Research Article Received: 23 March 2016

Revised: 17 June 2016

Accepted article published: 22 June 2016

Published online in Wiley Online Library:

(wileyonlinelibrary.com) DOI 10.1002/jctb.5043

Optimization of enzymatic sugar beet hydrolysis in a horizontal rotating tubular bioreactor Martina Andlar,a Iva Rezi´c,b Damir Oros,a Daniel Kracher,c Roland Ludwig,c Tonˇci Rezi´ca* and Božidar Šanteka Abstract BACKGROUND: Sugar beet pulp (SBP) is a promising feedstock for the production of 2nd generation biofuels, but efficient enzymatic hydrolysis remains a key challenge; therefore, new process designs and/or bioreactor designs are crucial to overcome this hurdle. In this regard, horizontal rotating tubular bioreactors (HRTB) offer the advantage of high substrate loadings while minimizing the space and energy demand compared with conventional stirred tank reactors. Here, a statistical approach is used to optimize the hydrolysis of sugar beet pulp in laboratory experiments, and it is shown that such a process can be implemented in a HRTB. RESULTS: Using the design of experiments (DOE) method, the reaction conditions of four commercial enzyme mixtures (Ultrazym AFP-L, Viscozyme L, Pectinase and Cellulase) was optimized for the degradation of SBP in small-scale experiments. Using Ultrazym AFP-L as the most efficient mixture, a 10 L scale conversion was performed in a HRTB. At a substrate loading of 135 g L−1 and optimized conversion parameters (enzyme load, pH and rotating speed of the reactor), 0.525 W dm−3 were needed to achieve solubilisation of 30% of the total mass of initial SBP after 24 h. CONCLUSION: DOE was found to be an easy-to-apply method that allowed optimizing the conditions for enzymatic hydrolysis of SBP, resulting in a higher sugar yield. The results could be transferred to an HRTB, which is a suitable system for enzymatic conversion and efficient saccharification of semi-solid or solid substrates with relatively low energy consumption. © 2016 Society of Chemical Industry Keywords: horizontal rotating tubular bioreactor; sugar beet pulp; commercial enzymes; design of experiment; enzymatic hydrolysis

NOTATION CSBP CG CGA CES EES mG mGA mES w T VE 𝛼

concentration of sugar beet pulp (g dm−3 ) concentration of glucose (g dm−3 ) concentration of galacturonic acid (g dm−3 ) enzyme loading (mg enzyme g−1 SBP) efficiency ratio of enzyme solution (g released sugars mg−1 enzymes) mass of released glucose in the batch reaction volume (mg) mass of released galacturonic acid in the batch reaction volume (mg) mass of applied enzyme solution in the batch reaction volume (mg) released (solubilized) sugar percent calculated on the total sugar beet pulp mass (%) temperature (∘ C) volume of enzyme solution (dm3 ) axial point of the design (1.4)

∗

Correspondence to: T Rezi´c, Faculty of Food Technology and Biotechnology University of Zagreb, Pierottijeva 6, 10000 Zagreb, Croatia. Email: [email protected]

a Department of Biochemical Engineering, Faculty of Food Technology and Biotechnology, University of Zagreb, 10000, Zagreb, Pierottijeva 6, Croatia b Department of Applied Chemistry, Faculty of Textile Technology, University of Zagreb, 10000, Zagreb, Prilaz Baruna Filipovi´ca 28a, Croatia

INTRODUCTION Sugar beet pulp (SBP) is an abundant lignocellulosic by-product from industrial sugar production and represents a promising J Chem Technol Biotechnol (2016)

resource for biomass-to-ethanol processes. In 2014, approx. 17 million tons of SBP (dry weight) accumulated globally after sugar extraction from sugar beets.1 Theoretically, about 115 kg of ethanol per ton of SBP can be obtained after enzymatic hydrolysis of the cellulose fraction and subsequent microbial fermentation, neglecting other fermentable sugars such as fructose, galactose, arabinose and xylose. This makes SBP a promising feedstock for the production of second generation bioethanol. Enzymatic hydrolysis of lignocellulosic raw materials into fermentable sugars is a key step in the production of bioethanol. Previously, approaches towards a fast and complete hydrolysis of SBP involved initial physical pre-treatment methods (ultrasound,

c Department of Food Sciences and Technology, University of Natural Resources and Life Sciences, Vienna, Muthgasse 18, 1190, Vienna, Austria

