Production of Some Higher Alcohols and Acetate Esters from Rice Bran by Yeasts Metabolisms of Kluyveromyces Marxianus and Debaryomyces

The aim of this study was to evaluate the biosynthesis of avor compounds from rice bran by yeast metabolisms. The microbial growth and avor biosynthesis of Kluyveromyces marxianus and Debaryomyces hansenii in rice bran by microbial fermentation were investigated. Growth of both yeasts was assessed by the calculation of specic growth rates and doubling time. Their aroma biosynthesis was evaluated by gas chromatography-olfactometry (GC-O), gas chromatography-mass spectrometry (GC-MS) and Spectrum™ sensory analysis. The specic growth rate (µ max ) and doubling time (t d ) of K. marxianus were calculated as 0.16/h and 4.21 h respectively, D. hansenii had 0.13/h of µ max and 5.33h of t d . K. marxianus and D. hansenii signicantly produced higher alcohols and acetate esters from rice bran at high levels. Results showed that K. marxianus can produce 827.27 µg/kg of isoamy alcohol, 169.77 µg/kg of phenyl ethyl alcohol and 216.08 µg/kg phenyl ethyl acetate during 24 h batch fermentation. Isovaleric acid was also synthesed by K. marxianus at high level (4013 µg/kg) in the batch fermentation of 96 hours. The highest concentration of of isoamyl alcohol and phenyl ethyl acetate was determined as 415.64 µg/kg and 135.77 µg/kg, respectively at 24 h fermentation of D.hansenii. Fermented cereals and rose were dened as characteristic avor descriptors for the fermented rice bran samples. Rose avor term in fermented rice bran samples were found to associate with phenyl ethyl alcohol, phenyl ethyl acetate, isoamyl acetate and guaiacol. The valorization of rice bran can be achieved with the production of natural avor compounds by yeast metabolisms. to hydrolytic and oxidative rancidity to its rich oil content. For this reason, processing of rice bran has high cost to produce bioactive compounds. However, microbial fermentation had high eciency and eco-friendly process. Valorization of rice bran to produce avor compounds by yeast fermentation are considered rst time with this study. Results showed that higher alcohols and acetate ester can be produced from rice bran with a high yield by the yeast metabolism each microbial solution. The asks were incubated at 30°C for 72 h in a shaking incubator (Sartorius-Certomat IS, Goettingen, Germany) at 120 rpm. The rice bran solution without microorganism was used as control group and applied the same preparation procedure of the samples inoculated with microorganism. Batch fermentation was performed by using stirred tank bioreactor (STR) (Biostat A-plus®, Sartorius, Melsungen, Germany) with a 4 L working volume. The STR has two six-blade impellers, pH probe (Hamilton, Easyferm K8/325) and PT 100 temperature sensor. Based on the shake ask fermentation, the batch fermentation condition was set as 0.325 vvm of the aeration rate, 120 rpm of the agitation speed, and 30°C of fermentation temperature for 120 h. The inoculation rate of the microbial solution for batch fermentation was 10 6−7 cell per mL rice bran solution. Both the shake ask and batch fermentation experiments were performed in duplicate. bioreactor


Introduction
Flavor compounds in foods either are formed naturally by the results of several chemical and biochemical reactions or are added as food additives during the manufacturing stages of foods. They are composed of several chemical compounds including acids, alcohols, aldehydes, esters, ketones, lactones, sulfur compounds, terpenes, and their derivatives. Over 6500 avor compounds are identi ed and well characterized in food. But, about 300 of them are widely used as food additives [1][2]. While most of avor compounds were produced synthetically by certain chemical reactions, a few of them are obtained from natural plant and animal sources such as mints, vanilla pods and beeswax by using conventional extraction and distillation methods [3].
Microbial fermentation has been gaining avor scientists and manufacturers attention for decades because of avor compounds produced enzymatic and microbiological processes are de ned as "natural avors" (European Council Directive, Directive 88/388/88 EEC). The growing of perception and desire for natural products by consumer also drawn on the development biotechnological process for the natural bio-products [4]. Today, agricultural wastes or by-products are emphasized as favorable sources in terms of economic and environmental impact to obtain a high added value product by microbial fermentation. Most of them have rich content of carbohydrates, protein, oil and other nutrients that are usable for microbial metabolisms [5][6]. Thus, several researches on using agro wastes or byproducts for microbial production of avor compounds have been conducted [7][8][9][10][11][12].
