Pseudomonas putida—a versatile host for the production of natural products

1For the biotechnological production of natural products, engineered bacteria generally offer several advantages over the original producers. As opposed to many natural producers, typically applied engineered bacteria are characterized by easy handling regarding laboratory cultivation which is the prerequisite for biotechnological applications. In addition, independence of natural regulation systems usually immanent in the original producer permits controlled biosynthesis and the construction of hyper-production strains. Furthermore, usage of non-harmful generally recognized as safe (GRAS)-certified strains such as Pseudomonas putida KT2440 allows studies in many laboratories as well as industrial-scale production. Applying well-established and genetically accessible laboratory work horses moreover allows for genetic manipulation of biosynthetic modules in order to direct biosynthesis to desired compounds.

2P. putida has emerged as one of the laboratory work horses matching the abovementioned advantages and offering specific features of particular interest beyond that. In recent decades, this Gram-negative soil bacterium has been virtually “domesticated” by means of synthetic biology, as excellently reviewed by Nikel et al.. Plenty of tools for genetic manipulation and gene expression are available including numerous inducible promoter systems (and corresponding inducer compounds) such as native Pm/XylS (m-toluate) (de Lorenzo et al. ), PsaI/NahR (salicylate) (de Lorenzo et al. ), and PalkB/AlkS (short-chain alkanes) (Panke et al. ) as well as the non-native systems Plac (IPTG) (Baumberg et al. ), Ptac (IPTG) (Bagdasarian et al. ), PT7 (IPTG via Plac) (Troeschel et al. ), PT7 (m-toluate via Pm/XylS) (Herrero et al. ), NagR/PnagAa (salicylate) (Hüsken et al. ), rhaPBAD (Rhamnose) (Jeske and Altenbuchner ), and Ptet (tetracycline) (Chai et al. ). The genomes of important strains such as P. putida KT2440 (Nelson et al. ) or S12 (Kuepper et al. ) are fully sequenced, providing the basis for understanding metabolic networks (Nelson et al.; Puchałka et al.; Wu et al. ) and sophisticated strain development approaches (Martínez-García et al. ). Due to its relatively high guanine-cytosine (GC) content (61.5 %), P. putida is suitable for heterologous expression of genes from GC-rich bacterial clades like actinobacteria or myxobacteria that are especially rich in secondary metabolite biosynthesis gene clusters. On the level of biosynthesis, P. putida offers a wealth of cofactors especially for oxidoreductases (Blank et al.; Tiso et al. ) and a versatile metabolism with diverse intrinsic enzymatic capacities for production purposes (Nelson et al. ), while at the same time, a rather “clean” background simplifies the detection of many heterologously synthesized metabolites (Martinez et al.; Stephan et al. ). Moreover, the bacterium exhibits a high tolerance towards xenobiotics including antibiotics and organic solvents. This extraordinary feature is the result of complex adaptations such as effective efflux systems which are typically activated in presence of xenobiotics (Fernández et al.; Simon et al. ), rendering it an ideal producer of such compounds and an especially suitable organism for production processes in two-phase system (Heipieper et al. ).

3This review covering the literature which has appeared until March 2015 aimed to summarize applications of P. putida as a host organism for recombinant natural product biosynthesis by means of heterologous gene expression and strain engineering. The focus is on bioconversion of basic nutrients for de novo synthesis of complex natural products from intrinsic building blocks (Fig. 1). Biotransformations, i.e., the turnover of preformed substrates into products by one or few recombinant biocatalysts (Berger; Tiso et al. ), using P. putida as a host are important applications as well which have been recently excellently reviewed elsewhere (Poblete-Castro et al.; Tiso et al. ) and are thus not covered by this review.

4Besides the utilization of P. putida for recombinant natural product biosynthesis, which is the main focus of this review, it is worth mentioning that strains of this organism naturally produce secondary metabolites of biotechnological interest. Particularly noteworthy are the polymers alginate (Chang et al.; Tiso et al. ) and medium-chain length polyhydroxyalkanoates (PHA) (Valentin et al. ). The latter attracted special attention due to their physical and material properties; they are thermoplastic and/or elastomeric, are insoluble in water, are enantiomerically pure, are non-toxic, exhibit a high degree of polymerization, and are therefore discussed as alternative biodegradable polymers for biotechnological industry (Steinbüchel and Lütke-Eversloh ). This polyester of hydroxy fatty acids is synthesized by PHA synthase, an α/β-hydrolase-like enzyme, from R-3-hydroxyacyl-CoA which is supplied by several metabolic pathways (Rehm et al.; Verlinden et al. ). The monomeric R-3-hydroxy fatty acids are in turn obtained from PHA hydrolysis to provide a chiral building block of pharmaceutical relevance (Lee et al.; Yuan et al.; O’Connor et al. ). Due to the material’s interesting properties, PHA synthesis was object of several studies concerning yield optimization and tailoring the polymer by metabolic engineering (Huijberts and Eggink; Steinbüchel and Lütke-Eversloh; Tiso et al.; Vo et al. ).

5Very recently, a rhizosphere soil bacterium identified as P. putida was shown to produce the antibiotic and antitumor phenazine derivate 5-methyl-phenazine-1-carboxylic acid (Kennedy et al. ).