www.soci.org

© 2016 Society of Chemical Industry

www.soci.org thermal)2,3 or acid/base pre-treatment1,4 of the substrate, which makes the well protected polysaccharides within the lignocellulosic matrix accessible for enzymatic conversion. For enzymatic hydrolysis of SBP commercial enzymes mixtures5 or fungal crude extracts have been employed.6 The incorporation of the design of experiment (DOE) methodology into this process can additionally help to enhance the hydrolysis efficiency by optimizing the enzyme-to-substrate loadings. The result of DOE methodology typically is an optimized enzymatic saccharification process, for which critical parameters affecting the efficiency of the enzymes (e.g. temperature or pH) are optimized to achieve maximum performance. Several case studies highlight the benefit of the DOE methodology for enzymatic hydrolysis of various lignocellulosic raw materials. Suwannarangsee et al.7 used DOE to optimize the efficiency of an experimental enzyme mixture designed for the hydrolysis of rice straw. Their work demonstrated that DOE can be successfully used for the optimization of complex multi-enzyme systems. Murphy et al.8 emphasized that a precise design of the underlying research plan determines the quality of generated wet-lab data used for the analysis, modeling and derivation of crucial parameters influencing the studied bioprocess.8,9 Importantly, DOE methodology has also been reported as an indispensable tool to reduce the amount of labour and the number of cost-intensive experiments in bioprocesses research.10,11 The implemented multi-objective optimization approach enables the combination of many different optimization criteria. For example, response surface modeling (RSM) was utilized to examine parameters that have an impact on the production of chitinolytic and 𝛽-1,3-glucanolytic enzymes in the Trichoderma atroviride strain P1. These experiments were based on a five factor of three level general full factorial face-centered design. The enzyme activity was modelled by a partial least-squares regression (PLS) method using the MODDE 6.0 software.12 The results of this research showed that the enzyme activity was strongly affected by the concentration of glucose and the pH of the medium, but unfortunately the effect of temperature was not studied. The factorial DOE method was also applied to optimize the composition of the synthetic medium used for Candida bombicola growth and enzyme production.13 The optimization of pectin hydrolysis by enzymes from Aspergillus niger was done by using different mathematical methods, including the DOE.14 In previous research the DOE methodology was applied for the optimization of different processes, including prediction of the surface tension of surfactant mixtures for detergent formulation and optimization of corrosion processes of stainless steel coating during cleaning in steel brewery tanks.15,16 There have been several reports on construction improvements of bioreactors for semi-solid or solid-state bioprocesses and their application for the hydrolysis of lignocellulosic raw materials.17 – 20 Horizontal rotating tubular bioreactors (HRTBs) offer several advantages for such applications. A HRTB is typically constructed as a tubular bioreactor placed on bearings that enable axial rotation of the whole bioreactor. In the HRTB and in similar bioreactor designs, efficient mixing of lignocellulosic raw materials at high solid loadings can be achieved. Furthermore, this bioreactor type features both liquefaction and saccharification of lignocellulosic raw materials at comparatively low energy consumption. These bioreactor designs were reported to lower the downstream processing costs due to the higher product concentrations in the fermentation broth.21 Disposal and treatment costs may also be reduced due to the low water content in the broth.22,23 Because of the high initial viscosity of these media, efficient liquefaction has

wileyonlinelibrary.com/jctb

M Andlar et al.

to be achieved before the saccharification process in order to lower the energy demand for mixing. High solid loadings additionally complicate the scale-up of bioprocesses, and overall processing costs are likely to be affected by the solid matter concentration.24 In stirred tank bioreactors, solid loadings of approx.12–15% (w/v) represent the upper limit at which solid matter can be effectively mixed.25 In addition, the low enzyme dosages per unit of solid lignocellulosic raw materials (due to the cost reduction) also reduce the overall hydrolysis efficiency. Therefore, mixing is a crucial parameter that allows high product concentrations in the media with high solid contents of lignocellulosic raw materials. For the hydrolysis of lignocellulosic raw materials, a sufficient mixing capacity, low energy consumption and effective enzyme loading have to be considered key parameters for bioreactor design.6 The main goal of this research was to establish an efficient enzymatic hydrolysis process employing SBP in a HRTB for subsequent bioethanol production. For this purpose, a DOE approach was employed to identify the optimal enzyme-to-substrate loadings, as well as pH- and temperature-optima of four commercial enzyme mixtures (Cellulase, Pectinase, Ultrazym AFP-L and Viscozyme L). An optimized process for SBP hydrolysis was successfully implemented in a HRTB at near-industrial conditions.

MATERIALS AND METHODS Sugar beet pulp and enzymes Dried sugar beet pulp (SBP) with a moisture content of 9% was acquired from the Sladorana d.o.o. sugar factory (Županja, Croatia) and ground to a final particle size of 0.3–0.5 mm (Moulinex mill; type-505). SBP contains cellulose (25–30%), hemicelluloses (24–32%), pectin (38–62%) and a low lignin content (∼1%).3 Cellulase from Trichoderma reesei ATCC 26921 and Pectinase from Aspergillus aculeatus were purchased from Sigma-Aldrich (Steinheim, Germany). Ultrazym AFP-L and Viscozyme L (multi-enzyme complexes) were obtained from Novozymes (Bagsværd, Denmark). Enzyme stock solutions had the following protein concentrations: Ultrazym AFP-L, 14.0 ± 0.1 g dm−3 (enzyme solution with polygalacturonase and cellulase activities; 100 U mL−1 ); Viscozyme L, 20.3 ± 0.3 g dm−3 (100 FGB g−1 ); Pectinase from A. aculeatus,11.3 ± 0.2 g dm−3 (3000 FDU mL−1 ); Cellulase from T. reesei, 42.0 ± 0.1 g dm−3 (700 EGU g−1 ). Batch hydrolysis experiments Enzymatic batch reactions were performed at 1 × 10−3 dm3 scale in 50 mmol L−1 sodium citrate buffer containing 20 mg SBP (ground, weighted and sterilized) and defined enzyme stock solution volume (VE = 20–80 × 10−6 dm3 ) (Table 1). Corresponding enzyme loadings (calculated as mg enzymes g−1 SBP) are presented in Table 1. Reactions were incubated for 24 h on a temperature controlled shaker operated at 900 min−1 . In these experiments, the effect of various temperatures (T = 20–50 ∘ C), pH values (pH = 3–7) and enzyme loadings (VE = 20–80 × 10−6 dm3 ) on the hydrolysis efficiency of SBP was tested (Table 2). The impact of temperature alteration on pH was initially tested with a pH microelectrode and found to deviate insignificantly (0.05 pH units) from room temperature (23 ∘ C) to 40 ∘ C. The efficiency of individual enzyme solutions (EES ) for hydrolysis of SBP was evaluated based on the efficiency ratio, which is defined by the following equation:

© 2016 Society of Chemical Industry

EES = mG ormGA ∕mES

(1)

J Chem Technol Biotechnol (2016)

Enzymatic sugar beet hydrolysis in a horizontal rotating tubular bioreactor

Table 1. Enzyme loadings (mg g−1 SBP) used in the design of experiment (for general full factorial design) Enzyme types Cellulase from T. reesei

