Biotechnological production of xylitol from agroindustrial residues

Batch, fed-batch, and semicontinuous fermentation processes were used for the production of xylitol from sugarcane bagasse hemicellulosic hydrolysate. The best results were achieved by the semicontinuous fermentation process: a xylitol yield of 0.79 g/g with an efficiency of 86% and a volumetric productivity of 0.66 g/L/h.


Introduction
Xylitol is a polyalcohol of five carbon atoms which is widely used in the food and chemical industry. Its interest is due to the sweetening power similar to that of sucrose. However, it has been shown that the use of xylitol as a sweetener is better due to its anticariogenecity, tooth rehardening and remineralization properties. It is suitable as sugar substitute for diabetics, and it limits the tendency to obesity when is continuously supplied in diet (Salgado et al., 2010a). Table 1 shows the properties of the food additives that are used as sweeteners and are permitted in the European Union (EU).
Xylitol can be found naturally in various fruits and vegetables such as strawberries, raspberries, yellow plum, lettuce and cauliflower (Prakasham et al., 2009). The extraction of xylitol from these products is low, for this reason is not a good source. It is produced mainly by chemical processes. This process is the hydrogenation of the five-carbon sugar D-xylose in the presence of nickel catalyst at elevated temperature and pressure (Prakasham et al., 2009). The chemical process has some drawbacks such as high cost of purification processes. To avoid aggressive stages of the chemical processes, biotechnological processes were studied for xylitol production. This alternative production is bioconversion of D-xylose to xylitol by microorganisms. To reduce costs and environmental problems renewable biomass from agro-industrial waste can be used as source of D-xylose.
The food, agricultural and forestry industries produce large volumes of wastes annually worldwide, causing a serious disposal problem (Rodríguez-Couto, 2008). In the EU waste hierarchy and legislation, prevention and minimization of waste is given the highest priority (Staniskis & Stasiskiene, 2005). The List of Wastes (formerly the European Waste Catalogue), is a catalogue of all waste types generated in the EU. This list of wastes is periodically reviewed on the basis of new knowledge and, in particular, of research results, and if necessary revised in accordance with Article 18 of Waste Framework Directive (75/442/EEC). The wastes studied in the current work, listed in Table 2, are included in paragraph 2 and 20 of the European List of Waste "Wastes from agricultural, horticultural, hunting, fishing and aquacultural primary production, food preparation and processing" and "Municipal wastes and similar commercial, industrial and institutional wastes including separately collected fractions", respectively. Their reutilization, as well as other wastes, is of great interest since, due to legislation and environmental reasons, the industry is increasingly being forced to find an alternative use for its residual matter, at the same time that the use of these wastes considerably reduces the production costs (Rodríguez-Couto, 2008).
Lignocelluloses in nature derive from wood, grass, agricultural residues, forestry wastes and municipal solid wastes (Pérez et al., 2002). They constitute a renewable resource from which many useful biological and chemical products can be derived. Accumulation of lignocellulose in large quantities in places where agricultural residues present a disposal problem results not only in deterioration of the environment but also in loss of potentially valuable material that can be used in paper manufacture, biomass fuel production, composting, human and animal feed among others (Sánchez, 2009). These lignocellulosic materials are renewable sources of energy where approximately 90 % of the dry weight of most plant materials is stored in the form of cellulose, hemicelluloses, lignin, and pectin (Kumar et al., 2009). Some authors have reported their fractionation to obtain a variety of marketable chemicals from the polymeric fractions of the raw materials , including the xylitol production by Debaryomyces hansenii. This chapter deals with the study of different agroindustrial wastes through their characterization, fractionation by acid pre-hydrolysis and further analysis of the hemicellulosic fraction. The pre-hydrolysis of pistachio hulls was optimized; and finally, the liquors obtained under the optimal conditions, after supplementation, where assayed to carry out the xylose to xylitol bioconversion by D. hansenii.

Quantitative acid hydrolysis
Aliquots from the homogenized lot were submitted to moisture determination and to quantitative hydrolysis in a two-stage acid treatment (the first stage with 72 wt % sulfuric acid at 30 ºC for 1 h, the second stage after dilution of the media to 4 wt % sulfuric acid at 121 ºC for 1 h) (Bustos et al., 2004). The solid residue after hydrolysis was considered as Klason lignin meanwhile hydrolyzates were analyzed by HPLC as described bellow.

