1 , 3-propanediol production by Escherichia coli expressing genes of dha operon from Clostridium butyricum 2 CR 371 . 5

1,3-propanediol is used as a monomer in the production of some polymers e.g. polytrimethylene terephthalate used in the production of carpets and textile fibers and in the thermoplastics engineering. However, the traditional chemical synthesis is expensive, generates some toxic intermediates and requires a reduction step under high hydrogen pressure. Biological production of 1,3-propanediol could be an attractive alternative to the traditional chemical methods. Moreover, crude glycerol which is a by-product of biodiesel production, can be used. We constructed a recombinant Escherichia coli strain producing 1,3-propanediol from glycerol by introducing genes of the dha operon from Clostridium butyricum 2CR371.5, a strain from our collection of environmental samples and strains. The E. coli strain produced 3.7 g of 1,3-propanediol per one litre of culture with the yield of 0.3 g per 1 g of glycerol consumed.


INTRODUCTION
1,3-propanediol (1,3-PDO) is used as a monomer in the production of several types of polymers, one of which is polytrimethylene terephthalate (PTT), produced by polycondensation of 1,3-PDO with terephtalic acid or its esters.PTT is a type of polyester used in the production of carpets and textile fibers and in thermoplastics engineering (Liu et al., 2010).
1,3-PDO can be produced in many ways i.a. from ethylene oxide over a catalyst in the presence of phosphine, water, carbon monoxide, hydrogen and an acid.However, the traditional chemical synthesis is expensive, generates some toxic intermediates and requires a reduction step under high hydrogen pressure (Raynaud et al., 2003;Emptage et al., 2009).Biological production of 1,3-PDO could be a useful alternative to the traditional chemical methods because of its numerous advantages, including a lower environmental impact.Moreover, crude glycerol, which is a by-product of biodiesel production, can be used as precursor.The conversion of glycerol to highervalue products, such as 1,3-PDO should decrease the cost of biofuel production (Zeng & Biebl, 2002;Yazdani & Gonzalez, 2007).
Glycerol fermentation by glycerol-fermenting microorganisms is a two-branched pathway.The 1,3-PDO production by C. butyricum is the reductive branch catalysed by two enzymes, (i) glycerol dehydratase encoded by dhaB1 and (ii) NADH-dependent 1,3-PDO oxidoreductase (encoded by dhaT), with 3-hydroxypropionaldehyde as an intermediate (Fig. 1).Glycerol dehydratase requires for activity the presence of a protein encoded by the dhaB2 gene.All three genes are located in the dha operon, the expression of which is induced in the presence of dihydroxyacetone or glycerol.On the other hand, in the oxidative branch, glycerol is dehydrogenated by NAD-dependent glycerol dehydrogenase to dihydroxyacetone (DHA).DHA is then converted sequentially to glyceraldehyde-3-phosphate, phosphoenolpyruvate, and pyruvate (Saint-Amans et al., 2001;Raynaud et al., 2003;Daniel et al., 2005;Gupta et al., 2009;Marçal et al., 2009).
Here we report construction of a recombinant E. coli strain producing 1,3-PDO from glycerol by introducing dhaB1, dhaB2, and dhaT genes from the dha operon from C. butyricum 2CR371.5, a strain from our collection of environmental samples and strain.The strain was previously isolated and tested as a 1,3-PDO producer and the results were described in the accompanying article (Dąbrowski et. al, 2012).