6Furthermore, different P. putida strains are known to produce non-ribosomal peptides (Gross and Loper ), which were, however, not yet intensely studied in terms of biotechnological applications. Very common among pseudomonads is the synthesis of the fluorescent siderophore pyoverdine, which is discussed for different applications such as plant growth promotion (Glick; Saha et al. ). P. putida strains 267, PCL1445, and W15Oct28 or strain RW10S2 are moreover reported to release lipopeptide biosurfactants with antimicrobial properties, namely putisolvins and a viscosin-like peptide (Kuiper et al.; Kruijt et al.; Rokni-Zadeh et al.; Ye et al. ). In silico genome mining of additional secondary metabolite clusters for P. putida KT2440 with antiSMASH 3.0 (Weber et al. ) revealed a gene cluster comprising another non-ribosomal peptide synthetase gene (cluster location 4,798,235–4,851,188 nt, NRP gene PP4243) and two clusters for biosynthesis of bacteriocins (1,126,477–1,137,310 and 2,738,633–2,749,466 nt), members of PiPPs (ribosomally synthesized and post-translationally modified peptides). An analogous analysis of the P. putida S12 chromosome and megaplasmid pTTS12 likewise unveiled the presence of two clusters for bacteriocin biosynthesis (951,794–962,627 and 4,313,737–4,324,570 nt), furthermore a type I polyketide synthase (3,653,938–3,701,437 nt, polyketide synthase (PKS) gene RPPX16405), but no further non-ribosomal peptide synthesis-related genes besides the cluster for pyoverdine synthesis.

7In the following, different groups of natural products synthesized in recombinant P. putida will be discussed, namely rhamnolipids, terpenoids, polyketides and non-ribosomal peptides as well as hybrids thereof, and further amino acid-derived compounds (Fig. 1). All products together with relevant characteristics of the respective studies are summarized in Table 1; selected examples are shown in Fig. 2 to illustrate diversity of compounds and architecture of cognate biosynthetic gene clusters.

8Rhamnolipids are currently considered as the best studied representatives of bacterial biosurfactants (Müller et al. ), a class of metabolites with most various chemical compositions produced by different bacteria and fungi (Hausmann and Syldatk ). Due to excellent surfactant properties, low toxicity, high biodegradability, and antimicrobial effects, rhamnolipids are discussed for various applications, e.g., in cleaning agents, cosmetics, food industry, biocontrol, and soil remediation (Fracchia et al. ). Rhamnolipids consist of a hydrophobic domain with generally two molecules of hydroxy fatty acids forming 3-(hydroxyalkanoyloxy)alkanoic acid (HAA) and a hydrophilic part of one or two molecules of the sugar rhamnose, thus forming mono-rhamnolipids and di-rhamnolipids, respectively (Abdel-Mawgoud et al. ). At least two enzymes are required for the biosynthesis of rhamnolipids from the precursor metabolites dTDP-rhamnose and hydroxy fatty acid-ACP: (i) the acyltransferase RhlA for generation of HAA and (ii) the rhamnosyltransferase RhlB for glycosidic bond formation. For synthesis of di-rhamnolipids, a second molecule of activated rhamnose is added to mono-rhamnolipids by the rhamnosyltransferase RhlC (Abdel-Mawgoud et al. ).

9The opportunistic human pathogen Pseudomonas aeruginosa is the best-studied organism for rhamnolipid production. Although nowadays different non-pathogenic bacteria are described to be capable of rhamnolipid formation (Toribio et al. ), among them the β-proteobacterium Burkholderia glumae (Costa et al.; Voget et al. ), the marine Thermus sp. (Rezanka et al. ), and some strains from the P. putida group (Tuleva et al.; Martínez-Toledo et al. ), most studies concerning optimization of rhamnolipid production are performed with P. aeruginosa (Müller et al.; Müller et al. ). Recombinant expression of rhamnolipid biosynthesis pathways in closely related P. putida offers a promising alternative.

10Mono-rhamnolipid (Fig. 1) production in P. putida strains by expression of rhlAB from diverse strains of P. aeruginosa and the product’s extracellular accumulation was reported by several groups at gram scale. Different promoters were evaluated for expression of the ca. 2-kb operon, namely the synthetic hybrid promoter Ptac, (Ochsner et al.; Wittgens et al.; Setoodeh et al. ), a set of completely synthetic promoters in comparison to Ptac (Blank et al. ), or the native regulation system via coexpression of the cognate autoinducer-dependent transcription factor/autoinducer synthase pair RhlR/RhlI from P. aeruginosa (Cha et al.; Cao et al. ). Metabolic network analysis identified the PHA formation (see above) via PHA synthases PhaC1/C2 as a competitor for the precursor hydroxy fatty acid-ACP. Consequential deletion of phaC1 was proven to be beneficial for rhamnolipid product accumulation (Wittgens et al. ).

11Remarkably, a rhamnolipid-producing P. putida strain with genome-integrated rhlABRI could be applied for remediation of soil contaminated with PAH (polycyclic aromatic hydrocarbons), whereat interestingly P. putida itself was shown to be unable to degrade PAH, but the secreted biosurfactants vastly increased the degradation by native soil microorganisms (Cao et al. ).

12Mono-rhamnolipids can be easily converted to di-rhamnolipids (Fig. 2, 1), as described above, via the activity of the second rhamnosyltransferase RhlC to form more hydrophilic biosurfactants. Nonetheless, there are only two reports describing the heterologous expression of all three RL-synthesis genes in P. putida. Expression of rhlABC from P. aeruginosa yielded up to 113 mg/l/OD600 using PHA-deficient P. putida GPp104 (Schaffer et al. ). Notably, di-rhamnolipid concentration in the culture supernatant was increased when PA1131, a putative major facilitator superfamily transporter that is organized in one operon with rhlC, was coexpressed. Thus far, the mechanism of rhamnolipid secretion is described neither for heterologous production strains nor for the native producers, and it remains unclear if PA1131 contributes actively to the secretion of the products or if there are currently unknown regulatory effects. Blank et al. demonstrated that rhamnolipid properties may vary with the origin of the biosynthetic genes applied. They reported the production of both mono- and di-rhamnolipids expressing rhlAB(C) genes from B. glumae PG1 controlled by Ptac, yielding 80 mg/l of pure mono-rhamnolipids and 50 mg/l of a mixture, respectively. Compared to P. aeruginosa, B. glumae naturally produces more hydrophobic rhamnolipids with higher average fatty acid chain length. Remarkably, comparable fractions were produced upon expression of the respective genes in P. putida KT2440.

