Enhancement of poly(3-hydroxybutyrate-co-3-hydroxyvalerate) accumulation in Arxula adeninivorans by stabilization of production

Background In recent years the production of biobased biodegradable plastics has been of interest of researchers partly due to the accumulation of non-biodegradable plastics in the environment and to the opportunity for new applications. Commonly investigated are the polyhydroxyalkanoates (PHAs) poly(hydroxybutyrate) and poly(hydroxybutyrate-co-hydroxyvalerate) (PHB-V). The latter has the advantage of being tougher and less brittle. The production of these polymers in bacteria is well established but production in yeast may have some advantages, e.g. the ability to use a broad spectrum of industrial by-products as a carbon sources. Results In this study we increased the synthesis of PHB-V in the non-conventional yeast Arxula adeninivorans by stabilization of polymer accumulation via genetic modification and optimization of culture conditions. An A. adeninivorans strain with overexpressed PHA pathway genes for β-ketothiolase, acetoacetyl-CoA reductase, PHAs synthase and the phasin gene was able to accumulate an unexpectedly high level of polymer. It was found that an optimized strain cultivated in a shaking incubator is able to produce up to 52.1% of the DCW of PHB-V (10.8 g L−1) with 12.3%mol of PHV fraction. Although further optimization of cultivation conditions in a fed-batch bioreactor led to lower polymer content (15.3% of the DCW of PHB-V), the PHV fraction and total polymer level increased to 23.1%mol and 11.6 g L−1 respectively. Additionally, analysis of the product revealed that the polymer has a very low average molecular mass and unexpected melting and glass transition temperatures. Conclusions This study indicates a potential of use for the non-conventional yeast, A. adeninivorans, as an efficient producer of polyhydroxyalkanoates.


Background
Currently plastic products cause many problems with disposal and recycling is not sufficient to prevent accumulation. One of the solutions is the production of biodegradable polymers made from renewable substrates. The polyhydroxyalkanoates (PHA) polymers, which are naturally produced by bacteria as a storage material [1] are a well-known examples. These polymers may be modified for a variety of applications from packaging and agriculture through to medical implants and drug delivery devices [2]. PHA may be composed of more than 150 different monomers which give them a variety of physical and chemical properties [3]. The most well-known PHA, poly(hydroxybutyrate) (PHB), has some disadvantages, e.g. it is brittle, stiff and highly crystalline [4]. In contrast, poly(hydroxybutyrate-co-hydroxyvalerate) (PHB-V) has better flexibility, toughness and has lower melting and

