Integrated bioprocess for the stereospecific production of linalool oxides from linalool with Corynespora cassiicola DSM 62475
- First Online:
- Cite this article as:
- Bormann, S., Etschmann, M.M.W., Mirata, M. et al. J Ind Microbiol Biotechnol (2012) 39: 1761. doi:10.1007/s10295-012-1181-2
- 300 Views
Linalool oxides are of interest to the flavour industry because of their lavender notes. Corynespora cassiicola DSM 62475 has been identified recently as a production organism because of high stereoselectivity and promising productivities [Mirata et al. (2008) J Agric Food Chem 56(9):3287–3296]. In this work, the stereochemistry of this biotransformation was further investigated. Predominantly (2R)-configured linalool oxide enantiomers were produced from (R)-(−)-linalool. Comparative investigations with racemic linalool suggest that predominantly (2S)-configured derivatives can be expected by using (S)-(+)-configured substrate. Substrate and product inhibited growth even at low concentrations (200 mg l−1). To avoid toxic effects and supply sufficient substrates, a substrate feeding product removal (SFPR) system based on hydrophobic adsorbers was established. Applying SFPR, productivity on the shake flask scale was increased from 80 to 490 mg l−1 day−1. Process optimisation increased productivity to 920 mg l−1 day−1 in a bioreactor with an overall product concentration of 4.600 mg l−1 linalool oxides.
KeywordsBiotransformationLinaloolLinalool oxideSubstrate feeding product removalHydrophobic adsorber
Stereochemistry can have a significant impact on the olfactorial properties of chiral molecules. According to Wüst and Mosandl , the olfactorial reception of linalool oxides depends solely on the configuration of the stereocenter at the C2-position. While (2R)-configured linalool oxides have an earthy, slightly leafy smell, (2S)-configured molecules are perceived as floral, creamy and sweet. Odour reception is equal for both furanoid and pyranoid linalool oxides. Thus, production of pure (2R)- or (2S)- configured products is desired, since they have distinctly different smells.
The postulated microbial biosynthetic pathway suggests that pure (2R)- or (2S)- configured products can be produced from enantiomerically pure (S)-(+)- or (R)-(−)-linalool, respectively (Fig. 1). Fortunately, linalool is available from several natural sources, often with high enantiomeric excess. The (S)-(+)-form can be found in coriander oil (Coriandrum sativum) with an enantiomeric excess of 60–70 %. Linalool from lavender (Lavandula angustifolia) has an even higher enantiomeric excess of more than 98 % (R)-(−) [1, 2, 27]. This is why natural (R)-(−)-linalool (>80 %) is available for as little as 90 US$ kg−1 (Sigma–Aldrich online catalogue Germany, accessed 17 October 2011) and might be even cheaper when obtained in bulk. In contrast, natural linalool oxide has a price of about 750 US$ kg−1 (Advanced Biotech, Paterson, NJ, personal correspondence). The value of enantiopure natural linalool oxide would be even higher. This illustrates the market potential of this biotransformation.
In order to produce natural linalool oxide according to US and EU legislation, natural linalool has to be oxidised by biological means. Several Aspergillus niger, Botrytis cinerea and Streptomyces albus strains are capable of performing this biotransformation [5, 7, 9]. Unfortunately, all these microorganisms oxidise linalool at a slow pace. Gatfield et al.  reported much higher volumetric productivities when linalool was oxidised in the presence of a Candida antarctica lipase, but did not describe the enantiomeric composition of the product.
Recently, Mirata et al.  identified Corynespora cassiicola DSM 62475 to be capable of performing this biotransformation. Compared to several A. niger and B. cinerea strains, product was formed at a higher rate. Moreover, C. cassiicola performed this biotransformation in a highly stereospecific manner, producing preferentially four of eight isomers.