13The here reviewed studies from both academia and industry illustrate that P. putida has emerged as an excellent platform for recombinant rhamnolipid production and may thus replace the pathogenic P. aeruginosa in future rhamnolipid production processes. An additional advantage of P. putida is its resistance to very high concentrations of rhamnolipids showing little change in growth rate if exposed to concentrations up to 90 g/l (Wittgens et al. ), suggesting that even higher yields are feasible. Crucial factors for optimization of recombinant rhamnolipid production in P. putida include the choice of an appropriate promoter and the increase in biosynthetic flux toward the product, e.g., by deletion of the competing PHA pathway.

14Terpenoids constitute one of the most diverse groups of secondary metabolites in nature, and many of them find applications in areas such as human health and nutrition. They are synthesized from universal C5 isoprene precursors isopentenyl diphosphate (IPP) and its isomer dimethylallyl diphosphate (DMAPP). Terpene synthases cyclize poly-isoprene units or produce linear compounds thereof. Typically, the synthase reaction is followed by decoration of the molecule, carried out by terpene-modifying enzymes such as cytochrome P450 monooxygenases (Bouvier et al.; Cane and Ikeda ). Much effort has been devoted toward their production in microbial hosts (Kirby and Keasling; Li and Pfeifer ), mostly employing Escherichia coli and yeast.

15The production of terpenes using P. putida started with biotransformation approaches, aiming to yield oxidation products of the plant monoterpene limonene which was supplemented to the process. These products are valuable for their flavoring, antibiotic, and anticancer properties (Schrader; Garcia et al. ). In one setup, a heterologously expressed P450 was applied for the hydroxylation of (S)-limonene to (S)-perillyl alcohol (Van Beilen et al.; Cornelissen et al. ). Other studies exploited the bacterium’s intrinsic enzymes that naturally degrade p-cymene via p-cumate. This three-step degradation involves enzymes also active towards limonene and its respective oxidation products perillyl alcohol and perillaldehyde, thereby converting limonene to the monoterpenoid perillic acid (Speelmans et al.; Mars et al.; Mirata et al. ).

16Full de novo biosynthesis of a monoterpenoid was demonstrated recently (Mi et al. ). In this study, P. putida strain DSM12264 was used for the production of geranic acid (Fig. 1) which has received attention for potential applications in fragrance and flavor industries (Schrader ). It was achieved by expression of geraniol synthase from Ocimum basilicum (basil) whose product geraniol was converted to geranic acid by an intrinsic enzyme. The required C10 substrate for the synthase is geranyl pyrophosphate (GPP). It is provided by P. putida via the MEP (2-C-methylerythritol-4-phosphate) isoprenoid biosynthesis pathway which starts using pyruvate and glyceraldehyde-3-phosphate from central metabolism and is the common pathway in most bacteria. Yields were improved significantly by increasing the isoprene precursor pool by coexpression of the six genes of the MVA (mevalonate) isoprenoid biosynthesis pathway from the myxobacterium Myxococcus xanthus DSM16526 (hmgs, hmgr, mvk, pmvk, mvd, and idi). This way, a second route from central metabolism (acetyl-CoA) to isoprenoid building blocks was installed. All genes were assembled into a synthetic operon (of about 8.5 kb) under control of the rhaPBAD promoter from E. coli in one plasmid. In a lab-scale bioreactor setup, 193 mg/l of geranic acid could be produced. Notably, this is the first report showing feasibility of microbial production of this compound in general. The product was for the most part found in the supernatant, whereby downstream processing is facilitated. Moreover, in associated toxicity tests, the authors could show that P. putida exhibited tolerance to significantly higher concentrations of the product than E. coli and yeast. This underlines the potential of P. putida as production host for monoterpenoids.

17Besides these examples for biosynthesis of monoterpenoids, recombinant production of tetraterpenes was also established using P. putida. The production of the yellow-colored xanthophyll carotenoid zeaxanthin (Fig. 2, 2), which is relevant as food and feed additive and offers potential in pharmaceutical applications (Baiao et al.; Nishino et al.; Abdel-Aal et al. ), was reported by Beuttler et al.. Expression of five carotenoid biosynthesis genes (crtEBIYZ) from the enterobacterium Pantoea ananatis DSM30080 established the conversion of the C15 isoprenoid precursor farnesyl pyrophosphate (FPP) to zeaxanthin. In addition, three genes from the E. coli MEP pathway (idi, ispA, and dxs) were coexpressed in order to enhance metabolic flux from C5 isoprenoid building blocks via GPP to FPP. All genes were assembled in an expression vector as a synthetic operon of about 8.5 kb under control of the rhaPBAD promoter and expressed in P. putida KT2440. After optimization of cultivation conditions and media additives, 239 mg/l of zeaxanthin could be produced. The same crt genes were used in another study for heterologous zeaxanthin biosynthesis in P. putida KT2440 (Loeschcke et al. ). Here, the applied transfer and expression system named TREX enabled the direct expression of the bidirectional natural gene cluster by convergent transcription from two T7 promoters, thereby circumventing the need for genetic reengineering. Chromosomal integration of different deletion versions of the cluster and subsequent T7 RNA polymerase-dependent expression led to production of zeaxanthin and β-carotene. Notably, this setup without engineered enhancement of the precursor pool only generated yields at low milligram scale.