Acid hydrolysis
Hydrolyzates were obtained in autoclave at 130 °C with 3 % H 2 SO 4 solutions during 30 min, using a liquid/solid ratio of 8 g g -1 (Portilla-Rivera et al., 2007). The liquid phase from the acid hydrolysis was neutralized with CaCO 3 to a final pH of 5.8-6.0, and the CaSO 4 precipitated was separated from the supernatant by filtration.

Charcoal adsorption
Powdered charcoal (Probus, Madrid, Spain) was activated with hot water and dried at room temperature. Charcoal detoxification of hydrolyzates was carried out by contacting hydrolyzates and charcoal (mass ratio: 10 g g -1 ) at room temperature under stirring for one hour (Rivas et al., 2002). The liquid phase was recovered by filtration and used for making culture media.

Experimental design and statistical analysis of the pistachio shells prehydrolysis
Pistachio shells were treated with solutions containing 1-3 % H 2 SO 4 during 15-45 min. at 130 ºC, according to a statistical experimental design (Design Expert version 5.0, Stat-Ease Inc., Minneapolis, USA), to optimize the prehydrolysis stage, and evaluated by the Response Surface Methodology using Statistica version 5.0 (Statsoft, USA) software considering the SS residual to evaluate the significance of the effects and the model. The influence of two operational variables (concentration of catalyser and reaction time) was tested on three levels in a 3**(2-0) full factorial design. All experiments were carried out in duplicate in randomized run order. Liquors were separated from the solid fraction by vacuum filtration through common laboratory paper filters and analyzed the concentration of xylose, glucose, arabinose, acetic acid, formic acid and furfural, which were the dependent variables.
This design allowed the estimation of the significance of the parameters and their interaction using Student's t-test. The interrelationship between dependent and operational variables was established by Eq.1, a model including linear, interaction and quadratic terms: y = b 0 + b 1 ·x 1 + b 11 ·x 1 2 + b 2 ·x 2 + b 22 ·x 2 2 + b 12 ·x 1 ·x 2 (1) where y represents the dependent variables, b denotes the regression coefficients (calculated from experimental data by multiple regression using the least-squares method), and x denotes the independent variables.

Microorganism and culture conditions
Debaryomyces hansenii NRRL Y-7426 was kindly provided by the National Center for Agricultural Utilization Research (Peoria, Illinois, USA). Freeze-dried cells were grown on a basal medium containing 30 g L -1 commercial xylose, 3 g L -1 yeast extract, 3 g L -1 malt extract, and 5 g L -1 peptone. The microorganism was maintained in agar slant tubes containing a medium formulated with the same components and concentrations as the previous one plus 20 g L -1 agar. Inocula were prepared by solubilization of cells with sterile water and underwent growth during 24 hours in the previous medium without agar. Biomass in inocula was measured by optical density at 600 nm and adjusted by dilution with water, and added to fermentation broth to reach a final concentration in the vicinity of 0.04 g L -1 .
Shake flask fermentation experiments were carried out under microaerophilic conditions in 250 mL Erlenmeyer flasks containing 100 mL of culture media (sterilized in autoclave at 100 ºC during 60 min) prepared with 85 mL hydrolyzates and 10 mL of the nutrients indicated in Table 3.
After inoculation (5 mL), fermentations were carried out in orbital shakers (New Brunswick, Edison, NJ, USA) at 100 rpm and 31 ºC for 96 hours. Samples (2 mL) were taken at given fermentation times and centrifuged at 6,000 rpm for 3 min. The supernatants were stored for analyses.

Analytical methods
Glucose, xylose, arabinose, xylitol, glycerol, acetic acid, ethanol, furfural, and hydroxymethylfurfural (HMF) were measured by a high-performance liquid chromatographic system (Agilent, model 1200, Palo Alto, CA) equipped with a refractive index detector and an Aminex HPX-87H ion exclusion column (Bio Rad) eluted with 0.003 M sulfuric acid at a flow rate of 0.6 mL min -1 at 50 ºC. Biomass concentration in experiments was measured by centrifugation, washing of the cells, and oven-drying to constant weight at 100 ºC.

