Metabolic engineering of Pseudomonas putida for production of docosahexaenoic acid based on a myxobacterial PUFA synthase

Metabolic Engineering ∎ (∎∎∎∎) ∎∎∎–∎∎∎

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Q1 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 Q2 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66

Contents lists available at ScienceDirect

Metabolic Engineering journal homepage: www.elsevier.com/locate/ymben

Original Research Article

Metabolic engineering of Pseudomonas putida for production of docosahexaenoic acid based on a myxobacterial PUFA synthase Katja Gemperlein a, Gregor Zipf b, Hubert S. Bernauer b, Rolf Müller a, Silke C. Wenzel a,n a Department of Microbial Natural Products, Helmholtz Institute for Pharmaceutical Research Saarland, Helmholtz Centre for Infection Research and Department of Pharmaceutical Biotechnology, Saarland University, D-66123 Saarbrücken, Germany b ATG:biosynthetics GmbH, D-79249 Merzhausen, Germany

art ic l e i nf o

a b s t r a c t

Article history: Received 9 August 2015 Received in revised form 22 September 2015 Accepted 17 November 2015

Long-chain polyunsaturated fatty acids (LC-PUFAs) can be produced de novo via polyketide synthase-like enzymes known as PUFA synthases, which are encoded by pfa biosynthetic gene clusters originally discovered from marine microorganisms. Recently similar gene clusters were detected and characterized in terrestrial myxobacteria revealing several striking differences. As the identified myxobacterial producers are difficult to handle genetically and grow very slowly we aimed to establish heterologous expression platforms for myxobacterial PUFA synthases. Here we report the heterologous expression of the pfa gene cluster from Aetherobacter fasciculatus (SBSr002) in the phylogenetically distant model host bacteria Escherichia coli and Pseudomonas putida. The latter host turned out to be the more promising PUFA producer revealing higher production rates of n-6 docosapentaenoic acid (DPA) and docosahexaenoic acid (DHA). After several rounds of genetic engineering of expression plasmids combined with metabolic engineering of P. putida, DHA production yields were eventually increased more than threefold. Additionally, we applied synthetic biology approaches to redesign and construct artificial versions of the A. fasciculatus pfa gene cluster, which to the best of our knowledge represents the first example of a polyketide-like biosynthetic gene cluster modulated and synthesized for P. putida. Combination with the engineering efforts described above led to a further increase in LC-PUFA production yields. The established production platform based on synthetic DNA now sets the stage for flexible engineering of the complex PUFA synthase. & 2015 International Metabolic Engineering Society. Published by Elsevier Inc. All rights reserved.

Keywords: Polyunsaturated fatty acids Polyketide synthase PUFA synthase Heterologous expression Synthetic gene cluster Pathway engineering

1. Introduction Long-chain polyunsaturated fatty acids (LC-PUFAs), especially eicosapentaenoic acid (EPA, 20:5, n-3) and docosahexaenoic acid (DHA, 22:6, n-3), show advantageous effects on human health, like prevention and treatment of cardiovascular diseases, obesity, and diabetes (Lorente-Cebrian et al., 2013). As continuously more people strive to benefit from these positive effects, the demand of n-3 LC-PUFAs as dietary supplements has increased intensely over the past years-with upward tendency for the future. Nowadays, most of EPA and DHA are obtained from oceanic fish and fish oil, but these natural sources are depleting and often contaminated with environmental toxins. In order to satisfy the demand for high-quality LC-PUFAs, the quest for alternative, sustainable sources is indispensable (Lenihan-Geels et al., 2013). Fermentation of prokaryotic and eukaryotic microorganisms capable of n

Corresponding author. E-mail address: [email protected] (S.C. Wenzel).

producing EPA or DHA in high amounts might have the potential to permit industrial-scale production of LC-PUFAs (Ratledge, 2004). Until recently, it was thought that PUFA-producing microbes are exclusively of marine origin. Some of them synthesize LC-PUFAs de novo from acyl-CoA precursors by iteratively acting polyketide synthase (PKS)-like enzymes known as PUFA synthases. These multienzyme complexes are encoded by PUFA (pfa) biosynthetic gene clusters (Kaulmann and Hertweck, 2002; Napier, 2002; Wallis et al., 2002) (Fig. 1). Initially, the occurrence of pfa gene clusters was described in the marine γ-Proteobacteria Shewanella pneumatophori SCRC-2738 and Photobacterium profundum SS9 (EPA producers) (Allen and Bartlett, 2002; Metz et al., 2001), Moritella marina MP-1 (DHA producer) (Morita et al., 2000), and in the marine microalga Schizochytrium sp. ATCC 20888 (DHA and n-6 docosapentaenoic acid (DPA, 22:5) producer) (Hauvermale et al., 2006; Metz et al., 2001). Establishment of optimal fermentation conditions for these marine microorganisms (Chang et al., 2013; Ren et al., 2009, 2014), treatment with the fatty acid synthase inhibitor cerulenin (Fang et al., 2004; Morita et al., 2005),

http://dx.doi.org/10.1016/j.ymben.2015.11.001 1096-7176/& 2015 International Metabolic Engineering Society. Published by Elsevier Inc. All rights reserved.

Please cite this article as: Gemperlein, K., et al., Metabolic engineering of Pseudomonas putida for production of docosahexaenoic acid based on a myxobacterial PUFA synthase. Metab. Eng. (2015), http://dx.doi.org/10.1016/j.ymben.2015.11.001i

67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91

2

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66

K. Gemperlein et al. / Metabolic Engineering ∎ (∎∎∎∎) ∎∎∎–∎∎∎

Fig. 1. Anaerobic biosynthesis of polyunsaturated fatty acids (PUFAs) by iterative type I fatty acid synthase (FAS)/polyketide synthase (PKS)-like PUFA synthases encoded by a pfa gene cluster. The primer molecule (acetyl-CoA) is extended by several rounds of decarboxylative Claisen condensation reactions, resulting in the elongation of the fatty acyl chain by two carbons (derived from malonyl-CoA) per cycle. Following each round of elongation, the β-keto group is either fully reduced or only reduced to the trans double bond which is then isomerized. Finally, an acyl chain with methylene-interrupted cis double bonds is synthesized. AT ¼ acyltransferase, ACP¼ acyl carrier protein, KS¼ketosynthase, KR¼ ketoreductase, DH ¼dehydratase/isomerase, ER¼ enoylreductase.

or transposon mutagenesis (Amiri-Jami et al., 2006) gave rise to an improved production of LC-PUFAs under laboratory conditions. In order to correlate the biosynthetic pathways with their products, to reduce the cultivation time, and/or to study LC-PUFA biosynthesis, several marine PUFA biosynthetic gene clusters were transferred and expressed into suitable host organisms. Thereby, recombinant PUFA production could be accomplished with the pfa gene clusters from Shewanella sp. in Escherichia coli (3.370.7 mg EPA/g cell dry weight (CDW), Synechococcus sp. (0.6 70.3 mg EPA/ g CDW) (Takeyama et al., 1997), and in Lactococcus lactis ssp. cremoris (0.17 0.04 mg EPA/g CDW and 1.4 70.5 mg DHA/g CDW) (Amiri-Jami et al., 2014), from M. marina MP-1 in E. coli (5% DHA of total fatty acids) (Orikasa et al., 2006) or from Schizochytrium sp. in E. coli (10% n-6 DPA þDHA of total fatty acids) (Hauvermale et al., 2006). Besides intracellular accumulation of a high-performance catalase in E. coli expressing the pfa gene clusters from Shewanella sp., which resulted in up to 7.3 mg EPA/g CDW (Orikasa et al., 2007), recombinant EPA production could be enhanced by optimizing the cultivation conditions for the latter host harboring the pfa gene cluster from Shewanella to 18 mg/g CDW (Amiri-Jami and Griffiths, 2010). Recently, pfa gene clusters for de novo LC-PUFA biosynthesis were also identified and characterized from several terrestrial myxobacteria. Intriguingly, the myxobacterial pathways differ significantly from the marine systems in terms of gene organization, catalytic domain arrangement, and sequence identity of the encoded PUFA synthases. Two types of pfa gene clusters were deciphered from genomes of linoleic acid (LA, 18:2, n-6) producing Sorangium cellulosum species and of novel myxobacterial isolates described as Aetherobacter spp., the latter of which turned out to be prolific producers of EPA and DHA (Garcia et al., 2011; Gemperlein et al., 2014). However, these native producer strains grow very slowly, are not easy to handle, and genetic modifications are difficult to implement. Cloning, transfer and heterologous expression of the pfa genes in the myxobacterial model strain Myxococcus xanthus allowed to reduce cultivation times compared to the slow growing native producer, but production rates still have to be improved to yield a commercially viable production system (Gemperlein et al., 2014).

In the present work, we aimed to evaluate alternative host strains for the expression of myxobacterial PUFA synthases. Besides E. coli, already used as expression strain for marine pfa gene clusters, we focussed on Pseudomonas putida KT2440, which has several advantageous features including a higher GC content and a codon usage similar to the myxobacteria. Due to these reasons, this strain has already been used for heterologous expression of several complex myxobacterial natural product pathways directing the production of polyketide/nonribosomal peptide hybrid metabolites (Chai et al., 2012; Fu et al., 2008; Gross et al., 2006; Wenzel et al., 2005). As P. putida is generally recognized as safe (GRAS) for industrial production, highly robust under extreme environmental conditions, genetically well accessible, and grows rapidly, it represents an attractive host for recombinant LC-PUFA production. Here, we describe our work on establishing the heterologous expression of the native pfa gene cluster from A. fasciculatus (SBSr002) in both hosts, E. coli and P. putida. We also report on the successful application of metabolic engineering and synthetic biology approaches to further improve LC-PUFA production in P. putida.