18The suitability of P. putida as terpenoid producer is hard to evaluate based on the low number of available studies. However, substantial yields of geranic acid and zeaxanthin suggest a promising potential. Engineering of the supply of isoprene precursors via the MEP or MVA pathway appears to be the crucial factor for success here.

19Polyketides and non-ribosomal peptides include a large and extremely diverse group of natural compounds with various highly valuable bioactivities such as antibiosis and cytotoxicity. They share features in biosynthesis and often cooccur in hybrid assembly systems (Wang et al. ). Briefly, the biosynthetic proteins of both assembly machineries catalyze the condensation of simple building blocks, i.e., carboxylic or amino acids, to produce polymer chains that can be cyclized and decorated to form numerous natural products. Three different types of polyketide synthases (PKSs) produce carbonyl polymers by condensing activated acyls (typically acetyl-CoA and malonyl-CoA) (Shen; Cummings et al. ): Type I PKSs are large, highly modular proteins containing domains which catalyze biosynthesis steps, whereas type II PKSs are complexes of multiple individual proteins with dedicated functions. In both types, the elongating polymer is handed from one polymer extending module to the next, where it is linked to the protein by thioester bonds. Type I and II systems share features and nomenclature with fatty acid synthases. Type III PKSs, also referred to as chalcone synthase-like PKSs, are homodimeric enzymes that catalyze condensation and cyclization reactions producing phenolic products. Non-ribosomal peptide synthases (NRPSs) produce peptidyl polymers by adenylation and subsequent condensation of amino acids. Like in polyketide synthesis, the elongating polymer is bound to the enzyme by a thioester bond. Similar to PKSs, there are both modular multidomain NRPS enzymes and NRPS enzyme complexes (Finking and Marahiel; Strieker et al. ).

20Heterologous production of polyketides and non-ribosomal peptides has been demonstrated employing a diverse set of hosts, including fungi, Gram-positive bacteria, and Gram-negative bacteria such as M. xanthus, E. coli, and P. putida (Fujii; Zhang et al.; Ongley et al. ). For heterologous expression of the described systems, it is important to consider that with exception of type III PKSs, PKSs and NRPSs require their acyl carrier protein (ACP) and peptidyl carrier protein (PCP) domains, respectively, to be posttranslationally modified by a phosphopantetheinyl transferase (PPTase) in order to function. The genes for cognate PPTases are often not part of the PKS/NRPS gene cluster. Notably, the predominantly used P. putida strain KT2440 provides a broad substrate range PPTase which is able to activate both ACP and PCP domains (Gross et al.; Owen et al. ) and is often more suitable than, e.g., E.coli PPTase, which circumvents the constraints of additional introduction of foreign PPTase genes.

21The first polyketide heterologously produced in P. putida was 2,4-DAPG (2,4-diacetylphloroglucinol) (Fig. 2, 3) which has received interest for its activity against plant pathogens (Bakker et al.; Haas and Défago ) and methicillin-resistant Staphylococcus aureus (Kamei and Isnansetyo ). A 6.5-kb DNA fragment from Pseudomonas fluorescens ATCC 49323 containing the respective gene cluster was inserted into the genome of P. putida KT2440 (Martinez et al. ). It encompasses five unidirectionally organized genes (phlACBDE), where PhlD is a type III polyketide synthase catalyzing the synthesis of monoacetylphloroglucinol from three molecules of malonyl-CoA, which is in turn converted to 2,4-DAPG by the action of the other phl-encoded enzymes (Bangera and Thomashow; Achkar et al. ). The expression of the genome-integrated phl cluster in P. putida was presumably driven by its native promoters or by chromosomal promoters adjacent to the insertion site. The accumulation of 2,4-DAPG could be detected but was not quantified.

22Another type III PKS product, the UV protective pigment (Zeng et al. ) flaviolin (2,5,7-trihydroxy-1,4-naphthoquinone), could be synthesized in P. putida KT2440 (Gross et al. ). Here, the 1.1-kb rppA gene from myxobacterium Sorangium cellulosum So ce56 was expressed from a plasmid using the m-toluate-inducible Pm promoter from P. putida. RppA, a 1,3,6,8-tetrahydroxynaphthalene synthase (THNS), utilizes five malonyl-CoA to synthesize THN, which is converted by autooxidation to flaviolin. About 6 mg of the red colored product could be purified from 1 l cell-free supernatant of culture broth.

23The earliest study showing the heterologous biosynthesis of a non-ribosomal peptide in P. putida reports on the production of the β-lactam antibiotic deacetyl-cephalosporin C (DAC). P. putida IFO14164 was used for expression of the clustered genes pcbABCcefEFD from the actinomycete Lysobacter lactamgenus YK90 (Kimura et al. ). The 16-kb gene region comprising five unidirectionally oriented genes was expressed from a plasmid under control of Plac from E. coli. The pcbAB gene encodes one protein comprising three NRPS modules, which enabled the synthesis of LLD-ACV (δ-(l-α-aminoadipyl)-l-cysteinyl-d-valine). The non-canonic amino acid l-α-aminoadipic acid is derived from lysine and provided by P. putida. By the action of the other cluster-encoded enzymes, LLD-ACV is converted to the antibiotic DAC via penicillin N. DAC is naturally further differentiated to various β-lactam products (Demirev et al.; Hamed et al. ). The L. lactamgenus specific end product cephabacin was not produced, since the hitherto required enzymes were excluded in the study. The study did not focus on production yield; however, about 2 mg of DAC could be extracted from cell material obtained from 1-l broth.

