Optimization of limonene biotransformation for the production of bulk amounts of α-terpineol

Bioresource Technology 294 (2019) 122180

Contents lists available at ScienceDirect

Bioresource Technology journal homepage: www.elsevier.com/locate/biortech

Optimization of limonene biotransformation for the production of bulk amounts of α-terpineol

T

⁎

Gustavo Molinaa,b,c, , Marina G. Pessôaa, Juliano L. Bicasa, Pierre Fontanillec,d, Christian Larrochec,d, Gláucia M. Pastorea Laboratory of Bioflavors, Department of Food Science, School of Food Engineering – University of Campinas, Campinas, São Paulo, Brazil Laboratory of Food Biotechnology, Food Engineering, Institute of Science and Technology – UFVJM, Diamantina, Minas Gerais, Brazil c Université Clermont Auvergne, Institut Pascal, TSA 60026, F-63178 Aubière cedex, France d CNRS, UMR 6602, IP, F-63178 Aubière cedex, France a

b

G R A P H I C A L A B S T R A C T

A R T I C LE I N FO

A B S T R A C T

Keywords: Aroma Bioflavors Limonene Sphingobium sp. Response surface methodology

The biotransformation of R-(+)-limonene into high concentrations of R-(+)-α-terpineol by Sphingobium sp. was investigated in order to optimize the main process variables (pH, biocatalyst concentration, substrate concentration, temperature and agitation). This strategy comprised the screening of variables by a Plackett-Burman design followed by a Central Composite Design. The statistical analysis showed that the optimal α-terpineol production were at 28 °C and pH 7.0, with a limonene concentration of 350 g/L of organic phase agitation of 200 rpm and a biocatalyst concentration of 2.8 g/L of aqueous phase (OD600 = 8). Further trials showed that the R-(+)-α-terpineol concentration was higher (240 g/L after 96 h) when using a ratio of 1:3 (v.v−1) of organic:aqueous phases. However, the total production and yield (in terms of biomass) of α-terpineol would be maximized for an aqueous:organic ratio of 1:1. The experimental design optimization adopted herein was an effective tool for this type of study.

1. Introduction α-Terpineol is one of the most commercially important monoterpene alcohols in the flavor industry. This compound has an aroma

threshold of 280 to 350 ppb, and its estimated annual consumption is approximately 9.2 tons (Burdock and Fenaroli, 2010). R-(+)-α-terpineol has a floral typically lilac odor, while its S-(−) couterpart has a coniferous odor character (Boelens et al., 1993). α-Terpineol is found in

⁎ Corresponding author at: Laboratory of Bioflavors, Department of Food Science, School of Food Engineering – University of Campinas, Campinas, São Paulo, Brazil. E-mail address: [email protected] (G. Molina).

https://doi.org/10.1016/j.biortech.2019.122180 Received 28 June 2019; Received in revised form 17 September 2019; Accepted 18 September 2019 Available online 21 September 2019 0960-8524/ © 2019 Elsevier Ltd. All rights reserved.

Bioresource Technology 294 (2019) 122180

G. Molina, et al.

2010a). R-(+)-Limonene (SAFC, ≥ 93% purity), dissolved in soybean oil (purchased in local market) as organic phase, was used as substrate. 1-Pentanol (Sigma-Aldrich, ≥99% purity) was employed as internal standard. Ethanol (Ecibra, 99.5% purity) was used to extract the aroma products from soybean oil (organic phase). The α-terpineol standard (SAFC, ≥96% purity) was used for constructing the calibration curve (quantification).