Open Access
Microbial Cell Factories *Correspondence: kunzeg@ipk-gatersleben.de 1 Leibniz Institute of Plant Genetics and Crop Plant Research (IPK), Corrensstr. 3, 06466 Gatersleben, Saxony-Anhalt, Germany Full list of author information is available at the end of the article glass transition temperatures, depending on PHV content [5].
Synthesis of PHB-V is carried out by at least three enzymes: β-ketothiolase, (R)-specific NADPH-dependent acetoacetyl-CoA reductase and PHA synthase [6]. This pathway has been primarily investigated in bacteria, e.g. Cupriavidus necator for PHB-V production with range of accumulation level between 50 and 90% of DCW containing up to 24% PHV [1,7]. Additionally yeast and plants have been employed for polymer production [8,9]. Nevertheless, the entire industrial production of PHB-V is currently based on bacteria, which may be associated with certain problems such as phage contamination of Escherichia coli process [10] or possible presence of lipopolysaccharides in the product, which excludes its use in medical applications [11]. On the other hand yeast have a higher contamination resistance, broad substrate spectrum including industrial by-products and cultivation may be carried out in harsh environments, e.g. acidic or high sugar concentration [10]. The majority of yeast studies have been conducted using baker's yeast (Saccharomyces cerevisiae), however, the obtained results were insufficient to compete with bacterial systems. Kocharin et al. [12] showed that a S. cerevisiae strain harbouring PHA pathway genes and additional genes for engineering acetyl-CoA metabolism is able to accumulate 0.25 g L −1 of PHB. In another study, overexpression of targeted PHA synthase to peroxisomes in S. cerevisiae, lead to the synthesis of up to 7% of DCW polymer composed of C 4 -C 8 monomers [13]. Some non-conventional yeast also exhibit PHA synthesis, for example, Kloeckera spp. accumulated 7.03% DCW PHB-V [14] and Pichia pastoris grown on oleic acid was able to produce 1% of medium-chain-length PHA [15]. Recently Li et al. [16] presented a genetically engineered Yarrowia lipolytica, which accumulated PHB up to 10.2% DCW (7.35 g L −1 ) when grown on glucose and acetate, which is the highest level of PHA accumulation in yeast reported to date. However, production strategies in yeast have concentrated only on direct PHA synthesis and optimization of metabolism, and not on stabilization of accumulation. Phasins are a group of low-molecular-weight proteins with amphiphilic properties. In bacteria, where these proteins were originally found, they play a regulatory and stability role during PHA synthesis and cell division [17]. Moreover, C. necator phasin PhaP1 gene, unexpectedly decreased cells stress when overexpressed in a non-phasin strain of E. coli [18].
Another non-conventional yeast, Arxula adeninivorans, was previously used for PHB-V production by Terentiev et al. who described an Arxula strain harbouring PHA pathway genes from C. necator which accumulated 0.019% PHB and 2.2% PHV using controllable ethanol fed-batch fermentation [19]. Since then the A. adeninivorans expression platform has been optimized and successfully used for production of a number of recombinant proteins [20][21][22][23]. Based on the improved platform, novel genetic modification techniques and codon optimization gene synthesis, it was proposed that it would be possible to improve PHB-V production in Arxula. Moreover Arxula does not have endogenous genes responsible for intracellular PHA degradation and because of its oleaginous character, has a high concentration of CoA and NADPH reduction power, which are necessary for PHA synthesis. Finally, the groundwork for this study was our previous research in which demonstrated that A. adeninivorans, with overexpressed thl thiolase from Clostridium acetobutylicum and phaB reductase from C. necator, is able to secret enantiomerically pure (R)-3-HB [24].
In present study we describe the optimized production and stable accumulation of PHB-V copolymer by overexpression of PHA pathways and phasin genes in the yeast, A. adeninivorans.
The wild-type strain, A. adeninivorans LS3, originally isolated from wood hydrolysate in Siberia and deposited as A. adeninivorans SBUG 724 in the strain collection of the Department of Biology of the University of Greifswald [25] was used as a control strain. The auxotrophic mutant, A. adeninivorans G1216 [aleu2 ALEU2::aade2] [26] and double auxotrophic mutant, MS1006 [aleu2 atrp1::ALEU2 aade2::ALEU2] [27] were used as recipient strains. All strains were cultivated at 30 °C, 180 rpm in 50 mL of broth in a 100 mL flask. The medium was either a selective yeast minimal medium supplemented with 20 g L −1 glucose and 43 mM NaNO 3 (YMM-glc-NO 3 ) [28,29] or a non-selective yeast complex medium containing 20 g L −1 glucose (YPD).

Co-substrate feeding
To check an influence of different co-substrates on polymer and copolymer synthesis, several of C-sources were trialled. Different concentrations of ethanol/1-propanol/ propionic acid/valeric acid/sodium propionate were added to cultures after 48 and 96 h of cultivation.

Fed-batch cultivation
Fed-batch cultures were performed in a 5-L bioreactor (Sartorius, Germany) with conditions set to optimize growth and PHB/PHB-V production. The temperature was maintained at 30 °C and a pH of 6.0 was maintained by the addition of 2.5 M NaOH or 1 M H 2 SO 4 . The level of oxygen was varied and maintained by the stirring rate. The culture was started in modified YPD followed by the addition of glucose or ethanol/1-propanol (1:1 v/v) and nitrogen to prevent C-source depletion and maintain the metabolism of the organism. Controlled addition of a silicone-based anti-foam agent (Strunktol SB 304, Schill+Seilacher GmbH, Germany) was employed to prevent foaming.
All variants of the final plasmids were linearized with AscI or SbfI restriction enzymes and separately transformed into the auxotrophic and double auxotrophic strains, A. adeninivorans G1216 and MS1006 (Fig. 1). The latter strain's growth medium was supplemented with 20 mg L −1 tryptophan to maintain growth after transformation (ATRP1 marker not yet complemented).