Ultrazym AFP-L

Viscozym L

Pectinase from Aspergillus aculeatus

𝛼 - axial point of the design is 1.4. CES - enzyme loading (mg enzyme g−1 SBP).

where mG is the mass of released glucose in the batch reaction volume (1 × 10−3 dm3 ), mGA is the mass of released galacturonic acid in the batch reaction volume and mES is the mass of applied enzymes in the batch reaction volume. HRTB hydrolysis experiments The HRTB is constructed as a 0.6 m long stainless steel tube with an inner diameter of 0.25 m, resulting in a total volume of approx. 30 dm3 . The interior of the HRTB contains two paddles (0.04 m height and 0.6 m length) mounted to the bioreactor wall in order to improve mixing and homogenization of the bioreactor contents. The HRTB is placed on rolling bearings that enable axial rotation of the whole bioreactor at a maximum rotation speed of 60 min−1 . In the present study, enzymatic hydrolysis of SBP in the HRTB was optimized in terms of SBP concentration and bioreactor rotation speed (5–15 min−1 ). Initially, the HRTB and SBP (in 50 mmol L−1 sodium citrate buffer) were separately sterilised by steam (121 ∘ C, 20 min). Next, the HRTB was filled with defined SBP loadings (60–180 g dm−3 ). Due to the fact that Ultrazym AFP-L showed the highest products release efficiency in the optimization experiments it was added to the HRTB in the concentration range 0.84–2.52 g per dm3 of medium. The final reaction volume was 10 dm3 . Ultrazym AFP-L was added in five aliquots at 5 min intervals to ensure a homogenous distribution in the slow rotating HRTB. Between additions the HRTB was mixed at a rotation speed of 15 min−1 . The hydrolysis was performed at pH 3.5 and 39 ∘ C. Samples were withdrawn and analysed after 24 h. The power consumptions of the HRTB and a stirred tank bioreactor were calculated based on their individual power numbers. For the stirred tank bioreactor, the power number value was calculated based to the impeller type and the Reynolds number at different impeller speeds. The power consumption of the HRTB was calculated using the equation for determination of power consumption in rotating tubular reactors.21 SBP rheological characteristics and mixing properties in the stirred tank bioreactor were estimated according to Rezi´c et al.3 J Chem Technol Biotechnol (2016)

Table 2. Experimental set-up based on the general full factorial design

Run

T (∘ C)

pH -

VE × 10−6 (dm3 )

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40

20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 30 30 30 30 30 30 30 30 30 30 30 30 30 30 30 30 30 30 30 30

3 4 5 6 7 3 4 5 6 7 3 4 5 6 7 3 4 5 6 7 3 4 5 6 7 3 4 5 6 7 3 4 5 6 7 3 4 5 6 7

20 20 20 20 20 40 40 40 40 40 60 60 60 60 60 80 80 80 80 80 20 20 20 20 20 40 40 40 40 40 60 60 60 60 60 80 80 80 80 80

CES (mg g−1 SBP) 42 84 126 168 14 28 42 56 20.3 40.6 60.9 81.2 11.3 22.6 33.9 45.2

www.soci.org

Run

T (∘ C)

pH -

VE × 10−6 (dm3 )

41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80

40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50

3 4 5 6 7 3 4 5 6 7 3 4 5 6 7 3 4 5 6 7 3 4 5 6 7 3 4 5 6 7 3 4 5 6 7 3 4 5 6 7

20 20 20 20 20 40 40 40 40 40 60 60 60 60 60 80 80 80 80 80 20 20 20 20 20 40 40 40 40 40 60 60 60 60 60 80 80 80 80 80

Analytical procedures Concentrations of released reducing sugars (glucose or galacturonic acid) were measured by the 3,5-dinitrosalicylic acid (DNSA) assay according to a previously published protocol.26 Briefly, 600 × 10−6 dm3 of the properly diluted sample solution was mixed with 600 × 10−6 dm3 of DNSA reagent containing 10 g dm−3 DNSA, 0.5 g dm−3 sodium sulfite, 10 g dm−3 sodium hydroxide and 2 g dm−3 phenol. The mixture was incubated for 15 min at 95 ∘ C to induce color formation, which was stopped by adding 200 × 10−6 dm−3 of a 40 g dm−3 ice-cold potassium sodium tartrate solution. Absorbance was immediately read at 575 nm and compared with calibration curves generated with galacturonic acid (for pectinase activity) and glucose (for all other enzymes). Enzyme activities are defined as the amount of reducing sugars released per minute

© 2016 Society of Chemical Industry

wileyonlinelibrary.com/jctb

www.soci.org under the defined assay conditions. Sugars released after SBP enzymatic hydrolysis in the HRTB, were quantified by HPLC on a Supelcogel C-610H column using a refractive index detector (RID, Shimadzu 10 A VP; Kyoto, Japan). Analytes were separated at a flow rate of 0.5 mL min−1 with 0.1% H3 PO4 as eluent at a constant temperature of 30 ∘ C. Before analysis, all samples were mixed with ZnSO4 to a final concentration of 10% to induce protein precipitation. Solid debris was removed by centrifugation (4500 min−1 for 20 min). Before column application, sample solutions were passed through a 0.20 μm filter. Design of experiments and statistical analysis of design Enzyme loading (for reasons of convenience indicated as the enzyme volume), pH and temperature of the particular experiment were predetermined by the DOE methodology (software Design

M Andlar et al.