While linalool inhibited growth of all investigated microorganisms at higher concentrations, C. cassiicola was one of the most tolerant species. No growth inhibition occurred up to a concentration of 150 mg l−1 . For these reasons, C. cassiicola DSM 62475 is a promising microorganism for industrial scale oxidation of linalool to linalool oxide.
Both linalool and linalool oxide are hydrophobic substances with a mildly hydrophilic character (log P values 3.5 and 2.4, respectively . Substances with log P values between 1 and 5 are generally considered as cytotoxic . While the toxicity of linalool has already been shown, linalool oxide is most likely to be cytotoxic as well. To reduce inhibitory effects in microbial biotransformations, several techniques can be applied. Substrate can be added by a feedback controlled pump to maintain a subinhibitory level. As online measurement of terpene concentrations is difficult, feedback control is unfeasible for the time being. To prevent accumulation of toxic product in the immediate aqueous microenvironment of the cells, in situ product removal (ISPR) techniques have been applied successfully [4, 10, 20]. Two-phase liquid–liquid and liquid–solid systems, as well as membrane separation are the most commonly used ISPR techniques. These methods work most successfully when substrate and product have considerably different physicochemical properties (e.g. charge, volatility, considerable log P change). Otherwise there is a risk of removing substrate from the process.
However, in processes where substrate and product have similar physicochemical properties, a combined substrate feeding product removal (SFPR) approach is favourable . To achieve this effect, adsorber is preloaded with substrate and introduced into the process. Substrate will desorb into the aqueous phase until equilibrium is established. In the presence of biocatalyst, substrate will be transformed to product, which is adsorbed. To reestablish equilibrium conditions, more substrate will desorb into the liquid phase, thus constantly supplying substrate according to the metabolic demand of the catalyst and removing toxic product from the aqueous phase [6, 14–17, 25].
While the same can be done with a two-phase liquid–liquid system [13, 24], the use of hydrophobic liquids has disadvantages. Organic solvents are often toxic to microorganisms while hydrophobic adsorbers are usually biocompatible. Moreover, many organic phases tend to build emulsions, especially when organisms produce surfactants, which complicates downstream processing [22, 28]. Adsorbers, in contrast, can be separated easily from the culture broth by filtration, eluted with a suitable solvent and reused.
This work will further elucidate the stereochemistry of the biotransformation by investigating the conversion of linalool to linalool oxides with C. cassiicola DSM 62475. The substrates chosen are racemic linalool and the commercially available (R)-(−)-linalool. Application of a combined SFPR approach based on hydrophobic adsorbers, transfer from shake flask to bioreactor conditions and process optimisation will be described.
Materials and methods
Strain and maintenance
Corynespora cassiicola DSM 62475 was obtained from DSMZ (Braunschweig, Germany) and grown on malt extract agar (MEA) at 24 °C .
The experiments were executed with the commercially available (R)-(−)-linalool and racemic linalool. All chemicals were purchased from Sigma-Aldrich (Schnelldorf, Germany), Fluka (Ulm, Germany) or Carl Roth (Karlsruhe, Germany). Purity of (±)-linalool and racemic linalool oxide was >97 %, purity of (R)-(−)-linalool was >98.5 %. Oxide standards for stereochemical discrimination per GC-FID were kindly provided and synthesised by the Mosandl group (Johann Wolfgang Goethe University Frankfurt/Main, Germany) according to Weinert et al.  and Askari and Mosandl .
Preparation of preculture
Spore suspension (1 % v/v) and 200 ml malt yeast broth (MYB) (both described by Mirata et al. ) were combined in a 1,000 ml Erlenmeyer flask and incubated at 24 °C and 200 rpm. All shake flask experiments were executed at an amplitude of 25 mm.
Determination of linalool/linalool oxide toxicity
Glass vials for solid phase microextraction (40 ml) were filled with 9.5 ml MYB and inoculated with 500 μl of a 7 day old preculture (homogenised with an Ultra-Turrax, IKA, Staufen, Germany, for 30 s). Linalool and linalool oxide were added from stock solutions in ethanol (3 g l−1 for 0.5 and 1 mM, 30 g l−1 for 2, 4 and 8 mM). Cell dry weight was determined after inoculation and after 24 h.