Xplor2.2
A similar multiple-step cloning procedure was used to prepare Xplor2 The Xplor2.2 expression platform is similar to Xplor2.4 with the only difference being that the auxotrophic marker is tryptophan instead of adenine [31].

Designation
Oligonucleotide sequence Source Stabilization of the yeast transformants was performed by passaging on selective and non-selective media as described by Klabunde et al. [32]. Plasmid DNA isolation and DNA restrictions were performed as described by Wartmann et al. [33].

Gas chromatography/mass spectrometry (GC/MS) analysis
Gas chromatography/mass spectrometry analysis was used to determine the concentration of PHA in dried cells. 3-4 mg of lyophilized cells was subjected to propanolysis as described by Riis and Mai [34] but with modification. 1 mL 1,2-DCE and 1 mL n-propanol-HCl solution (4:1 v/v) was added to cells and incubated at 90 °C for 4 h. After cooling to room temperature the reaction mixture was extracted with 2 mL of triple distilled H 2 O and the lower organic phase was taken for analysis in a Clarus ® 680 GC combined with Clarus ® SQ 8 S MS (PerkinElmer, USA) equipped with Elite-624 column (PerkinElmer, 30 m × 0.5 mm, 1.4 µm). The GC program was set up as follows: initial temperature 80 °C for 5 min, ramping at 10 °C min −1 to 235 °C and held for 10 min.
Commercial PHB-V (Sigma-Aldrich, 12%mol of PHV) was used to construct a standard curve. The analysis was conducted in triplicate and results were analyzed by Tur-boMass 6.1 software.

Polymer extraction and analysis
Poly(3-hydroxy)butyrate co-polymer produced by Arxula was isolated by solvent extraction method [35] with further modification. Yeast were harvested, washed twice with water and processed in a French Press. 30 mg L −1 of trypsin was added to the disrupted cell suspension and incubated in 37 °C for 1 h. The resulting cell debris was freeze-dried overnight and 5 g of lyophilized material was boiled for 6 h in 100 mL of chloroform. The excess of solvent was evaporated making the solution viscous which was then mixed dropwise with 10-times volume cold ethanol to precipitate final product. Finally the polymer was collected on a paper filter (2.7 µm, Whatman, USA) and dried in 55 °C.

Cell imaging
Imaging of PHA granules inside Arxula cells was performed using BODIPY 493/503 (Thermo Fisher, USA), a green fluorescent agent, which binds to neutral lipids and similar compounds. It was shown to be more suitable for PHA staining than Nile Red [36]. The cell material for microscopic analysis was collected after 240 h and twice washed with distilled water. 200 µL of cells suspension was mixed with 20 µL of 0.01% solution of BODIPY 493/503 in DMSO and incubated at room temperature for 5 min. The suspension was then centrifuged and resuspended in an equal volume of distilled water. Freshly prepared material was then analyzed using a confocal laser-scanning microscope (Zeiss LSM 780; Zeiss, Jena, Germany). The BODIPY signal was detected using an argon 488-laser equipped with a 560-615 nm band pass filter.
To enable ultrastructural examination using transmission electron microscopy (TEM), high pressure freezing followed by freeze substitution and resin infiltration were performed according to the method described by Daghma et al. [37]. Ultrathin sections of ~70 nm thickness were prepared as described previously [38] and visualised with a Tecnai Sphera G2 transmission electron microscope (FEI Company, Eindhoven, The Netherlands) set at 120 kV.

Statistics
Two cultures were grown as independent experiments and GC/MS analysis was performed in triplicate. The final results are average values of the data.