Expert v 7.0 by State-Ease, USA and Statistica v. 7.0 by State-Soft, USA). The experimental domain (including minimum and maximum values of independent variables) was designed on the basis of preliminary research. Levels of the investigated factors (independent variables) are presented in Table 2. An experimental research plan (80 experiments) for SBP hydrolysis was established by using a general full factorial statistical design for the study of linear, quadratic and mixed term effects of the three investigated factors (temperature, pH and enzyme loading), each on four to five levels (Table 2). Temperatures ranged from 20 to 50 ∘ C, pH values from pH 3 to 7 and enzyme loadings from 20 to 80 × 10−6 dm3 . In all tested designs, response surface models were fitted to the response variable (glucose or galacturonic acid concentration) for each enzyme. The software Design of Experiments (v. 7.0,

Table 3. Glucose (or galacturinic acid; g dm−3 ) concentrations obtained as a consequence of Viscozyme L (Visc.), Ultrazym AFP-L (Ultr.), Cellulase (Cell) and Pectinase (Pect.) activity Run

Visc.

Ultr.

Cell.

Pect.

Run

Visc.

Ultr.

Cell.

Pect.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40

3.39 3.06 2.02 1.54 1.21 4.33 4.09 3.01 1.96 1.21 5.62 4.77 3.30 2.51 1.80 3.89 3.77 3.19 3.07 2.55 5.21 4.85 3.19 1.11 1.18 6.18 5.71 4.35 1.75 1.46 7.57 7.24 4.71 7.19 4.35 8.71 8.49 6.51 4.56 2.59

2.88 2.44 1.24 1.01 0.76 4.55 4.19 2.59 1.34 0.76 5.17 3.65 2.29 1.60 1.00 3.72 3.49 3.03 1.17 1.02 4.66 3.81 1.96 1.23 0.97 5.21 5.60 3.46 1.56 0.94 6.75 6.01 3.90 5.16 3.63 8.04 8.41 5.35 2.87 1.61

0.07 0.09 0.08 0.16 0.08 0.06 0.17 0.21 0.21 0.24 0.01 0.07 0.17 0.16 0.17 0.11 0.10 0.08 0.21 0.22 0.07 0.17 0.16 0.23 0.38 0.06 0.22 0.22 0.70 0.64 0.00 0.04 0.17 0.19 0.51 0.01 0.21 0.26 0.26 0.25

2.87 2.20 1.67 1.09 0.99 3.38 2.96 2.08 1.43 0.99 3.84 3.76 3.41 1.59 1.24 2.87 3.17 2.96 1.10 1.00 4.19 3.48 2.01 1.07 1.12 4.83 4.34 3.72 1.85 1.07 5.91 5.38 3.93 5.46 3.02 7.09 6.44 5.33 2.45 1.76

41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80

6.37 6.77 4.13 1.76 1.28 8.06 8.79 6.42 2.67 1.99 8.89 9.06 7.16 4.34 2.38 5.99 7.74 7.78 3.91 2.25 2.61 4.76 2.80 0.97 0.39 4.42 7.72 5.61 2.25 1.39 4.38 7.32 5.38 2.21 1.00 6.13 8.40 4.55 2.77 2.23

6.31 6.10 3.06 1.21 0.84 7.17 7.07 4.43 1.64 1.64 7.09 7.44 6.21 2.49 1.97 2.35 7.51 5.78 2.72 2.79 4.33 5.37 2.59 0.95 0.59 3.86 5.34 4.71 1.92 1.04 5.98 6.72 5.27 2.04 0.45 7.64 9.42 3.90 2.57 2.24

0.15 0.28 0.25 0.30 0.40 0.20 0.31 0.26 0.34 0.10 0.12 0.38 0.30 0.34 0.25 0.12 0.13 0.39 0.44 0.09 0.09 0.28 0.23 0.25 0.00 0.05 0.33 0.47 0.33 0.09 0.01 0.13 0.25 0.65 0.27 0.00 0.00 0.00 0.00 0.00

4.76 4.98 3.44 1.43 1.19 6.14 6.60 4.82 2.14 1.63 7.09 7.44 6.21 2.49 1.97 6.10 6.72 6.16 3.36 2.95 2.32 3.59 2.94 1.17 0.28 5.56 6.79 4.62 1.69 1.55 3.47 5.82 4.55 1.91 0.55 5.14 7.80 3.98 2.52 2.23

wileyonlinelibrary.com/jctb

© 2016 Society of Chemical Industry

J Chem Technol Biotechnol (2016)

Enzymatic sugar beet hydrolysis in a horizontal rotating tubular bioreactor State-Ease, USA) was used for calculation and fitting the values of the independent variables to the proposed model:

y = b0 +

k ∑ i=1

bi xi +

k ∑

i≺j ∑ ∑ bii xi2 + bij xi xj + e

i=1

i

(2)

Table 4. Relations between investigated factors and response variable (glucose or galacturonic acid concentration) for different enzymes Enzyme

Equation No.

Equation

j

where xi and xj are the investigated factors (independent variables), b0 , bi , bii and bij are regression coefficients, e is the error and y is the response-dependent variable. In addition, the lack-of-fit test was used to determine if the models developed adequately describe the experimental data. If the F-test of the model was significant (≥5% level) the variation in the response variable can be explained and verified by the established model. Two- and three-dimensional surface plots illustrate the main and the interactive effects of the independent variables on the dependent ones. The response surface methodology (RSM) was used for establishing and observing the most desirable response values and operating bioprocess conditions in all tested designs in the same manner: the regression models were used to predict the response for different combinations of investigated factors that were not tested during the model building in preliminary experiments. Response surface plots were generated as a two-dimensional contour plot with the response surface drawn in a two-dimensional plane or as a three-dimensional surface plot that provides a clearer picture of the response and consequently the overall bioprocess behaviour. In order to evaluate the accuracy and repeatability of the galacturonic acid determination, six pectinase assays were performed at pH 4.0 (50 mmol L−1 sodium citrate buffer) and 40 ∘ C using an enzyme loading of 92 mg pectinase g−1 sugar beet pulp. Sugar beet pulp hydrolysis without external enzymes addition was also performed under the same conditions, but release of reducing sugars was not observed. This observation is in agreement with our previous study.6 All samples were measured in two independent experiments. Galacturonic acid was not detected in the absence of Pectinase. Results obtained show that the standard deviation of all measurements was in the range of experimental error (below 4.6%). Glucose-releasing assays were performed with all other enzymes under the same conditions and showed similar standard deviations below 5% (data not shown).

