Integrated Process of Arabinose Biopurification and Xylitol Fermentation Based on the Diverse Action of Candida boidinii

Agricultural and agro-industrial residues (agro-residues) containing high amounts of carbohydrates, such as cellulose and hemicellulose, are produced in large amounts annually worldwide.1,2 Agro-residues are attractive raw materials of biorefinery processes producing value-added products due to their relatively low commercial value, renewability and abundant availability.3 Biorefining of agro-residues for biofuel and bioproducts does not compete with food production and is considered to be advantageous from the environmental point of view, as it contributes to waste management, and its products can replace fossil-derived ones.4,5 One of the most important steps in lignocellulose-based biorefining technologies is the fractionation of the lignocellulosic biomass into its core constituent (hemicellulose, cellulose and lignin). Several studies proved the applicability of dilute acid-catalysed hydrolysis to selectively solubilise the hemicellulose fraction, therefore it is widely used as part of lignocellulose fractionation.2,3 Hemicellulosic hydrolysates containing high amounts of xylose have great potential as a raw material for microbial xylitol production1, while hemicellulosic hydrolysates with considerable amounts of arabinose might be suitable for arabinose production.6 L-Arabinose can be used as a non-caloric sweetener in the food industry; moreover, it strongly inhibits intestinal sucrase uncompetitively and consequently inhibits the absorption of sucrose in the small intestine, which makes it a promising sugar substitute in treatment of diabetes.6,7 L-Arabinose can be utilized for bacterial diagnostics,8 as a starting material in the synthesis of non-ionic surfactants9 and as intermediate for antiviral drug synthesis.10 Commercial production of L-arabinose is a complex process consisting of acid hydrolysis of gum arabic, followed by multiple procedures of purification.6 The relatively high cost of gum arabic and the expensive purification steps result in the high cost of pure arabinose,11 which has evoked an increasing research effort to develop new cost-effective methods of arabinose production from agro-residues rich in hemicellulose or pectin, for example from sugar beet pulp,11 corn hull,12 xylose-mother liquor,6 wheat bran13 and corn fibre.14 Biopurification of hemicellulosic hydrolysate is an interesting and inexpensive strategy to produce a pure arabinose solution through the depletion of other sugars (e.g. glucose, xylose, galactose) using Integrated Process of Arabinose Biopurification and Xylitol Fermentation Based on the Diverse Action of Candida boidinii


Introduction
Agricultural and agro-industrial residues (agro-residues) containing high amounts of carbohydrates, such as cellulose and hemicellulose, are produced in large amounts annually worldwide. 1,2gro-residues are attractive raw materials of biorefinery processes producing value-added products due to their relatively low commercial value, renewability and abundant availability. 3Biorefining of agro-residues for biofuel and bioproducts does not compete with food production and is considered to be advantageous from the environmental point of view, as it contributes to waste management, and its products can replace fossil-derived ones. 4,5ne of the most important steps in lignocellulose-based biorefining technologies is the fractionation of the lignocellulosic biomass into its core constituent (hemicellulose, cellulose and lignin).Several studies proved the applicability of dilute acid-catalysed hydrolysis to selectively solubilise the hemicellulose fraction, therefore it is widely used as part of lignocellulose fractionation. 2,3Hemicellulosic hydrolysates containing high amounts of xylose have great potential as a raw material for microbial xylitol production 1 , while hemicellulosic hydrolysates with considerable amounts of arabinose might be suitable for arabinose production. 6Arabinose can be used as a non-caloric sweetener in the food industry; moreover, it strongly inhibits intestinal sucrase uncompetitively and consequently inhibits the absorption of sucrose in the small intestine, which makes it a promising sugar substitute in treatment of diabetes. 6,7L-Arabinose can be utilized for bacterial diagnostics, 8 as a starting material in the synthesis of non-ionic surfactants 9 and as intermediate for antiviral drug synthesis. 10Commercial production of L-arabinose is a complex process consisting of acid hydrolysis of gum arabic, followed by multiple procedures of purification. 6The relatively high cost of gum arabic and the expensive purification steps result in the high cost of pure arabinose, 11 which has evoked an increasing research effort to develop new cost-effective methods of arabinose production from agro-residues rich in hemicellulose or pectin, for example from sugar beet pulp, 11 corn hull, 12 xylose-mother liquor, 6 wheat bran 13 and corn fibre. 14iopurification of hemicellulosic hydrolysate is an interesting and inexpensive strategy to produce a pure arabinose solution through the depletion of other sugars (e.g.glucose, xylose, galactose) using the adequate microorganisms. 15The drawbacks of the biopurification process are wasting of the potential of other sugars by converting them into value-added products and the formation of a considerable amount of cell mass as by-product.
Xylitol has a potential use in the pharmaceutical, odontological and food industries as an alternative sweetener. 16Besides its anticariogenic properties, xylitol can be used for diabetes treatments, since its metabolism is partially independent of insulin. 17On an industrial scale, xylitol is produced through the chemical reduction of xylose derived from hemicellulosic hydrolysate of birchwood or other xylose-rich materials. 1The purification and separation steps required to remove other by-products from xylose and xylitol makes the process relatively expensive. 1As an alternative method, microbial production of xylitol is becoming more attractive, since the downstream processing is expected to be cheaper and, unlike the chemical route, mild reaction conditions are required. 16Many studies have been conducted to produce xylitol from the hemicellulose portion of agro-residues, like rice straw, 18 corn cob, 19 brewer's spent grain, 20 sugarcane bagasse, 21 corn stover, 22 barley bran 23 and corn fibre 24 using microbial processes.Xylitol can be produced by some bacteria and filamentous fungi, but the best producers are yeasts, especially species of genus Candida, such as C. guilliermondii, C. pelliculosa, C. parapsilosis, C. tropicalis and C. boidinii. 25Microbial production of xylitol from hemicellulose hydrolysate is influenced by several factors including the strain, the fermentation conditions (pH, dissolved oxygen concentration, initial cell concentration, temperature) employed in the process and the composition of the fermentation medium (initial xylose concentration, concentrations of other sugars and the ratios of these sugars, concentrations of inhibitor compounds and nutrients). 16,26he use of high initial cell concentration results in increased volumetric productivity and reduces the toxic effect of inhibitors, 26 however, a considerable amount of carbon source is required to obtain an inoculum of high cell density.
In this study, arabinose biopurification and xylitol fermentation were investigated on a semidefined medium and real hemicellulosic hydrolysate derived from corn fibre, and an integrated process was developed, in which the cell mass obtained as by-product in the biopurification step was utilized to inoculate the xylitol fermentation step.This process resulted in more effective carbon utilization, as the cell propagation for the xylitol fermentation did not require an additional carbon source or it did not consume xylose convertible into xylitol in the xylitol fermentation.