24Recently, the biosynthesis of serrawettin W1 (Fig. 2, 4), another NRPS-derived compound, could be established in P. putida. This cyclic lipopeptide has gathered attention for its biosurfactant properties as well as antimicrobial and antitumor activities (Matsuyama et al.; Kadouri and Shanks ). In order to produce the compound, the 4-kb swrW gene from the enterobacterium Serratia marcescens DSM12481 was cloned in a vector under control of Ptac (Thies et al. ). The SwrW protein comprises the NRPS module required to produce the symmetrical serrawettin W1 molecule consisting of two serine residues attached to two β-hydroxy fatty acids via ester and amide bonds. Expression of swrW in P. putida KT2440 enabled serrawettin W1 recovery from culture supernatant at milligram scale (S Thies, unpublished).

25The first product synthesized by a heterologously expressed type I PKS/NRPS hybrid system in P. putida was myxochromide S (Fig. 2, 5) (Wenzel et al. ). To this end, the 30-kb mchABC cluster from Stigmatella aurantiaca DW4/3-1 was introduced into P. putida KT2440. The PKS module which utilizes acetyl-CoA, malonyl-CoA, and propionyl-CoA is encoded in mchA, while mchB encodes a two-module NRPS and mchC a four-module NRPS (one of those being skipped in biosynthesis), together forming the peptide part from alanine, glutamine, threonine, and leucine. The unidirectional gene cluster was inserted into the host chromosome at the anthranilate synthase gene trpE by homologous recombination. Expression of mch genes and thus accumulation of the yellow-orange colored myxochromide S in P. putida cells was implemented using the m-toluate-inducible Pm promoter. Notably, lowering of expression temperature from 30 to 16 °C after induction resulted in a 1000-fold increase of production to 40 mg/l. Supply of malonyl-CoA was identified as yield-limiting bottleneck that might be addressed in future studies to increase myxochromide production (Stephan et al. ).

26Another myxobacterial type I PKS/NRPS hybrid system from S. aurantiaca DW4-3/1 expressed in P. putida resulted in the production of myxothiazol A (Fig. 2, 6) (Perlova et al.; Gross et al. ), an inhibitor of the respiratory chain with antifungal and insecticidal activities (Clough ). The 60-kb mta gene cluster (mtaBCDEFG) was inserted into the chromosome of P. putida by the same method as the abovementioned mch cluster in gene trpE. While MtaB, MtaE, and MtaF are PKS parts and MtaC and MtaG are NRPS modules, the gene mtaD encodes a hybrid protein containing both PKS and NRPS modules. The PKS parts employ isovaleryl-CoA, which is provided by P. putida from the degradation of leucine, as the starter unit, and acetyl-CoA as well as methylmalonyl-CoA (mm-CoA) as extender units. The NRPS modules incorporate the amino acids cysteine and glycine (Perlova et al. ). Since P. putida does not intrinsically synthesize mm-CoA, the bacterium was engineered to provide the necessary precursor from its succinyl-CoA pool. This was achieved by expression of an operon from S. cellulosum So ce56 encoding mm-CoA epimerase (epi), mm-CoA mutase (mcm), and an MCM complex protecting protein (meaB) that was integrated in the chromosome. Expression of the mm-CoA operon was executed either by coexpression with the host genes at the integration site and/or driven by the promoter of a neomycin resistance gene which was coinserted. Expression of the myxothiazol cluster using the Pm promoter resulted in minor production levels. Product formation was increased to 0.6 mg/l by feeding of leucine and vitamin B12. Both the vitamin, required as cofactor for mm-CoA mutase, and the precursor leucine are intrinsically synthesized by P. putida but nevertheless appear to be limiting factors in product formation.

27Via expression of the tubulysin gene cluster from the myxobacterium Cystobacter sp. SBCb004, a further type I PKS/NRPS hybrid system, was successfully installed in P. putida (Chai et al. ). The applied ~40 kb gene cluster comprised 11 genes, namely the unidirectional core assembly line part tubBCDEF as well as tubA, tubZ, orf2, orf1, orf17, and orf18 organized around it (Sandmann et al. ). While TubB, TubC, and TubE are NRPS parts and TubF is a PKS part, TubD is a PKS/NRPS hybrid protein. First, TubZ produces pipecolic acid from lysine, which is then methylated and used as starter unit N-methyl-pipecolic acid by TubB. By extension with isoleucine, valine, cysteine, and tyrosine or phenylalanine as well as two molecules of acetyl-CoA, the core pretubulysin is formed, which is naturally further decorated by oxidation and acylation reactions. Similar to tubulysin, pretubulysin exhibits also highly valuable activities inducing apoptosis and inhibiting cancer cell migration and tubulin assembly in vitro and in vivo (Herrmann et al.; Braig et al. ). The tub cluster was inserted into the chromosome of P. putida using the MycoMar transposon and was expressed relying on the native promoter(s) and by additional insertion of the Ptet promoter from transposon Tn10 in front of tubCDEF. This resulted in the production of the two variants pretubulysin A (Fig. 1) and tyrosine pretubulysin A (a product of module skipping). Yields could be pushed eightfold to 1.76 μg/l by supplementing pipecolic acid. Interestingly, in contrast to observations made with myxochromide S, yields of pretubulysin A were doubled at 30 °C compared to 16 °C cultivation temperature.