The rice bran is one of the main by-products in the rice milling. Rice bran has about 12-22% oil, 11-17% protein, 6-14% ber, 10-15% moisture, and 8-17% ash, and comprises 10% of raw rice grain. It is also rich in some minor compounds including phytosterols, tocopherols, tocotrienols, B vitamins, phosphorus, and potassium. It is stated that the composition of rice bran was changed depending on many factors including rice variety, pretreatments for milling, degree of milling. China is the largest producer with a production of 10 million tons per year [13][14][15]. Due to higher nutritional value of rice bran, it is widely discussed for producing value added products such as rice brain oil, protein hydrolysate, γ-oryzanol, α-amylase and dietary ber. But, it has not been fully utilized and also the researches on the potential usage of rice bran as raw material in the biotechnological process to produce high value added bioproducts are scarce [9,13,16]. In this context, Kaur and Chakraborty [9] optimized the production of vanillin from culture medium contained 15% of rice bran and %0.005 ferulic acid by fermentation of Pediococcus acidilactici. It was found that 1.269 g/L of vanillin can be obtained within 24 h fermentation at pH 5.6, at 37°C under shaking conditions at 180 rpm. Zheng et al. [16] reported that the highest yield reached 2.2 g/L of vanillic acid can be produced from waste residue of rice brain oil by the fermentation of Aspergillus niger CGMCC0774 in a 25 L fermenter. Rice bran has promising potential to use for producing natural avor compounds by biotechnological process. This study aimed to determine avor production capabilities of Kluyveromyces marxianus and Debaryomyces hansenii from rice bran. The avor production behavior of these yeasts was assessed based on the state of art avor chemistry and bioprocess technique.

Material
Rice bran used in the present was obtained from local rice mill factory in Canakkale. The rice bran was stored at -18°C. Strains of Kluyveromyces marxianus NRRL YB-155 and Debaryomyces hansenii NRRL YB-6373 were obtained from ARS Culture Collection (NRRL collection, Peoria, Illinois). Analytical and chromatographic grades of chemicals are used in the all analysis in this study (Merck, Darmstadt, Germany, and Sigma-Aldrich, St Louis, USA).

Preparation of rice bran solution
Solution of (10%) of rice bran was used for both shake ask and bioreactor scale fermentation. In brief, the rice bran was grinded by knife miller Restch GM 200 (Haan, Germany) for 15 min. Then, 10% of rice bran solution was prepared with distilled water. The solution was homogenized at 24.000 rpm by Ultraturrax (IKA-WERKE GmbH, Germany) [11].

Preparation of microbial cultures
Adaptation and preparation of cell suspensions with K. marxianus and D. hansenii for rice bran fermentation were followed by the method mentioned by Guneser et al. [11]. Yeast extract Peptone Dextrose (YEPD) medium and YEPD medium with 0.3M NaCl were used for the cultivation of D. hansenii and K. marxianus, respectively. The cell concentration of K. marxianus and D. hansenii in microbial solution was determined by using Thoma slide under light microscope [17].

Fermentation experiments
For shake ask fermentation, 100 mL of the rice bran solution was poured into 250 mL Schott bottle with non-vented cap and a conventional sterilization procedure (121°C for 15 min) was applied by using autoclave (Hirayama, Saitama, Japan). Afterward, the rice bran solutions were inoculated using ca. 10 6−7 cell per mL of each microbial solution. The asks were incubated at 30°C for 72 h in a shaking incubator (Sartorius-Certomat IS, Goettingen, Germany) at 120 rpm. The rice bran solution without microorganism was used as control group and applied the same preparation procedure of the samples inoculated with microorganism.
Batch fermentation was performed by using stirred tank bioreactor (STR) (Biostat A-plus®, Sartorius, Melsungen, Germany) with a 4 L working volume. The STR has two six-blade impellers, pH probe (Hamilton, Easyferm K8/325) and PT 100 temperature sensor. Based on the shake ask fermentation, the batch fermentation condition was set as 0.325 vvm of the aeration rate, 120 rpm of the agitation speed, and 30°C of fermentation temperature for 120 h. The inoculation rate of the microbial solution for batch fermentation was 10 6−7 cell per mL rice bran solution. Both the shake ask and batch fermentation experiments were performed in duplicate.