All adsorbers were conditioned by washing with about two volumes of methanol and deionized H2O and dried for 24 h at 105 °C. All weights given are dry weights.
Adsorption isotherms of linalool
To determine adsorption isotherms, 20 mg adsorber were added to 20 ml of an aqueous linalool solution (100–1,000 mg l−1) and incubated in a 20 ml screw cap vial at 24 °C. After 24 h, a sample was taken and the remaining aqueous linalool concentration was determined by GC-FID analysis. Investigated adsorbers were Amberlite XAD2, XAD4, XAD7 and XAD16 (Rohm and Haas, Philadelphia, PA), Lewatit VP OC 1163 (Lanxess, Leverkusen, Germany) and Diaion HP-2MG (Mitsubishi Chemical, Tokyo, Japan).
Linalool oxide affinity
Adsorber (0.1 % w/v) was added to 20 ml of an aqueous solution containing both linalool and linalool oxide (500 mg l−1 each) and incubated in a 20 ml screw cap vial at 24 °C for 24 h.
Fed-batch biotransformation on shake flask scale
In a 2,000 ml Erlenmeyer flask, 50 ml of a 7-day-old homogenised preculture was added to 450 ml MYB and cultivated at 24 °C and 200 rpm. Every 24 h, 150 mg l−1 linalool (30 g l−1 stock solution in ethanol) and 5 g l−1 glucose (750 g l−1 stock solution) were added before a sample was taken. Both quantitative and stereochemical analysis were done by GC-FID.
SFPR biotransformation on shake flask scale
Lewatit VP OC 1163 (dried for 1 h at 105 °C to prevent contamination) was loaded with 250 mg linalool (load 0.56 g g−1) in 45 ml MYB for 24 h in a 300 ml Erlenmeyer flask. A 7-day-old homogenised preculture (10 % v/v) was added to start the biotransformation (24 °C, 200 rpm). Glucose and terpene concentrations were measured every 24 h and glucose was accordingly adjusted to 10 g l−1.
Biotransformation in a small scale bioreactor
The biotransformation was carried out in a 4 × 1 l parallel bioreactor fedbatch-pro (DASGIP, Jülich, Germany), where 450 ml MYB were inoculated with 50 ml of a 7-day-old homogenised preculture. The reactor was operated at 24 °C, 500 rpm and aerated at 0.3 vvm with air. The air was saturated with linalool prior to injection into the reactor by passing it through a fritted washing flask filled to one-third with linalool. Glucose was added at a rate of 15 g l−1 day−1. For SFPR-cultures, 8.9 g adsorber were loaded with 5 g linalool in 100 ml MYB (load 0.56 g g−1) for 24 h, filtered, added to the bioreactor and stirred for 1 h at 500 rpm prior to inoculation to ensure equilibrium conditions. For fed-batch-cultures, 150 mg l−1 linalool were added every 24 h.
Elution of adsorber
The adsorber was separated from the culture broth by filtration (TE 38 membrane filter, 5 μm, Whatman Schleicher & Schuell, Maidstone, UK). It was eluted with 10 and 50 ml ethanol (shake flask and bioreactor scale, respectively) for 1 h. This procedure was repeated five times. The eluates were combined and analysed by GC-FID. Terpene concentrations given are always based on culture volume.
Sample preparation for GC analysis
Aqueous samples were extracted 1:2 with MTBE. The extract was dried over sodium sulfate and analysed by GC-FID. Prior to extraction, 2-octanol (5 % v/v of a 2 g l−1 stock solution in ethanol) was added as an internal standard. Ethanol based samples were analysed without further preparation after addition of internal standard.