28Two members of the syrbactin group that are synthesized by type I PKS/NRPS hybrid systems (Dudler ) were produced in P. putida P3. The compounds have received particular attention for their ability to inhibit proteases and induce apoptosis in different cancer cell lines such as neuroblastoma (Archer et al. ). Syringolin A (Fig. 2, 7) was produced by expression of the sylABCDE cluster from Pseudomonas syringae pv. syringae (Ramel et al. ). The biosynthesis core is constituted by sylC encoding an NRPS module and sylD encoding two NRPS modules and one PKS module. Syringolin A is synthesized from valine which is joined to a second molecule valine by an unusual ureido group, 3,4-dehydrolysine, another valine and malonyl-CoA. SylB is thought to act as desaturase converting lysine to 3,4-dehydrolysine, while sylA and sylE encode a putative transcription activator and the exporter of syringolin. The 22-kb gene region was expressed from a cosmid relying on the original promoters. Produced syringolin A was recovered from the medium, corroborating functionality of the syringolin exporter encoded in the cluster. Glidobactin A was produced by expression of the homologous plu1881–1877 gene cluster from Photorhabdus luminescens subsp. laumondii TT01 (Dudnik et al. ). The biosynthesis core in the 18.3-kb cluster consists of NRPS encoding plu1878 (sylC homolog) and NRPS/PKS encoding plu1880 (sylD homolog). In glidobactins, the ureido-valyl moiety typical for syringolin is replaced by a fatty acid tail attached to the starter amino acid threonine. Glidobactin A biosynthesis proceeds by incorporation of lysine, which is modified by 4′-hydroxylation, alanine, and malonyl-CoA. Remarkably, heterologous production in P. putida was successful in unagitated cultures, grown for 5 days. Determination of product yields was not in focus of the study.

29Prodigiosin (Fig. 2, 8) represents a metabolite derived from a biosynthesis pathway very different from the typical NRPS/PKS assembly lines but still related via some shared motives in the involved enzymes. The red colored tripyrrolic pigment with highly valuable bioactivities such as antibiosis and cytotoxicity (Hassankhani et al.; Lapenda et al. ) is synthesized from 2-octenal, pyruvate, proline, malonyl-CoA, and serine in a complex bifurcated pathway, whose two products are condensed in a final reaction to form prodigiosin (Williamson et al. ). In S. marcescens W838, the pathway is encoded in the 22-kb pigABCDEFGHIJKLMN cluster. PigA, PigG, PigH, PigI, and PigJ act together as an NRPS/PKS assembly line producing a chain from proline, malonyl-CoA, and serine (Garneau-Tsodikova et al. ). Application of the TREX expression system enabled identification of P. putida KT2440 as a promising producer by chromosomal integration and PT7-dependent expression of the pig gene cluster, yielding 0.5 mg/g cell dry weight (CDW) of prodigiosin (Loeschcke et al. ). Based on that, the group evaluated a different expression setup, pushing yields by 2 orders of magnitude (Domröse et al., unpublished) and proving the capability of P. putida for high-level production of the interesting metabolite. Here, the expression strategy appeared to be crucial for increasing yields.

30The general capability of P. putida to synthesize PKS/NRPS products is undisputed as documented by the here provided list of successful examples. As mentioned above, the bacterium’s features such as its PPTase and xenobiotic tolerance render it especially suitable for the production of this group of compounds. Thus, use of P. putida for expression of PKS/NRPS systems can be considered as a valuable option for the identification of high-value compounds and the elucidation of their biosynthetic pathways. However, yields are often low and compounds are too diverse to recommend a general strategy for improvement. Considering the large gene clusters encoding biosynthesis, the expression strategy is surely a dominant bottleneck. On the level of biosynthesis, it is worth underlining that in some cases, cultivation conditions significantly influenced production, apparently in a highly product-specific manner. Optimal conditions thus have to be evaluated experimentally. Since several studies report increased yields if precursors are supplemented, engineering approaches to enhance precursor supply may in the future improve production.

31The extraordinary tolerance of P. putida S12 towards organic solvents (Weber et al.; Kuepper et al. ) enabled its application in the production of various aromatic compounds valuable as building blocks for bioactive small molecules, resins, and polymers (Meijnen et al. ). Here, shikimate pathway-derived amino acids are utilized as precursors (Gosset ).

32In contrast to many of the compounds described in this review, introduction of only one or two enzymes instead of whole operons or gene clusters into P. putida is sufficient for the production of aromatics. Accumulation of diverse products was implemented by additionally targeted control of the host’s intrinsic metabolic inventory by deletion of genes at different stages of aromatics catabolism. This way, several aromatics producers were generated and developed from one another. An excellent overview on the genealogy of the strains applied for production of aromatics is given by Tiso et al.. The relevant studies are summarized briefly in the following.

33First, t-cinnamate and phenol producer strains were generated by introduction of the phenylalanine/tyrosine ammonia lyase gene pal from Rhodosporidium toruloides ATCC 64815 and the tyrosine phenol lyase gene tpl from Pantoea agglomerans AJ2985, respectively (Nijkamp et al; Wierckx et al ). Heterologous expression of these key biosynthetic enzymes was controlled by the salicylate-inducible promoter system NagR/PnagAa from Comomonas testosteroni (tpl) or Ptac (pal) and resulted after further strain improvement by chemical mutagenesis in production strains that accumulated extracellularly 5.4 mM (0.8 g/l) t-cinnamate and 9.2 mM (0.9 g/l) phenol, respectively.