PHA shaking flask screening
Overexpression of ß-ketothiolase gene (origins from C. acetobutylicum or C. necator H16) and acetoacetyl-CoA reductase and PHA synthase genes from C. necator H16 in A. adeninivorans led to synthesis of PHA by this yeast. Additionally, the phasin gene from C. necator H16, which has been reported to be a stabilizer of PHA production [41], was overexpressed. The plasmids Xplor2.4-thl-phaB-phaC and Xplor2.2/Xplor2.4-thl/bktB-phaB-phaC-phaP1 were created with each containing all three production-genes and an additional gene to stabilize synthesis and flanked with the strong constitutive TEF1 promoter and PHO5 terminator. The endogenous auxotrophic AADE2 and/or ATRP1 markers, which restore adenine or tryptophan synthesis pathways in A. adeninivorans respectively, were used to select positive transformants. After linearization with AscI (YRC-homologous recombination) or SbfI (YIC-non-homologous recombination) plasmids were transformed into the corresponding auxotrophic mutants, A. adeninivorans G1216 or MS1006 and selected on media that does not contain adenine and/or tryptophan (Table 2).

Cells without phasin gene
As the first screening procedure, strain AAG_thl was tested for PHA production. Before stabilization, measurements of PHA revealed that most of transformants were able to produce PHB after 96 h, with the highest polymer content 7.47% of DCW. On the other hand after stabilization only one of the transformants retained PHB production ability with 1.72% of DCW after 96 h which is only 23.1% of the initial production. The aim of the stabilization procedure is to exclude false positive transformants and also to check the sensitivity of the cells to the polymer. The results showed that the production of PHA by A. adeninivorans has a strong negative influence and only the cells with low or zero polymer production can survive. This effect eliminates these yeast for industrial polymer production.

Cells with phasin gene
To try and counter loss of production ability during the stabilization of the strains, an additional gene-phasin-was expressed. The resulting strains, AAG_thlp and AAG_bktBp, were analysed and the results showed that overexpression of phasin gene had a strong positive influence on PHA synthesis. Most of transformants retained the ability to synthesise PHA and the average loss of production was only 12.0 and 17.3% for AAG_thlp and AAG_bktBp respectively (Fig. 1). The highest content of PHB for AAG_thlp #7 was 9.58% of DCW and 6.71% in AAG_bktBp #14 after 96 h cultivation. Based on these results, these transformants were used in further experiments.

Time-course experiments
Changes of PHA content of the cells are dependent on cultivation time and carbon source. Both factors were explored in simple shake flask experiments without additional feeding. In the first experiment strain AAG_thlp #7 was cultivated in YMM-glc-NO 3 and YPD medium. After 120 h, the polymer content was 13.3% and 18.7% of DCW and 1.02 and 2.61 g L −1 of PHB for YMM and YPD medium respectively. Additionally there was no decrease in polymer level over the cultivation period. Based mainly on the final polymer yield per volume of culture medium, the rich medium was chosen for further experiments. Strains AAG_thlp #7 and AAG_bktBp #14 were compared in rich media without additional C-sources. The two strains differ only in the type of thiolase they have, either thl or bktB. After 240 h, the strain with thl thiolase reached 26.7% of DCW of PHB compared to 19.8% of DCW for strain with bktB thiolase (Table 3). This difference in PHB production was predicted based the findings of Wang et al. [4] and our previous results [24].