2. Materials and methods 2.1. Strains, culture conditions and transformation procedures E. coli DH10B (Grant et al., 1990) or SCS110 (Stratagene) were used for cloning experiments. E. coli HB101/pRK2013 (Figurski and Helinski, 1979) was used as helper strain for conjugation experiments. E. coli BL21(DE3) (Studier and Moffatt, 1986) was used for heterologous expression experiments. E. coli HS996/pSC101-BADgbaA (tetR) (Wang et al., 2006) and GB05-red (Fu et al., 2012) were used for plasmid modification via Red/ET recombineering. The cells were grown in LB medium or on LB agar (1% tryptone, 0.5% yeast extract, 0.5% NaCl, (1.5% agar)) at 30–37 °C (and 200 rpm) overnight. Antibiotics were used at the following concentrations: 100 mg/ml ampicillin, 50 mg/ml kanamycin, 34 mg/ml chloramphenicol, 20 mg/ml gentamicin, and 6–12.5 mg/ml tetracycline.

Please cite this article as: Gemperlein, K., et al., Metabolic engineering of Pseudomonas putida for production of docosahexaenoic acid based on a myxobacterial PUFA synthase. Metab. Eng. (2015), http://dx.doi.org/10.1016/j.ymben.2015.11.001i

67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132

K. Gemperlein et al. / Metabolic Engineering ∎ (∎∎∎∎) ∎∎∎–∎∎∎

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66

Transformation of E. coli strains was achieved via electroporation in 0.1-cm-wide cuvettes at 1250 V, a resistance of 200 Ω, and a capacitance of 25 μF. P. putida KT2440 (Bagdasarian et al., 1981) was cultivated in liquid LB medium at 30 °C and 200 rpm overnight. Transformation of P. putida was achieved via triparental conjugation according to an established procedure (Hill et al., 1994). Transformants were selected on PMM agar plates (0.61% K2HPO4, 0.5% KH2PO4, 0.1% (NH4)2SO4, 0.66% disodium succinate, and 1.5% agar; pH adjusted to 7.0; supplementation with 1.2 mM MgSO4 after autoclaving). Antibiotics were used at the following concentrations: 60 mg/ml kanamycin, 30 mg/ml tetracycline, and 20 mg/ml gentamicin. Detailed information on the construction of P. putida mutant strains is provided in the Supplementary data. 2.2. DNA isolation, manipulation and analysis Isolation of chromosomal DNA from P. putida KT2440 and mutants thereof for subsequent PCR verification experiments was performed using the Gentra Puregene Yeast/Bact. Kit (Qiagen). An alkaline lysis method (Sambrook and Russell, 2001) was used to isolate and purify plasmid DNA from E. coli and P. putida strains. Restriction endonucleases, alkaline phosphatase (FastAP) and T4 DNA ligase were obtained from Thermo Scientific and were applied according to standard protocols (Sambrook and Russell, 2001). PCRs were performed using either Taq DNA polymerase or Phusion DNA polymerase (Thermo Scientific) according to the manufacturer`s protocol and were carried out in an Eppendorf

3

Mastercycler. Oligonucleotides for PCR were obtained from SigmaAldrich and are listed in Table S1. For cloning experiments, PCR products were purified by agarose gel electrophoresis and subsequent gel extraction with NucleoSpin Extract II Kit (MachereyNagel). Red/ET recombineering experiments for plasmid modifications were performed according to previously established procedures (Zhang et al., 2000). Experimental details on the construction of plasmids and the genotypic verification of mutant strains are provided in the Supplementary Data.

2.3. Sequence analysis and design of the synthetic gene clusters The sequence of the pfa gene cluster of A. fasciculatus (SBSr002) was analyzed and compared to the genome sequence of P. putida KT2440 (Nelson et al., 2002) retrieved from NCBI Genome RefSeq NC_002947. Based on this, relevant parameters for constructional and functional sequence design were defined to generate artificial pathway versions using the evoMAGis software package (ATG: biosynthetics GmbH) as described previously (Osswald et al., 2012). The sequence design process included engineering of restriction sites, adaptation of the codon usage, elimination of sequence repeats as well as rare codon clusters, engineering of Shine-Dalgarno (SD)‒ anti-SD interactions, and introduction of hidden stop codons in unused frames. Details on the sequence design are provided in the Supplementary data.

Table 1 Constructed expression strains of P. putida and achieved production yields of n-6 DPA and DHA after cultivation at 16 °C and 200 rpm for 24 h. CDW ¼ cell dry weight; n.d. ¼ not determined. The engineered P. putida strains did not show obvious variations in their phenotypes. All plasmids constructed in this study are described in detail in Table S2. Strain

Characteristics

n-6 DPA production

DHA production

P. putida KT2440 þpJBPfaAf1

Pseudomonas putida KT2440 with replicative plasmid pJBPfaAf1; KanR P. putida KT2440 with chromosomal integration of plasmid pPptAf1 into the trpE locus; TetR P. putida KT2440::pPptAf1 with replicative plasmid pJBPfaAf1; TetR, KanR P. putida KT2440::pPptAf1 with replicative plasmid pJBPfaAf1* (Pm*); TetR, KanR P. putida KT2440::pPptAf1 with replicative plasmid pJBPfaAf1** (Pm**); TetR, KanR P. putida KT2440::pPptAf1 with replicative plasmid pJB*PfaAf1** (trfA*, Pm**); TetR, KanR P. putida KT2440 with chromosomal knockout of fadH via single crossover; GentR P. putida KT2440::pfadH_KO with replicative plasmid pJB*PfaAf1** (trfA*, Pm**); GentR, KanR P. putida KT2440::pfadH_KO þpJB*PfaAf1** (trfA*, Pm**) with chromosomal integration of plasmid pPptAf1 into the trpE locus; GentR, TetR, KanR P. putida KT2440::pfadH_KO þpJB*PfaAf1** (trfA*, Pm**) with chromosomal integration of plasmid pME3 (zwf-1) into the trpE locus; GentR, TetR, KanR P. putida KT2440::pfadH_KO þpJB*PfaAf1** (trfA*, Pm**) with chromosomal integration of plasmid pME2 (accB, accC-1, accA, accD) into the trpE locus; GentR, TetR, KanR P. putida KT2440::pfadH_KO/pME2 þ pJB*PfaAf1** (trfA*, Pm**) cured of plasmid pJB*PfaAf1**; GentR, TetR P. putida KT2440::pfadH_KO/pME2 with replicative plasmid pPm**SynPfaAf1 (artificial pfa gene cluster version 1, trfA*, Pm**); GentR, TetR, KanR P. putida KT2440::pfadH_KO/pME2 with replicative plasmid pPm**SynPfaAf2 (artificial pfa gene cluster version 2, trfA*, Pm**); GentR, TetR, KanR P. putida KT2440::pfadH_KO/pME2 with replicative plasmid pPm**SynPfaAf2a (artificial pfa gene cluster version 2 a, trfA*, Pm**); GentR, TetR, KanR

0.03 7 0.007 mg/g CDW 0.05 7 0.01 mg/l n.d.

0.2 7 0.007 mg/g CDW 0.3 7 0.02 mg/l n.d.

0.077 0.006 mg/g CDW 0.17 0.006 mg/l 0.09 7 0.004 mg/g CDW 0.2 7 0.009 mg/l 0.17 0.002 mg/g CDW 0.2 7 0.01 mg/l 0.17 0.01 mg/g CDW 0.2 7 0.04 mg/l n.d.

0.3 7 0.007 mg/g CDW 0.5 7 0.02 mg/l 0.5 7 0.01 mg/g CDW 0.8 7 0.06 mg/l 0.5 7 0.02 mg/g CDW 0.8 7 0.05 mg/l 0.6 7 0.05 mg/g CDW 1.0 70.2 mg/l n.d.

n.d.

n.d.

0.17 0.02 mg/g CDW 0.2 7 0.02 mg/l

0.7 7 0.1 mg/g CDW 0.9 7 0.1 mg/l

0.2 7 0.004 mg/g CDW 0.3 7 0.01 mg/l

0.8 7 0.04 mg/g CDW 1.17 0.05 mg/l

0.17 0.02 mg/g CDW 0.2 7 0.07 mg/l

0.9 7 0.04 mg/g CDW 1.3 70.2 mg/l

n.d.

n.d.