RESULTS AND DISCUSSION General full factorial design of experiment The experimental design was performed as follows: (i) defining the problem; (ii) choosing the system variables and their ranges; (iii) selecting an objective function and experimental design method (general full factorial design, two factorial design, Box Behnken design or D-optimal design); (iv) performing the necessary conversion experiments and measuring the responses (glucose or galacturonic acid concentration); (v) analyzing the data; (vi) experimental confirmation of the calculated data; and (vii) deriving final conclusions on the optimal experimental setup to achieve the highest performance of enzymes. Data acquired from 80 experiments following a general full factorial design established by DOE (Table 3) were fitted to mathematical models in order to relate the response variables to the experimental data. These allowed the calculation of statistical parameters, including factor coefficients, interactive terms and probability values. After the response variables were predicted in many points of the experimental domain, a graphical representation of the process model J Chem Technol Biotechnol (2016)

www.soci.org

Cellulase

3

Viscozyme L

4

Ultrazyme AFP-L

5

Pectinase

6

CG = − 1.53 + 0.04 • T + 0.30 • pH + 0.01 • VE – 1.04 × 10−3 • T • pH −8.22 × 10−5 • T • VE + 1.23 × 10−4 • pH • VE – 4.75 × 10−4 • T 2 – 2.3 × 10−2 • pH2 – 8.75 × 10−5 • VE 2 CG = − 11.91 + 0.69 • T + 1.85 • pH – 0.02 • T • pH + 2.06 × 10−4 • T • VE – 5.01 × 10−4 •pH • VE – 8.38 × 10−3 • T 2 – 0.24 pH2 – 8.40 × 10−4 • VE 2 CG = − 5.56 + 0.47 • T + 0.70 • pH + 0.07 • VE – 0.02 • T • pH + 4.50 × 10−4 • T • V – 1.6 × 10−3 • pH • VE – 5.27 × 10−3 • T 2 – 0.12 • pH2 – 4.97 × 10−4 • VE 2 CGA = − 9.20 + 0.52 • T + 1.50 • pH + 0.09 • VE – 0.01• T • pH + 5.19 × 10−4 • T • VE – 3.37 × 10−3 • pH • VE – 6.12 × 10−3 • T 2 – 0.18 • pH2 – 6.03 × 10−4 • VE 2

was generated. Since the exclusion of non-significant parameters is an important step to improve the adequacy of the model, the impact of parameters was not considered to be significant when the probability values were higher than 0.05. Modeling and response surface methodology of general full factorial design The design of experiment (DOE) was performed to evaluate the impact of process variables (investigated factors) on the efficiency of enzymatic SBP hydrolysis. The results of the DOE approach were correlations between response variables (released glucose (CG ) or galacturonic acid (CGA ) concentration) and the investigated factors (enzyme loading, pH and temperature). In order to validate the estimated polynomial equations (and thus the experimental model), calculated data were compared with experimental results. For example, the highest release of galacturonic acid by Pectinase was measured at 45 ∘ C after addition of 80 × 10−6 dm3 of enzyme solution at pH 4. A comparison with data obtained by the prediction equation (Table 4, Equation (6)) showed very good agreement between the measured galacturonic acid concentration (7.62 g dm−3 ) and the predicted concentration (7.54 g dm−3 ). The same trend was observed for all other enzymes, which is underpinned by generally high R2 values (0.900–0.959). Response surface model plots for concentrations of released glucose or galacturonic acid as a function of temperature and pH were fitted to obtain response variables for Cellulase, Pectinase, Viscozyme L and Ultrazym AFP-L (Fig. 1(a)–(d)). A correlation between temperature and enzyme loading (volume of enzyme solution, VE ) was found based on the experimental data shown in Fig. 2. A plot of glucose or galacturonic acid concentrations as a function of enzyme loading and pH for all enzymes is shown in Fig. 3. From these data, the optimal ranges of the investigated factors (enzyme loading rate, pH and temperature) for each of the studied

© 2016 Society of Chemical Industry

wileyonlinelibrary.com/jctb

www.soci.org

M Andlar et al.

Figure 1. Response surface model plots for glucose (or galacturonic acid) concentration as a function of temperature (T) and pH: (a) Cellulase, (b) Pectinase, (c) Viscozyme L and (d) Ultrazym AFP-L.

Figure 2. Response surface model plots for glucose (or galacturonic acid) concentration as a function of temperature (T) and enzyme loading rate (VE ): (a) Cellulase, (b) Pectinase, (c) Viscozyme L and (d) Ultrazym AFP-L.

wileyonlinelibrary.com/jctb

© 2016 Society of Chemical Industry

J Chem Technol Biotechnol (2016)

Enzymatic sugar beet hydrolysis in a horizontal rotating tubular bioreactor

www.soci.org

Figure 3. Response surface model plots for glucose (or galacturonic acid) concentration as a function of enzyme loading (VE ) and pH: (a) Cellulase, (b) Pectinase, (c) Viscozyme L and (d) Ultrazym AFP-L.