a large number of essential oils, such as the oils of Cupressaceae, Pinaceae, and Lavandin; it also exists in many other leaves, herbs, and flowers. This monoterpene alcohol may also arise during the fractional distillation of pine oils. Nevertheless, it is commonly synthesized by acid hydration of α-pinene or turpentine followed by partial dehydration (Burdock and Fenaroli, 2010; Surburg and Panten, 2006). Its R(+)- enantiomer was the main enantiomer in mango flavor (ee 37.2%) while S-(−)-α-terpineol was the predominant (ee 78%) in litchis flavor and a racemic mixture was found yellow passion fruit (Werkhoff et al., 1993). The traditional commercial uses of α-terpineol include household products, cosmetics, pesticide and flavor preparations (Bauer et al., 2001), but the increasing discoveries on its bioactivities, such as antioxidant, anticancer, antinoiceptive, anticonvulsant and sedative activities (Khaleel et al., 2018), may potentially open new markets in the future (de Oliveira Felipe et al., 2017). The microbial biotransformation of limonene, α-pinene, or β-pinene has also been used to produce αterpineol (Bicas et al., 2008a.; Sales et al., 2018). Various authors have reported that microorganisms are able to convert limonene into α-terpineol, especially Penicillium digitatum (Tan and Day, 1998a,b; Tan et al., 1998; Adams et al., 2003), Fusarium oxysporum (Molina et al., 2015) and other fungi (Siddhardha et al., 2012; Lee et al., 2015; Bier et al., 2017). However, one of the most viable biotransformation processes described so far reached almost 130 g/L of α-terpineol from limonene in a biphasic medium comprising of vegetable oils as organic phase and the bacterium Sphingobium sp. as biocatalyst (Bicas et al., 2010a). Classical optimization methods consist of varying the parameters one at a time while maintaining the other variables constant. This strategy is usually time-consuming, requires a large number of experiments, does not consider whether the interactions between factors affect the reaction, and rarely aids full understanding of the process (Sen and Swaminathan, 1997). The Plackett–Burman design is a screening approach that helps to statistically select the significant variables of numerous factor-experiments, aiming to reduce the number of trials in the final design. The central composite design is a statistical methodology that analyzes how the studied variables and their interaction impact a process. This technique culminates in the proposal of a mathematical model that describes the behavior of the analyzed factors and establishes their optimal values (Rodrigues and Iemma, 2014). Some reports of such approach are available in literature for optimizing the production of natural aroma compounds (Çelik et al., 2004; Melo et al., 2005), including the biotransformation of R-(+)-limonene into R(+)-α-terpineol (Bicas et al., 2008b; Rottava et al., 2011). The main incentive for using an organic phase in terpene biotransformations is associated to the improvement of the biocatalyst performance, since this strategy decreases the volatility and toxicity of both substrate and product towards the microorganism, besides facilitating the recovery of both product and biocatalyst (De Carvalho and Da Fonseca, 2006). Considering that Sphingobium sp. produces high amounts of R(+)-α-terpineol from R-(+)-limonene even in non-optimized conditions (Bicas et al., 2010a) this work aimed at defining how the medium composition (pH) and the cultivation conditions (temperature and agitation, and substrate and inoculum concentration) affect α-terpineol production. Therefore, applying a Plackett–Burman matrix with 12 assays (PB-12) for the variables screening, followed by the central composite design methodology, the ideal conditions for such process could be defined.

2.2. Pre-culture preparation Three loopfuls of a 48-hour-old culture from a Petri dish were transferred to a 500-mL conical flask containing 0.68 g of glucose, 0.17 g of (NH4)2SO4, 3.4 mL of Hutner solution (Fontanille et al., 2002), 6.8 mL of solution A (5.2 g of K2HPO4 and 6.62 g of KH2PO4 in 200 mL of distilled water) and 159.8 mL distilled water. This pre-culture was incubated at 30 °C and 200 rpm for 24 h, to reach an optical density at 600 nm (OD600) close to 5.0. OD600 was correlated to biomass concentration according to the following equation (Bicas et al., 2010a): Biomass (g/L) = 0.35 × OD600. 2.3. Biocatalyst production The biocatalyst was produced in a bioreactor with a working volume of 4.8 L (Bioflo 310, Bioflo & Celligen, New Brunswick, USA), using 170 mL of the pre-culture as inoculum, 4 L of distilled water, 170 mL of solution A, 85 mL of Hutner solution, 4.25 g of (NH4)2SO4, 212.5 mL of n-hexadecane, and 15 g of R-(+)-limonene (70 g/L, organic phase). Temperature, agitation and aeration were kept at 30 °C, 500 rpm, and 0.5 slpm, respectively. The pH was set at 6.5; the OD600 of the liquid phase and the composition of the organic and aqueous phases (GC-FID, see Section 2.7) were monitored periodically (Bicas et al., 2010a). After 48 h, the biomass was recovered from the bioreactor by centrifuging the culture medium at 10,000 g for 10 min (5 °C). The supernatant was eliminated, and the resulting biomass was re-suspended in phosphate buffer 20 mM pH 7.0 to reach three different optical densities (OD600 = 8, 13,18). Those biomass suspensions were kept at −18 °C until its use. 2.4. General biotransformation procedure The concentrated biomass (biocatalyst) at a defined optical density and a certain volume of organic phase were transferred to a 250-mL conical flask with 50 mL of total working volume. The substrate was then added, to reach the final concentration per liter of organic for each trial. The flasks were incubated under varying temperature and incubation conditions. The detailed conditions of each assay will be given in the following section and may also be seen on Tables 1–4. Blank experiments (without limonene addition) excluded the de novo synthesis of α-terpineol by Sphingobium sp. and control experiments (without biomass), at different pH values, excluded hypothesis of formation of αterpineol by limonene abiotic oxidation. 2.5. Optimization experiments 2.5.1. Plackett-Burman design A Plackett–Burman design (Rodrigues and Iemma, 2014) with 12 experiments (PB-12) and three center points (Table 2) was run to estimate the experimental error and select the main parameters involved in R-(+)-limonene biotransformation into R-(+)-α-terpineol using Sphingobium sp. The center points for this screening design were: substrate concentration of 120 g per liter of organic phase, biomass concentration of 4.55 g per liter of aqueous phase (DO600 = 13), pH 7 and incubation at 30 °C and 150 rpm. The other conditions followed the combination required for each assay (Table 2). For biotransformations at the extreme pH values, the biomass was resuspended on acetate buffer 20 mM pH 5.3 or on Tris-HCl buffer 20 mM pH 8.7, until the

2. Materials and methods 2.1. Microorganism and chemicals The strain employed in this work was Sphingobium sp. (formerly misclassified as Pseudomonas fluorescens NCIMB 11671) (Bicas et al., 2

Bioresource Technology 294 (2019) 122180

G. Molina, et al.

3. Results and discussion

desirable OD600 was reached. All experiments used a ratio of 1:1 between aqueous phases and organic (soybean oil) phases (25 mL of each).