Corn fibre
Corn fibre was kindly donated by Hungrana Starch and Isosugar Manufacturing and Trading Co. Ltd. (Szabadegyháza, Hungary).It was dried and stored at room temperature.

Yeast strains and inoculum preparation
Candida boidinii NCAIM Y.01308, Candida parapsilosis NCAIM Y.01011, Candida guilliermondii (Pichia guilliermondii) NCAIM Y.01050, Hansenula anomala (Pichia anomala) Y.01499 were purchased from the National Collection of Agricultural and Industrial Microorganisms (Budapest, Hungary).Yeast strains were stored and maintained on agar slants containing 1 % (w/w) glucose, 1 % (w/w) peptone, 0.3 % (w/w) yeast extract and 2 % (w/w) agar at room temperature and at 4 °C, respectively.The medium used for inoculum preparation (pH = 6) contained 10 g L -1 yeast extract, 15 g L -1 KH 2 PO 4 , 1 g L -1 MgSO 4 • 7H 2 O, 3 g L -1 (NH 4 ) 2 HPO 4 and 30 g L -1 xylose.The solutions containing xylose and the other components were sterilized separately at 120 °C for 15 minutes in an autoclave.The cells were cultivated in 750-mL cotton-plugged Erlenmeyer flasks containing 150 mL of inoculum medium at 220 rpm rotation speed in a rotary shaker at 30 °C for 72 h, subsequently recovered by centrifugation (1000 × g, 5 min), washed with sterile distilled water, and the adequate amount of cell mass was directly resuspended in the xylitol fermentation and biopurification media.

Two-step acidic fractionation of corn fibre
The fractionation process of corn fibre includes two sequential hydrolyses catalysed by sulphuric acid (Fig. 1).The first acidic hydrolysis was carried out in 1000-mL closed glass-flasks containing 800 g of corn fibre suspensions at 90 °C for 51 minutes (plus 15 minutes warm-up period) without agitation in water bath.The corn fibre suspensions contained 10 % (w/w) dry matter and 1.1 % (w/w) sulphuric acid.Subsequently, the flasks were cooled in coldwater (10 °C) for 1 minute and the solid fractions were separated by vacuum filtration through a 150 µm-pore size nylon filter (Rimóczi és Társa Ltd., Hungary).The supernatants were collected and stored at -10 °C (first hydrolysate).The solid fractions were washed with distilled water until neutral pH, collected, then dried and stored at 40 °C (first solid residue).The first hydolysate was treated at 120 °C for 60 minutes to decompose the oligosaccharides.The solution obtained is referred to as glucose-and arabinose-rich hydrolysate.The first solid residue was used in the second acidic hydrolysis step, which was performed in 500-mL glass-flasks containing 300 g suspensions at 120 °C for 30 minutes in an autoclave.The suspensions of the first solid residue contained 10 % (w/w) dry matter and 1.1 % (w/w) sulphuric acid.After the flasks were removed from the autoclave, they were cooled in coldwater (10 °C) for 1 minute, and the solid fractions were separated by vacuum filtration through a nylon filter (150 µm).The supernatants were collected and stored at -10°C, and called as xylose-rich hydrolysate.The solid fractions were washed with distilled water until neutral pH, collected, then dried and stored at 40 °C (cellulose-rich solid fraction).