Characterization of different agro-industrial wastes
Lignocellulosic materials consists of three types of polymers -cellulose, hemicelluloses, and lignin -that are strongly intermeshed and chemically bonded by non-covalent forces and by covalent cross-linkages. Cellulose and hemicelluloses are macromolecules from different sugars; whereas lignin is an aromatic polymer synthesized from phenylpropanoid precursors (Pérez et al., 2002). They also contain smaller amounts of proteins, pectin, acids, salts, minerals, ashes and extractives, including soluble non-structural sugars, nitrogenous material, chlorophyll, and waxes.
Cellulose is a linear polymer of D-glucose subunits linked to each other by β-(1,4)-glucosidic bonds with fibrous structure with a degree of polymerization of up to 10,000 or higher (Jørgensen et al., 2007). The long-chain cellulose polymers are linked together by hydrogen and van der Waals bonds, which cause the cellulose to be packed into microfibrils. Hemicelluloses and lignin cover the microfibrils (Kumar et al., 2009). Cellulose can appear in crystalline form (crystalline cellulose) and, in smaller percentage, in non-organized cellulose chains (amorphous cellulose) (Pérez et al., 2002).
Hemicelluloses is a complex carbohydrate polymer consisting of short highly branched chains of different five-carbon sugars (xylose, rhamnose and arabinose) and six-carbon sugars (glucose, galactose and mannose) and smaller amounts of non-sugars, mainly acetyl groups, but also uronic acids such as 4-o-methylglucuronic, D-glucuronic, and D-galactouronic acids. The backbone of hemicelluloses is either a homopolymer or a heteropolymer with short branches linked by β-(1,4)-glycosidic bonds and occasionally β-(1,3)-glycosidic bonds, and also can have some degree of acetylation. These polymers do not aggregate, even when they cocrystallize with cellulose chains (Kumar et al., 2009). Glucuronoxylan is the principal component of hardwood hemicelluloses whereas glucomannan is predominant in softwood (Pérez et al., 2002). The degree of polymerization is below 200 (Jørgensen et al., 2007).
Finally, lignin is a complex network formed by polymerization of phenyl propane units and constitutes the most abundant non-polysaccharide fraction in lignocellulose (Jørgensen et al., 2007). Lignin is an amorphous heteropolymer complex. The large molecular structure contains cross-linked polymers of phenolic monomers linked by alkyl-aryl, alkyl-alkyl, and aryl-aryl ether bonds (Kumar et al., 2009). Lignin encrusts the cell walls and cements the cells together (Hamelinck et al., 2005), conferring structural support, impermeability, and resistance against microbial attack and oxidative stress (Pérez et al., 2002).   Demirbas (2003) reported the highest percentage of lignin (53.0 %) in hazelnut kernel shells, although this percentage varies widely depending on the material. Papers and herbaceous plants show in general the lowest contents; meanwhile softwoods and nuts have the highest contents (see Table 4).