0.17 0.03 mg/g CDW 0.2 7 0.01 mg/l

0.5 7 0.2 mg/g CDW 0.8 7 0.1 mg/l

0.3 7 0.08 mg/g CDW 0.6 7 0.3 mg/l

1.2 70.3 mg/g CDW 2.4 7 0.8 mg/l

0.2 7 0.06 mg/g CDW 0.4 7 0.2 mg/l

1.4 70.3 mg/g CDW 3.0 7 1.0 mg/l

P. putida KT2440::pPptAf1 P. putida KT2440::pPptAf1 þpJBPfaAf1 P. putida KT2440::pPptAf1 þpJBPfaAf1* P. putida KT2440::pPptAf1 þpJBPfaAf1** P. putida KT2440::pPptAf1 þpJB*PfaAf1** P. putida KT2440::pfadH_KO P. putida KT2440:: pfadH_KOþ pJB*PfaAf1** P. putida KT2440::pfadH_KO/ pPptAf1þpJB*PfaAf1** P. putida KT2440::pfadH_KO/ pME3 þpJB*PfaAf1** P. putida KT2440::pfadH_KO/ pME2 þpJB*PfaAf1** P. putida KT2440::pfadH_KO/pME2 P. putida KT2440::pfadH_KO/ pME2 þpPm**SynPfaAf1 P. putida KT2440::pfadH_KO/ pME2 þpPm**SynPfaAf2 P. putida KT2440::pfadH_KO/ pME2 þpPm**SynPfaAf2a

Please cite this article as: Gemperlein, K., et al., Metabolic engineering of Pseudomonas putida for production of docosahexaenoic acid based on a myxobacterial PUFA synthase. Metab. Eng. (2015), http://dx.doi.org/10.1016/j.ymben.2015.11.001i

67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 Q3 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132

K. Gemperlein et al. / Metabolic Engineering ∎ (∎∎∎∎) ∎∎∎–∎∎∎

4

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66

pPfaAf2 Promoter deletion by Red/ET recombination

bla p15A oriV

pfa1

pfa2

pfa3

bla cat pfa1 p15A oriV

pfa2

pfa3

pfa2

pfa3

pfa2

pfa3

pPfaAf6 Promoter insertion by conventional cloning

Pm/Pm*/Pm**

pPfaAf7.1, pPfaAf7.1*, or pPfaAf7.1** Vector backbone exchange by conventional cloning

bla xylS p15A oriV

pfa1

Pm/Pm*/Pm** pJBPfaAf1, pJBPfaAf1*, pJBPfaAf1**, or pJB*PfaAf1**

pPfaAf7.1

trfA/ aphI xylS trfA* RK2 RK2 oriV oriT

pfa1

pPptAf1

pPptAfA Exchange of pfa genes for ppt by conventional cloning

Pm bla xylS p15A oriV

ppt

Vector backbone modification by conventional cloning

Pm trpE tetR tetA oriT bla xylS p15A oriV

ppt

Fig. 2. Cloning strategy for the construction of plasmids for recombinant LC-PUFA production in Pseudomonas putida KT2440 and strain improvement by genetic engineering. (A) Cloning of the pfa gene cluster from Aetherobacter fasciculatus (SBSr002) into replicative plasmids. (B) Cloning of the gene encoding the PPTase AfPpt from A. fasciculatus (SBSr002), into an integrative plasmid. oriV¼ origin of replication, trfA ¼gene encoding replication initiation protein in RK2 replicons (trfA* ¼cop271C mutation included), oriT¼origin of transfer, bla¼ ampicillin resistance gene, cat ¼chloramphenicol resistance gene, aphI ¼ kanamycin resistance gene, tetA¼ tetracycline resistance gene, Pm¼ Pm promoter (Pm* ¼ mutated 5'-UTR of version LII-11 (Berg et al., 2009) included; Pm** ¼mutated 5'-UTR of version LII-11 (Berg et al., 2009) plus the mutated core promoter region of version ML2-2 (Bakke et al., 2009) included), xylS¼ gene encoding the transcriptional regulator for Pm promoter, tetR ¼gene encoding the tetracycline transcriptional regulator, trpE ¼gene encoding the anthranilate synthase component 1 from P. putida KT2440.

2.4. Heterologous LC-PUFA production in E. coli and P. putida For heterologous LC-PUFA production two different host strains, E. coli BL21(DE3) and P. putida KT2440, were evaluated. As summarized in Table 1, different expression strains were constructed (more detailed information on expression constructs and strain engineering is provided in the Supplementary data). In most cases, the pfa biosynthetic gene cluster was co-expressed with a suitable phosphopantetheinyl transferase from A. fasciculatus (AfPpt) and for P. putida additional strain engineering was performed. For heterologous LC-PUFA production in E. coli, cultivations were carried out in 50 ml LB medium containing 50 mg/ml kanamycin and 100 mg/ml ampicillin. The medium was inoculated with an overnight culture (1:100) and incubated at 37 °C. Expression of the Pfa proteins and AfPpt was induced at OD600 of 0.8 by addition of m-toluic acid to a final concentration of 2 mM in case of pJBPfaAf1 plus pPptAfA and pJBPfaAf1* plus pPptAfA or of L-arabinose to a final concentration of 2% in case of pJBPfaAf4 plus pPptAfD. After induction, the cells were cultivated at 16 °C and 200 rpm for 24 h and then harvested by centrifugation at 8000 rpm for 5 min. For heterologous LC-PUFA production in P. putida, cultivations were carried out in 50 ml LB medium containing 60 mg/ml kanamycin, and if necessary 30 μg/ml tetracycline, and 20 mg/ml gentamicin. The medium was inoculated with an overnight culture (1:100) and incubated at 30 °C for 4 h. Expression of the Pfa proteins or AfPpt was induced by addition of either m-toluic acid to a final concentration of 2 mM or of L-arabinose to a final concentration of 2%, depending on the heterologous promoter system used. After induction, the cells were cultivated at 16 °C and 200 rpm for 24 h and then harvested by centrifugation at 8000 rpm for 5 min.

2.5. Extraction of cellular fatty acids and analysis of fatty acid methyl esters by GC–MS The cellular fatty acids were extracted using the FAME method (Bode et al., 2006; Gemperlein et al., 2014) and analyzed via GC– MS (Gemperlein et al., 2014).

3. Results and discussion In continuation of our previous studies on the identification of myxobacterial PUFA synthases and their heterologous expression in the myxobacterial model strain M. xanthus (Gemperlein et al., 2014), we aimed to establish alternative heterologous expression platforms in hosts which show better growth characteristics and for which a considerable set of genetic tools is available. Besides E. coli, the GRAS strain P. putida KT2440 was intensively investigated as expression host for the DPA/DHA-type PUFA biosynthetic pathway from A. fasciculatus (SBSr002). 3.1. Initial heterologous expression studies in P. putida The initial strategy to achieve heterologous expression of the DPA/DHA-type pfa gene cluster from A. fasciculatus (SBSr002) in Pseudomonas was designed similarly to a previously described approach applied for heterologous expression of a myxobacterial lipopeptide pathway (Wenzel et al., 2005). It included the chromosomal integration of the expression construct into the trpE locus of the genome, encoding the anthranilate synthase component 1, and the expression of pathway genes under control of the inducible xylS-Pm promoter system (details not shown). This procedure did not result in a detectable heterologous LC-PUFA production in P. putida. To evaluate if the pfa gene dosage might be

Please cite this article as: Gemperlein, K., et al., Metabolic engineering of Pseudomonas putida for production of docosahexaenoic acid based on a myxobacterial PUFA synthase. Metab. Eng. (2015), http://dx.doi.org/10.1016/j.ymben.2015.11.001i

67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132

K. Gemperlein et al. / Metabolic Engineering ∎ (∎∎∎∎) ∎∎∎–∎∎∎ 1.0 0.9

0.8 0.7 0.6 0.5

+ 327%

mg / g cell dry weight

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66

5

n-6 DPA

0.4 DHA 0.3 0.2 0.1 0.0 P. putida KT2440 + pJBPfaAf1

P. putida KT2440:: pPptAf1 + pJBPfaAf1

P. putida KT2440:: pPptAf1 + pJBPfaAf1*

P. putida KT2440:: pPptAf1 + pJBPfaAf1**

P. putida KT2440:: pPptAf1 + pJB*PfaAf1**

P. putida P. putida P. putida KT2440:: KT2440:: KT2440:: pfadH_KO/ pfadH_KO/ pfadH_KO/ pPptAf1 + pME3 + pME2 + pJB*PfaAf1** pJB*PfaAf1** pJB*PfaAf1**

Fig. 3. Recombinant production of n-6 DPA and DHA in engineered Pseudomonas putida KT2440 strains based on native pfa genes from Aetherobacter fasciculatus (SBSr002). The indicated values are means and standard deviations of three biological samples.