3.1. Screening design According to Krings and Berger (1998), the main challenges involved in the biotransformation of terpenes are: (i) the chemical instability of both substrate and product, (ii) the low water solubility of substrate, (iii) the high volatility of both substrate and product, (iv) the high toxicity of both substrate and product, (v) the low yields. The use of biphasic systems is an efficient strategy to deal with these difficulties: such systems facilitate product recovery and may increase yields by reducing substrate and product toxicity and their losses by volatilization (Bicas et al., 2013; Cabral, 2001). The organic phase typically consists of hydrocarbon solvents (e.g., n-decane, n-hexadecane) with an octanol/water partition coefficient (log P) higher than 4, a commonly accepted requisite for good tolerance by whole microbial cells (Fontanille and Larroche, 2003). Authors have already applied this approach to produce isonovalal from α-pinene oxide by Pseudomonas rhodesiae (Fontanille and Larroche, 2003) and patented it to bioconvert some terpenes (Muller et al., 2007). However, in some cases this strategy has not led to good results, such as in the fungal biotransformation of limonene to limonene-1,2-diol (Sales et al., 2019a). Here, it was employed vegetable oil, a cheaper alternative to organic solvents which has been previously considered as organic phase for limonene biotransformation (Bicas et al., 2008a; Bicas et al., 2010a). Moreover, simple changes in the bioprocess conditions (temperature, pressure, pH, substrate concentration, inoculums size etc.) can significantly improve the productivity of a biotransformation. Therefore, response surface methods are considered important tools to overcome the low yields obtained for the biotransformation of terpenes, since this approach help in cutting down the number of experiments required in multi-factorial systems (Berger, 2009). Table 1 describes the levels of the tested variables in the screening design, and these codified values were applied in the PB-12 Plackett–Burman matrix (Table 2). In this case, a p value of 0.1 was used since it is more conservative and lowers the risk of falsely excluding statistically significant parameters. To screen the tested variables, limonene was added to the organic phase at concentrations ranging from 40 (level −1) to 200 (level +1) grams per liter of organic phase. Table 3 shows that the substrate concentration positively affected (p < 0.1) the response (α-terpineol concentration). Therefore, the limonene concentration was increased on the subsequent trials (central composite design). In a similar study, Bicas et al. (2008b) observed that limonene exerted a negative effect on α-terpineol concentration, probably because R-(+)-limonene was toxic to Fusarium oxysporum. Besides being apparently less susceptible to limonene toxicity, the biocatalyst used in the present study was also present in a biphasic system, which is known to reduce the toxic effects of hydrophobic substrates.

2.5.2. Central composition design To determine the influence of the process conditions on α-terpineol production, a 23central composite design was performed, considering the parameters selected during the screening design (Section 2.5.1). Six replicates were conducted at the central conditions, totaling 20 experiments, whose conditions are shown in Fig. 1 and Table 1. The response observed was the concentration of α-terpineol obtained at different intervals (6, 21, 30, 46, 70, and 94 h). For biotransformations at different pH values, the biomass was resuspended on acetate buffer 20 mM mM (pH 5.3, 4.3, and 3.6) or on phosphate buffer 20 mM (pH 6.3 and 7.0), until OD600 of 8.0 (2.8 g biomass per liter of aqueous phase) was reached. All experiments were incubated at 28 °C and used a ratio of 1:1 between aqueous phases and organic (soybean oil) phases (25 mL of each). A second-order model (Eq. (1)) was adopted, to fit the response variables:

Y = β0 +

∑ βi Xi + ∑ βi2 Xi2 + ∑ βij Xij

(1)

where Y is the dependent variable (α-terpineol concentration), Xi and Xj are the coded independent variables, β0 is the constant, βi is the linear coefficient, βi2 is the quadratic coefficient, and βij is the interaction coefficient.

2.5.3. Aqueous/organic phase proportion After optimizing the other process conditions (Sections 2.5.1 and 2.5.2), an experiment was performed to define the ideal proportion of aqueous and organic phases. Three different ratios were tested, while keeping the final volume at 50 mL: (i) 25 mL of organic phase and 25 mL of aqueous phase (1:1), (ii) 12.5 mL of organic phase and 37.5 mL of aqueous phase (1:3) and (iii) 37.5 mL of organic phase and 12.5 mL of aqueous phase (3:1). All the other conditions were maintained under the optimized conditions (28 °C, pH 7.0, limonene at 350 g per liter of organic phase, biomass concentration of 2.8 g per liter of aqueous phase and agitation of 200 rpm).

2.6. Data analysis The results were analyzed using the Statistica 7.0 software (StatSoft Inc, Oklahoma, USA). A significance level of 10 (p < 0.1) and 5% (p < 0.05) was considered for the screening and the central composite design, respectively.

