. In a study by Liu et al., the heterologous expression of the thailandepsin pathway regulator tdpR in B. thailandensis MSMB43, increased yields of romidepsin from 115.9 mg/L to 168.5 mg/L in a 20-L fermentor, highlighting B. thailandensis MSMB43 as a potential alternative source of romidepsin [56]. Marrone Bio inventors reported another method for increasing romidepsin production by adding copper sulfate to fermentation media reaching about 400 mg/L using both Chromobacterium spp. and Burkholderia rinojensis A396 [60].

Taken together, genetic engineering of native producers and media optimizations have enabled the application of Burkholderia strains to industrial-scale production of NPs.

Conclusions

Burkholderia bacteria are a promising source of structurally diverse NPs [3]. Genetic engineering of native producers has been widely applied to study NP biosynthesis, to connect NPs to BGCs, and to activate silent BGCs. Approaches for targeted mutagenesis vary from host-based [23,29] and phage-based homologous recombination [29,32,33] to CRISPR-Cas9 engineering [36]. Screening methods applied for untargeted transposon mutagenesis vary from loss/increase in an NP activity that can be observed directly from colonies such as a loss in antibacterial activity [42,43] or an increase in surfactant activity [47], respectively, to more elaborate approaches using reporter genes [45] or mass spectrometry [46]. All examples highlighted involved nonribosomal peptides or polyketides.

In contrast, only three Burkholderia strains have been tested as heterologous hosts, two of which successfully. Both B. gladioli pv agaricola HKI676 [50] and Burkholderia sp. FERM BP-3421 [52] were tested with RiPP lasso peptide BGCs leading to yields of 1 to 116 mg/L respectively. The 116 mg/L capistruin yields with FERM BP-3421 were obtained with a serendipitously isolated outlier clone, the molecular basis of which is currently being investigated.

Two Burkholderia NPs have advanced to bioreactor scale fermentation, e.g., PK-NRPs thailanstatin A, an antitumor agent that entered preclinical development, and romidepsin, an approved anticancer drug. To reach industrial scale yields, genetic engineering of native producers and/or media optimization were performed. The best reported yields of thailanstatin A were 2.4 g/L by Burkholderia sp. FERM BP-3421 with deleted dioxygenase and overexpressed cytochrome P450 genes [53]. The best reported yields of romidepsin were 400 mg/L with wild-type B. rinojensis A396 [60].

In sum, at least two Burkholderia NPs have reached industrial-scale bioprocess development. The varied biotechnology approaches recently applied to native producers and the emerging Burkholderia heterologous hosts promise to facilitate further development of Burkholderia natural products.

Table 2.

Natural products produced with Burkholderia heterologous hosts, and the yields obtained.

Burkholderia Host	Natural Product	Class	Yield (mg/L)	Ref.
B. graminis C4D1M	N/A	Environmental DNA	N/A	[49]
B. gladioli pv agaricicola HKI0676	Burhizin-23	RiPP (lasso peptide)	1	[50]
B. sp. FERM BP-3421	Capistruin	RiPP (lasso peptide)	Up to 116	[52]

RiPP, ribosomally synthesized and post-translationally modified peptide.