The growth curves of K. marxianus and D. hansenii Growths of K. marxianus and D. hansenii were followed during shake ask and batch fermentations for evaluating the microbial growth behavior of the strains in rice bran. Speci c growth rate (μ max ) and doubling time (t d ) of both yeasts were calculated for batch fermentation. The rice bran sample was taken from shake ask and bioreactor intermittently (24 h) in aseptic conditions. Pour plate counting technique was used in this study. YEPD with 0.3M NaCl agar for K. marxianus and YEPD agar for D. hansenii were used. Plates were incubated at 30°C for 72 h.
Gas chromatography-olfactometry (GC-O) and Gas chromatography-mass spectrometry (GC-MS) Analysis Extraction of volatile compounds from the fermented rice bran samples were achieved by using the solid-phase microextraction (SPME) method combined with GC-MS and GC-O analysis. For this purpose, about 3 g sample was weighed into a 40-mL vial, and 1 g NaCl and 10 µL internal standard (a mixture of 2-methyl valeric acid and 2-methyl 3-heptanone) were added. The samples were vortexed (Biosan V-1, Riga, Latvia) for 1 min. The vials were maintained at 40°C in a water bath for 20 min to equilibrate the volatile compounds in the headspace of the vials. Afterward, SPME needle (2 cm-50/30 µm Divinylbenzene/Carboxen/Polydimethylsiloxane stable ex, Bellafonte, USA) was inserted into each vial. The SPME ber was exposed at a depth of 2 cm in the headspace of the vial for 20 min. The SPME Gas chromatography olfactometry procedure was achieved in duplicate by two experienced panelists by using the method of Guneser et al. [11]. The National Institute of Standards and Technology (NIST) and Wiley Registry of Mass Spectral Data bases were used for the identi cation of volatile compounds. Quantitative analysis of the volatile compounds determined by GC-MS was performed as described by Avsar et al. [18].

Sensory analysis
Spectrum™ method was used to determine the sensory properties and changes in aroma pro les of the rice bran samples. Sensory evaluation was conducted with seven trained panelists (4 female and 3 were male, aged from 24 to 45 years). Panelists quanti ed the attributes using 15-point product-speci c scale anchored to the left with 'not' and to the right with 'very' [19].

Statistical analysis
Two-way analysis of variance analysis (ANOVA) was performed to evaluate the growth of both yeasts in rice bran during the shake ask fermentation. One-way ANOVA analysis was also performed to determine the differences in the intensities of avor compounds in the rice bran samples during fermentation. Tukey's test was applied to compare statistically signi cant differences at a signi cance level of p≤0.05 [20]. For all statistical analyses, SPSS for Windows (version 15.0) packed program was used. Principal component analysis (PCA) was performed by using the XLSTAT statistical program (trial version, 2015, Addinsoft, Inc., New York, NY, USA) to evaluate the relationship between the intensities of produced avor compounds and their sensory impacts or perceptions.

Results And Discussion
Growth behavior ofK. marxianusandD. hanseniiin rice bran The changes in number of cells of K. marxianus and D. hansenii in rice bran during shake ask fermentation are presented in Fig. 1.
During shake ask fermentation, signi cant increases were observed in the number of the cells for both yeast (p ≤ 0.05). The number of K. marxianus cells increased slightly during 48 h fermentation, and the number of K. marxianus reached to the level of 2.47 log CFU/mL at the end of the fermentation. Unlike K. marxianus, the cell numbers of D. hansenii steady increased from 7.0 log CFU/mL to 10.69 CFU/mL. In case of bioreactor fermentation, the same growth behavior was observed for both yeasts. The numbers of K. marxianus and D. hansenii were found to increase in the level of 2.30 log CFU/mL and 3.28 log CFU/mL, respectively during 120 h bioreactor fermentation. While the speci c growth rate (µ max ) and doubling time (t d ) of K. marxianus were calculated as 0.16/h and 4.21 h, D.
hansenii had 0.13/h of µ max and 5.33h of t d . When calculated values of µ max and t d were compared for both yeasts, it can be concluded that the growth of K. marxianus was faster than D. hansenii in the rice bran fermentation (Fig. 2).