a bottleneck, we aimed to analyze PUFA production based on a replicative plasmid. A medium-copy number replicative expression construct was generated in a similar way as described for heterologous expression of a myxobacterial chalcone synthase in P. putida (Gross et al., 2006). Starting from a construct harboring the entire DPA/DHA-type pfa gene cluster of A. fasciculatus (pPfaAf2, (Gemperlein et al., 2014), a replicative expression construct was generated by a combination of conventional cloning techniques and Red/ET recombineering (Zhang et al., 2000) (Figs. 2 and S1). Flanking DNA upstream of the pfa gene cluster including the native promoter region was replaced by the inducible xylS-Pm promoter system, and the vector backbone was exchanged against the broad-host-range expression vector pJB861 (Blatny et al., 1997). After transfer of the resulting plasmid pJBPfaAf1 into P. putida KT2440, pfa gene cluster expression (driven by the xylS-Pm promoter system) was induced with toluic acid. The transgenic strain P. putida KT2440/pJBPfaAf1 was shown to produce 0.03 70.007 mg n-6 DPA/g CDW and 0.2 7 0.007 mg DHA/g CDW (Fig. 3), demonstrating the applicability of P. putida as expression host for PUFA synthases. As functional expression of the PUFA synthase requires posttranslational activation by a phosphopantetheinyl transferase (PPTase), this result shows that the only PPTase from P. putida KT2440, PP1183, is obviously able to activate (at least partially) the tandem acyl carrier protein (ACP) domains of Pfa2, which correlates with the observation of a broad substrate specificity for PP1183 in previous studies, e.g. (Gross et al., 2005). However, as (efficient) posttranslational activation might be a bottleneck in the expression strain P. putida KT2440/pJBPfaAf1, we aimed to improve LC-PUFA production by co-expression of the native PPTase from A. fasciculatus (SBSr002). A corresponding gene, afppt, which encodes a Sfp-type PPTase that likely catalyses phosphopantetheinylation of PUFA synthases, was identified in the myxobacterial PUFA producer strain and subcloned into an integrative expression plasmid under control of the xylS-Pm promoter system (Figs. 2 and S1). After chromosomal integration of the PPTase expression construct pPptAf1 into the host and transformation with the (previously used) pfa gene cluster expression construct pJBPfaAf1, the obtained transgenic strain, P. putida KT2440::pPptAf1/pJBPfaAf1, turned out to produce significantly higher amounts of LC-PUFAs (0.0770.006 mg n-6 DPA/g CDW and 0.3 mg70.007 DHA/g CDW) than P. putida KT2440/pJBPfaAf1 (Fig. 3). This result indicates that the intrinsic PPTase PP1183 from P. putida KT2440 does not completely activate the acyl carrier protein (ACP) domains of the overexpressed PUFA synthase and that this bottleneck can (at least partially) be overcome by coexpression of the authentic PPTase AfPpt.

In order to further optimize heterologous LC-PUFA production in P. putida, different cultivation conditions were evaluated, including the use of several complex or minimal liquid media in addition to the established production in LB medium (details not shown). Bacterial growth was boosted in rich, complex media but LC-PUFA production yields were lower than in LB medium. Using minimal media, growth was significantly impaired and absolute amounts of produced LC-PUFAs were not elevated. Based on these observations, LB medium seems to be the best choice for heterologous LC-PUFA production in P. putida, providing good growth characteristics for the bacteria and considerable LC-PUFA production titres. The production yields in the heterologous expression systems described above were achieved in LB medium and cultivation at 16 °C after induction of AfPpt and/or Pfa protein expression. When heterologous expression was carried out at 30 °C, no detectable LC-PUFA production could be observed. Investigations on the LC-PUFA production kinetics revealed that n6 DPA and DHA continually accumulated within 24 h of cultivation at 16 °C and that the yield of these LC-PUFAs per culture volume only slightly increased after additional 24 h of cultivation. 3.2. Genetic engineering of expression constructs to optimize LCPUFA production in P. putida To further increase LC-PUFA production rates in P. putida, we aimed to engineer the pfa gene dosage and to analyze different promoter systems for pfa (and afppt) gene expression. As alternative to the established xylS-Pm promoter system from the replicative pfa gene cluster expression construct pJBPfaAf1 and the integrative PPTase expression construct pPptAf1, derivatives harboring the L-arabinose inducible araC-PBAD promoter system (pJBPfaAf4 and pPptAf4) were generated and transferred into P. putida KT2440. The obtained transgenic strain P. putida KT2440:: pPptAf4/pJBPfaAf4 produces n-6 DPA and DHA in amounts comparable to the yield obtained with P. putida KT2440::pPptAf1/ pJBPfaAf1 (details not shown). This result indicates that under the applied conditions the araC-PBAD promoter system is not superior to the xylS-Pm promoter system, although any downstream bottlenecks possibly masking an effect in either direction cannot be excluded. In our further engineering efforts we focussed on the xylS-Pm promoter system for which variants in the 5'-untranslated region (UTR) (Berg et al., 2009) as well as in the promoter core region (Bakke et al., 2009) were described from random mutagenesis studies in E. coli. It has been shown that the 5'-UTR mutated Pm promoter version LII-11 allowed for up to 20-fold increase of reporter protein activity and seven-fold increase of

Please cite this article as: Gemperlein, K., et al., Metabolic engineering of Pseudomonas putida for production of docosahexaenoic acid based on a myxobacterial PUFA synthase. Metab. Eng. (2015), http://dx.doi.org/10.1016/j.ymben.2015.11.001i

67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132

6

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66

K. Gemperlein et al. / Metabolic Engineering ∎ (∎∎∎∎) ∎∎∎–∎∎∎

reporter gene transcript amount compared to the performance of the wild-type Pm promoter in E. coli (Berg et al., 2009). These data encouraged us to introduce the mutation described for LII-11 in the Pm promoter region of the pfa gene cluster to drive the expression via a modified xylS-Pm* promoter system. The resulting expression construct pJBPfaAf1* was transferred into P. putida KT2440::pPptAf1 to yield P. putida KT2440::pPptAf1/pJBPfaAf1*, which was shown to produce significantly higher amounts of LCPUFAs (0.09 70.004 mg n-6 DPA/g CDW and 0.5 70.01 mg DHA/g CDW) than P. putida KT2440::pPptAf1/pJBPfaAf1 (Fig. 3). Obviously, the 5'-UTR mutation of version LII-11 described from an E. coli reporter gene expression system does also have a positive effect on PUFA synthase expression in P. putida. Additional exchange of the wild-type Pm promoter upstream of the PPTase encoding gene afppt for the modified Pm* promoter did, however, not result in further production enhancement (data not shown), indicating that either posttranslational activation does not represent a bottleneck in the present expression system or that the 5’-UTR mutation does not mediate a positive effect on PPTase expression, e.g. due to unfavorable secondary structures on the mRNA level. In a second round of engineering, we aimed to additionally introduce mutations in the Pm promoter core region. In the random mutagenesis studies mentioned before the mutated Pm promoter version ML2-2 was shown to increase reporter protein activity up to 13.5-fold and reporter gene transcript amount up to ten-fold in E. coli (Bakke et al., 2009). In order to combine the beneficial effects observed for ML2-2 and LII-11, the described mutations within the Pm promoter core region of ML2-2 were introduced into the xylS-Pm* promoter system (already harboring the LII-11 mutations). Based on the resulting xylS-Pm** promoter system the expression construct pJBPfaAf1** was generated and transferred into P. putida KT2440::pPptAf1. Compared to P. putida KT2440::pPptAf1/pJBPfaAf1*, the amounts of LC-PUFAs produced by the obtained expression strain P. putida KT2440::pPptAf1/ pJBPfaAf1** (0.17 0.002 mg n-6 DPA/g CDW and 0.5 7 0.02 mg DHA/g CDW) are only slightly higher (Fig. 3). Thus, either the additional implementation of mutations within the core promoter region had only a minor positive effect on pfa gene expression in P. putida or other limitations in downstream cellular processes do not allow for higher LC-PUFA production yields. Apart from targeting the promoter region, genetic engineering was extended to the RK2 replicon, which consists of an origin of replication (oriV) and gene trfA encoding a helper protein for initiation of replication (Durland and Helinski, 1987). It has been shown that the copy number of RK2 based vectors can be raised by introducing a R271C point mutation into TrfA (Blatny et al., 1997). Assuming that this copy-up mutation can increase the pfa gene dosage which might correlate with higher LC-PUFA production the vector backbone of expression construct pJBPfaAf1** was exchanged for a trfA-mutated (R271C) version. Transfer of the resulting expression construct pJB*PfaAf1** into P. putida KT2440::pPptAf1 yielded P. putida KT2440::pPptAf1/pJB*PfaAf1**, which showed slight improvements of LC-PUFA production (0.1 70.01 mg n-6 DPA/g CDW and 0.6 70.05 mg DHA/g CDW) compared to P. putida KT2440::pPptAf1/pJBPfaAf1** (Fig. 3). As copy numbers and stability of plasmids pJBPfAf1** and pJB*PfaAf1** in the P. putida expression strains have not been analyzed in detail, it remains to be shown if PUFA production yields directly correlate to the pfa gene dosage or if there are any other limitations in downstream cellular processes. However, this engineering step had a positive effect on PUFA production in P. putida. 3.3. Heterologous expression studies in E. coli Besides P. putida, E. coli was established as expression host for the DPA/DHA-type myxobacterial PUFA synthase from A. fasciculatus