2.7. Analytical conditions

Table 1 Variables and levels used on the Placket-Burman design (PB-12) (Table 2) and in the 23 central composite design (CCD) (Fig. 1).

Samples from the organic phase were periodically withdrawn from the reaction system to monitor substrate consumption and product formation. Before being injected (1 µL) into the gas chromatograph, the oil had to be extracted (vortexing for 40 s) with the same volume of ethanol (1:1, v/v) containing 0.2% (v/v) 1-pentanol as internal standard. The products were analyzed on a HP-7890A (Agilent Technologies) gas chromatograph with flame ionization detector (GCFID). A HP-5 capillary column with 30 m length × 0.250 mm i.d. × 0.25-µm film thickness was employed. Helium was the carrier gas with a constant flow of 1 mL.min−1. Limonene and α-terpineol were quantified based on a calibration curve made after extracting these monoterpenes from soybean oil at knows concentrations. Terpene concentrations were always expressed as mass of product per liter of organic phase.

Variables

3

Units

Level −1.68

−1

0

+1

+1.68

PB-12 pH Biomass concentration Limonene concentration Temperature Agitation

g/Laq. phase g/Lorg. phase o C rpm

– – – –

5.3 2.80 40 21 0

7.0 4.55 120 28 150

8.7 6.30 200 35 300

– – – –

CCD pH Limonene concentration Agitation

– g/Lorg rpm

3.6 100 50

4.3 200 113

5.3 350 200

6.3 500 287

7.0 600 350

Bioresource Technology 294 (2019) 122180

G. Molina, et al.

Table 2 Placket-Burman matrix with codified values (PB-12) and α-terpineol concentration (g/L) for each experiment. The conversion of coded values to real values is found on Table 1. pH

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

1 1 −1 1 1 1 −1 −1 −1 1 −1 −1 0 0 0

Biomass

−1 1 1 −1 1 1 1 −1 −1 −1 1 −1 0 0 0

Limonene

Temperature

−1 1 −1 1 1 −1 1 1 1 −1 −1 −1 0 0 0

1 −1 1 1 −1 1 1 1 −1 −1 −1 −1 0 0 0

α-Terpineol Concentration (g/L)

Agitation

−1 −1 1 −1 1 1 −1 1 1 1 −1 −1 0 0 0

Time (h)

Effect

SE

t (9)

P value

Mean

30 46 70 94 30 46 70 94 30 46 70 94 30 46 70 94 30 46 70 94 30 46 70 94

22.31 24.73 26.11 27.76 −45.64 −51.45 −54.40 −58.00 1.87 4.85 5.70 4.09 34.86 40.80 43.83 44.80 −1.25 −3.87 −4.65 −2.86 44.22 49.55 52.04 55.23

7.83 9.21 9.83 10.05 17.50 20.59 21.99 22.47 17.50 20.59 21.99 22.47 17.50 20.59 21.99 22.47 17.50 20.59 21.99 22.47 17.50 20.59 21.99 22.47

2.851 2.686 2.656 2.763 −2.608 −2.499 −2.474 −2.581 0.107 0.236 0.259 0.182 1.992 1.981 1.994 1.994 −0.072 −0.188 −0.211 −0.127 2.526 2.406 2.367 2.459

0.0191 0.0250 0.0262 0.0220 0.0284 0.0339 0.0353 0.0296 0.9173 0.8191 0.8011 0.8596 0.0775 0.0789 0.0774 0.0773 0.9445 0.8552 0.8373 0.9016 0.0324 0.0395 0.0421 0.0362