enzyme cocktails were determined. Generally, optimum performances of all investigated enzymes were measured between 35 and 40 ∘ C and at acidic pH values around 3.7. The only exception was cellulase, which showed a less acidic optimum at pH 5.9. Under the experimental conditions, the optimum enzyme volumes (loadings) were 45, 80, 70 and 75 × 10−6 dm3 for Cellulase, Pectinase, Viscozyme L and Ultrazym AFP-L, respectively. Data presented in Figs 1–3 show that all enzymes have large global optima. Furthermore, the effects of pH and temperature are independent, which is more pronounced for Pectinase and Viscozyme L than for Ultrazym AFP-L and Cellulase. Optimization of product yields In a next step we aimed to maximize the hydrolysis efficiency of the enzyme mixtures by keeping all independent parameters (temperature, pH and enzyme loading) within the previously determined limits while maximizing the function of glucose or galacturonic acid concentration. Following this approach, the experimental set-up for DOE resulted in equations 3–6 shown in Table 4. The optimum temperatures and pH ranges (Table 5) were almost identical to the previously determined optima. In general, the multi-enzyme complexes Pectinase, Ultrazym AFP-L and Viscozyme L showed comparable efficiencies with efficiency ratios (EES ) ranging from 6.6 to 8.3 g g−1 . On the basis of EES , Pectinase was the most efficient enzyme mixture and solubilized 38% (or 7.54 mg) of the total SBP present in the assay. Viscozyme L solubilized 45% (or 9.07 mg) of the initial SBP, but due to the higher enzyme loading EES was slightly lower than for the other enzyme J Chem Technol Biotechnol (2016)

mixtures. The Cellulase preparation from T. reesei was least efficient, showing approximately 40-times lower EES compared with all other enzymes. These results are in agreement with a previous study which showed that a combined action of several enzymes with different substrate specificities is required to access and hydrolyze the cellulose fraction in SBP.6 Optimization of the enzyme loading The second part of the optimization procedure focused on the minimization of enzyme loadings, while keeping the hydrolysis efficiency at a maximum (Table 6). Further fine-tuning of experimental conditions of DOE resulted in slightly altered temperature optima of the enzymes that were marginally higher for Cellulase (38.1 ∘ C), Viscozyme L (37.9 ∘ C) and Pectinase (40.4 ∘ C) and lower for Ultrazym AFP-L (39.1 ∘ C) when compared with the initial optimization. At the same time, pH optima for all enzymes were slightly reduced (maximum change: 0.2 pH units). In all experiments, the concentrations of released reducing sugars from SBP were lowered due to the overall decreased enzyme loadings. Concurrently, EES -values increased approximately 2-fold for Pectinase and Viscozyme L, and 3-fold for Ultrazym AFP-L. In addition, the performance of Cellulase was improved, but no substantial hydrolysis was achieved since only 1.6 % (or 0.32 g) of the total SBP was degraded. The highest EES was observed for Ultrazym AFP-L. In a next step, we tried to verify the data obtained from the previously described DOE-approach by performing a lab-scale saccharification process in a HRTB.

© 2016 Society of Chemical Industry

wileyonlinelibrary.com/jctb

www.soci.org

M Andlar et al.

Table 5. Optimization of bioprocess parameters to maximize the glucose (or galacturonic acid) concentration calculated by general full factorial design for different enzymes solutions Enzyme

T (∘ C)

pH

Cellulase Viscozym L Ultrazym AFP-L Pectinase

35.8 35.6 39.7 39.7

5.94 3.66 3.68 3.70

VE × 10−6 (dm3 ) 45.2 67.7 74.0 80.0

CES (mg g−1 )

mES (mg)

94.92 68.71 51.8 45.2

1.898 1.374 1.036 0.904

mG (mGA ) (mg) 0.37 9.07 7.70 7.54*

w (%)

EES (g g−1 )

1.8 45.6 38.5 37.7

0.195 6.601 7.432 8.341

*galacturonic acid.

Table 6. Optimization of bioprocess parameters to maximize the glucose (or galacturonic acid) concentration and keeping the lowest level of the enzyme concentration calculated by using General full factorial design for different enzymes solutions Enzyme

T (∘ C)

pH

Cellulase Viscozym L Ultrazym AFP-L Pectinase

38.1 37.9 39.1 40.4

5.81 3.56 3.46 3.59

VE × 10−6 (dm3 ) 20.0 20.0 20.0 20.0

CES (mg g−1 )

mES (mg)

42.0 20.3 14.0 11.3

0.840 0.406 0.280 0.262

mG (mGA ) (mg) 0.32 6.26 5.54 4.85*

w (%)

EES (g g−1 )

1.6 31.3 27.7 24.3

0.381 15.419 19.786 18.653

*galacturonic acid.

Effect of rotation speed and sugar beet pulp concentration on the hydrolysis in the HRTB Hydrolysis of SBP by Ultrazym AFP-L was selected as a model bioprocess to study the impact of various bioprocess parameters (enzyme loading, pH, bioreactor rotation speed and sugar beet pulp concentration) on the hydrolysis efficiency in the HRTB. The main goal was to maximize the outlet reducing sugars concentration using Ultrazym AFP-L, which showed the highest efficiency in the previous optimization approach. Our previous attempts to study enzymatic SBP conversion in the HRTB resulted in non-effective hydrolysis due to insufficient pulp/enzyme ratios and non-optimal pH (data not shown). Data from the DOE optimization allowed us to optimize the process; consequently we used a final enzyme load of 1.89 g dm−3 Ultrazym AFP-L and adjusted the pH to 3.5. Response surface model plots show the effect of HRTB rotation speed (n) and SBP concentration (CSBP ) on the glucose concentration (CG ) after 24 h of hydrolysis (Figure 4.). The SBP concentration (CSBP ) had a more pronounced effect on enzymatic hydrolysis (expressed as reducing sugar concentration) than the bioreactor rotation speed. This observation can be explained by the impact of water in the substrate, and the mass diffusion limitation in the concentrated substrate. Furthermore, an increase in HRTB rotation speed (n) also led to a slight increase in reducing sugar concentration, probably as a consequence of better broth homogenization in the HRTB. At a SBP concentration of 135 g dm−3 and a bioreactor rotation speed of 11 min−1 , a maximum sugar concentration of 39.5 g L−1 was measured, corresponding to 29.6% of the initial amount of SBP, which is in the range of degradation found in the optimization procedure. However, a comparison between SBP hydrolysis in a stirred tank bioreactor (STR) and the HRTB showed that the released sugar concentrations were 3-fold higher in the HRTB (Table 7). This can be explained by the fact that SBP concentrations above 60 g dm−3 result in insufficient mixing in a stirred tank bioreactor.3 In addition, a further increase of SBP concentration would decrease the water content in the fermentation broth, which negatively affects