Page 5/15
The ndings of the present study are in good agreement with the results of previous studies [11,[21][22]. Zafar and Owais [22] reported that the maximum growth rate of K. marxianus was 0.157/h in the production of ethanol by crude whey fermentation. In another study by Guneser et al. [11], the speci c growth rates for K. marxianus and D. hansenii were calculated as 0.08/h and 0.177/h in the tomato pomace fermentation, respectively. Goshima et al. [23] reported higher growth rate for the strains of K. marxianus NBRC177 and DMB1 than those of the present study. The researchers determined that the growth rates varied between 0.8/h and 1.06/h depending on the fermentation temperature for these strains in the bioethanol production from lignocellulosic hydrolysates. Compared to our ndings, lower growth rates for K. marxianus and D. hansenii were also reported by previous studies [24,25]. In a study by Madawati et al. [24], the growth rate for D. hansenii was determined between 0.03/h and 0.06/h in arabitol production from by-products of Reutalis trisperma biodiesel. Similarly, Sharma et al. [25] reported 0.02/h of growth rate for K. marxianus NIRE-K3 (MTCC 5934) in the YEPX medium with xylose (3%) medium at 45°C. Overall, it can be concluded that variation in growth rates of K. marxianus and D. hansenii results from strain type, utilization behavior of substrate by the microbial strains, agro waste types and its physicochemical structure and fermentation conditions. In general, several pre-treatment processes such as acidic or enzymatic hydrolysis are addressed for promoting of microbial growth on agro waste. But, the cost of pre-treatment should be taken into careful consideration in fermentation processes [26,27].
Flavor production characteristics ofK. marxianusandD. hanseniiin rice bran In the present study, avor production behaviors of K. marxianus and D. hansenii in rice bran were determined by using GC-O and GC-MS techniques. Aroma-active compounds, which produced by both yeasts, and their intensities were determined in the shake ask fermentation stage by GC-O analysis. Total 33 aroma-active compounds were identi ed in the fermented and unfermented rice bran samples. Identi ed aroma-active compounds composed of acids, alcohols, aldehydes, esters, ketones, and pyrazines ( Table 1). As seen in Table 1, higher alcohols and acetate esters among the aroma-active compounds were the most abundant compounds in the fermented rice bran samples. The intensities of isovaleric acid (sour, dirty), isoamyl acetate (fermented fruit), phenyl ethyl alcohol (rose) and phenyl ethyl acetate ( oral) in rice bran samples fermented by both yeasts were found to be higher than the unfermented samples. It was also found that the rice bran samples fermented by K. marxianus had higher intensities of guaiacol (cotton candy) and 2,4-nonedienal (oxide oil) than others, whereas 2,5-dimethyl-3-ethylpyrazine (dirty), 2-heptanone (metallic), 2-heptanal (oxide), 1-octen-3-ol (mushroom) was perceived at the highest intensities in rice bran samples fermented by D. hansenii. Gas chromatography olfactometry expresses the relative intensities of avor compounds and their odor impacts. Therefore, the ndings GC-O should be reviewed with the results GC-MS that indicate the quantities of the avor compounds in the aroma research [12] [28].
Based on the results of GC-MS analysis, it was observed that six avor compounds including isoamyl alcohol, isovaleric acid, isoamyl acetate, phenyl ethyl alcohol, phenyl ethyl acetate and guaiacol were produced from rice bran by K. marxianus and D. hansenii at higher concentrations in 72 h shake ask fermentation ( Table 2). The concentration of 2,4-nonedienal and 2,5-dimethyl-3-ethylpyrazine determined in the fermented rice bran samples did not increase. These ndings could be related to the odor thresholds and concentrations of these compounds [29,30]. Basically, the odor threshold of aroma compounds was affected by their vapor pressures and molecular interactions of other substances which existed in the medium. The odor threshold value was determined by sni ng avor compound and de ned as the lowest concentration of avor compound which is required for the recognition of its odor. The intensity of avor compound was directly related with the odor threshold. Therefore, even if the changes in the concentrations of avor compounds is low, the intensities of them could be perceived as high [31].