pH

Biomass concentration

Limonene concentration

Temperature

Agitation

21 h

30 h

46 h

70 h

94 h

0 0 36.70 0 0 0 1.10 47.34 22.35 0 0.50 0.39 7.85 9.24 8.02

0 0 104.43 0 0.31 0 2.00 98.17 29.26 0 0.94 0.93 16.35 18.16 17.41

0 0 138.87 0 0.33 0 2.47 100.50 30.19 0 1.18 0.95 20.14 20.05 20.02

0 0 163.74 0 0.48 0 3.49 110.01 29.26 0 1.68 1.02 21.58 19.40 20.35

0 0 174.02 0 0.47 0 3.96 117.21 28.10 0 2.34 1.27 21.70 21.28 21.33

0 0 178.82 0 0.47 0 4.55 125.47 35.39 0 2.88 1.33 22.49 23.31 21.68

example, found that biocatalyst content influenced the maximum product recovery positively in the case of α-pinene oxide biotransformation into isonovalal. Similarly, Sales et al. (2019b) found that inoculum load was positively correlated to the amount of limonene-1,2-diol produced by limonene biotransformation; however, the increase in the biomass content decreased the yield of product per biomass (YP/X). In the present study, biomass content in a range of 2.8–6.3 g per liter of aqueous phase presented no effect (p > 0.1) on α-terpineol concentration (Table 3). Therefore, for the subsequent design, all experiments used a biomass concentration of 2.8 g/L. A similar behavior was also observed for the biotransformation of limonene to α-terpineol by F. oxysporum, whose biomass content was not statistically significant to improve product concentration (Bicas et al., 2008b). This was also observed for high biomass contents (13.2 and 26.4 g/L) in the aforementioned biotransformation of limonene to limonene-1,2-diol (Sales et al., 2019b). Finally, the screening design indicated how culture conditions (temperature and agitation) affected α-terpineol production. It is widely known that temperature directly influences biological reactions and that medium agitation promotes mass transfer, homogenization and cell-substrate interaction. However, in addition to increasing the process energy costs, high temperatures and agitation rates might enhance substrate and product loss as well as side reactions. Thus, it is essential to seek an ideal balance, to obtain the best results (Bicas et al., 2008b). In the case of agitation speed, this variable presented a positive effect (p < 0.1, Table 3) on α-terpineol concentration, which means that an increase in agitation would promote limonene conversion, probably due to increased mass transfer and cell-substrate contact. Thus, optimal agitation may be situated at values above the maximum tested in the screening design (300 rpm), suggesting the use of a wider range for this variable in the central composite design. Considering a confidence interval of 90%, temperature did not impact the biotransformation process significantly (Table 3), which means that this process is quite robust in terms of temperature control, since yields would be roughly constant in the range of 21–35 °C. For the central composite design, temperature was kept constant at 28 °C (close to ambient temperature). This was close to the optimum temperature for the biotranformation of limonene to limonene-1,2-diol by Colletotrichum nymphaeae CBMAI 0864 (Sales et al., 2019b). Authors have reported that for Penicillium digitatum NRRL 1202 (Tan et al., 1998), Pseudomonas putida (Chatterjee and Bhattacharyya, 2001), and F. oxysporum 152B (Bicas et al., 2008b, 2010b) the bioconversion decreased dramatically at temperatures above 32, 30, and 28 °C, respectively.

Table 3 Estimates of the effects of the parameters analyzed after 30, 46, 70 and 94 h of biotransformation. Factor

6h

SE Standard error. Parameter in bold are statistically significant for the response (P < 0.1), considering the residual SS.

For establishing the effect of pH on limonene biotransformation, three different pH values were investigated. Preliminary experiments indicated that neither acetate (20 mM) nor Tris-HCl (20 mM) influenced the biotransformation at pH 7.0 (data not shown), therefore, the buffers employed were phosphate buffer 20 mM pH 7.0, acetate buffer 20 mM pH 5.3, and Tris-HCl buffer pH 8.7. The results presented in Table 3 demonstrated that pH had a negative effect on α-terpineol concentration, i.e. an increase in pH would decrease limonene biotransformation. Indeed, no biotransformation occurred at maximum level (pH 8.7). Therefore, lower pH values were considered in the central composite design. In a previous study, a pH variation in a range of 5.2–8.2 has not resulted in statistically different responses on limonene biotransformation (Bicas et al., 2008b). The screening design also examined the effect of biomass (biocatalyst) concentration. In general, the highest the biocatalyst content, the highest is the conversion rate. Fontanille and Larroche (2003), for 4

Bioresource Technology 294 (2019) 122180

G. Molina, et al.

Fig. 1. Time course for α-terpineol concentration of each one of the incubation conditions (Cond.) in coded values (according to Table 1), i.e. pH, limonene concentration ([Lim]) and agitation speed (Agit). The vertical error bars represent the standard deviation of the six replicates of the center points.

3.2. Central composite design

Table 4 Results for regression coefficients and analysis of variance (ANOVA) for the αterpineol concentration after 70 h of the central composite design.

Based on the former results, medium pH, substrate (limonene) concentration and the agitation speed were optimized using a central composite design considering new tested levels (Table 1). Independently on the experimental conditions, a similar production profile was observed, with a maximal α-terpineol concentration generally occurring around 70 h of process (Fig. 1). Hence, the statistical evaluation was conducted only for 46, 70 and 94 h of bioconversion time, considered the probable optimum interval. The data for these three bioconversion times were fitted to Eq. (1) considering only the statistically significant parameters, giving rise to Eqs. (2)–(4).

Y1 = 87.00 + 29.62 X1 + 28.38 X3 − 14.65 X12 − 12.15 X22 − 13.76 X32 + 20.95 X1 X3 Y2 = 102.43 + 35.82 X1 + 30.32 X3 − 14.82

− 17.13

X22

− 17.98

X32 + 21.55 X1 X3

Coefficient

α-Terpineol Concentration (g/L) – 70 h

Mean

β0

102.43

Linear X1 X2 X3

β1 β2 β3

35.82 – 30.32

Quadratic X1 X2 X3

β12 β22 β32

−14.82 −17.13 −17.98

β12 β13 β23

– 21.55 – 14.05 2.92 4.81 < 0.0001 0.87

Interaction X1X2 X1X3 X2X3 Fcalculated Flistedb Fcalculated:Flisted p-value R2 c

(2)

X12

Parametera

(3)