wileyonlinelibrary.com/jctb

Figure 4. Response surface model plot of the effect of HRTB rotation speed (n) and SBP concentration (CSBP ) on the sugars concentration (CG ) after 24 h of Ultrazyme AFP-L hydrolysis in the HRTB.

the enzymatic activity (hydrolysis process). On the basis of our results, it is evident that the most important parameters for efficient hydrolysis in the HRTB are the concentration of lignocellulosic raw material, the water-to-solid ratio and the enzyme loading. Literature data suggests that homogenous enzyme distribution in the bioreactor is the most important parameter.27 Conversely, some authors noted that the degree of mixing was of less importance for the hydrolysis efficiency.28,29 An overview of accessible literature data for power comsumption during enzymatic hydrolysis in different bioreactors was done in order to make comparison between them as well as to show the expected range of power consumption for this purpose in different bioreactors. As can be seen in Table 7, a comparable degree of hydrolysis (30% conversions after 24 h) was achieved in the HRTB at a lower power input of 0.525 W dm−3 . In contrast to the HRTB, stirred tank bioreactors typically operate at higher rotation speeds and higher shear forces. Therefore, energy inputs are also higher (1.387 W dm−3 measured

© 2016 Society of Chemical Industry

J Chem Technol Biotechnol (2016)

Enzymatic sugar beet hydrolysis in a horizontal rotating tubular bioreactor

www.soci.org

Table 7. Comparison of lignocellulosic (LC) feedstocks enzyme pretreatment in the HRTB and other bioreactors Bioreactor Type Characteristics

STR3

HRTB

HRB29

Feedstock Pretreatment

Sugar beet pulp Autoclave 121 ∘ C 20 min Batch 22 h 1 dm3 60 g dm−3 1.387 W dm−3 Cellulase from A. niger 42 mg g−1 40 Co No 0.045 %

Sugar beet pulp Autoclave 121 ∘ C 20 min Batch 24 h 10 dm3 135 g dm−3 0.525 W dm−3 UltrazymAFP-L 14 mg g−1 39 ∘ C No 0.144 %

Corn stover Acid steam exploded 170 ∘ C 5 min Batch or fed-batch 84 h 1 dm3 200 g dm−3 0.855 W dm−3 Cellic CTec 19–29 mg g−1 50 Co Yes 0.086 %

Operational conditions Working volume LC loading Power consumption Enzyme loading Temperature Chemical consumption Yields of glucose (in w/w %)

STR-helix29 Corn stover Acid steam exploded 170 ∘ C 5 min Batch or fed-batch 84 h 1 dm3 200 g dm−3 0.721 W dm−3 Cellic CTec 19–29 mg g−1 50 Co Yes 0.073 %

STR: stirred tank bioreactor (two turbine impellers); STR-helix: stirred tank bioreactor (helical ribbon impeller); HRB: horizontal rotating bioreactor; HRTB: horizontal rotating tubular bioreactor.

for the stirred tank bioreactor equipped with two impellers). For example, a stirred tank bioreactor equipped with helical ribbon impellers (STR-helix) showed considerably lower shear forces and a power input of 0.721 W dm−3 , although higher concentrations of different substrates (200 g dm−3 ; corn stover) were used (Table 7). A HRTB is operated at relatively low rotation speeds and relies on the mixing obtained when material falls to the bottom of the tube. The shear forces in the HRTB are also relatively low, and independent of the rotation speed, which could explain the minor impact of rotation speed on the SBP hydrolysis. A direct comparison between the HRTB and the horizontal rotating bioreactor (HRB) is difficult due to the different experimental set-ups (different substrate types and concentration as well as different enzyme mixtures), the wall effect and the impact of the area/volume ratio, which is reduced with increasing bioreactor volume.30 However, similar bioreactor designs appear to have very different energy consumptions. As can been seen in Table 7, HRTB generally has a lower energy demand compared with the HRB. Furthermore, a comparison between bioprocesses in HRB and HRTB also shows that these bioprocesses were performed at different scales (1 dm3 HRB; 10 dm3 HRTB), which additionally highlights the potential of the HRTB where lower energy consumption was observed (Table 7). Finally, the choice of process parameters and bioreactor type will be determined by the overall bioprocess efficiency and economy.31 In our previous research, bioethanol production in the HRTB from raw sugar beet cossettes (sugar rich substrate) was studied using different initial quantities of yeast suspensions (9.1–23.7% V/m = volume of inoculum / mass of raw sugar beet cossettes) in order to define the minimal liquid content required. Based on these results it is clear that the liquid (or solid) content of the broth had a pronounced effect on the bioprocess performance in the HRTB.32 Especially at high solid loadings, and therefore high product concentrations, the saccharification process can be hampered due to the inhibition of hydrolytic enzymes, which lowers the overall efficiency of the bioprocess.33 Several reasons for this have been proposed: (i) inhibitors released during the pre-treatment process may directly inhibit enzymes; (ii) end-product inhibition of J Chem Technol Biotechnol (2016)

hydrolases at high sugar concentrations can slow down hydrolysis; (iii) non-productive adsorption of enzymes to lignin may prevent efficient cleavage of polysaccharides; or (iv) mass transfer limitations may occur.34,35 However, the detailed mechanism underlying the reduction of hydrolysis efficiency still remains elusive.