Prehydrolysis of lignocellulosic materials
Due to the robust structure of lignocellulosic biomass, pretreatment is a prerequisite for hydrolysis into fermentable sugars to be completed within an industrially acceptable time frame (Jørgensen et al., 2007). Although fermentable D-glucose could be produced from cellulose through the action of either acid or enzymes breaking the β-(1,4)glycosidic linkages (Kumar et al., 2009), the structure of cellulose along with the intermolecular hydrogen bonds gives cellulose high tensile strength, makes it insoluble in most solvents and is partly responsible for the resistance of cellulose against microbial degradation (Ward & Moo-Young 1989). On the contrary, hemicelluloses, because of their branched, amorphous nature, are relatively easy to hydrolyze (Hamelinck et al., 2005), particularly, in contrast to cellulose, the polymers present in hemicelluloses are easily hydrolyzable (Kumar et al., 2009).
Over the years a number of different technologies have been successfully developed for pretreatment of lignocellulose (Jørgensen et al., 2007), including concentrated or dilute-acid hydrolysis. Concentrated acids such as H 2 SO 4 and HCl have been used to treat lignocellulosic materials. Nevertheless, although they are powerful agents for cellulose hydrolysis, concentrated acids are toxic, corrosive, hazardous, and thus require reactors that are resistant to corrosion, which makes the pretreatment process very expensive. In addition, the concentrated acid must be recovered after hydrolysis to make the process economically feasible (Kumar et al., 2009).
Among the chemichal hydrolysis methods, dilute-acid hydrolysis is probably the most commonly applied (Taherzadeh & Karimi, 2007). Therefore, dilute-acid hydrolysis (prehydrolysis) where a mineral acid is added to the reaction media to break down polymers, can be a possible first processing step for an integral use of biomass, since hemicelluloses can be almost completely solubilized, whereas little alteration is caused in both lignin and cellulose, which are recovered in the solid phase . Among them, sulfuric acid has been of the most interest in such studies as it is inexpensive and effective (Kumar et al., 2009).
The lignocellulosic materials were subjected to prehydrolysis treatments under the same conditions (3 % H 2 SO 4 , 130 ºC, 30 min., liquid to solid ratio = 8 g/g) previously reported by other authors (Portilla-Rivera et al., 2007). Table 5 reports the sugars content and other compounds solubilized during the treatment. Xylose was the sugar released in highest concentration, reaching a maximal xylose content of 41.8 g L -1 in pistachio shells. Additionally, glucose and arabinose appear in the lowest amounts of 0.88 g L -1 and 0.67 g L -1 , respectively, showing a xylose to glucose ratio of 47.6. Nevertheless, some possible toxins of further fermentable broths were generated, including acetate from the deacetylation of xylan and furan dehydration products (furfural from xylose and hydroxymethylfurfural from glucose). Using pistachio shells, 8.0 g L -1 of acetic acid and 1.7 g L -1 of furfural were released, meanwhile the highest amount of HMF, 1.2 g L -1 , was observed after the acid hydrolysis of white grape stems. Other compounds were also present in hydrolyzates, including citric, tartaric and lactic acids, and glycerol. Tartaric acid concentrations were relatively elevated in viticulture wastes (white and red grape stems and vine leafs) with concentrations up to 9.8 g L -1 . Nevertheless, these acids are not toxic for further fermentations stages.  Table 5. Composition (g/L) of hydrolyzates after prehydrolysis at 130 °C with 3 % H 2 SO 4 during 30 min. using a liquid/solid ratio = 8 g g -1

Optimization of the prehydrolysis of pistachio shells
Considering that pistachio shells provides the highest amount of xylose, the acid hydrolysis of this material was optimized using a 3**(2-0) full factorial design, to produce the highest level of hemicelluloses hydrolysis with the lowest level of degradation of the cellulosic fraction. The use of experimental designs to optimize the hydrolysis conditions is a technology widely used in bibliography (Bustos et al., 2005). In the present work, the independent variables considered and their variation ranges were: H 2 SO 4 concentration (1-3 %) and duration of treatments (15-45 min). The standardized (coded) adimensional variables employed, having variations limits (-1,1), were defined as x 1 (coded H 2 SO 4 concentration) and x 2 (coded time). The correspondence between coded and uncoded variables was established by linear equations deduced from their respective variation limits: x 2 = (t -30)/15 The composition of liquors was measured by dependent variables: y 1 (glucose, g L -1 ), y 2 (xylose, g L -1 ), y 3 (arabinose, g L -1 ), y 4 (acetic acid, g L -1 ), y 5 (formic acid, g L -1 ) and y 6 (furfural, g L -1 ). Table 6 shows the set of experimental conditions assayed expressed in terms of coded variables, as well as the experimental data obtained for variables y 1 to y 6 . The sequence for the experimental work was randomly established to limit the influence of systematic errors on the interpretation of results.

Experiment
x 1 x 2 y 1 y 2 y 3 y 4 y 5 y 6  Table 6. Operational conditions considered in this study (expressed in terms of the coded independent variables dimensionless H 2 SO 4 concentration x 1 and dimensionless time x 2 and experimental results achieved for the dependent variables y 1 (glucose concentration, g/L); y 2 (xylose concentration, g/L); y 3 (arabinose concentration, g/L); y 4 (acetic acid concentration, g/L); y 5 (formic acid concentration, g/L) and y 6 (furfural concentration, g/L). Results are the average of two independent experiments. Standard deviations were in the range of 1-3 % of the mean Table 7 lists the regression coefficients and their statistical significance (based on a t-test) as well as the statistical parameters (R 2 , R 2 adjusted, F and the significance level based on the F test) measuring the correlation and the statistical significance of the models, respectively. It can be noted that all the models showed good statistical parameters for correlation and significance allowing a close reproduction of experimental data. In relation with the influence of independent variables, H 2 SO 4 concentration (x 1 ) caused the strongest effect on the variation of dependent variables considered, as it can be seen from the absolute value of the corresponding coefficients. From the coefficients of Table 7, Eq. 4 was deduced for y 2 (xylose concentration): y 2 = 41.06889 + 1.72 * x 1 -1.21333 * (x 1 ) 2 -2.16333 * (x 2 ) 2 -1.4975 *x 1 * x 2 (4) Using the "solver" application of Microsoft Excell the maximum xylose concentration predicted for the model (y 1 = 41.8 g L -1 ) was achieved when x 1 = 0.90 (2.9 %) and x 2 = -0.31 (25.3 min.). This value corresponds to 90.32 % of the theoretical xylose present in the raw material.