Co-substrates
Poly(hydroxyvalerate) was not detected in any of the above experiments and in an attempt to increase PHV synthesis additional substrates were trialled. Four substrates: ethanol, 1-propanol, propionic acid and valeric acid were added at a final concentration of 1% (EtOH, 1-PrOH after 48 h) and 0.1% (PrCOOH, VrCOOH after 48, 72 and 96 h). The results showed that EtOH increases PHB production substantially while a small amount of PHV synthesis occurred when 1-PrOH, PrCOOH and VrCOOH (Table 3; Fig. 2) were added. The maximum level of PHB was obtained for AAG_thlp #7 when ethanol was used (42.9% of DCW of PHB after 240 h of cultivation). On the other hand in the sample with 1-propanol, PHB decreased to 26.0% of DCW but with an additional 2.35% of DCW of PHV. For strain AAG_bktBp #14 the highest content of PHB was also obtained for ethanol (30.5% of DCW), however, because of increased growth on glucose, the highest production was for glucose (4.21 g L −1 ). Use of 1-propanol resulted in cells producing the copolymer (20.3% of DCW) with 22.5%mol of PHV (compared to 7.30%mol for thl strain). Use of propionic and valeric acid for both strains gave lower yields and created problems because of their toxicity and difficult handling.
Sodium propionate was also used as a substrate. For this cultivation the concentration of the substrates were calculated so as to be the same molar carbon level which were 0.45, 0.38 and 0.50% for glucose, 1-propanol and sodium propionate respectively. These substrates were added at 48 and 96 h. The results shown in Table 3 revealed a similar co-polymer content and PHV contribution for both 1-propanol and sodium propionate. However, when the growth behaviour was compared, cells fed with sodium propionate grew more slowly (sodium Table 3 PHA production using different co-substrates  Negative control 0.00 0.00 -propionate is commonly used as a food preservative). Thus, while use of ethanol as a substrate is the best way to increase PHB, 1-propanol increases PHV production.

Double mutant strains
Production of polymers by yeast cells can be dependent on several factors, e.g. C-sources, precursor and cofactor availability and growth conditions. Additionally, gene expression levels also may have an influence on PHB-V synthesis. To increase enzyme levels, a double auxotrophic strain A. adeninivorans MS1006 [aleu2 atrp1::ALEU2 aade2::ALEU2] was transformed twice with the same expression plasmid. The resulting strains, AAMS_thlp and AAMS_bktBp, were tested for PHB-V synthesis using rich media supplemented after 48 and 96 h with ethanol/1-propanol mixture (Fig. 3). After 240 h strain AAMS_bktBp #1 accumulated 52.1% of its DCW as PHB-V (10.8 g L −1 ) of which 12.3%mol was PHV. This was the maximal level of polymer obtained by A. adeninivorans in shaking flasks cultures. In the second strain (with thl thiolase) 43.6% of its DCW (7.80 g L −1 ) was PHB-V of which 4.08% was PHV. For comparison, strains transformed once with the expression plasmid, had lower levels of productivity; 38.0% DCW was PHB-V (7.67 g L −1 ; 3.04%mol PHV) and 26.8% DCW was PHB-V (4.12 g L −1 ; 10.9%mol PHV) for AAG_thlp #7 and AAG_ bktBp #14 respectively (Table 3). Figure 3b shows the trend of PHV accumulation over the period of the experiment. The %mol of PHV for strain with thl thiolase is nearly constant through whole cultivation (except at the first measurement point before EtOH/1-PrOH feeding) while the bktB thiolase strain PHV contribution increased after both supplementation points. This demonstrates that using bktB thiolase and 1-propanol feeding can increase PHV content in the polymer produced by A. adeninivorans.

Bioreactor results
Optimization of culture conditions for improving polymer synthesis was conducted by a controlled fed-batch fermentation. At first, optimized conditions for growth and protein synthesis (aerobic, glucose and nitrogen feeding) was used with strain AAG_thlp #7. After 52 h cells produced only 0.961% of DCW of PHB (0.981 g L −1 ). In similar conditions that were hypoxic instead of aerobic, production was slightly higher with 4.03% DCW PHB (1.58 g L −1 ) after 111 h (longer period of cultivation is due to the slower consumption of glucose). The next step was to use ethanol as an additional carbon source without nitrogen feeding (the same basic medium). Catabolism of ethanol in yeast is an aerobic process and maximal aerobic conditions in the bioreactor were maintained. After 144 h, strain AAG_thlp #7 was able to accumulate up to 10.7% of DCW of PHB (10.7 g L −1 ). To induce synthesis of the PHV fraction, a mixture of equal volumes of ethanol and 1-propanol was added. The three strains selected for PHB-V production were finally grown on the basic modified rich medium with ethanol/1-propanol, without nitrogen feeding and with pO 2 at 40%. As expected, PHV was found in all cultivation conditions (Fig. 3). Strain AAG_thlp #7 accumulated 6.95% of DCW of PHB-V (7.36 g L −1 ) after 113 h but with very low PHV contribution (maximum 1.18%mol after 46 h). For both strains with bktB thiolase, the results were improved with the highest level obtained by strain AAMS_bktBp #1 with 15.3% of DCW of PHB-V (11.6 g L −1 ) after 113 h growth, compared to 11.0% (10.4 g L −1 ) for strain AAG_bktBp #14 at 113 h. Also PHV concentrations were significantly higher for both strains. At the end of the cultivation, 23.1%mol of PHV was found for AAMS_bktBp #1 in contrast to 21.3%mol of PHV for AAG_bktBp #14. Figure 4c shows that, while the maximum accumulation for AAMS_bktBp #1 was 38.3%mol of PHV after 30 h, the concentration decreased until end of the cultivation. However PHV production in AAG_bktBp #14 rose continuously until the end of the cultivation. Polymer synthesis can have a negative influence on growth and the strain with the lowest polymer production had the highest rate of growth (106 g L −1 after 113 h), while in the best PHB-V producer growth rate was the lowest (75.7 g L −1 after 113 h).