(SBSr002). As expression of Pfa proteins alone did not lead to detectable LC-PUFA production in E. coli (details not shown), the native PPTase from A. fasciculatus (AfPpt) was always co-expressed in the following experiments. To evaluate PUFA production in E. coli under control of the Pm wild-type, the mutated Pm* and the PBAD promoter pfa gene cluster expression constructs from the previous studies with P. putida were used. E. coli BL21(DE3) strains harboring the following combinations of expression constructs were generated: pJBPfaAf1/ pPptAfA, pJBPfaAf1*/pPptAfA, or pJBPfaAf4/pPptAfD. In contrast to P. putida, the employed PPTase expression constructs do not integrate into the chromosome, but do allow for replication besides the pfa expression vectors. Comparative LC-PUFA production analysis revealed the highest yields in E. coli BL21(DE3)/pJBPfaAf4/pPptAfD from which 0.1 mg DHA/g CDW and trace amounts of n-6 DPA were detected. Thus, in E. coli the araC-PBAD promoter system was superior to the xylSPm and xylS-Pm* promoter systems whose induction with m-toluic acid impaired cell growth. Overall, using the same expression constructs, higher LC-PUFA production yields could be achieved with P. putida, which seems to be the better host for expressing myxobacterial pfa gene clusters. It is likely that efficient translation of the GC-rich pfaand afppt-mRNAs is impaired in E. coli, whose codon usage differs significantly from those of myxobacteria. In contrast, the higher GCcontent of the genome sequence and the codon usage of P. putida certainly make it more suitable for the expression of the GC-rich myxobacterial genes, which might explain the observed higher LCPUFA production rates. However, additional factors might also play a significant role, e.g. mRNA stability, translational helper factors, precursor supply and product stability, just to name a few. Accordingly, subsequent engineering approaches were focussed on P. putida and extended towards optimization of product stability and precursor supply. 3.4. Metabolic engineering of P. putida to optimize LC-PUFA production Interestingly, in some cultures of transgenic P. putida or E. coli strains expressing the myxobacterial DPA/DHA-type PUFA biosynthetic pathway, trace amounts of arachidonic acid (AA, 20:4, n6) and EPA (20:5, n-3) could be also detected. As it is hypothesized that the PUFA synthase of Aetherobacter spp. may only produce DHA and some minor amounts of DPA as by-products, AA and EPA could result from degradation of these C22 PUFAs to C20 products (Gemperlein et al., 2014). This speculation is supported by our finding that feeding of 5 mg DHA to a 50 ml culture of P. putida KT2440 and cultivation for 4 h at 30 °C continued by 24 h at 16 °C resulted in a conversion of 4 74% of DHA into EPA. The conversion of DHA into EPA or n-6 DPA into AA via β-oxidation requires the activity of the house-keeping enzymes 2,4-dienoyl-CoA reductase (EC 1.3.1.34) and Δ3-Δ2-enoyl-CoA isomerase (EC 5.3.3.8). In order to block this catabolic route, expression of a functional 2,4-dienoyl-CoA reductase was prevented by insertion of the suicide plasmid pfadH_KO-gmR (see Supplementary data) via homologous recombination into the encoding gene locus (fadH) in the P. putida chromosome. The gentamycin selection marker of this plasmid does not interfere with the kanamycin and tetracycline resistance genes used for selection of the pfa gene cluster and PPTase expression constructs. Feeding the generated mutant strain P. putida KT2440::pfadH_KO with DHA under the same conditions as described for the wild-type still resulted in a conversion of DHA into EPA, which is however reduced to only 31% compared to 474% in the wild-type. The observed remaining ‘degradation activity’ might result from some residual activity of the mutated fadH or from complementation of FadH by other enzymes from the host. After transformation of the P. putida KT2440::pfadH_KO mutant with pPptAf1 and pJB*PfaAf1**, higher amounts of LCPUFAs (0.17 0.02 mg n-6 DPA/g CDW and 0.7 70.1 mg DHA/g

Please cite this article as: Gemperlein, K., et al., Metabolic engineering of Pseudomonas putida for production of docosahexaenoic acid based on a myxobacterial PUFA synthase. Metab. Eng. (2015), http://dx.doi.org/10.1016/j.ymben.2015.11.001i

67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132

K. Gemperlein et al. / Metabolic Engineering ∎ (∎∎∎∎) ∎∎∎–∎∎∎

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66

7

Native pfa gene cluster sequence pfa1

pfa2

Restriction sites engineering

pfa3

In silico sequence design SwaI MfeI NheI EcoRV

Artificial gene cluster sequence pfa-V1 (71 out of 5396 codons modified)

MluI SgrDI ScaI BglII

FseI AclI

ApaLI SnaBI FspI

PacI BspEI

pfa-V1

Additional sequence optimization Artificial gene cluster sequence pfa-V2 (2792 out of 5396 codons modified)

pfa-V2 • • • • • •

Gene synthesis

Codon adaptation to P. putida Local CAI modulation Elimination of rare codon clusters Reduction of SD-anti-SD interactions within CDS Increase of stop codons in unused frames Optimization of TIRs

SwaI MfeI NheI

BB1

Building blocks (BB) 1-7 of V1* or V2 (and TR-pfa2-pfa3_V2 for V2a)

BB5

MluI BglII

BB2

SnaBI

AclI

BB3

MfeI EcoRV

Assembly in pACYC-derived plasmid

ApaLI

MluI SgrDI ScaI BglII

FseI AclI

BB4

PacI BspEI

bb7 ApaLI SnaBI FspI

FseI AclI

BB6

TR

pACYC_BB1-7_V1* or V2(a) aphI Vector backbone exchange

BB1

BB2-4

TR

BB5-7

BB2-4

TR

BB5-7

p15A oriV Pm**

pPm**SynPfaAf1* or pPm**SynPfaAf2(a)

trfA* aphI xylS RK2 RK2 oriV oriT

BB1

Fig. 4. Design, synthesis, and cloning of synthetic pfa gene clusters originating from Aetherobacter fasciculatus (SBSr002) for recombinant LC-PUFA production in Pseudomonas putida KT2440. oriV ¼origin of replication, trfA* ¼gene encoding replication initiation protein in RK2 replicons, cop271C mutation included, oriT¼ origin of transfer, aphI ¼kanamycin resistance gene, Pm** ¼ Pm promoter, mutated 5'-UTR of version LII-11 (Berg et al., 2009) plus the mutated core promoter region of version ML2-2 (Bakke et al., 2009) included, xylS ¼gene encoding the transcriptional regulator for Pm promoter; * ¼not illustrated.

CDW) could be detected in P. putida KT2440::pfadH_KO/pPptAf1/ pJB*PfaAf1** compared to P. putida KT2440::pPptAf1/pJB*PfaAf1** (Fig. 3). In addition to the reduction of PUFA catabolism, optimization of the precursor supply for LC-PUFA biosynthesis represents a promising approach to increase production yields. De novo PUFA synthesis requires sufficient supply of malonyl-CoA to build the carbon chain and NADPH as cofactor for numerous reduction steps involved (Metz et al., 2001). To increase malonyl-CoA supply, we aimed to overexpress the intrinsic acetyl-CoA carboxylase (EC 6.4.1.2) catalysing the first committing step in fatty acid biosynthesis by a two-step mechanism. In the first reaction, biotin carboxylase is involved in the ATP-dependent carboxylation of biotin with bicarbonate. The carboxyl group is then transferred to acetyl-CoA by the carboxyl transferase, yielding malonyl-CoA in the second reaction step (Cronan and Waldrop, 2002). The acetyl-CoA carboxylase from P. putida is encoded by the genes accA and accD plus an operon comprising genes accB-accC-1. Genes accA and accD encode the carboxyl transferase subunits α and β, whereas accC-1 encodes the biotin carboxylase and accB encodes the biotin carboxyl carrier protein of the multienzyme complex. Each gene or operon was

subcloned from P. putida under control of the tac promoter to assemble a four-gene expression box flanked by the tonB transcription terminator. Insertion of Ptac-accB-accC-1-Ptac-accA-PtacaccD-TtonB into the PPTase expression plasmid pPptAf1 resulted in construct pME2 (see Supplementary Data). Subsequent transfer into P. putida KT2440::pfadH_KO/pJB*PfaAf1** yielded the expression strain KT2440::pfadH_KO/pME2/pJBPfaAf1**. In a parallel approach, we aimed to engineer the host for improved supply of the reductant NADPH by overexpression of the intrinsic glucose-6-phosphate dehydrogenase (EC 1.1.1.49) from the Entner-Doudoroff pathway, the predominant pathway for glucose catabolism in P. putida (del Castillo et al., 2007). The enzyme catalyses the oxidation of D-glucose-6-phosphate into 6-phospho-D-glucono-1,5-lactone under consumption of one equivalent of NADP þ , which is reduced to NADPH. Gene zwf-1 encoding the glucose-6-phosphate dehydrogenase from P. putida was subcloned under the control of the tac promoter and the tonB transcription terminator. The transcription unit was inserted into the PPTase expression plasmid pPptAf1 to generate construct pME3 (see Supplementary Data), which was transferred into P.