Strains E. coli BL21(DE3) and Rosetta2(DE3) (Novagen) were used as hosts for expression of the dhaB1, dhaB2 and dhaT genes.They were grown at 37°C in LB medium containing glycerol (10 g • l -1 ) and eventually also glucose (10 g • l -1 ).The solid medium and liquid medium for non-induced overnight culture did not contain glucose or glycerol.LB medium for E. coli Rosetta2(DE3) was additionally supplemented with 50 μg • ml -1 of chloramphenicol and media for culturing transformants were also supplemented with 50 μg • ml -1 ampicillin.
DNA manipulation.DNA manipulations were carried out according to standard procedures (Sambrook & Russel, 2001) or manufacturer's recommendations.Restriction endonucleases were purchased from Fermentas, T4 DNA ligase was from A&A Biotechnology.To purify DNA after enzymatic reactions DNA Clean-up AX and DNA Gel-out kits (A&A Biotechnology) were used.To isolate plasmid and genomic DNA Plasmid Mini AX and Genomic Mini AX Bacteria Spin (A&A Biotechnology), respectively, were used.
Sequence similarity analysis was done with the BLASTN and BLASTX programs using the server at the National Center of Biotechnology Information (http:// blast.ncbi.nlm.nih.gov/Blast.cgi).
Construction of expression plasmid.Plasmid pUC57 (Fermentas) was used for construction of the expression vector.
For amplification of the dhaB1 and dhaB2 genes as a single fragment from C. butyricum 2CR371.5,DNA primers ForDhaB1SphI 5′-GCG GCG GCA TGC GTG ATT GAA GGA GTA AAA ATG ATA AG-3′ and RevDhaB2SalI 5′-GCG GCG GTC GAC GTA AAG CTA CTA TTA CTC AGC TCC-3′ were used.The obtained PCR product of 3360 bp was cloned into SphI and SalI sites of pUC57 vector which resulted in pUC-dhaB1B2 plasmid.
To clone the dhaT gene, primers ForDhaB2BamHI 5′-GCG GCG GGA TCC AAG GAG ATA AAA GTA ATG AGT AAG G-3′ and RevDhaTKpnI 5′-GCG GCG GGT ACC TTT TAC TTT GAA TCC TTT AAA TAG-3′ were used.The PCR product of 2201 bp was then digested with AflII and KpnI and cloned into pUCdhaB1B2.The resulting plasmid was named pUCd-haB1B2T (Fig. 2).Sequence correctness of the constructed plasmid was confirmed by sequencing (Macrogen) with the use of universal primers M13F-pUC and M13R-pUC (Macrogen).Sequences of specific recognition sites in the primers used are underlined.
The PCR reaction profile was: (i) initial denaturation 94°C for 2 min; (ii) 30 repeats: 94°C for 30 s, 65°C for 30 s, 72°C for 2 min; (iii) final extension 72°C for 5 min.The reaction was performed with the use of 2×PCR Master Mix Plus High GC (A&A Biotechnology).
Cells were grown aerobically for 16 h at 37°C in LB medium containing antibiotics.The preculture was then inoculated (1%) into 100 ml of fresh LB medium with antibiotics, glycerol or glycerol plus glucose.The cultivation was at 37°C until optical density at 600 nm of 0.5 was reached.The culture was then supplemented with 0.5 mM isopropyl-β-d-thiogalactopyranoside (IPTG) and grown without agitation for 2 weeks.To limit the oxygen concentration, the induced cultures were grown in anaerobic flasks.
To test the influence of iron ion on the 1,3-PDO production the liquid medium was supplemented with FeCl 2 up to 100 µM.
Analytical methods.HPLC analyses were performed using a HPLC Agilent 1200 Series system with RID and DAD detectors, a Phenomenex Rezex ROA column (300 × 7.80 mm; 8 microns) with a 3 mM H 2 SO 4 as the eluent (0,6 ml • min -1 ) and the column temperature of 60°C.
For quantitative determination of 3-HPA a colorimetric method described by Krauter et al. (2012) was used.The absorbance was measured at 650 nm (Perkin Elmer Lambda EZ 150).