34Based on the abovementioned two chassis strains, producers of further aromatic compounds were constructed. Nijkamp et al. enhanced metabolic flux towards tyrosine and introduced phenylalanine auxotrophy by random mutagenesis of the t-cinnamate producer strain. Additional deletion of the feruloyl-CoA synthetase gene fcs whose product catalyzes the first step of p-coumarate degradation led to an effective production strain accumulating p-coumarate (Fig. 1) instead of t-cinnamate in amounts of up to 1.7 g/l.

35A p-hydroxystyrene producer was constructed by additional introduction of the p-coumaric acid decarboxylase gene pdc from Lactobacillus plantarum DSM20174 into the fcs-deficient pal expressing chassis strain with yields up to 21 mM (2.5 g/l) with expression of pal and pdc under NagR/PnagAa control (Verhoef et al. ). Accumulation of p-hydroxybenzoate up to 1.8 g/l was achieved by Verhoef et al. via replacement of the tpl expression vector in the phenol producer strain by a pal expression vector and deletion of pobA in the host’s genome. The latter gene product catalyzes the first step in p-hydroxybenzoate degradation. As a result, Pal-synthesized p-coumarate was converted by P. putida S12 intrinsic metabolic pathways to p-hydroxybenzoate, which was not further degraded. Comparison of transcriptome data from the producer strain with the wild-type P. putida S12 identified the multidrug efflux MFS transporter PP1271–PP1273 to play an important, but not essential, role in p-hydroxybenzoate secretion. Furthermore, the 4-hydroxyphenylpyruvate dioxygenase encoding gene hpd as a target for further strain improvement. Deletion of hpd enhanced metabolic flux towards the product and led to an increase in yield from 0.25 to 0.31 g/l (2.3 mM) in the applied experimental setup by Verhoef et al.. Targeted engineering of precursor supply for aromatic molecules via the pentose phosphate pathway was achieved by introduction of the E. coli DH5α xylose degradation genes xylAB_FGH. The respective enzymes channeled xylose into the pathway which resulted in yields increased by further 22 % and moreover enabled production of these compounds from plant biomass (Meijnen et al., ).

36The listed examples and convincing yields show the potential for P. putida-based industrial production processes. Besides engineered enhancement of the precursor pool, random chemical mutagenesis, and subsequent application of feasible screening strategies proved to be excellent tools for simple generation and identification of strains with improved aromatics production (Tiso et al. ). In some of the mentioned studies, optimal yields of aromatics were obtained by application of two-phase systems which combine the culture medium with an organic solvent for immediate removal of toxic aromatic products, a technique for which P. putida S12 is especially suited due to its remarkable solvent tolerance (Heipieper et al., Verhoef et al. ).

37Aromatic amino acid-derived metabolites other than phenolic compounds that were produced in P. putida are N-acyl aromatic amino acids, a class of secondary metabolites with antimicrobial activity that was initially discovered by screening of metagenomic libraries for compounds capable of inhibiting Bacillus subtilis. Active expression of the biosynthetic machinery with the key enzymes from the N-acyl amino acid synthase family from a 29-kb environmental DNA fragment and secretion of the product was reported for P. putida KT2440 (Craig et al. ). In the same study, introduction of environmental DNA was sufficient to induce increased intrinsic production of porphyrine-derived pigments in clones of a metagenomic library in P. putida. Therefore, the authors conclude that P. putida has a high potential in metagenome-based studies concerning the detection of novel secondary metabolites. Considering that P. putida does not intrinsically produce antibiotics but exhibits extraordinary tolerance to producing such compounds, it may prove especially useful in functional screenings for novel antimicrobials. This potential is yet to be exploited. Further reports support the idea of P. putida as an alternative host for functional screening of metagenomics libraries (Troeschel et al.; Liebl et al. ).

38Likewise an aromatic amino acid derivative is the bacterial pigment deoxyviolacein (Fig. 2, 9). It is formed from two molecules of tryptophan and described to show dyeing and antimicrobial properties (Xing and Jiang; Jiang et al. ). This derivative of violacein was accumulated in P. putida mt-2 at levels of 1.5 g/l dependent on expression of vioABCE, an engineered variant of the violacein biosynthesis operon vioABCDE from Duganella sp. B2, controlled by the alkane-inducible alKB promoter from P. putida (Xing and Jiang ).

39In bacteria, the shikimate pathway provides precursors like chorismate not only for mentioned aromatic amino acids but also for different carbocyclic aromatic secondary metabolites like phenazines. These often antimicrobial redox mediators fulfill different functions, e.g., to support virulence and competitive fitness of the producing organisms. Due to their antibacterial and antifungal properties, phenazine derivatives are of interest for pharmaceutics and biocontrol (Pierson and Pierson; Jayaseelan et al. ). Glandorf et al. modified the plant growth promoting P. putida strain WCS358r to produce the yellow pigment PCA (phenazine1-carboxylic acid) by Ptac-controlled expression of the genome-integrated phzABCDEFG operon from P. fluorescens 2-79. Thereby, the host strain was turned into an improved fungal growth-inhibiting biocontrol strain (Glandorf et al.; Bakker et al.; Viebahn et al. ). Recently, production of PCA and its subsequent conversion to pyocyanin (Fig. 2, 10) was established in KT2440 by simultaneous expression of phzA1B1C1D1E1F1G1 and phzMS from P. aeruginosa PAO1 on two compatible plasmids (Schmitz et al. ), yielding 45 mg/l pyocyanin. Remarkably, accumulating the recombinant redox mediator in the culture medium enabled P. putida to sustain strongly oxygen-limited culture conditions by redox balancing via an anode as electron acceptor.