Biopurification, medium preparation
Biopurifications were carried out at 30 °C in a rotary shaker (220 rpm) for 96 hours in 100-mL cotton-plugged Erlenmeyer flasks containing 20 mL of the semidefined medium or glucose-and arabinose-rich hydrolysate of corn fibre, and monitored through daily sampling.The semidefined biopurification medium (pH = 6) contained 10 g L -1 yeast extract, 15 g L -1 KH 2 PO 4 , 1 g L -1 MgSO 4 • 7H 2 O, 3 g L -1 (NH 4 ) 2 HPO 4 , 15 g L -1 arabinose, 7.5 g L -1 xylose and 7.5 g L -1 galactose.The solution of the sugars and that of the other components were sterilized separately at 120 °C for 15 minutes in an autoclave.The glucose-and arabinose-rich hydrolysate was sterilized using the same conditions.Before sterilization, the pH of the glucose-and arabinose-rich hydrolysate (pH = 1) was adjusted to 6 by the addition of calcium hydroxide (powder).The precipitated gypsum was removed by filtration with a folding filter.

Xylitol fermentation, medium preparation
Xylitol fermentations were performed on a semidefined medium or xylose-rich hydrolysate of corn fibre at 30 °C in a rotary shaker for 96 hours, and monitored through daily sampling.The semidefined fermentation medium (pH = 6) contained 10 g L -1 yeast extract, 15 g L -1 KH 2 PO 4 , 1 g L -1 MgSO 4 • 7H 2 O, 3 g L -1 (NH 4 ) 2 HPO 4 , and 30 g L -1 or 70 g L -1 xylose.The solution of the xylose and that of the other components were sterilized separately at 120 °C for 15 minutes in an autoclave.The xylose-rich hydrolysate was sterilized using the same conditions.Before sterilization, pH adjustment and clarification of the xylose-rich hydrolysate were performed.The pH of xylose-rich hydrolysate (pH = 1) was adjusted to 6 by adding calcium hydroxide (powder).The precipitated gypsum was removed by filtration with a folding filter.After pH adjustment, the xylose-rich hydrolysate was clarified by activated carbon (0.05 g/100 g hydrolysate) at room temperature for 30 minutes with continuous agitation, subsequently, the activated carbon was removed by filtration with a folding filter.Activated carbon (Norit DX ULTRA 8005.3) was kindly donated by Cabot Norit Activated Carbon (Amersfoort, The Netherlands).Fermentations were carried out in 100-mL Erlenmeyer flasks closed with cotton plugs.In order to simulate different aeration conditions, three levels of medium volume (35, 50 and 65 mL) were used at two levels of rotation speed (125 and 220 rpm).Xylitol yields were considered to evaluate the experiments.Xylitol yield was calculated from the highest xylitol concentration obtained during the fermentation and expressed as percentage of theoretical.Theoretical xylitol yield was calculated from the initial xylose concentration by assuming a complete (stoichiometric) conversion.Xylitol volumetric productivities were also calculated at the time of the maximum xylitol concentration.

Compositional analysis
Composition of the carbohydrate fractions of corn fibre and the solid residues (first solid residue and cellulose-rich solid fraction) derived from fractionation of corn fibre were determined using the method of the National Renewable Energy Laboratory with minor modifications. 27Half a gram of dry matter was mixed with 2.5 mL of 72 % (w/w) sulphuric acid and the mixture was kept at room temperature for 2 hours.Then, 75 mL of distilled water were added and the suspension was treated at 120 °C in the autoclave for 1 hour in 100-mL glassflasks.The supernatant was analysed using high-performance liquid chromatography (HPLC) to determine its monosaccharide content.The starch content of the corn fibre was determined using thermostable a-amylase.Ground, air-dried corn fibre was suspended in a sodium acetate buffer (pH = 4.8, 100 mmol L -1 ) at 3 % (w/w) dry matter content, and then treated by thermostable a-amylase (20 mmol min -1 kg -1 dry matter) in 1-L closed glassflasks at 90 °C for 3 hours with continuous agitation (250 rpm) in a water bath.The thermostable a-amylase enzyme preparation with a volumetric enzyme activity of 4 mmol min -1 mL -1 ) (defined on starch substrate at 25 °C) was donated by Hungrana Starch and Isosugar Manufacturing and Trading Co. Ltd.The supernatant was separated by vacuum filtration through a nylon filter (150 μm), mixed with 8 % (w/w) sulphuric acid at a volume ratio of 1:1, and treated at 120 °C in an autoclave for 15 minutes to decompose the oligosaccharides.Then it was analysed for glucose by HPLC.The cellulose content was calculated as the difference of the total glucan and starch content.The compositional analysis was carried out in triplicate.
Cell concentration in the inoculum and fermentation samples was calculated from the optical density of the sample using a calibration curve based on the relationship of optical density and cell dry weight.Cell dry weight was determined gravimetrically after separation of a certain volume of inoculation broth (48 hours) by centrifuge (1000 × g, 5 minutes), washing it with distilled water, and drying the cells at 105 °C.Optical density was determined by spectrophotometer (Ultrospec III, Pharmacia LKB, Uppsala, Sweden) at a wavelength of 600 nm.
Total phenolics content was estimated using Folin-Ciocalteu reagent according to the method described by Guo et al. 28 Gas-liquid mass transfer coefficients (K L a) of oxygen from the headspace of the flask to the media of xylitol fermentation (semidefined medium and xylose-rich hydrolysate) were determined by using a non-fermentative gassing-out method. 29A 100-mL Erlenmeyer flask was equipped with an optical oxygen sensor (VisiFerm DO 120, HAMILTON Bonaduz AG, Switzerland) to measure dissolved oxygen concentration (C).After gassing out of the fermentation media with nitrogen, the increase in the dissolved oxygen concentration due to the shaking of the flask was measured until constant level of the dissolved oxygen concentration (C * ).The measurements were performed at 30 °C by using different levels of medium volume (35, 50 and 65 mL) and rotation speed (125 and 220 rpm).The values of the expression -ln(1-(C/C * )) were plotted as a function of time.The slope of the fitted linear curve gave the value of K L a (h -1 ). 29The maximum oxygen transfer rate (OTR) was calculated by multiplying K L a (h -1 ) and C * (mmol L -1 ).