Y3 = 99.89 + 42.73 X1 + 32.20 X3 − 17.41 X22 − 20.08 X32 + 25.04 X1 X3 (4) where Y is the dependent variable (α-terpineol concentration at different times of the fermentation process, Y1 = 46 h, Y2 = 70 h, and Y3 = 94 h), X1, X2 and X3 are pH, limonene (g/L) and agitation (rpm), respectively. As the analysis of equations (2)–(4) indicate an optimum bioconversion time of 70 h, only these data were submitted to an analysis of variance (ANOVA) at 95% confidence level. The data was treated using the software Statistica 7.0, which generated the regression coefficients and the respective statistical analysis of the examined parameters. Table 4 summarizes the results of the ANOVA test, including the regression coefficients for the coded second order polynomial equation, the coefficients of determination (R2), and the F and p values. The results suggest that the fitted model was suitable (significant and predictive) and led to significant regression, low residual value, low lack of fit, an acceptable coefficient of determination (0.87) and a p value lower than 0.0001. The Fcalculated/Flisted ratio was also considered high enough to indicate the validity of this model (Table 4). In other words, the ANOVA results evidenced that the quadratic model adjusted for the process responses was satisfactory. Therefore, contour curves of this model (Eq. (3)) was constructed to illustrate the correlation between αterpineol concentration (in grams per liter of organic phase) and pH, limonene concentration and agitation after 70 h of bioconversion at 28 °C with 2.8 g of biomass per liter of aqueous phase (Fig. 2). Fig. 2a and c showed that the optimal limonene concentration was close to the central value (350 g per liter of organic phase). The optimal

a b c

X1 = pH, X2 = limonene (g/L), X3 = agitation (rpm). Values of Flisted at p < 0.05. R2 = coefficient of determination.

pH for this process is close to neutrality (pH 7.0), while higher agitation speeds (> 300 rpm) tended to increase limonene biotransformation. The use of agitation speed ≤ 113 rpm or pH ≤ 4.3 considerably reduced α-terpineol production. An analysis of Eq. (3) showed that the limonene concentration, agitation speed and pH could oscillate in the range of 260–440 g/L, 300 to 350 rpm and 6.7–7.0, respectively, keeping the α-terpineol concentration at least 90% of its maximum predicted value (182 g/L). The experimental validation was conducted in triplicate using the conditions near the optimal range: pH 6.5, limonene concentration of 350 g/L and agitation speed of 272 rpm, while keeping temperature of 28 °C and a biomass concentration of 2.8 g/L. At these conditions, the α-terpineol concentration was approximately 131 g/L after 70 h, while the predicted value (Eq. (3)) was 137 g/L after the same biotransformation time, representing a relative deviation of −4.6%. In conclusion, the experimental values presented some variation, considering that there is a great difficulty in the control of a biological system, however it were reasonably similar to the expected ones. Other authors have used multi-response analysis to establish the optimal process conditions for other systems of limonene biotransformation. For example, Bicas et al. (2008b) obtained the best

5

Bioresource Technology 294 (2019) 122180

G. Molina, et al.

Fig. 2. Contour curves of α-terpineol concentration after 70 h of biotransformation process as a function of (a) limonene concentration × pH (agitation = 200 rpm), (b) agitation speed × pH (limonene concentration = 350 g/L), (c) agitation speed × limonene concentration (pH = 5.3).

results using 0.5% (v/m) (aprox. 4 g/L) R-(+)-limonene, inoculum/ medium ratio of 0.25 (m/m), and 72 h cultivation at 26 °C and 240 rpm when they employed a Fusarium oxysporum strain. In these optimized conditions, they attained 2.4 g/L product (Bicas et al., 2008b). In another study, Rottava et al. (2011) reported production of approximately 1.7 g/L α-terpineol in optimized conditions, comprising of substrate concentration of 1.75%, mass of inoculum of 2 g, and substrate-toethanol volume ratio of 1:1. The biotransformation of limonene to limonene-1,2-diol by Colletotrichum nymphaeae CBMAI 0864 in shake flasks presented the best behavior when 13.2 g/L biomass was incubated in the presence of 20 g/L of substrate at 27 °C, 250 rpm and pH of 6.0 (Sales et al., 2019b).

3.3. Organic phase proportion

Fig. 3. Biotransformation of R-(+)-limonene into R-(+)-α-terpineol varying the proportion of organic phase in the system, where: ●: 25 mL of organic phase for 25 mL of aqueous phase, ▲: 37.5 mL of organic phase for 12.5 mL of aqueous phase and ■: 12.5 mL of organic phase for 37.5 mL of aqueous phase. Process conditions: pH 7.0, 350 g/L limonene concentration, 200 rpm of agitation speed and 28 °C.