Please cite this article as: Gemperlein, K., et al., Metabolic engineering of Pseudomonas putida for production of docosahexaenoic acid based on a myxobacterial PUFA synthase. Metab. Eng. (2015), http://dx.doi.org/10.1016/j.ymben.2015.11.001i

67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132

8

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66

K. Gemperlein et al. / Metabolic Engineering ∎ (∎∎∎∎) ∎∎∎–∎∎∎

putida KT2440::pfadH_KO/pJB*PfaAf1** to yield P. putida KT2440:: pfadH_KO/pME3/pJBPfaAf1**. The engineered expression strains P. putida KT2440::pfadH_KO/ pME2þpJBPfaAf1** and P. putida KT2440::pfadH_KO/pME3þ pJBPfaAf1** were shown to produce higher amounts of LC-PUFAs than P. putida KT2440::pfadH_KO/pPptAf1þ pJB*PfaAf1** (Fig. 3). Overexpression of the glucose-6-phosphate dehydrogenase via plasmid pME3 resulted in the production of 0.270.004 mg n-6 DPA/g CDW and 0.870.04 mg DHA/g CDW, whereas overexpression of the acetyl-CoA carboxylase via plasmid pME2 yielded 0.170.02 mg n-6 DPA/g CDW and 0.970.04 mg DHA/g CDW. Hence, it can be deduced that limited supply of malonyl-CoA or NADPH represented a bottleneck for LC-PUFA biosynthesis, which could, at least in part, be eliminated by overexpression of acetyl-CoA carboxylase or glucose-6phosphate dehydrogenase. As rational consequence of this result, collective overexpression of accB-accC-1, accA, and accD plus zwf-1 was attempted by assembly of the entire gene set onto one expression plasmid, which was transferred and integrated into the genome

of P. putida (details not shown). Unfortunately, the resulting expression strain produces unexpectedly low amounts of LC-PUFAs (0.170.01 mg n-6 DPA/g CDW and 0.770.05 mg DHA/g CDW). A possible reason for this finding is the instability of the expression plasmid (e.g. due to constitutive overexpression of five genes) as stable and complete genomic integration of the intact expression construct was not verified. 3.5. Design and construction of artificial pfa gene clusters for expression in P. putida In continuation of our efforts on expression construct and strain engineering, we intended to explore the potential of synthetic pfa genes to further optimize the heterologous expression system. Using the native pfa pathway sequence from A. fasciculatus (SBSr002) as input, two different versions of artificial pfa gene cluster sequences were calculated. For version 1 (pfa-V1), only constructional sequence requirements were implemented by engineering of restriction sites

Fig. 5. Codon usage and local CAI. (A) Plot of synonymous codon fractions of the complete Pseudomonas putida codon usage table (black bars) and the native sequence of all pfa gene cluster CDS (gray bars) from Aetherobacter fasciculatus (SBSr002). (B) Plot of normalized synonymous codon fractions of the reduced P. putida codon usage table (black bars) and the optimized sequence of all pfa gene cluster CDS (gray bars). (C) Comparison of the course of the local codon adaptation index (CAI) of the 16 kb native and artificial pfa biosynthetic gene cluster sequences. The native sequence is shown in the background (gray), and the artificial sequence version 2 (pfa-V2) is shown in black. Window width is 25 codons. In pfa-V2, the gradient between start and stop codon of each coding DNA sequence (CDS) was adjusted to 0.05. Untranslated regions between the protein coding regions were excluded from the graph.

Please cite this article as: Gemperlein, K., et al., Metabolic engineering of Pseudomonas putida for production of docosahexaenoic acid based on a myxobacterial PUFA synthase. Metab. Eng. (2015), http://dx.doi.org/10.1016/j.ymben.2015.11.001i

67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132

K. Gemperlein et al. / Metabolic Engineering ∎ (∎∎∎∎) ∎∎∎–∎∎∎

1.8

1.6

1.4

1.2

mg/g cell dry weight

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66

1.0

n-6 DPA DHA

0.8

0.6

0.4

9

secondary structural interactions were attenuated during the codon adaptation procedure to design the synthetic gene cluster pfa-V2 (Fig. S3). Another important factor to be considered is the metabolic cost of translational miselongation upon a frameshift (Itzkovitz and Alon, 2007). In order to cause early termination of translation of incorrect reading frames, the ribosome must encounter an off-frame stop codon. Hence, stop codons were accumulated in the two unused frames of the artificial cluster sequence pfa-V2 (Fig. S3). Furthermore, the emergence of homopolymeric stretches was suppressed and the GC-content was kept within preselected limits during sequence optimization (Fig. S3). In contrast to pfa-V1, the resulting artificial gene cluster sequence (pfa-V2) differs significantly from the native sequence (2792 out of 5396 codons were modified, Table S4). The pfa-V2 design also included the optimization of pfa2 and pfa3 translational initiation regions (TIRs, Fig. S5). Due to the natural overlap of pfa2/pfa3, which indicates a translational coupling, two different variants (V2 and V2a) have been generated. In the artificial gene cluster pfa-V2a, the overlap of pfa2 and pfa3 was eliminated by introducing a 40 bp intergenic transition region, comprising a ShineDalgarno sequence 7 bp upstream of the start codon of pfa3 (Fig. S5). According to the analyses of the minimum free energies of the mRNA secondary structures (Table S4), this modification should result in an optimization of the translation initiation rate. 3.6. Construction of artificial pfa gene clusters for heterologous PUFA production in P. putida

0.2

0.0 P. putida KT2440:: pfadH_KO/ pME2 + pJB*PfaAf1**

P. putida KT2440:: pfadH_KO/ pME2 + pPm**Syn PfaAf1

P. putida KT2440:: pfadH_KO/ pME2 + pPm**Syn PfaAf2

P. putida KT2440:: pfadH_KO/ pME2 + pPm**Syn PfaAf2a

Fig. 6. Recombinant production of n-6 DPA and DHA by metabolically engineered Pseudomonas putida KT2440 with native or synthetic pfa gene clusters originating from Aetherobacter fasciculatus (SBSr002). The indicated values are means and standard deviations of three biological samples.

for pathway cloning and modification (Fig. 4, Fig. S2). The resulting pfa-V1 sequence is highly similar to the native pfa gene cluster sequence from A. fasciculatus, as only 71 codons out of 5396 were affected by synonymous substitutions (Table S4). Retaining the customized restriction sites pattern of pfa-V1, a second artificial pfa gene cluster sequence (pfa-V2) was generated by additional gene optimization with particular emphasis on parameters affecting translational elongation (see also SI). As illustrated in Fig. 5, the codon usage was adapted to a reduced and renormalized codon usage table of P. putida (excluding rare codons), and the local codon adaptation index (CAI) (Sharp and Li, 1987) was modulated. Compared to the native sequence, the local CAI of pfa-V2 shows a clearly smoothed shape without any distinct peaks. The generated slight gradient along each coding DNA sequence is intended to enhance the ribosome occupancy to shield the transcript from degradation (Pedersen et al., 2011). In general, the speed of ribosomal protein synthesis and the resulting translational elongation rate depends on several factors, including the availability of aminoacyl-tRNAs, but also on certain sequence features of the mRNA template, e.g. rare codon clusters (Clarke and Clark, 2008) or internal Shine-Dalgarno sequences (Li et al., 2012). These sequence features are often assigned to temporal separation of translation of segments within the peptide chain and proper co-translational folding of proteins with direct impact on their solubility and activity (Zhang et al., 2009). However, the search for relevant rare codon clusters within the native pfa genes did not reveal any conspicuous candidates. Thus, rare codon clusters were eliminated and Shine-Dalgarno (SD)⎔3 anti-SD interfering mRNA

The 16.2 kb artificial pfa gene cluster sequences (pfa-V1 and pfa-V2) were split into seven building blocks of moderate size (BB1-BB7) by calculating all possible combinations of commercially available restriction enzymes. The selected cryptic restriction sites were activated in the variable space of the codon table, unwanted sites were transformed to cryptic ones. Then the building blocks were generated via gene synthesis (Fig. 4). A pACYC-derived assembly plasmid was constructed by insertion of a suitable multiple cloning site, enabling stepwise stitching of the synthetic fragments via unique restriction sites to yield plasmids pACYC_BB1-7_V1 and pACYC_BB1-7_V2 (see Fig. SI). Gene cluster version pfa-V2 was further modified to pfa-V2a by incorporation of an artificial intergenic transition region between genes pfa2 and pfa3 as described above. The resulting three artificial gene cluster versions pfa-V1, pfa-V2, and pfa-V2a were subsequently subcloned onto the expression vector backbone of pJB*PfaAf1** (see Section 3.2). The resulting plasmids pPm**SynPfaAf1, pPm**SynPfaAf2, or pPm**SynPfaAf2a were transferred into P. putida KT2440:: pfadH_KO/pME2, which was engineered for improved LC-PUFA production (see Section 3.4). PUFA production was analyzed in comparison to strain P. putida KT2440::pfadH_KO/pME2 þpJB*PfaAf1** (see Section 3.4) expressing the native pfa gene cluster from A. fasciculatus (SBSr002) under the same conditions to yield 0.1 70.02 mg n-6 DPA/g cell dry weight (CDW) and 0.9 70.04 mg DHA/g CDW (Fig. 6). Surprisingly, strain P. putida KT2440:: pfadH_KO/pME2 þ pPm**SynPfaAf1, expressing the artificial gene cluster pfa-V1, turned out to produce LC-PUFAs in significantly lower amounts (0.1 70.03 mg n-6 DPA/g CDW and 0.5 70.2 mg DHA/g CDW) (Fig. 6). The only differences between these two strains are silent mutations located in 71 codons within the pfa genes modified accordingly during constructional sequence design, which was not expected to have a negative impact on their functional expression. However, reproducible lower production yields from several clones of P. putida KT2440::pfadH_KO/ pME2þpPm**SynPfaAf1 indicated that one or more of these point mutations significantly affect pfa gene expression on the transcriptional and/or translational level. On the contrary, after applying additional changes to the DNA sequence to better adapt the pfa genes for expression in P. putida with special emphasis on

Please cite this article as: Gemperlein, K., et al., Metabolic engineering of Pseudomonas putida for production of docosahexaenoic acid based on a myxobacterial PUFA synthase. Metab. Eng. (2015), http://dx.doi.org/10.1016/j.ymben.2015.11.001i

67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 Q4 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132

10

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66

K. Gemperlein et al. / Metabolic Engineering ∎ (∎∎∎∎) ∎∎∎–∎∎∎

optimization of translational elongation, a clear positive effect on recombinant LC-PUFA production was observed. Strain P. putida KT2440::pfadH_KO/pME2 þ pPm**SynPfaAf2 containing the artificial gene cluster pfa-V2 produces 0.3 7 0.08 mg n-6 DPA/g CDW and 1.2 70.3 mg DHA/g CDW. Further engineering of pfa-V2 into pfa-V2a included efforts towards the optimization of the translation initiation rate on pfa3-mRNA. The corresponding strain P. putida KT2440::pfadH_KO/pME2/pPm**SynPfaAf2a shows only a slight but not statistically significant improvement of DHA production (0.2 mg7 0.06 n-6 DPA/g CDW and 1.4 mg7 0.3 DHA/g CDW). From these data it can be deduced that the outcome of heterologous expression of artificial gene cluster sequences rationally designed by introducing silent point mutations is still far from being easily predictable, as many cellular processes affecting gene expression and stability of foreign DNA are not yet sufficiently understood. However, the application of additional functional sequence design approaches on the first synthetic gene cluster (pfa-V1) led to the development of optimized pathways (pfa-V2/2a) that allow for much higher PUFA production with yield improvements of approximately 180%. Interestingly, all analyzed strains harboring artificial pfa gene clusters show a high standard deviation in production between distinct clones (Fig. 6).