Arabinose biopurification on semidefined medium
The capability of the investigated yeast strains for arabinose biopurification was tested on a semidefined medium containing sugars typically present in hemicellulosic hydrolysates, such as xylose, arabinose and galactose.The aim of the biopurification is the selective depletion of other sugars beside the arabinose by using the adequate microorganism, which results in a pure arabinose solution.The absence of other sugars enables the effective crystallization of pure arabinose. 15During aerobic biopurification, the undesired sugars are utilized mainly to produce cell mass, which occurs as a by-product of the process.Biopurification offers an inexpensive method to purify arabinose from hemicellulosic hydrolysates, however due to the loss of the other valuable sugars, this method might be reasonable only in the case of utilization of arabinose-rich hemicellulosic hydrolysates.Hence, the concentration of arabinose in the semidefined medium used in this study was two times higher than the concentrations of xylose and galactose.Four xylitol-fermenting yeast strains, namely C. boidinii, C. guilliermondii, C. parapsilosis and H. anomala were cultured in shake flasks under aerobic conditions to investigate their sugar consumption from the mixed sugar solution.
C. guilliermondii, C. parapsilosis and H. anomala utilized xylose, galactose and arabinose simultaneously to produce cell mass.C. parapsilosis consumed all of the sugars in two days, C. guillier mondii and H. anomala depleted xylose and galactose in two days, and arabinose in three days (data not shown).Hence, these strains are found to be inappropriate for arabinose biopurification from hemicellulosic hydrolysate.C. boidinii metabolized almost all of the xylose in one day (Fig. 2).The galactose concentration continuously decreased, 85 % of the initial was consumed in two days and 7 % of the initial (0.4 g L -1 ) remained after four days (Fig. 2).C. boidinii did not consume arabinose, even if the other carbon sources were present in negligible quantities (Fig. 2).The cell concentration increased from 0.5 g L -1 to 5.5 g L -1 during the biopurification (Fig. 2), which corresponded to a cell yield of 0.38 g dry cell mass/g sugar consumed.
The purity of the arabinose solution was defined as the ratio of arabinose content to total sugar content, in percentage.After four days, the biopurified medium contained 13 g L -1 arabinose and 0.4 g L -1 galactose, which resulted in a purity of 97 %.As C. boidinii was found to be appropriate for arabinose biopurification on semidefined media, it was selected for further investigations.