After optimizing the reaction conditions, a new set of experiments were done to evaluate the best ratio between organic and aqueous phases, while keeping constant the limonene concentration (350 g per liter of organic phase), the biomass concentration (2.8 g per liter of aqueous phase) and the total reaction volume (50 mL). Three different conditions were considered, as follows: (a) 25 mL of organic phase and 25 mL of aqueous phase (the condition used in the optimization experiments), (b) 37.5 mL of organic phase and 12.5 mL of aqueous phase and (c) 12.5 mL of organic phase and 37.5 mL of aqueous phase. All the other optimized system conditions were kept near the optimal conditions (pH 7.0, 200 rpm and 28 °C). Fig. 3 shows that higher aqueous/ organic ratios significantly increased α-terpineol concentration, but not

the total α-terpineol production: when using a ratio of 3:1 (37.5 mLaq:12.5 mLorg), α-terpineol concentration reached approximately 240 g per liter of organic phase after 96 h of process, however the total mass of α-terpineol was c.a. 3.0 g (in 12.5 mL), which is approximately 2/3 of the total mass obtained with a ratio of 1:1 (182 g/L, i.e. 4.5 g in 25 mL). Considering the α-terpineol yield in terms of biomass, the optimal value of c.a. 65 g of α-terpineol per gram of biomass was achieved 6

Bioresource Technology 294 (2019) 122180

G. Molina, et al.

for the ratio of 1:1, while lower yields were obtained for ration of 1:3 (20 g/g) and 3:1 (29 g/g). Therefore, although not providing the highest α-terpineol concentration, the ratio of 1:1 was considered more suitable for this biotransformation.

Perfum. Flavor. 18, 1993. Burdock, G.A., Fenaroli, G., 2010. Fenaroli’s Handbook of Flavor Ingredients, sixth ed. CRC Press, Boca Raton. Cabral, J.M.S., 2001. Biotransformations. In: Ratledge, C., Kristiansen, B. (Eds.), Basic biotechnology, second ed. Cambridge University Press, Cambridge, pp. 471–501. Çelik, D., Bayraktar, E., Mehmetoglu, U., 2004. Biotransformation of 2-phenylethanol to phenylacetaldehyde in a two-phase fedbatch system. Biochem. Eng. J. 17, 5–13. Chatterjee, T., Bhattacharyya, D.K., 2001. Biotransformation of limonene by Pseudomonas putida. Appl. Microbiol. Biotechnol. 55, 541–546. De Carvalho, C.C.R., Da Fonseca, M.M.R., 2006. Biotransformations of terpenes. Biotechnol. Adv. 24, 134–142. de Oliveira Felipe, L., de Oliveira, A.M., Bicas, J.L., 2017. Bioaromas-Perspectives for sustainable development. Trends Food Sci. Technol. 62, 141–153. Fontanille, P., Larroche, C., 2003. Optimization of isonovalal production from alphapinene oxide using permeabilized cells of Pseudomonas rhodesiae CIP 107491. Appl. Microbiol. Biotechnol. 60, 534–540. Fontanille, P., Le Flèche, A., Larroche, C., 2002. Pseudomonas rhodesiae PF1: a new and efficient biocatalyst for production of isonovalal from α-pinene oxide. Biocatal. Biotransformation. 20, 413–421. Khaleel, C., Tabanca, N., Buchbauer, G., 2018. α-Terpineol, a natural monoterpene: a review of its biological properties. Open Chem. 16, 349–361. Krings, U., Berger, R.G., 1998. Biotechnological production of flavors and fragrances. Appl. Microbiol. Biotechnol. 49, 1–8. Lee, S.Y., Kim, S.H., Hong, C.Y., Park, S.Y., Choi, I.G., 2015. Biotransformation of (-)-αpinene and geraniol to α-terpineol and p-menthane-3, 8-diol by the white rot fungus, Polyporus brumalis. J. Microbiol. 53, 462–467. Melo, L.L.M.M., Pastore, G.M., Macedo, G.A., 2005. Optimized synthesis of citronellyl flavour esters using free and immobilized lipase from Rhizopus sp. Process Biochem. 40, 3181–3185. Molina, G., Bution, M.L., Bicas, J.L., Dolder, M.A.H., Pastore, G.M., 2015. Comparative study of the bioconversion process using R-(+)-and S-(–)-limonene as substrates for Fusarium oxysporum 152B. Food Chem. 174, 606–613. Muller, M., Dirlam, K., Wenk, H.H., Berger, R.G., Krings, U., Kaspera, R., 2007. Method for the production of flavor-active terpenes. US Patent Application n. 20070172934. 07.07.26. Rodrigues, M.I., Iemma, A.F., 2014. Experimental Design and Process Optimization. CRC Press. Rottava, I., Cortina, P.F., Martello, E., Cansian, R.L., Toniazzo, G., Antunes, O.A.C., Oestreicher, E.G., Treichel, H., Oliveira, D., 2011. Optimization of α-terpineol production by the biotransformation of R-(+)-limonene and (−)-β-pinene. Appl. Biochem. Biotechnol. 164, 514–523. Sales, A., Paulino, B.N., Pastore, G.M., Bicas, J.L., 2018. Biogeneration of aroma compounds. Curr. Opin. Food Sci. 19, 77–84. Sales, A., Moreira, R.C., Pastore, G.M., Bicas, J.L., 2019a. Establishment of culture conditions for bio-transformation of R-(+)-limonene to limonene-1, 2-diol by Colletotrichum nymphaeae CBMAI 0864. Process Biochem. 78, 8–14. Sales, A., Pastore, G.M., Bicas, J.L., 2019b. Optimization of limonene biotransformation to limonene-1,2-diol by Colletotrichum nymphaeae CBMAI 0864. Process Biochem. https://doi.org/10.1016/j.procbio.2019.07.022. Sen, R., Swaminathan, T., 1997. Application of response-surface methodology to evaluate the optimum environmental conditions for the enhanced production of surfactin. Appl. Microbiol. Biotechnol. 47 (4), 358–363. Siddhardha, B., Kumar, M.V., Murty, U.S.N., Ramanjaneyulu, G.S., Prabhakar, S., 2012. Biotransformation of α-pinene to terpineol by resting cell suspension of Absidia coerulea. Indian J. Microbiol. 52, 292–294. Surburg, H., Panten, J., 2006. Common Fragrance and Flavor Materials. Preparation, Properties and Uses, fifth ed. WILEY-VCH Verlag, Weinheim. Tan, Q., Day, D.F., 1998a. Bioconversion of limonene to α-terpineol by immobilized Penicillium digitatum. Appl. Microbiol. Biotechnol. 49, 96–101. Tan, Q., Day, D.F., 1998b. Organic co-solvent effects on the bioconversion of (R)(+)-Limonene to (R)-(+)-α-Terpineol. Process Biochem. 33, 755–761. Tan, Q., Day, D.F., Cadwallader, K.R., 1998. Bioconversion of (R)-(+)-limonene by P. digitatum (NRRL 1202). Process Biochem. 33 (1), 29–37. https://doi.org/10.1016/ S0032-9592(97)00048-4. Werkhoff, P., Brennecke, S., Bretschneider, W., Giintert, M., Hopp, R., Surburg, H., 1993. Chirospecific analysis in essential oil, fragrance and flavor research. Zeitschrift für Lebnsmittel-Untersuchung und Forschung 196, 307–328.