Polymer analysis and microscopy imaging
Poly(hydroxybutyrate-co-hydroxyvalerate) co-polymer produced by strain AAMS_bktBp #1 was extracted using an optimized extraction procedure, which resulted in 71.9% polymer recovery with 99.2% purity.
The extracted material consisted of 73.1%wt of PHB and 26.1%wt of PHV. GPC analysis revealed that M N (number average molar mass) and M W (weight average molar mass) values were 8630 and 17,300 respectively which is considered to be very low. The calculated polydispersity index is 2.00 (M W /M N ). The melting point (T m ) and glass transition temperature (T g ) were analysed using the DCS method. The first and second T m was 163.3 and 138.1 °C respectively, which are unexpectedly higher than other PHB-V co-polymers with similar PHV fractions [39]. T g is also higher with a value 55.3 °C compared to −5 °C for a co-polymer with 20%mol PHV. Thermogravimetric analysis allows the thermal degradation temperature to be calculated. Degradation of the co-polymer started at 260 °C with a sharp decrease of M W at 300 °C with a final loss of 99.2% of the M W . These values are similar to other PHB-V co-polymers [40]. The TOF-SIMS study confirmed the results obtained by GC/MS, especially the proportion of PHB and PHV. Furthermore, a trace of polydimethylsiloxane was found which probably come from anti-foam agent using to stop foaming during growth in the bioreactor.
Microscopic analysis of cells using BODIPY 493/503 staining agent was performed. Due to cytoplasmic targeting of PHAs pathway proteins, polymer granules were supposed to be located in cytoplasm. Microscopic images suggested that may be the case (Fig. 5a) and TEM analysis confirmed it (Fig. 5b). PHAs in Arxula exist as many granules, in contrast to the few granules seen in bacteria, however, the granule size is similar in yeast and bacteria [17].