The concentrations of higher alcohols and acetate esters in the fermented rice bran samples signi cantly increased during the shake ask fermentation (p ≤ 0.05) ( Table 2). Rice bran fermented by K. marxianus had 522 µg/kg of isoamyl alcohol and 111 µg/kg of phenyl ethyl alcohol while the concentrations of isoamyl alcohol and phenyl ethyl alcohol were determined as 445 µg/kg, and 108 µg/kg in rice bran fermented by D. hansenii respectively, which were higher than those of unfermented rice bran samples ( Table 2) Apart from the primary metabolites (as carbon dioxide, ethanol and glycerol), acids, higher alcohols and esters are produced as the main secondary metabolites in the yeast metabolism. These secondary metabolites, which are produced by different metabolic pathways of yeast, contributed the avor pro le of several fermented foods and beverages such as bread and wine [32,33]. In yeast metabolism, the synthesis of higher alcohols including isoamyl alcohol and phenyl ethyl alcohol are achieved by Ehrlich pathway, which contains several biochemical reactions of branched chain, aromatic and sulfur-containing amino acids such as transamination, decarboxylation, oxidation, and reduction. In the Ehrlich pathway, L-phenylalanine and L-leucine are the main substrate for the biosynthesis for isoamyl alcohol and phenyl ethyl alcohol, respectively.
Phenyl ethyl alcohol can also be synthesized from carbohydrate via Shikimate pathway in yeast cells [34][35][36]. Most knowledge of the Ehrlich and Shikimate pathways was obtained from studies on the fermentation behavior of Saccharomyces cerevisiae, several transaminases, decarboxylase, dehydrogenase enzymes are responsible for these pathways in the metabolism of S. cerevisiae, but the secondary metabolites production pathways of K. marxianus and D. hansenii and their enzymes systems are still being investigated [36,37]. There are two types of ester can be produced by yeast metabolism. They are acetate esters and medium-chain fatty acid ethyl esters.
When the examined our ndings, K. marxianus and D. hansenii are seen to produce especially acetate esters including isoamyl acetate and phenyl ethyl acetate from rice bran ( Table 2). The concentration of phenyl ethyl acetate (205.93 µg/kg) in rice bran fermented by K. marxianus was found remarkably higher than that of D. hansenii (35.99 µg/kg) (p ≤ 0.05). However, no signi cant difference was observed between isoamyl acetate production rates of both yeasts (p ≥ 0.05). While acetate esters composed of the acid and alcohol groups, medium-chain fatty acid ethyl esters contain the alcohol and medium-chain fatty acids. Both esters can be produced through lipid and acetyl-CoA metabolism and an acyl transferase or ester synthase enzymes are play important role in the formation of esters in yeast. The synthesis of acetate esters were catalyzed by the alcohol acetyltransferases I and II encoded by the genes ATF1 and ATF2 in yeast [32,33]. Moreover, the concentration of acetyl-CoA and a fusel alcohol and the activities of enzymes affected the rate of acetate ester formation. It was emphasized that the expression levels of the ATF1 and ATF2 during the fermentation are the most important factor for the concentration levels of the acetate ester as well as the substrate concentration in the medium [32,38]. Therefore, higher phenyl ethyl acetate concentration in rice bran fermented by K. marxianus could be attributed the higher ATF1 and ATF2 expression compared to D. hansenii.
Concentrations of isovaleric acid and guaiacol increased signi cantly in the fermented rice bran during shake ask fermentation (p ≥ 0.05). The concentrations of isovaleric acids in the rice bran samples fermented by K. marxianus and D. hansenii were found as 210.17 and 285.60 µg/kg, respectively ( Table 2). Determination of isovaleric acid at higher concentrations in fermented rice bran samples could be attributed to the usage of the leucine, isoleucine and valine found in rice bran protein by both yeasts and their aminotransferase enzymes activities. Similarly, the synthesis of higher alcohols, it was emphasized that α-keto acids such as isovaleric acid can be also synthesized in the rst step of the Ehrlich pathway. In this step, speci cally leucine, isoleucine and valine can be converted to α-keto acids and the aminotransferase enzymes catalyzed this reaction [39].