Xylitol fermentation on semidefined medium
Xylitol fermentations were performed on a semidefined medium to determine the most favourable conditions for xylitol production by C. boidinii.One of the most important factors affecting microbial production of xylitol is the aeration. 16In order to simulate different aeration conditions in shake flasks, the fermentations were performed at three levels of filling ratio (0.35, 0.5 and 0.65), each of them at two levels of rotation speed (125 and 220 rpm) (Table 1).The filling ratio is defined as the ratio of the medium volume to the flask volume.A higher filling ratio results in lower oxygen supply, while a higher rotation speed intensifies the dissolution of oxygen, during shake flask fermentations.In order to confirm that, OTR values of the different settings of filling ratio and rotation speed were determined (Table 1).The effects of initial cell concentration, initial xylose concentration, and methanol addition were also investigated.The different sets of the variables are shown in Table 1.Fermentations were carried out for four days.The investigation of different aeration conditions was performed with 1 g L -1 initial cell concentration and 30 g L -1 initial xylose concentration (Table 1).Xylitol fermentations under these initial conditions using C. boidinii followed the same trends.One example is demonstrated in Fig. 3a.Xylitol concentration keeps increasing as long as xylose is available in the fermentation broth.After xylose is depleted, C. boidinii starts to consume xylitol, hence xylitol concentration has a maximum value during the fermentation.Therefore, in the case of industrial implementation, it is crucial to terminate the process at the highest xylitol concentration, for which continuous monitoring of xylitol and xylose concentrations is necessary.Small amounts of ethanol is always produced simultaneously with xylitol, however, after xylose depletion the ethanol is also consumed.The cell mass continuously increased, which indicated that xylitol and ethanol were used to form cell mass after xylose depletion.Similar behaviour was observed by Walther et al. using C. tropicalis. 30t the rotation speed of 220 rpm, the OTR value decreased from 6.6 mmol L -1 h -1 to 5.2 mmol L -1 h -1 by increasing the filling ratio from 0.35 to 0.65 (Table 1).At the filling ratios of 0.35, 0.5 and 0.65 xylitol yields of 22 %, 25 % and 28 % were obtained in two days (Table 1), which resulted in volumetric productivities of 0.14 g L -1 h -1 , 0.16 g L -1 h -1 and 0.18 g L -1 h -1 , respectively.In the case of 125 rpm rotation speed, the highest xylitol yield, 44 %, was achieved at the filling ratio of 0.5, where an OTR value of 2.8 mmol L -1 h -1 was measured (Table 1).
The maximum xylitol concentration was observed in two days, which resulted in 0.28 g L -1 h -1 volumetric productivity.At 0.35 and 0.65 filling ratios, 4.2 mmol L -1 h -1 and 1.9 mmol L -1 h -1 OTR was observed, respectively (Table 1).At 0.35 filling ratio, a xylitol yield of 36 % was obtained in one day (Ta-Ta b l e 1 -Fermentation conditions, maximal xylitol yields achieved on semidefined media and times required to achieve the maximal yield.Standard deviations are calculated from triplicates and indicated in parenthesis.

Rotation speed (rpm)
Filling ratio OTR (mmol L -1 h -  ble 1), resulting in a volumetric productivity of 0.46 g L -1 h -1 .At the filling ratio of 0.65, the xylitol yield was 27 %, which was obtained in two days (Table 1).The volumetric productivity was 0.17 g L -1 h -1 .
In terms of the xylitol yields achieved, the most favourable aeration condition in shake flask fermentation using C. boidinii was obtained at the filling ratio of 0.5 with 125 rpm rotation speed, hence such conditions were set for the subsequent experiments.These conditions of aeration can be considered as microaerobic conditions according to Walther et al. 30 The xylitol fermentation performed at 1 g L -1 initial cell concentration and 30 g L -1 initial xylose concentration under microaerobic conditions is referred to as the base case in the following sections.
The effect of high initial xylose concentration for the fermentative capacity of C. boidinii was investigated using 70 g L -1 initial xylose concentration (Fig. 3b).The high initial xylose concentration resulted in a xylitol yield of 40 % (Table 1), which is slightly lower compared to that obtained in the base case (44 %).The xylitol concentration continuously increased until the end of the fermentation (Fig. 3b), resulting in a volumetric productivity of 0.3 g L -1 h -1 .The continuous increase in xylitol concentration through four days was due to the presence of xylose during the entire fermentation.After four days, 10 g L -1 xylose remained in the medium, hence the cells did not consume any accumulated xylitol (Fig. 3b).The ethanol formation showed a similar trend to that of xylitol (Fig. 3b).The ethanol concentration increased until 6 g L -1 (Fig. 3b).The cell mass also continuously increased during the fermentation, which resulted in a final cell concentration of 4.5 g L -1 (Fig. 3b).Although it has been published in many studies that high initial xylose concentration increases the xylitol yield, 1,16 it was not observed in this study.The relatively high amount of ethanol accumulated during the fermentation could negatively affect the xylitol production. 30Vandeska et al. investigated the effect of initial xylose concentration on xylitol production of C. boidinii NRRL Y-17213 by varying xylose concentration from 20 g L -1 to 200 g L -1 , and it was found that 150 g L -1 initial xylose concentration is the most favourable, as it results in a xylitol yield of 0.47 g g -1 xylose consumed. 31In our study, similar xylitol yield (0.46 g g -1 xylose consumed) was obtained using 70 g L -1 initial xylose concentration and 1 g L -1 initial cell concentration.However, Vongsuvanlert and Tani reported that increasing xylose concentration up to 150 g L -1 resulted in lower xylitol production than that with 100 g L -1 of xylose, using C. boidinii no.2201. 32High initial xylose concentration negatively affects the xylitol production, which might be due to the osmotic stress on the cells of C. boidinii. 32e effect of high cell density on the fermentation process was investigated using 5 g L -1 initial cell concentration.High initial cell density resulted in significantly higher xylitol yield within shorter fermentation time compared to the base case, as 58 % xylitol yield was achieved in one day (Table 1).The volumetric productivity of the fermentation was 0.73 g L -1 h -1 .The cell concentration was almost constant throughout the fermentation, less than 1 g L -1 was formed (Fig. 3c), in contrast with the cases of low (1 g L -1 ) initial cell density (Fig. 3a and 3b).Due to the reduced cell formation, more xylose was available in the cells for xylitol production, which probably contributed to the higher xylitol yield.The ethanol formation was also reduced, only 1 g L -1 was produced in two days (Fig. 3c), which is half of that obtained in the base case (Fig. 3a).Hence, high initial cell concentration was found to be favourable in terms of the xylitol yield, the xylitol productivity, and the ratio of xylitol to ethanol.The improvement of xylitol yield (g xylitol/g xylose consumed) by increasing the initial cell density of C. boidinii NRRL Y-17213 was reported by Vandeska et al. 31 Methanol addition is considered to be favourable for polyols production using methylotrophic yeasts, since the oxidation of methanol results in the formation of NADH, which is needed for the reduction of sugars. 1 However, the addition of methanol as a co-substrate did not result in a significant increase in xylitol yield (Fig. 3d vs. 3c).A xylitol yield of 60 % was obtained when 12 g L -1 of methanol was added to the fermentation broth containing 30 g L -1 of initial xylose and 5 g L -1 of initial cell concentrations (Table 1).However, a longer time was needed (2 days) to reach the highest xylitol concentration compared to that without methanol addition (1 day) (Fig. 3d vs. 3c), which resulted in a volumetric productivity of 0.38 g L -1 h -1 .The methanol concentration decreased until 10 g L -1 in four days (Fig. 3d).The cell concentration slowly increased during the fermentation, and 6 g L -1 of final cell concentration was obtained (Fig. 3d).Only 0.5 g L -1 ethanol, half of that obtained without methanol addition, was produced simultaneously with xylitol, which indicates that methanol addition can repress the formation of ethanol (Fig. 3d).In contrast, Vongsuvanlert and Tani reported that xylitol production using C. boidinii no.2201 had significantly increased by adding methanol, and the methanol addition resulted in increased ethanol formation. 32ccording to our investigation of xylitol production on semidefined media using C. boidinii, aeration and initial cell density have the greatest effects on xylitol yield.Xylitol production is favoured under microaerobic conditions (2.8 mmol L -1 h -1 OTR) using increased (5 g L -1 ) initial cell concentration.