40These examples suggest P. putida to be a suitable host for production of aromatic pigments with antimicrobial activity in future studies.

41Besides these examples for utilization of aromatic amino acids or their precursor molecules, it was demonstrated that other amino acids provided by the bacterium’s metabolism can also serve as precursors for different natural products.

42Monoethanolamine(MEA) is an alkanolamine extensively used, e.g., to prevent corrosion, as a detergent, as a precursor for the production of ethylamines, or for CO2 capture (Aaron and Tsouris; Foti et al. ). Currently, most MEA is produced from petroleum feedstock. A sustainable alternative was offered by production of MEA in the solvent tolerant P. putida S12 strain via decarboxylation of serine (Foti et al. ). To this end, the authors introduced l-serine decarboxylase (sdc) genes from Arabidopsis thaliana (ecotype Columbia) and Vibrio carteri f. nagariensis. Efficient formation of MEA was only observed with a truncated version of A. thaliana Sdc. Strain improvement by deletion of genes for MEA breakdown (eutBC) and media optimization led to yields up to 0.2 g/l (2.6 mM in culture medium).

43Cyanophycin is a polymer consisting of arginine and aspartic acid, which was initially discovered in cyanobacteria (Frommeyer et al. ). Compounds derived thereof have gained interest for their applicability in technical processes and as pharmaceuticals (Mooibroek et al.; Steinbüchel and Sallam ). Consequently, several studies aimed at recombinant production to achieve improved biotechnological access to this polyamide (Frommeyer et al. ). P. putida KT2440 was first used for constitutive expression of the cyanophycin synthetase cphA from Synechocystis sp. strain PCC6308. In mineral salt medium supplemented with aspartic acid and arginine, the product could be accumulated to 11 % of CDW. However, the obtained polymer exhibited reduced length and polydispersity compared to the authentic material from cyanobacteria (Aboulmagd et al. ). Voss et al. expressed cphA from further cyanobacterial strains in KT2240 as well as GPp104, which resulted in the accumulation of cyanophycins to amounts between 6.8 and 24 % of CDW in mineral media. Highest yields were achieved by expressing the cyanophycin synthetase of Anabena sp. strain PCC72120. GPp104 being deficient for the production of the intrinsic polymer PHA (see above) accumulated in general slightly higher amounts than the wild type during expression of any CphA. Interestingly, substitution of arginine residues by lysine within the polymer up to 10 mol% was observed in case of cphA6308, if arginine was not supplemented to the mineral medium. Another modified variant of cyanophycin with citrulline partially substituting arginine was produced by choosing the citrulline accumulating strain P. putida ATCC 4359 as expression host (Wiefel et al. ). Optimization of culture conditions resulted in total amounts of citrulline-containing cyanophycin of 43.4 % of CDW. Partial substitution of arginine by other amino acids like lysine or citrulline was shown to lead to increased solubility of the naturally largely insoluble polymer (Frommeyer and Steinbüchel ).

44P. putida has been biotechnologically domesticated by concerted efforts of different fields of modern microbiology, especially over the last decade. The here given overview of natural products that could be synthesized by heterologous gene expression demonstrates the bacterium’s potential in this context. Synthetic biology further spurs these developments: Recent studies reporting a genome-edited P. putida strain provide a robust next-generation cell factory with enhanced features regarding genetic stability as well as energy state and availability of reduction equivalents for future studies (Martínez-García et al.; Lieder et al. ). Furthermore, liberation of P. putida from its obligate aerobic nature has been reported, providing the basis for bioreactor-based processes without aeration and allowing natural product formation by oxygen-sensitive biocatalysts (Nikel and de Lorenzo; Schmitz et al. ). Engineering the bacterium for efficient utilization of lignin-derived aromatics and xylose expanded the spectrum of applicable carbon sources and enables the utilization of plant biomass (Meijnen et al.; Meijnen et al.; Johnson and Beckham ). In such advanced cell factories (Fig. 3), novel cloning and expression technologies will allow implementation of various biosynthetic production pipelines, further expanding P. putida’s product portfolio (Zhang et al.; Fu et al., Gibson; Martínez-García and de Lorenzo; Fu et al. Martínez-García and de Lorenzo; Loeschcke et al.; Durante-Rodríguez et al.; Martínez-García et al. ). By in-depth analysis of the bacterium’s carbon metabolism as well as application of available “omics” tools, a systems biology perspective will provide the basis for knowledge-based metabolic engineering of P. putida in order to increase yields of natural products (Nogales et al.; Puchałka et al.; Wu et al.; Sudarsan et al.; Simon et al. ). This ever-accelerating development in methodology and knowledge gain paves the way for the research to come in the field of natural product biosynthesis with P. putida.

45Future opportunities for application of P. putida may include the production of valuable compounds, but also the use as delivery system for a variety of bioactive molecules (Fig. 3). In this context, the bacterium’s outer membrane vesicles are discussed for their potential as adjuvants or vaccine carriers due to their low pathological activity (Choi et al. ). Furthermore, based on release of intrinsically or recombinantly produced compounds, it was suggested to apply P. putida for plant growth promotion and protection of plants from disease as biocontrol agent (Bakker et al.; Glick ). Likewise, potential application of engineered P. putida in the context of soil remediation via delivery of suitable metabolites or enzymes was pointed out (de Lorenzo; Cao et al.; Chen et al. ).

46Such studies demonstrate the potential of P. putida in highly diverse fields of application and may inspire further exciting developments towards the establishment of P. putida as a platform for production of various natural products in the future.