Sample Analysis by GC-FID
Linalool and linalool oxides were analysed by gas chromatography (GC 17A equipped with FID, Shimadzu, Tokyo, Japan). For quantification, a DB-WAXetr column (30 m × 0.25 mm × 0.25 μm, Agilent, Santa Clara, CA) was used. Parameters were as follows: carrier gas, helium; split, 1:41, column flow, 0.8 ml min−1; temperature: 120 °C (7 min), to 250 °C at 30 °C min−1,250 °C (3 min). For stereochemical analysis, a Chiraldex B-DM (30 m × 0.25 mm × 0.12 μm, Sigma–Aldrich, Schnelldorf, Germany) was used with the following parameters: carrier gas, helium; split, 1:20, column flow, 1.1 ml min−1; temperature: isothermal 95 °C (30 min). Elution order on achiral DB-WAXetr was: trans-furanoid linalool oxide, cis-furanoid linalool oxide, linalool, trans-pyranoid linalool oxide, cis-pyranoid linalool oxide as determined by comparison with reference compounds. On the chiral Chiraldex B-DM column, the elution order was: trans-(2R,5R)-furanoid linalool oxide, trans-(2S,5S)-furanoid linalool oxide, cis-(2R,5S)-furanoid linalool oxide, cis-(2S,5R)-furanoid linalool oxide, (R)-(−)-linalool, (S)-(+)-linalool, trans-(2S,5R)-pyranoid linalool oxide, trans-(2R,5S)-pyranoid linalool oxide + cis-(2S,5S)-pyranoid linalool oxide (coelution), cis-(2R,5R)-pyranoid linalool oxide as determined by comparison with reference compounds. To quantify terpene concentrations, a response factor in relation to 2-octanol was determined.
Concentration of coeluted trans-(2R,5S)-pyranoid linalool oxide and cis-(2S,5S)- pyranoid linalool oxide was determined by substraction of trans-(2S,5R)-pyranoid linalool oxide or cis-(2R,5R)-pyranoid linalool oxide concentration from the respective total pyranoid diastereomer concentration, determined by GC analysis on achiral DB-WAXetr.
Cell dry weight determination
Culture broth samples (10 ml) were filtered through a preweighed 0.45 μm cellulose nitrate filter (Whatman, Maidstone, UK). The filter was dried overnight at 105 °C. Biomass was determined gravimetrically.
Glucose concentrations were determined enzymatically (2,700 Select, YSI, Yellow Springs, OH).
Results and discussion
Stereochemistry of biotransformation products
Kinetics of fed-batch biotransformation
Substrate and product toxicity
Since linalool and its oxides are structurally and therefore physicochemically similar, not only substrate inhibition but also product inhibition may interfere with the biotransformation. Hence, the influence of elevated substrate and product concentrations on the growth was investigated.
Nonetheless, 150 mg l−1 linalool was chosen as a suitable concentration for further experiments since it was in the range of the IC50 value. Lower concentrations might have been difficult to control due to the high volatility of the substrate.
It became apparent that linalool had to be dosed carefully to achieve constant substrate supply while maintaining low concentrations. Furthermore, accumulating product had to be removed constantly from the process to prevent growth inhibition. For this reason, a combined substrate delivery and product removal system based on hydrophobic adsorbers had to be established.
Screening of adsorbers
Moreover, the affinity of linalool oxide in comparison to linalool towards the adsorbers was explored by incubating the adsorbers in a solution with similar amounts of both linalool and linalool oxide (data not shown). Amberlite XAD2, XAD7 and Diaion HP20MG adsorbed only linalool. Amberlite XAD4, XAD16 and Lewatit VP OC1163 adsorbed both terpenes, but the amount of adsorbed linalool was always more than three times the amount of linalool oxide. Interestingly, the mildly hydrophilic methacrylate adsorbers (Diaion HP20MG, Amberlite XAD7) did not perform better at adsorbing the more hydrophilic linalool oxide compared to the unpolar adsorbers with a styrene/divinylbenzene matrix.