Fractionation of corn fibre
The fractionation process of corn fibre includes two sequential hydrolyses catalysed by sulphuric acid (Fig. 1).Through the fractionation steps, xylose and galactose were determined as one component, which is referred to as xylose (+galactose).
In the first step of the fractionation, corn fibre was treated under mild acidic conditions to solubilise the starch content and liberate most of the arabinose. 14After the first hydrolysis, a considerable amount of oligosaccharides was obtained in the supernatant, thus an oligomer hydrolysis step was required to recover the sugars in monomeric form (Fig. 1).According to preliminary results, additional acid was not required to decompose oligomers of the first hydrolysate, when the oligomer hydrolysis step was carried out at 120 °C for 1 hour (data not shown).After the first and oligomer hydrolyses, the glucose, arabinose and xylose (+galactose) were recovered in the supernatant with yields of 103 %, 74 % and 32 % of theoretical based on the raw material composition (Table 2), respectively.In this case, the glucose yield was calculated from the starch content of corn fibre as the cellulose was assumed to be intact under mild conditions applied in the first hydrolysis.Glucose yield of 103 % could be explained by the presence of a small amount of glucose in the hemicellulose fraction, which can be easily hydrolysed under mild acidic treatments. 14The glucose-and arabinose-rich hydrolysate contained 15 g L -1 glucose, 8.4 g L -1 xylose (+galactose), 10.5 g L -1 arabinose and 0.9 g L -1 acetic acid.It was utilized as the fermentation medium for biopurification after pH adjustment (pH = 6) and sterilization.
The solid residue of the first step was utilized in the second acidic hydrolysis to produce the xylose-rich hydrolysate (Fig. 1).Glucose, xylose (+galactose) and arabinose were recovered in the supernatant with the yields of 10 %, 88 % and 63 % of theoretical based on the composition of the first solid residue, respectively.The lower yield of arabinose than that of xylose might be due to arabinose degradation.The low glucose recovery during the second hydrolysis resulted in a cellulose-rich solid fraction (Fig. 1) containing 50 % cellulose based on dry weight.The xylose-rich liquid fraction contained 3 g L -1 glucose, 28 g L -1 xylose (+galactose), 6.6 g L -1 arabinose, and 2.3 g L -1 acetic acid.The total phenolics content was 5.3 g L -1 .The xylose-rich hydrolysate was utilized for xylitol production after pH adjustment (pH = 6), treatment with activated carbon and sterilization.