4. Conclusion This work investigated the ideal conditions for the microbial production of α-terpineol from limonene. The tests indicated that α-terpineol concentration in the organic phase could be significantly increased to c.a. 240 g/L (two times the highest concentration reported so far) when 12.5 mL of soybean oil containing 350 g/L of limonene and 37.5 mL of an aqueous phase (pH 7.0) containing 2.8 g/L biomass was kept at 200 rpm, and 28 °C for 96 h. However, the highest amount (4.5 g) and yield of α-terpineol in terms of biomass (65 g/g) was achieved with aqueous:organic ratio of 1:1. Acknowledgments Authors acknowledge National Counsel of Technological and Scientific Development (CNPq, Process numbers 473981/2012-2; 460897/2014-4 and 400411/2016-4), São Paulo Research Foundation (FAPESP, grant number 2016/21619-7) and “Fundação de Amparo à pesquisa de Minas Gerais” (Fapemig, Process number APQ-01056-17) for the financial support. This study was financed in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior – Brasil (CAPES) – Finance Code 001 and PROEX/CAPES 23038.000795/201861. References Adams, A., Demyttenaere, J.C.R., De Kimpe, N., 2003. Biotransformation of (R)-(+)- and (S)-(–)-limonene to alpha-terpineol by Penicillium digitatum – Investigation of the culture conditions. Food Chem. 80, 525–534. Bauer, K., Garbe, D., Surburg, H., 2001. Common Fragrance and Flavor Materials Preparation, Properties and Uses, fourth ed. Wiley-VCH, Weinheim, Germany, pp. 293. Berger, R.G., 2009. Biotechnology of flavours–the next generation. Biotechnol. Lett. 31 (11), 1651. Bicas, J.L., Maróstica Jr., M.R., Barros, F.F.C., Molina, G., Pastore, G.M., 2013. Bioadditives produced by fermentation. In: Soccol, C.R., Pandey, A., Larroche, C. (Eds.), Fermentation Processes Engineering in the Food Industry. CRC Press, Boca Raton, FL, pp. 371–403. Bicas, J.L., Barros, F.F.C., Wagner, R., Godoy, H.T., Pastore, G.M., 2008a. Optimization of R-(+)-α-terpineol production by the biotransformation of R-(+)-limonene. J. Ind. Microbiol. Biotechnol. 35, 1061–1070. Bicas, J.L., Fontanille, P., Pastore, G.M., Larroche, C., 2010a. A bioprocess for the production of high concentrations of R-(+)-α-terpineol from R-(+)-limonene. Process Biochem. 45, 481–486. Bicas, J.L., Fontanille, P., Pastore, G.M., Larroche, C., 2008b. Characterization of monoterpene biotransformation in two Pseudomonads. J. Appl. Microbiol. 105, 1991–2001. Bicas, J.L., Quadros, C.P., Neri-Numa, I.A., Pastore, G.M., 2010b. Integrated process for co-production of alkaline lipase and R-(+)-α-terpineol by Fusarium oxysporum. Food Chem. 120, 452–456. Bier, M.C.J., Medeiros, A.B.P., Soccol, C.R., 2017. Biotransformation of limonene by an endophytic fungus using synthetic and orange residue-based media. Fungal Biol. 121, 137–144. Boelens, M.H., Boelens, H., van Gemert, L.J., 1993. Sensory Properties of optical isomers.

7