Guaiacol is de ned as phenolic off avor in fermented beverages such as wine. Phenolic volatiles can be synthesized from the phenolic acids of the hydroxycinnamic series (e.g. 4-hydroxycinnamic, p-coumaric and 4-hydroxy-2-methoxycinnamic or ferulic acid) by decarboxylation reactions in yeast metabolism [40,41]. In the present study, guaiacol was identi ed as aroma-active compounds at lower intensity in the rice bran samples fermented by D. hansenii and unfermented rice samples by GC-O analysis (Table 1). However, it was determined by GC-MS in all rice bran samples ranged from 0.31-7.85 µg/kg ( Table 2). This difference may be attributed to the fact that molecular interactions of guaiacol with other substances which are found in the medium. Sensitivities of both analysis techniques may also lead to this difference [42].
The ndings of the present study are in good agreement with the results of several previous studies on the production of aroma compounds from different agro wastes by yeast metabolism [7,21,38,[43][44][45][46][47]. In a recent study by Kılmanoğlu et al. [43], it was reported that higher alcohols and acetate esters including isoamyl alcohol, phenyl ethyl alcohol, phenyl ethyl acetate and ethyl acetate could be produced from tomato pomace hydrolysate by the fermentation with K. marxianus similar to our ndings. The researchers also expressed that heat-treated dilute acid pretreatment and cellulolytic enzyme hydrolysis led to increase in the amount of higher alcohols and esters by increasing the fermentable sugars in tomato pomace. Similarly, Martinez et al. [48] showed that higher alcohols and esters can be produced from the mixture of sugarcane bagasse with sugar beet molasses by solid state fermentation of K. marxianus ATCC 10022. According to this, the researchers expressed that the maximum cumulative volatile production was achieved at 40°C, 35% molasses rate and speci c air ow rate of 0.14 L/h g and the main volatile components composed of 43% alcohols and 18% ester species including ethanol, isoamyl alcohol, isobutyl alcohol, ethyl acetate, isoamyl acetate and isobutyl acetate. It was also observed that production of esters can be increased up to 35% with the working conditions as 30°C of fermentation temperature, 25% molasses rate, and speci c air ow rate of 0.11 L/h g. Vong and Liu [44] investigated the avor production capabilities of ten yeasts originated from wine and dairy foods in the soybean residues (okara) which is a by-product soy milk. The researchers expressed that soybean residues fermented by dairy yeasts including Geotrichum candidum, Yarrowia lipolytica, Kluyveromyces lactis and D. hansenii have musty and moldy aromas at higher intensities owing to the higher content of aldehydes and methyl ketones which are different from our ndings about the fermentation of D. hansenii.
The changes in aroma compounds in the rice bran samples were also determined during batch fermentation in the present study. Table 3 shows the changes in the concentrations of aroma compounds in bioreactor conditions. It was observed that the concentration of the aroma compounds produced from rice bran changed signi cantly depends on the yeast types. In the K. marxianus fermentation, the synthesis of isoamyl alcohol and phenyl ethyl acetate was the highest at 24 h. Amount of both compounds signi cantly decreased from 48 h to 72 h of fermentation (p ≤ 0.05). Hence, it was observed that the amount of phenyl ethyl alcohol increased from 5.42 µg/kg to 202.01 µg/kg during 96 h of fermentation. In the same manner, the synthesis of isovaleric acid and guaiacol was found to increase during 96h of fermentation. It was observed that the concentration of isoamyl acetate produced by K. marxianus increased during 48 h, then a sharp decrease in the concentration of this compound determined at the end of fermentation.