Analysis of the dha operon
A genomic DNA fragment containing the 1,3-propanediol operon (dha operon) of C. butyricum 2CR371.5 was identified using the CLC Genomic Workbench.It was submitted to GenBank with the accession number JQ346526.The nucleotide sequence of this fragment shows the identity with 99% to the corresponding operon of C. butyricum DSM 2478.We found three distinct open reading frames encoding DhaB1, DhaB2 and DhaT, components of the dha operon of C. butyricum 2CR371.5.DhaB1, comprising of 2,361 nucleotides, encodes a protein of 787 amino acids.The deduced amino acid sequence shows the highest identity (100%) with the amino acid sequence of B 12 -independent glycerol dehydratase from C. butyricum.DhaB2, of 912 nucleotides, encodes a protein of 304 amino acids.The deduced amino acid sequence shows 100% identity with the amino acid sequence of glycerol dehydratase activator from C. butyricum.DhaT, of 1,155 nucleotides, encodes a protein of 385 amino acids.The deduced amino acid sequence shows 100% identity with the amino acid sequence of 1,3-propanediol dehydrogenase from C. butyricum.

Production of 1,3-PDO by E. coli
The growth of all the E. coli strains harbouring the pUCdhaB1B2BT plasmid was much slower comparing 1,3-propanediol production by Escherichia coli with control E. coli strains transformed with pUC57 plasmid.
The HPLC analysis performed two weeks after IPTG induction showed 1,3-PDO concentration of 0.34 g • l -1 in the case of E. coli BL21(DE3) and 3.7 g • l -1 in the case of E. coli Rosetta2(DE3) when glycerol was used as the sole carbon source (Fig. 3).
When a medium containing both glycerol and glucose was used, the concentration of 1,3-PDO two weeks after IPTG induction did not exceed 0.3 g • l -1 for either of the two E. coli strains (Fig. 3).
The yield of the 1,3-PDO production relative to the amount of glycerol consumed was up to 0.028 g per one gram of glycerol for eE. coli BL21(DE3) and up to 0.3 g for Rosetta2(DE3) (Fig. 4).
Simultaneously with 1,3-PDO the amount of 3-HPA was also measured, following induction with IPTG its production increased more than that of 1,3-PDO (Table 1).
Iron ions at concentration up to 100 μM had no effect on 1,3-PDO production (not shown).

DISCUSSION
There are several microorganisms naturally fermenting glycerol to 1,3-PDO, but their large-scale application is limited due to some difficulties.The main problems are (i) accumulation of by-products which are toxic and can inhibit 1,3-PDO production; (ii) some of these microorganisms are human pathogens; (ii) most of the 1,3-PDOproducing bacteria need B 12 vitamin as a cofactor for glycerol dehydratase (DhaB) which increases costs (Nakamura & Whited, 2003;O'Brien et al., 2004).Construction of genetically modified E. coli strains express B 12independent glycerol dehydratase gene could overcome these difficulties.
We constructed a system for heterologous expression in E. coli genes of the glycerol of metabolic pathway from C. butyricum 2CR371.5, from the A&A Biotechnology collection.The cloned genes dhaB1, dhaB2 and dhaT encode a vitamin B 12 -independent glycerol dehydratase, its activating factor and an NADHdependent 1,3-PDO oxidoreductase, respectively.
The first problem we came across was the poor growth of the E. coli strains harbouring the pUCdhaB1B2T plasmid.It can be

Table 1. 3-hydroxypropionaldehyde and 1,3-propanediol production by E. coli Rosetta2(DE3) cells expressing genes of dha operon of C. butyricum 2CR371.5 in LB medium supplemented with glycerol.
The increase of 1,3-PDO and 3-HPA was calculated as the ratio of the product amount at a given time to that at the previous time analysed.explained by the fact that the LB medium could contain a small amount of lactose to induce expression of the cloned genes under the lac promoter at a low level.That in turn would in accumulation of an intermediate (3-HPA), final product (1,3-PDO), or heterologous protein/-s (DhaB1, DhaB2 or DhaT).We suppose the agent which caused the growth inhibition was 3-HPA, because it was proved previously to be toxic to bacterial cells (Hao et al., 2008b).