Database linking The GenBank accession numbers for the artificial pfa biosynthetic gene cluster sequences V1, V2, and V2a reported in this paper are GenBank: KT734806; GenBank: KT734807; GenBank: KT734808.

Uncited references Bäck and Schwefel (1993), Goldberg (1989), Osswald et al. (2014).

Acknowledgments This work was supported by the Helmholtz-Initiative on Synthetic Biology.

Appendix A. Suplementary Information Supplementary data associated with this article can be found in the online version at: http://dx.doi.org/10.1016/j.ymben.2015.11.001.

4. Conclusions In this work, the pfa biosynthetic gene cluster from A. fasciculatus was successfully expressed in the non-related hosts E. coli and P. putida leading to higher DPA/DHA production yields in Pseudomonas. Various engineering steps were applied to address possible bottlenecks on different levels of PUFA biosynthesis: enhancement of pfa gene dosage (via replicon mutations) and pfa gene expression (via promoter mutations), optimization of PUFA synthase posttranslational activation (via co-expression of the authentic myxobacterial PPTase), improvement of precursor and cofactor supply for PUFA biosynthesis as well as enhancement of product stability (via metabolic engineering of P. putida). Further yield improvement could be achieved by pfa gene optimization and adaptation for expression in P. putida, e.g. to adapt codon usage. The constructed artificial pfa gene clusters represent to the best of our knowledge the first examples for polyketide biosynthetic gene clusters modulated and synthesized for P. putida. Although a lot remains to be learned regarding the limitations of heterologous expression of large foreign biosynthetic gene clusters encoding multifunctional megasynthetases, this approach holds great promise to better exploit complex natural product biosynthetic pathways, especially from genetically intractable or difficult to handle organisms or even from metagenomes. Our experiments establishing a heterologous PUFA expression platform now set the stage to address these challenges and thus to continuously improve our ability to harvest the enormous potential of bacterial natural product biosynthetic pathways. Currently, we also work on the redesign and expression of myxobacterial pfa gene clusters in the oleaginous yeast Yarrowia lipolytica. This strain was recently engineered to produce 56.6% of total fatty acids EPA (Xue et al., 2013) via the aerobic PUFA biosynthesis route. Here, saturated fatty acids are converted into PUFAs by a combination of several oxygen-dependent desaturases and elongases. It would interesting to evaluate the de novo biosynthesis approach in Yarrowia based on polyketide synthase-like biosynthetic machineries as described and employed in this study from myxobacteria.

References Allen, E.E., Bartlett, D.H., 2002. Structure and regulation of the omega-3 polyunsaturated fatty acid synthase genes from the deep-sea bacterium Photobacterium profundum strain SS9. Microbiology 148 (6), 1903–1913. Amiri-Jami, M., Griffiths, M.W., 2010. Recombinant production of omega-3 fatty acids in Escherichia coli using a gene cluster isolated from Shewanella baltica MAC1. J. Appl. Microbiol. 109 (6), 1897–1905. Amiri-Jami, M., Lapointe, G., Griffiths, M.W., 2014. Engineering of EPA/DHA omega3 fatty acid production by Lactococcus lactis subsp. cremoris MG1363. Appl. Microbiol. Biotechnol. 98 (7), 3071–3080. Amiri-Jami, M., Wang, H., Kakuda, Y., Griffiths, M.W., 2006. Enhancement of polyunsaturated fatty acid production by Tn5 transposon in Shewanella baltica. Biotechnol. Lett. 28 (15), 1187–1192. Bäck, T., Schwefel, H.P., 1993. An overview of evolutionary algorithms for parameter optimization. Evol. Comput. 1 (1), 1–23. Bagdasarian, M., Lurz, R., Ruckert, B., Franklin, F.C., Bagdasarian, M.M., Frey, J., Timmis, K.N., 1981. Specific-purpose plasmid cloning vectors. II. Broad host range, high copy number, RSF1010-derived vectors, and a host-vector system for gene cloning in Pseudomonas. Gene 16 (1–3), 237–247. Bakke, I., Berg, L., Aune, T.E., Brautaset, T., Sletta, H., Tondervik, A., Valla, S., 2009. Random mutagenesis of the Pm promoter as a powerful strategy for improvement of recombinant-gene expression. Appl. Environ. Microbiol. 75 (7), 2002–2011. Berg, L., Lale, R., Bakke, I., Burroughs, N., Valla, S., 2009. The expression of recombinant genes in Escherichia coli can be strongly stimulated at the transcript production level by mutating the DNA-region corresponding to the 5'untranslated part of mRNA. Microb. Biotechnol. 2 (3), 379–389. Blatny, J.M., Brautaset, T., Winther-Larsen, H.C., Karunakaran, P., Valla, S., 1997. Improved broad-host-range RK2 vectors useful for high and low regulated gene expression levels in gram-negative bacteria. Plasmid 38 (1), 35–51. Bode, H.B., Ring, M.W., Schwär, G., Kroppenstedt, R.M., Kaiser, D., Müller, R., 2006. 3-Hydroxy-3-methylglutaryl-coenzyme A (CoA) synthase is involved in the biosynthesis of isovaleryl-CoA in the myxobacterium Myxococcus xanthus during fruiting body formation. J. Bacteriol. 188 (18), 6524–6528. Chai, Y., Shan, S., Weissman, K.J., Hu, S., Zhang, Y., Muller, R., 2012. Heterologous expression and genetic engineering of the tubulysin biosynthetic gene cluster using Red/ET recombineering and inactivation mutagenesis. Chem. Biol. 19 (3), 361–371. Chang, G., Gao, N., Tian, G., Wu, Q., Chang, M., Wang, X., 2013. Improvement of docosahexaenoic acid production on glycerol by Schizochytrium sp. S31 with constantly high oxygen transfer coefficient. Bioresour. Technol. 142, 400–406. Clarke, T.F., Clark, P.L., 2008. Rare codons cluster. PLoS One 3 (10), e3412. Cronan Jr., J.E., Waldrop, G.L., 2002. Multi-subunit acetyl-CoA carboxylases. Prog. Lipid Res. 41 (5), 407–435. del Castillo, T., Ramos, J.L., Rodriguez-Herva, J.J., Fuhrer, T., Sauer, U., Duque, E., 2007. Convergent peripheral pathways catalyze initial glucose catabolism in Pseudomonas putida: genomic and flux analysis. J. Bacteriol. 189 (14), 5142–5152.

Please cite this article as: Gemperlein, K., et al., Metabolic engineering of Pseudomonas putida for production of docosahexaenoic acid based on a myxobacterial PUFA synthase. Metab. Eng. (2015), http://dx.doi.org/10.1016/j.ymben.2015.11.001i

67 68 69 70 71 72 73 74 75 Q5 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132

K. Gemperlein et al. / Metabolic Engineering ∎ (∎∎∎∎) ∎∎∎–∎∎∎

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Q6 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66