Since Lewatit VP OC 1163 showed superior adsorption characteristics for linalool and similar adsorption characteristics for linalool oxide compared to Amberlite XAD4 and XAD16, it was chosen for use in a SFPR-bioprocess.
Application of SFPR in the bioprocess
Nonetheless, product formation could be observed. After 1 day, the aqueous linalool concentration had decreased from 250 to 40 mg l−1 and an aqueous linalool oxide concentration of 480 mg l−1 was determined. The aqueous molar product concentration after 24 h exceeded the initial aqueous substrate concentration. This clearly indicated that substrate from the adsorber had desorbed into the liquid phase. After 3.7 days no more substrate was detected in the aqueous phase. Elution of the adsorber showed that 1,200 mg l−1 linalool oxide had been adsorbed during the biotransformation, which was twice the amount of product in the aqueous phase. Thus, it was possible to produce 1,800 mg l−1 linalool oxide in 3.7 days, which corresponds to a productivity of 486 mg l−1 day−1. This accounts for a molar yield of 33; 28 % of the substrate was recovered from the resin and 39 % was lost, most likely due to evaporation. Compared to the fed-batch approach, volumetric productivity was six times higher.
Because of differences in mass transfer, transition from shake flask to bioreactor can have a significant impact on a bioprocess. Preliminary experiments suggested that the volatility of linalool may limit substrate supply in an aerated bioreactor. Stripping experiments with an aeration rate of 0.3 vvm showed that more than 500 mg l−1 linalool could be blown out of the culture broth within 1 day. In contrast, less than 2 % of 300 mg l−1 linalool oxide were lost over the course of 3 days. To compensate for the loss of substrate via the air stream, the air supply was saturated by passing it through a linalool filled fritted washing flask. The bioprocess was carried out with SFPR and conventional fed-batch substrate supply in parallel (Fig. 6b, c). For both setups, a similar increase in aqueous linalool oxide concentration was observed. For the SFPR setup, the linalool concentration decreased from an initial 200 to 14 mg l−1 over the course of 5 days. For the fed-batch-setup, substrate concentrations (measured 24 h after every substrate addition) varied between 20 and 60 mg l−1. The constant presence of substrate in this setup is most likely due to substrate entry through aeration. After 5 days, the biotransformation was stopped and the SFPR resin was eluted. In addition to the 1,000 mg l−1 linalool oxide dissolved in the aqueous phase, 3,600 mg l−1 product were recovered from the adsorber. Unfortunately, linalool did not desorb completely into the aqueous phase. About 35 % linalool was eluted at the end of the process. The incomplete desorption of substrate and adsorption of product is understandable since the hydrophobic adsorber has a higher affinity for the more hydrophobic compound. Once again, it becomes obvious that the development of high specific area, mildly hydrophilic adsorbers is important to improve ISPR and SFPR application in microbial biotransformations, as outlined by Straathof . Nonetheless, 42 % of the substrate was oxidised over the course of 5 days, which corresponds to a productivity of 921 mg l−1 day−1. In comparison to the fed-batch process, this is an increase in productivity by factor 4.6.
Microbial oxidation products of (R)-(−)-linalool by C. cassiicola DSM 62475 are predominantly trans-(2R, 5R) furanoid and trans-(2R, 5S) pyranoid linalool oxides. Comparison with the results of the biotransformation of (±)-linalool shows that the corresponding (2S)-configured linalool oxides can be expected if pure (S)-(+)-linalool is fed. This biotransformation therefore allows stereospecific access to the desired pure products with uniform olfactorial properties. Application of a liquid–solid two-phase SFPR technique greatly reduced the toxic effects of substrate and product. At the same time, continuous substrate supply and facilitated downstream processing has been achieved, resulting in the highest product concentrations and productivities reported so far for this microbial biotransformation.