In case of fermentation by D. hansenii, isovaleric acid was not determined in the samples in bioreactor conditions. Like the fermentation by K. marxianus, the synthesis of isoamyl alcohol and phenyl ethyl acetate was the highest at 24 h and the concentration of these compounds were determined as 41.564 and 135.77 µg/kg, respectively. The highest phenyl ethyl alcohol concentration was determined at 48h. Moreover, it was observed that D. hansenii showed the similar synthesis behavior with K. marxianus in terms of guaiacol and isoamyl acetate production during bioreactor fermentation. However, D. hansenii produced lower amount of isoamyl alcohol than K. marxianus at the same fermentation time (Table 3). When the productivities of aroma compounds were examined during bioreactor fermentation, the highest productivity was calculated to produce isoamyl alcohol in both yeasts' fermentation followed by phenyl ethyl acetate and phenyl ethyl alcohol (Data not shown). Productivity of isoamyl alcohol for the fermentations of K. marxianus and D. hansenii were calculated as 34.46 µg/kg h and 17.31 µg/kg h. Hence, it was calculated that K. marxianus had a higher phenyl ethyl acetate productivity (9.00 µg/kg h) than that of D. hansenii (5.65 µg/kg h) while the productivity value of phenyl ethyl alcohol for both yeasts was calculated to be relatively close to each other. The ndings of the present study are consistent with previous studies on the production of avor compounds by fermentation [11,12,[49][50][51][52]. Mederios et al. [51] reported increasing concentration of ethyl acetate until 22 h of fermentation time, and then ethyl acetate concentration signi cantly decreased in the solid state fermentation of cassava bagasse by K. marxianus on a packed bed column bioreactor. In a study by Guneser et al. [11], the highest production of isoamyl alcohol and phenyl ethyl alcohol in tomato pomace fermented by K. marxianus and D. hansenii was reported at 24 h fermentation and the concentrations of isoamyl alcohol and phenyl ethyl alcohol was determined in the ranges between 0.23-126.72 µg/kg and 3.30-158.72 µg/kg. Similarly, Yılmaztekin et al. [52] reported that the production of isoamyl alcohol from sugar beet molasses fermented by Williopsis saturnus var. saturnus in 5L bioreactor increased gradually during 288 h, while the production of isoamyl acetate gradually increased until about 140 h fermentation. In our previous study [12], we also observed similar production changes for mushroom-like avors including 1-octen-3-ol, 1-octen-3-one and octanol which were produced by fungal metabolism of T. atroviride and A. sojae in tomato and pepper pomaces in bioreactor fermentation.
After a certain time of fermentation, loss of aroma compounds produced in bioreactor condition can be related to many factors such as nutrients/precursor concentration, secondary metabolite inhibitions and aeration of bioreactor. Depending on these factors, several systems have been reviewed for recovery of aroma compounds from bioreactor system [1,53,54]. For instance, Tay [55] performed in situ product removal techniques for the production of isoamyl acetate from sugar beet molasses in bioreactor with batch and feed batch systems to avoid limiting factors. The researcher found that the maximum production of isoamyl acetate was 308.1 mg/L in the presence of macroporous resin H103 where only 38.4 mg/L of isoamyl acetate was produced without resin in batch fermentation.
Unavoidably discharge of the aroma compounds from the bioreactor system with the exhaust gas de ned as stripping which is considered the main reason for losses of aroma compounds during fermentation in the aerated bioreactors [49]. Urit el al. [49] modelled the stripping process for the production of ethyl acetate by K. marxianus on whey. The researchers expressed that the stripping of ethyl acetate highly depended on the aeration rate but was independent of the phase-transfer coe cient and the cooling the exhaust gas condenser of bioreactor system did not in uence the stripping of the ethyl acetate.
Relationship between aroma pro le and sensory characteristics of fermented rice bran Total 13 descriptive avor terms were de ned in fermented and unfermented rice bran samples by the panelists and signi cant differences were observed between the rice bran samples in terms of boiled corn, sweet aromatic, fermented cereal, wet bran and rose aromas (Fig. 3). According to this, boiled corn was found to be the highest in the unfermented rice bran samples as control and no signi cant differences were observed between in rice bran samples fermented K. marxianus and D. hansenii regarding this aroma. In the same way, it was determined that both fermented rice bran samples had a higher fermented cereal avor than control samples. Rose aroma in rice bran fermented by K. marxianus was determined noticeably higher than others.
For revealing of the relationship between sensory properties and avor compounds in fermented rice bran solution, Principal Component Analysis (PCA) was performed based on the ndings of sensory evaluation and GC-O analysis. It was determined that two principal components expressed by PCA the variations of sensory and aroma pro les of the rice bran samples. While the component PC1 was 59.53%, 40.47% of variations was expressed by component PC2 (Fig. 4). It can be con rmed that the fermented and unfermented rice bran samples have different sensory and aroma pro les. While rice bran samples fermented by K. marxianus can be characterized by rose aroma. Volatile compounds including phenyl ethyl alcohol, phenyl ethyl acetate, isoamyl acetate and guaiacol are well associated with rose aroma in fermented rice bran samples by K. marxianus. On the other hand, fermented cereal, polish and wood aromas well characterized in rice bran samples fermented D. hansenii and these aromas was found to related with 2-heptanal, 2,5,dimethyl-3ethylpyrazine, methanethiol, acetic acid and 2-heptanone presented in the rice bran samples fermented by D. hansenii (PC2). In case of unfermented rice samples, boiled corn, boiled wheat porridge, wet bran and dough aromas well characterized in unfermented rice bran and they are associated with 1-octen-3-ol, isobutyl acetate, diacetyl, 2-acetyl-1-pyrroline, 2-acetyl-2-thiazoline and (E)-2-nonenal (PC1).