Fermentation under the optimal hydrolysis conditions of pistachio shells
Diluted acids were used under the optimal conditions predicted by the solver application to obtain hydrolyzates with high content of xylose. Fig. 2 and Table 8 show the effect of nutrient supplementation on xylitol production by D. hansenii. Under all the conditions assayed, a 24h lag phase was observed. Xylose started to be consumed after glucose was firstly depleted within 24h. Conversely to other works with the same yeast (Carvalheiro et al., 2007), it is interesting to note the pronounced diauxic growth pattern obtained for the   Table 8. Stoichiometric parameters, productivities and yields after 74h for hemicellulosic pistachio shells hydrolyzates bioconversion in assays carried out by D. hansenii. Results are the average of two independent experiments. Standard deviations were in the range of 2-4 % of the mean mixture of glucose and xylose. A similar pattern was observed for Candida guilliermondii (Canilha et al., 2005), and Candida tropicalis (Walther et al., 2001). Arabinose and acetic acid were scarcely consumed (see Table 8). Thus, meanwhile no arabinose was consumed in fermentation 4 (raw hydrolyzates without supplementation), the consumption in fermentation 5 (detoxified hydrolyzates supplemented with commercial nutrients) only reached 37 %; whereas acetic acid consumption ranged 39.1-58.8 % regardless of supplementation or detoxification treatment. Supplementation with economic nutrients (vinasses or corn steep liquor) led to an increase in the overall sugars consumption rate in comparison with the use of commercial nutrients.
Xylitol was the main product for all media except for fermentation 4 where no production was observed, meaning that these hydrolyzates require to be supplemented with additional nutrients to stimulate the xylitol production. The necessity to supplement hydrolyzates with nutrients seems to depend on the raw material (Carvalheiro et al., 2007), and while the supplementation was shown to be beneficial for Candida guilliermondii during the xylose-toxylitol bioconversion in Eucalyptus hydrolyzates (Canettieri et al., 2001), no addition of nutrients was necessary in rice straw hydrolyzates for the same yeast (Silva & Roberto, 1999). The highest concentration, 19.3 g L -1 after 74 hours, was obtained in hydrolyzates supplemented with economic wastes (vinasses and stream B), corresponding to fermentation 2 (Q P = 0.260 g L -1 h -1 ; Y P/S = 0.63 g g -1 ), followed by the use of another economic nutrient in fermentation 3, corn steep liquor, where xylitol reached up to 18.0 g L -1 (Q P = 0.244 g L -1 h -1 ; Y P/S = 0.56 g g -1 ). In fermentations supplemented with commercial nutrients (fermentations 1 and 5), conversely to the expected, the higher xylitol value, 16.2 g L -1 , was observed in fermentation without detoxification. Ethanol and glycerol were also produced although in small amounts.
Finally, it is important to mention the amount of biomass generated in all fermentations with the exception of fermentation 4 where only 2.3 g L -1 was achieved. Usually, the growth begins with a lag phase in which the microorganism adapts to the enzymatic systems in order to metabolize the new substrate, but without growing (Sánchez et al., 2008), however, the previous adaptation of D. hansenii to hydrolyzates minimize this phase, showing a linear growth during the total period of fermentation, producing up to 9.5 g L -1 after 96 hours in fermentation 3 (Q X = 0.099 g L -1 h -1 ; Y X/S = 0.29 g g -1 ).

Conclusions
In this study, the following conclusions can be drawn related to agro industrial wastes evaluation: 1. Based on the characterization and further diluted acid hydrolysis carried out to the agro industrial wastes evaluated, it can be concluded their great potential as feedstocks for the production of industrially relevant food additives. 2. In particular, pistachio shells show the higher xylan content. Optimizing the conditions of hydrolysis by an experimental design reaches 90% of the theorical xylose present in raw material. 3. Hydrolyzate medium obtained with nutrients supplementation can be effectively employed for the xylitol production by Debaryomyces hansenii.