Durland, R.H., Helinski, D.R., 1987. The sequence encoding the 43-kilodalton trfA protein is required for efficient replication or maintenance of minimal RK2 replicons in Pseudomonas aeruginosa. Plasmid 18 (2), 164–169. Fang, J., Kato, C., Sato, T., Chan, O., McKay, D., 2004. Biosynthesis and dietary uptake of polyunsaturated fatty acids by piezophilic bacteria. Comput. Biochem. Physiol. B Biochem. Mol. Biol. 137 (4), 455–461. Figurski, D.H., Helinski, D.R., 1979. Replication of an origin-containing derivative of plasmid RK2 dependent on a plasmid function provided in trans. Proc. Natl. Acad. Sci. U.S.A. 76 (4), 1648–1652. Fu, J., et al., 2012. Full-length RecE enhances linear-linear homologous recombination and facilitates direct cloning for bioprospecting. Nat. Biotechnol. 30 (5), 440–446. Fu, J., et al., 2008. Efficient transfer of two large secondary metabolite pathway gene clusters into heterologous hosts by transposition. Nucleic Acids Res. 36 (17), e113. Garcia, R., Pistorius, D., Stadler, M., Muller, R., 2011. Fatty acid-related phylogeny of myxobacteria as an approach to discover polyunsaturated omega-3/6 fatty acids. J. Bacteriol. 193 (8), 1930–1942. Gemperlein, K., Rachid, S., Garcia, R., Wenzel, S.C., Müller, R., 2014. Polyunsaturated fatty acid biosynthesis in myxobacteria: Different PUFA synthases and their product diversity. Chem. Sci. 5 (5), 1733–1741. Goldberg, D.E., 1989. Genetic algorithms in search, optimization and machine learning. Addison-Wesley. Grant, S.G., Jessee, J., Bloom, F.R., Hanahan, D., 1990. Differential plasmid rescue from transgenic mouse DNAs into Escherichia coli methylation-restriction mutants. Proc. Natl. Acad. Sci. U.S.A. 87 (12), 4645–4649. Gross, F., Gottschalk, D., Müller, R., 2005. Posttranslational modification of myxobacterial carrier protein domains in Pseudomonas sp. by an intrinsic phosphopantetheinyl transferase. Appl. Microbiol. Biotechnol. 68 (1), 66–74. Gross, F., Luniak, N., Perlova, O., Gaitatzis, N., Jenke-Kodama, H., Gerth, K., Gottschalk, D., Dittmann, E., Müller, R., 2006. Bacterial type III polyketide synthases: phylogenetic analysis and potential for the production of novel secondary metabolites by heterologous expession in pseudomonads. Arch. Microbiol. 185 (1), 28–38. Hauvermale, A., Kuner, J., Rosenzweig, B., Guerra, D., Diltz, S., Metz, J.G., 2006. Fatty acid production in Schizochytrium sp.: involvement of a polyunsaturated fatty acid synthase and a type I fatty acid synthase. Lipids 41 (8), 739–747. Hill, D.S., Stein, J.I., Torkewitz, N.R., Morse, A.M., Howell, C.R., Pachlatko, J.P., Becker, J.O., Ligon, J.M., 1994. Cloning of genes involved in the synthesis of pyrrolnitrin from pseudomonas fluorescens and role of pyrrolnitrin synthesis in biological control of plant disease. Appl. Environ. Microbiol. 60 (1), 78–85. Itzkovitz, S., Alon, U., 2007. The genetic code is nearly optimal for allowing additional information within protein-coding sequences. Genome Res. 17 (4), 405–412. Kaulmann, U., Hertweck, C., 2002. Biosynthesis of polyunsaturated fatty acids by polyketide synthases. Angew. Chem. Int. Ed. Engl. 41 (11), 1866–1869. Lenihan-Geels, G., Bishop, K.S., Ferguson, L.R., 2013. Alternative sources of omega-3 fats: can we find a sustainable substitute for fish? Nutrients 5 (4), 1301–1315. Li, G.W., Oh, E., Weissman, J.S., 2012. The anti-Shine-Dalgarno sequence drives translational pausing and codon choice in bacteria. Nature 484 (7395), 538–541. Lorente-Cebrian, S., Costa, A.G.V., Navas-Carretero, S., Zabala, M., Martinez, J.A., Moreno-Aliaga, M.J., 2013. Role of omega-3 fatty acids in obesity, metabolic syndrome, and cardiovascular diseases: a review of the evidence. J. Physiol. Biochem. 69 (3), 633–651. Metz, J.G., et al., 2001. Production of polyunsaturated fatty acids by polyketide synthases in both prokaryotes and eukaryotes. Science 293 (5528), 290–293. Morita, N., Nishida, T., Tanaka, M., Yano, Y., Okuyama, H., 2005. Enhancement of polyunsaturated fatty acid production by cerulenin treatment in polyunsaturated fatty acid-producing bacteria. Biotechnol. Lett. 27 (6), 389–393. Morita, N., Tanaka, M., Okuyama, H., 2000. Biosynthesis of fatty acids in the docosahexaenoic acid-producing bacterium Moritella marina strain MP-1. Biochem. Soc. Trans. 28 (6), 943–945.

11

Napier, J.A., 2002. Plumbing the depths of PUFA biosynthesis: a novel polyketide synthase-like pathway from marine organisms. Trends Plant Sci. 7 (2), 51–54. Nelson, K.E., et al., 2002. Complete genome sequence and comparative analysis of the metabolically versatile Pseudomonas putida KT2440. Environ. Microbiol. 4 (12), 799–808. Orikasa, Y., Ito, Y., Nishida, T., Watanabe, K., Morita, N., Ohwada, T., Yumoto, I., Okuyama, H., 2007. Enhanced heterologous production of eicosapentaenoic acid in Escherichia coli cells that co-express eicosapentaenoic acid biosynthesis pfa genes and foreign DNA fragments including a high-performance catalase gene, vktA. Biotechnol. Lett. 29 (5), 803–812. Orikasa, Y., Nishida, T., Yamada, A., Yu, R., Watanabe, K., Hase, A., Morita, N., Okuyama, H., 2006. Recombinant production of docosahexaenoic acid in a polyketide biosynthesis mode in Escherichia coli. Biotechnol. Lett. 28 (22), 1841–1847. Osswald, C., Zipf, G., Schmidt, G., Maier, J., Bernauer, H.S., Muller, R., Wenzel, S.C., 2014. Modular construction of a functional artificial epothilone polyketide pathway. ACS Synth. Biol. 3 (10), 759–772. Pedersen, M., Nissen, S., Mitarai, N., Lo, S.S., Sneppen, K., Pedersen, S., 2011. The functional half-life of an mRNA depends on the ribosome spacing in an early coding region. J. Mol. Biol. 407 (1), 35–44. Ratledge, C., 2004. Fatty acid biosynthesis in microorganisms being used for single cell oil production. Biochimie 86 (11), 807–815. Ren, L.J., Huang, H., Xiao, A.H., Lian, M., Jin, L.J., Ji, X.J., 2009. Enhanced docosahexaenoic acid production by reinforcing acetyl-CoA and NADPH supply in Schizochytrium sp. HX-308. Bioprocess. Biosyst. Eng. 32 (6), 837–843. Ren, L.J., Sun, L.N., Zhuang, X.Y., Qu, L., Ji, X.J., Huang, H., 2014. Regulation of docosahexaenoic acid production by Schizochytrium sp.: effect of nitrogen addition. Bioprocess. Biosyst. Eng. 37 (5), 865–872. Sambrook, J., Russell, D.W., 2001. Molecular cloning: a laboratory manual Cold Spring Harbor. Cold Spring Harbor Laboratory Press, NY. Sharp, P.M., Li, W.H., 1987. The codon adaptation index-a measure of directional synonymous codon usage bias, and its potential applications. Nucleic Acids Res. 15 (3), 1281–1295. Studier, F.W., Moffatt, B.A., 1986. Use of bacteriophage T7 RNA polymerase to direct selective high-level expression of cloned genes. J. Mol. Biol. 189 (1), 113–130. Takeyama, H., Takeda, D., Yazawa, K., Yamada, A., Matsunaga, T., 1997. Expression of the eicosapentaenoic acid synthesis gene cluster from Shewanella sp. in a transgenic marine cyanobacterium, Synechococcus sp. Microbiology 143 (8), 2725–2731. Wallis, J.G., Watts, J.L., Browse, J., 2002. Polyunsaturated fatty acid synthesis: what will they think of next? Trends Biochem. Sci. 27 (9), 467–473. Wang, J., Sarov, M., Rientjes, J., Fu, J., Hollak, H., Kranz, H., Xie, W., Stewart, A.F., Zhang, Y., 2006. An improved recombineering approach by adding RecA to lambda Red recombination. Mol. Biotechnol. 32 (1), 43–53. Wenzel, S.C., Gross, F., Zhang, Y., Fu, J., Stewart, F.A., Müller, R., 2005. Heterologous expression of a myxobacterial natural products assembly line in pseudomonads via red/ET recombineering. Chem. Biol. 12 (3), 349–356. Xue, Z.X., Sharpe, P.L., Hong, S.P., Yadav, N.S., Xie, D.M., Short, D.R., Damude, H.G., Rupert, R.A., Seip, J.E., Wang, J., Pollak, D.W., Bostick, M.W., Bosak, M.D., Macool, D.J., Hollerbach, D.H., Zhang, H.X., Arcilla, D.M., Bledsoe, S.A., Croker, K., McCord, E.F., Tyreus, B.D., Jackson, E.N., Zhu, Q., 2013. Production of omega-3 eicosapentaenoic acid by metabolic engineering of Yarrowia lipolytica. Nat. Biotechnol. 31 (8), 734–740. Zhang, G., Hubalewska, M., Ignatova, Z., 2009. Transient ribosomal attenuation coordinates protein synthesis and co-translational folding. Nat. Struct. Mol. Biol. 16 (3), 274–280. Zhang, Y., Muyrers, J.P.P., Testa, G., Stewart, A.F., 2000. DNA cloning by homologous recombination in Escherichia coli. Nat. Biotechnol. 18 (12), 1314–1317.

Please cite this article as: Gemperlein, K., et al., Metabolic engineering of Pseudomonas putida for production of docosahexaenoic acid based on a myxobacterial PUFA synthase. Metab. Eng. (2015), http://dx.doi.org/10.1016/j.ymben.2015.11.001i

67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132