Due to limited number of sensory studies in fermented rice bran for production of avor compounds, the ndings in the present study could not compared with the literature results. However, aroma production behaviors of K. marxianus and D. hansenii on tomato and pepper pomace were evaluated in our previous study [11]. Rose aroma was also determined in pepper pomace fermented by K. marxianus and D. hansenii and a higher wet bulgur aroma observed in tomato pomace fermented by D. hansenii. In a study by Kılmanoğlu et al. [43], rose aroma was also determined at high intensity in tomato pomace hydrolysate fermented by K. marxianus during 8.5 hours. Similar to our ndings, İşleten-Hoşoğlu [56] reported that K. marxianus produced higher alcohol and acetate esters including isoamyl alcohol, phenyl ethyl alcohol, isoamyl acetate and phenyl ethyl acetate in synthetic culture media supplemented with different carbon and nitrogen sources. The researcher determined that "sourdough", " ower" and "sweet aromatic" were characteristic descriptors for fermented culture media. Sweet aromatic and caramel aromas were found to associate with isoamyl acetate and phenylethyl propionate based on multidimensional scale analysis.
D. hansenii commonly found as natural oral yeast for some surface ripened cheese and dry meat products and it is considered as potential avor producer [57,58]. Sørensen et al. [58] reported that strain of D. hansenii D18335 primarily produced branched-chain aldehydes and their alcohol derivatives including 2-methylpropanal and 2/3-methylbutanal, 2-methyl-1-propanol and 2/3-methyl-1butanol in a cheese-surface model. While branched-chain aldehydes produced by D. hansenii are associated with green, malty, chocolate, and almond, alcohols are associated fruity, solvent, alcoholic, fusel and pomace aromas. In another study by Leclercq-Perlat et al. [59], methyl ketones with fruity, oral moldy, cheesy, wine aromas and 2-phenyl ethyl alcohol with rose aroma were determined in deacidi ed model cheese medium generated by D. hansenii. I the present study, 2-methylpropanal and 2/3-methylbutanal and their alcohols derivatives by the metabolism of D. hansenii in rice bran were not determined. This difference might be attributed to nature of raw materials used in the studies and strain differences. On the other hand, the production of phenyl ethyl alcohol associated with rose aroma in fermented rice bran were determined in the present study which was similar to the ndings of Leclercq-Perlat et al. [59].

Conclusion
Several researchers focused on the potential usage of agro wastes due to their sustainable utilizations for eco-friendly processes and low costs. The observed results in the present study showed that higher alcohols including isoamyl alcohol and phenyl ethyl alcohol and their acetate esters can be produced from rice bran by using K. marxianus and D. hansenii. The microbial growths of both yeasts in rice bran were determined to range between 2.30-3.69 log CFU/mL depending on fermentation conditions. In bioreactor fermentation, K. marxianus can produce 827.27 µg/kg of isoamyl alcohol, 169.77 µg/kg of phenyl ethyl alcohol and 216.08 µg/kg phenyl ethyl acetate during 24 h batch fermentation. Isovaleric acid was also found produced by K. marxianus at high level (4013 µg/kg) in the batch fermentation. The highest concentrations of isoamyl alcohol and phenyl ethyl acetate were determined as 415.64 µg/kg and 135.77 µg/kg, respectively at 24 h fermentation by D. hansenii. Overall, the rice bran could be valorized for the natural avor production by biotechnological process. Further studies are required for revealing the avor production of certain yeasts in different food by-products or agro wastes. Moreover, the studies on modeling of stripping effect in the bioreactor fermentation should also be performed for the microbial production of avor compounds from rice bran to obtain higher productivity rates.