Discussion
The overexpression of the genes, ß-ketothiolase, acetoacetyl-CoA reductase and PHAs synthase in A. adeninivorans, made it able to synthesize the PHB polymer and PHB-V co-polymer. The types of thiolase and reductase genes that were chosen were based on our previous results [24] but another thiolase-bktB gene was investigated for an increased proportion of the PHV fraction [4]. A lack of an endogenous PHA degradation system allows accumulation of polymer during prolonged incubation experiments and resulted in high levels of polymer in the cell.
Arxula adeninivorans was previously used for PHA production but the results were poor [19]. Our initial research showed that lack of stability in the transformants resulted in A. adeninivorans losing most of its capacity to synthesise the polymer. This phenomenon may be, in theory, explained by increased cell stress and the inability of cells to multiply. To counter this effect, the phasin gene, which is one of the factors for PHA granule stabilization in bacteria [41] and is known to decrease of cell stress in E. coli [18], was introduced. Overexpression of the phasin gene resulted in the stabilization of the production of the polymer, even during extended cultivation.
Results show that rich media (YPD) is preferable for PHA synthesis, possibly due to higher protein content and the availability of precursors and cofactors. Without co-substrates, Arxula is able to produce only PHB and to investigate the development of co-polymerization of PHB-V, different C-sources were investigated as cosubstrates. Ethanol proved to be the best co-substrate for PHB accumulation and has been shown to be the preferred substrate for synthesis of secondary metabolites in yeast due to its direct conversion to acetyl-CoA in the cytoplasm [42]. Furthermore other co-substrates also allowed Arxula to produce the PHB-V co-polymer. Although sodium propionate resulted in the highest proportion of PHB-V as a % of cell weight, the toxicity of the molecule reduced the DCW. The use of 1-propanol as the co-substrate gave the highest yield in g L −1 which led to its use in all subsequent experiments. Moreover, Arxula is able to use C 2 -C 6 alcohols and convert them to relevant CoA compounds which may theoretically led to synthesis of medium-chain-length PHAs [43]. Thus an equal mixture of ethanol and 1-propanol was chosen to maximise co-polymer production.
The strains used in our research differed only in the type of thiolase they contained. The strain with thl thiolase was the superior polymer producer, however because the product of this thiolase is acetoacetyl-CoA, which is precursor of the 3HB monomer, the resulting copolymer contained low levels of PHV (up to 8.38%mol). The strain with the bktB thiolase, produced β-ketovaleryl-CoA, which although it accumulated at a lower level, had almost 3-times as much PHV. In the strains transformed twice with the same expression plasmid, the strain with bktB thiolase accumulated the most PHB-V co-polymer. While the PHV fraction was half that in the single transformation strains, it was consistently 3-times higher than in the thl thiolase strain. The production of PHB-V to over 50% of DCW for the double transformed strain may be explained by an increase in the enzymes present (which was the aim of double transformation).
In theory, controllable fed-batch fermentation should increase production by for example, better aeration or pH control. However our study on Arxula cells gave disparate results. While cells were able to accumulate more than 50% PHB-V of DCW in the shake flasks experiment, growth in the bioreactor resulted in around 15% PHB-V of DCW, however the final polymer level calculated in g L −1 was 7.41% higher in the bioreactor experiments. This effect revealed a correlation between growth behaviour and polymer accumulation. Using the same cultivation conditions, strain AAG_thlp #7 accumulated 6.95% DCW of PHB-V with 106 g L −1 of total DCW in contrast to 15.3% DCW of PHB-V with 75.7 g L −1 of total DCW for AAMS_bktBp #1. Thus the highest production of PHB-V by Arxula seems to be a compromise between the level of growth and the accumulation process.
The study of the extracted polymer provided interesting information. Material accumulated by Arxula has a low to very low average molecular mass compared to other microbially produced PHAs which are up to 150times heavier [44]. The low-molecular-weight of the polymer cannot be explained by the size of the granules due to their similarity in yeast and bacteria (Fig. 5).
The low molecular weight has an influence on the other properties. Most unusual is the glass transition temperature, which is around 60 °C higher than that reported for other polymers with similar PHV fractions. This means that products made of the Arxula polymer will remain brittle at higher temperature, which is a property that undesirable in some applications, however the melting temperature is higher than usual, making this material more robust at higher temperatures. An additional advantage of a low molecular weight polymer may be faster biodegradation [45]. These differences may come from polydimethylsiloxane, a compound that is present in the polymer, or from other unknown impurities. It is also possible that the extraction method may have an influence on the polymer properties and this needs to be investigated.

Conclusion
In this paper we have demonstrated that the non-conventional yeast, A. adeninivorans, is able to produce 52.1% of DCW of PHB-V (10.8 g L −1 ) with 13.2%mol of PHV in shaking flask and 15.3% of DCW of PHB-V (11.6 g L −1 ) with 23.1%mol of PHV in fed-batch bioreactor cultivation, which are the highest amounts ever seen in yeast. These results show that A. adeninivorans could be a useful host organism for the production of PHAs. Future work to increase the production of the polymer will include shifting the flux of acetyl-CoA in direction of PHA synthesis by genetic engineering and further optimization of fed-batch growth conditions and scaling up of production process.