Days after induction with IPTG
The observed 1,3-PDO production by E. coli Rosetta2(DE3) was about 11-fold higher comparing with E. coli BL21(DE3) (Fig. 3).It can be explained by the fact that E. coli Rosetta2(DE3) contains a plasmid conferring chloramphenicol resistance that supplies rare tRNAs.Consequently, translation is not limited by the codon usage of E. coli (Kane, 1995).Analysis of the predicted amino acid sequences of DhaB1, DhaB2 and DhaT revealed that they contain 14, 12 and 14% of rare codons, respectively.
The low amount of 1,3-PDO produced in our expression system could be correlated with 3-HPA concentration.3-HPA, an intermediate of 1,3-PDO production, is an inhibitor of DhaB1 (Barbirato et al., 1996a;Hao et al., 2008a;2008b).The 3-HPA accumulation is correlated with a higher NAD/NADH ratio and it is known that NAD behaves as a competitive inhibitor of glycerol dehydratase (Barbirato et al., 1996b).The level of 3-HPA in culture medium could be controlled by the substrate (glycerol) concentration, and a lower level of glycerol could prevent 3-HPA accumulating to a high, lethal concentration (Hao et al., 2008).3-HPA accumulates when the first step of glycerol fermentation to 1,3-PDO is faster than the second one, e.g. when the activity of DhaT is lower than that of DhaB1.Such a situation often occurs in polycistronic expression systems where the mRNA lacks internal ribosome entry sites for the next cistrons (Nishizawa et al., 2010).That was not the case in our expression system, as we cloned the whole dha operon with all the sequences between individual genes.However, in a heterologous expression system the expression of individual genes may differ from that in the natural host (C. butyricum).An analysis of DNA sequences upstream of the dha genes has revealed some departures from the consensus Shine-Dalgarno sequence of E. coli.Thus, the recognition of the ribosome binding sites of the C. butyricum dha genes by E. coli translation machinery could be less effective (Makrides, 1996;Mironova et al., 1999;Shultzaberger et al., 2001).However, above the difficulty could probably be overcome by cloning a gene encoding a DhaT isoenzyme more active than DhaT from C. butyricum e.g.YqhD from E. coli, encoding a NADPH-dependent 1,3-PDO dehydrogenase (Emptage et al., 2009).YqhD utilizes NADPH rather than NADH and it is likely that the differences in the cofactor reduced/oxidized ratios contribute to the higher production of 1,3-PDO by this enzyme (Nakamura & Whited, 2003).
Media containing glycerol or glycerol-glucose mixture as the carbon source were tested.E. coli cells used glucose preferentially (not shown), which resulted in no or lower 1,3-PDO production (Fig. 3).Moreover, the dha operon is repressed by glucose (Sprenger et al., 1989).Therfore, when glycerol was used as sole carbon source, the obtained 1,3-PDO amount was 1.1 and 13.2-times higher than in glycerol-glucose medium for E. coli BL21(DE3) and E. coli Rosetta2(DE3), respectively (Fig. 3).
Fermentation by recombinant E. coli strains was performed in anaerobic flasks, without shaking, but the media used were not reduced before.Thus, it is possible that, in such conditions a part of glycerol is dehydrogenated by the glycerol dehydrogenase (DhaD) of E. coli (Sprenger et al., 1989).Consequently, low 1,3-PDO concentration and yield was observed -3.7 g • l -1 and 0.3 g per 1 g of glycerol consumed, respectively, for induced expression in E. coli Rosetta2(DE3) (Fig. 3, 4).On the other hand, culturing of the recombinant E. coli in strict anaerobic conditions failed to produce satisfying results because the strain was not able to grow (not shown).
One way to improve 1,3-PDO production in E. coli could be construction of a double-induced expression system, with a delayed dhaB1 and B2 induction.DhaT may be also cloned under control of a stronger promotor than that of the dhaB1B2 genes.
Is was observed that DhaT from C. freundii expressed in E. coli was more active in the presence of 50 µM Fe 2+ (Daniel et al., 1995).However, there is no such information for its isoenzymes from Clostridium sp.On the other hand, iron limitation causes higher 1,3-PDO production by Clostridium sp., probably by inhibiton of other dehydrogenases involved in the formation of butanol and ethanol (Dabrock et al., 1992;Raynaud et al., 2003;O'Brien et al., 2004).We therefore checked if supplementing the medium with Fe 2+ up to 100 μM could improve 1,3-PDO production, but no such effect was obtained.One of the reasons for this lack of improvement could be the absence of competing dehydrogenases in the heterologous system used.
Several papers describing construction of 1,3-PDOproducing E. coli strains have been published.Emptage et al. (2009) constructed an E. coli strain producing up to 130 g • l -1 1,3-PDO.However, in there expression system glucose was used as the sole carbon and energy source.Moreover, the recombinant E. coli strain harboured genes of the dha operon of K. pneumoniae and so B 12 vitamin had to be added to the medium to activate the glycerol dehydratase.Glycerol was used as the sole carbon source was obtained by others (Skraly et al., 1998) and 6.3 g • l -1 of 1,3-PDO was produced, again for E. coli expressing the K. pneumoniae dha genes.
For comparision, a single-stage culture of C. butyricum with raw glycerol produced up to 35-48 g of 1,3-PDO per litre with a yield of 0.55 g per one gram of glycerol (Papanikolaou et al., 2000;Chatzifragkou et al., 2011).Glycerol fermentation by the C. butyricum 2CR371.5, which was the source of the genes studied here, results in 0.57 g of 1,3-PDO per 1 g of glycerol consumed (Dąbrowski et al., 2012).
The recombinant E. coli strains expressing genes of the dha operon from C. butyricum 2CR371.5 did not give satisfying results due to the low 1,3-PDO production level and long time required for maximal production.We believe it should be possible to obtain much more efficient glycerol fermentation to 1,3-PDO by E. coli expressing genes of the 1,3-propanediol operon from C. butyricum 2CR371.5 if modifications discussed above are introduced.That effort should be worthwhile, because 1,3-PDO is a desired chemical and the costs of its synthesis are still high.