Arabinose biopurification on glucoseand arabinose-rich hydrolysate of corn fibre
The glucose-and arabinose-rich hydrolysate derived from the fractionation process of corn fibre was utilized to produce a pure arabinose solution through biopurification using C. boidinii.Biopurification was carried out for three days.The biopurification medium contained 13.5 g L -1 glucose, 9.2 g L -1 arabinose, 5.8 g L -1 xylose and 2.3 g L -1 galactose after the inoculation, and the initial cell concentration was 0.5 g L -1 (Fig. 4).It also contained 0.8 g L -1 acetic acid.During the biopurification of glucose-and arabinose-rich hydrolysate of corn fibre, the consumptions of glucose, xylose and galactose occurred simultaneously, however with different consumption rates.The consumption rate of glucose was the highest, followed by that of xylose, while the galactose consumption was the slowest (Fig. 4).Glucose and xylose were totally depleted in one day and two days, respectively (Fig. 4).Galactose was not consumed completely, 32 % and 57 % of the initial amount were removed in one day and four days, respectively (Fig. 4).Acetic acid was consumed within one day (data not shown).Most of the cell mass (~4.5 g L -1 ) was formed in one day; however cell concentration increased continuously until 6.5 g L -1 (Fig. 4).Beside the cell formation, a considerable amount of ethanol was produced (2 g L -1 ) in one day, however it was consumed within the next day (Fig. 4).In the case of biopurification of glucose-free semidefined medium, ethanol formation was not observed, which might indicate that the ethanol was formed from glucose during biopurification of the corn fibre hydrolysate.To find the exact explanation of ethanol production under aerobic biopurification, further investigation of the metabolism of C. boidinii is required.The arabinose concentration was constant through the whole biopurification process (Fig. 4).After three days of biopurification, the medium contained 9.2 g L -1 arabinose and 1 g L -1 galactose (Fig. 4), i.e., the purity of arabinose was 90 % of total sugars, which exceeds the purity requirement for crys-  6 The purity of arabinose obtained under optimized conditions of biopurification was 86 % of total sugars, the optimal time of biopurification was 75 hours.Park et al. developed a method to produce arabinose from purified corn fibre arabinoxylan by enzymatic hydrolysis and arabinose biopurification. 15Williopsis saturnus yeast strain was selected to perform the arabinose biopurification on enzymatic hydrolysate of corn fibre arabinoxylan.Almost all of the xylose was consumed in three days with no loss of arabinose however data of the concentrations of other components were not reported.Compared to these methods, arabinose biopurification from corn fibre hydrolysate using C. boidinii is a competitive strategy.

Xylitol fermentation on xylose-rich hydrolysate of corn fibre, integration of xylitol fermentation and arabinose biopurification
The xylose-rich hydrolysate derived from the fractionation of corn fibre was utilized to produce xylitol, however it contained a significant amount of phenolic compounds, which are strong inhibitors of microbial xylitol production. 26In order to remove phenolic substances, the xylose-rich hydrolysate was treated with activated carbon before xylitol fermentation (Fig. 1).Activated carbon treatment is widely considered as an inexpensive and efficient method to reduce the content of inhibitory compounds in hemicellulosic hydrolysates. 26After the activated carbon treatment, the total phenolic con-tent reduced to 0.3 g L -1 .Xylose concentration during activated carbon treatment did not change.The initial composition of the xylitol fermentation medium was the following: 2.2 g L -1 glucose, 24.1 g L -1 xylose (+galactose) and 6.1 g L -1 arabinose.The acetic acid concentration (2.3 g L -1 ) had not changed significantly through the treatments of the xylose-rich hydrolysate.
The fermentation was performed under microaerobic conditions (2.8 mmol L -1 h -1 OTR) using 5 g L -1 initial cell concentration.The C. boidinii cell mass used to inoculate the fermentation was harvested from the arabinose biopurification (Fig. 1), in contrast with the fermentations on semidefined medium, in which cell mass for inoculation was propagated on semidefined inoculum medium.This kind of integration of xylitol fermentation and arabinose biopurification (Fig. 1) enables the utilization of the by-product cell mass of biopurification and results in a more effective carbon utilization, as the cell propagation of xylitol fermentation does not require additional carbon source or it does not consume xylose convertible into xylitol in the fermentation step.Utilization of on-site by-products and integration of different production routes are crucial for developing a viable biorefinery process.
The profiles of utilization of sugars and formation of products through xylitol fermentation of the integrated process are shown in Fig. 5. Glucose was present in the medium in a small concentration and it was depleted within one day (Fig. 5).The arabinose concentration was unaltered throughout the fermentation (Fig. 5).The cell concentration was almost constant, less than 0.5 g L -1 was produced in four days (Fig. 5).Xylitol concentration increased until the third day, when it was 10.4 g L -1 (Fig. 5), resulting in a volumetric productivity of 0.14 g L -1 h -1 .This concentration corresponds to a xylitol yield of 43 % of theoretical based on the initial concentration of xylose (+galactose).However, according to our previous results, 80 % of the component measured as xylose (+galactose) in the xylose-rich hydrolysate was xylose, thus the initial xylose concentration can be assumed to be 19.3 g L -1 .By using this assumption during the calculation of the concentration of theoretically achievable xylitol, a xylitol yield of 53 % was obtained.This is only 5 % less, than the xylitol yield of the case performed under the same conditions (microaerobic aeration, 5 g L -1 initial cell concentration) on semidefined media, however, on semidefined media the xylitol yield was close to the maximum in one day (Fig. 3c).During the fermentation of xylose-rich hydrolysate, a higher amount of ethanol (3 g L -1 ) was produced in two days, than that obtained under the same conditions on semidefined medium (1 g L -1 ) (Fig. 5 vs Fig. 3c).The increased ethanol formation might contribute to the reduced yield and slow accumulation of xylitol.After the third day of xylitol fermentation on xylose-rich hydrolysate of corn fibre, the fermentation broth contained 10.4 g L -1 xylitol, 6.1 g L -1 arabinose, 4.1 g L -1 xylose (+galactose) and 2.7 g L -1 ethanol (Fig. 5).As the xylitol concentration is approximately the same as the concentration of the residual sugars, further purification steps might be required to enable xylitol crystallization from the broth.
Leathers and Dien developed a two-stage, sequential fermentation process for xylitol and arabitol production from neutralized and deionized corn fibre hydrolysate using P. guilliermondii. 33This strategy resulted in a xylitol yield of 0.27 g xylitol/g initial xylose within four days, which yield is around half of that achieved in our process.Rao et al. investigated xylitol production from corn fibre hydrolysate, which was neutralized, treated with activated charcoal and ion exchange resins. 24C. tropicalis cells were adapted by sub-culturing in hydrolyaste containing medium for 20 cycles.This method resulted in a xylitol yield of 0.58 g/g xylose utilized, within two days.In our process, a xylitol yield of 0.54 g/g xylose utilized was obtained within three days by assuming that all of the initial xylose was utilized, however the deionization step was not included.Buhner and Agblevor investigated different detoxification methods to produce xylitol from concentrated corn fibre hydrolysate by using C. tropicalis. 34The highest xylitol yield, 0.4 g/g xylose utilized, was obtained within four days in the case of the highest concentrations (three times the original hydrolysate) that had been partially neutralized by adding calcium hydroxide and treated with activated charcoal prior to the fermentation.Hence, xylitol production from corn fibre obtained in our integrated process is competitive with those methods found in the literature.

Conclusions
The two-step acidic fractionation of corn fibre was developed to produce a glucose-and arabinose-rich hydrolysate containing 15 g L -1 glucose, 10.5 g L -1 arabinose, 8.4 g L -1 xylose (+galactose), and a xylose-rich hydrolysate containing 28 g L -1 xylose (+galactose), 6.6 g L -1 arabinose, 3 g L -1 glucose.C. boidinii NCAIM Y.01308, a natural isolate yeast strain, was found to be suitable for producing an arabinose solution with a purity of 90 % from the glucose-and arabinose-rich hydrolysate through aerobic biopurification, as well as appropriate for xylitol production from the xylose-rich hydrolysate using high initial cell concentration (5 g L -1 ) under microaerobic conditions (2.8 mmol L -1 h -1 OTR).In order to eliminate the separate cell propagation step prior to xylitol fermentation and to solve the utilization of cell mass by-product obtained in biopurification, an integrated biorefinery process based on the diverse action of C. boidinii was developed, in which the cell mass produced in biopurification was used to perform the xylitol fermentation.Xylitol fermentation resulted in a xylitol yield of 53 % of theoretical in three days.Hence, an integrated biorefinery process was developed to purify arabinose and produce xylitol from hemicellulosic hydrolysates by using C. boidinii yeast strain.

F i g . 1 -
Process scheme of fractionation of corn fibre and integration of arabinose biopurification and xylitol fermentation.Process steps and material streams are indicated in capital letters and italics, respectively.

F i g . 3 -
Fermentation profiles of xylitol production on semidefined medium using Candida boidinii NCAIM Y.01308 under different conditions.Fermentations were carried out under microaerobic conditions (0.5 filling ratio, 125 rpm) using (a) 1 g L -1 initial cell concentration and 30 g L -1 initial xylose (base case), (b) 1 g L -1 initial cell concentration and 70 g L -1 initial xylose, (c) 5 g L -1 initial cell concentration and 30 g L -1 initial xylose and (d) 5 g L -1 initial cell concentration, 30 g L -1 initial xylose and 12 g L -1 methanol.Standard deviations are calculated from triplicates.

F i g . 4 -
Arabinose biopurification on glucose-and arabinose-rich hydrolysate of corn fibre using Candida boidinii NCAIM Y.01308.Standard deviations are calculated from triplicates.
6a b l e 2 -Carbohydrates and acetate content of corn fibre.Standard deviations are calculated from quintuplicate and indicated in parenthesis. of L-arabinose from a solution.6Theseresultsprove the applicability of C. boidinii to produce arabinose-rich solution from crude hemicellulosic hydrolysates through aerobic biopurification.Yeast mediated biopurification of arabinose from hemicellulosic hydrolysate was published by Cheng et al. (2011) and Park et al. (2001).Cheng et al. performed arabinose biopurification on xylose mother liquor using P. anomala. tallization