CONCLUSION In this research, the Design of Experiments method for the identification and optimization of critical parameters (pH, temperature and enzyme to substrate ratio) affecting the enzymatic hydrolysis of SBP was successfully employed. Among several tested commercial enzyme mixtures, the multi-enzyme cocktail Ultrazym AFP-L (contains polygalacturonase and cellulase activities) showed the highest release of reducing sugars in initial 1 mL scale reactions. On the basis of these results the efficient scale-up of SBP hydrolysis to a volume of 10 L was made in the HRTB. Aside from the enzyme-to-substrate ratio, it was identified that the overall lignocellulose concentration in the bioreactor is a crucial parameter affecting the final sugar yield. The HRTB design offers high mixing capacity, while the energy input is lower than with other bioreactor designs, such as the conventionally used stirred tank bioreactors. Furthermore, the HRTB has the capability to operate at higher substrate loadings compared with the stirred tank bioreactor, which is an important aspect regarding the economic viability of such bioprocesses.

ACKNOWLEDGEMENTS This work was supported financially by the Croatian Science Foundation under the project ‘Sustainable production of bioethanol and biochemicals from agricultural waste lignocelullosic raw materials (SPECH-LRM; No. 9158)’ and ‘Oxidative enzymes for removal of dyes from textile wastewater (WTZ2014/2015)’.

REFERENCES 1 Foster BL, Dale BE and Doran-Peterson JB, Enzymatic hydrolysis of ammonia-treated sugar beet pulp. Appl Biochem Biotechnol 91–93:269–282 (2011).

© 2016 Society of Chemical Industry

wileyonlinelibrary.com/jctb

www.soci.org 2 Kühnel S, Schols HA and Gruppen H, Aiming for the complete utilization of sugar-beet pulp: examination of the effects of mild acid and hydrothermal pretreatment followed by enzymatic digestion. Biotechnol Biofuels 4:14 (2011). 3 Rezi´c T, Oros D, Markovic I, Kracher D, Ludwig R and Šantek B, Integrated hydrolyzation and fermentation of sugar beet pulp to bioethanol. J Microbiol Biotechnol 23:1244–1252 (2013). 4 Zheng Y, Lee C, Yu C, Cheng Y-S, Zhang R, Jenkins BM et al., Dilute acid pretreatment and fermentation of sugar beet pulp to ethanol. Appl Energy 105:1–7 (2013). 5 Micard V, Renard CMGC and Thibault J-F, Enzymatic saccharification of sugar-beet pulp. Enzyme Microbiol Technol 19:162–170 (1996). 6 Kracher D, Oros D, Yao W, Preims M, Rezi´c I, Haltrich D et al., Fungal secretomes enhance sugar beet pulp hydrolysis. Biotechnol J 9:483–492 (2014). 7 Suwannarangsee S, Bunterngsook B, Arnthong J, Paemanee A, Thamchaipenet A, Eurwilaichitr L et al., Optimisation of synergistic biomass-degrading enzyme systems for efficient rice straw hydrolysis using an experimental mixture design. Bioresource Technol 119:252–261 (2012). 8 Murphy EF, Gilmou SG and Crable MJC, Effective experimental design enzyme kinetics in the bioinformatics. Era Drug Discov 7:187–191 (2002). 9 Murphy EF, Gilmour SG and Crable MJC, Efficient and accurate experimental design for enzyme kinetics: Bayesian studies reveal a systematic approach. J Biochem Biophys Meth 5:155–160 (2003). 10 Telen D, Logist F, Van Derlinden E, Tack I and Van Impe J, Optimal experiment design for dynamic bioprocesses: a multi-objective approach. Chem Eng Sci 78:82–97 (2012). 11 Hinkelmann K and Kemphtorne O, Design and Analysis of Experiments, 6th edn. John Wiley & Sons, New York (2007). 12 Donzelli BGG, Siebert KJ and Harman GE, Response surface modeling of the factors influencing the production of chitinolytic and beta-1,3-glucanolytic enzymes in Trichoderma atroviride strain P1. Enzyme Microbiol Technol 37:82–92 (2005). 13 Casas JA, Lara SG and García-Ochoa F, Optimization of a synthetic medium for Candida bombicola growth using factorial design of experiments. Enzyme Microbiol Technol 21:221–229 (1997). 14 Rodríguez-Nogalez JM, Ortega N, Perez-Mateos M and Busto MF, Experimental design and response surface modeling applied for the optimization of pectin hydrolysis by enzymes from A. niger CECT 2088. Food Chem 101:634–642 (2007). 15 Rezi´c I, Prediction of the surface tension of surfactants mixtures for detergent formulation using Design Expert software. Monatsh Chem Chem Mon 142:1219–1225 (2011). 16 Rezi´c T, Rezi´c I, Blaženovi´c I and Šantek B, Optimization of corrosion process of stainless steel coating during cleaning in steel brewery tanks. Mater Corros 64:321–327 (2013). 17 Wingren A, Galbe M and Zacchi G, Techno-economic evaluation of producing ethanol from softwood: comparison of SSF and SHF and identification of bottlenecks. Biotechnol Prog 19:1109–1117 (2003). 18 Varga E, Klinke HB, Rézey K and Thomsen AB, High solid simultaneous saccharification and fermentation of wet oxidized corn stover to ethanol. Biotechnol Bioeng 88:567–574 (2004).

wileyonlinelibrary.com/jctb

M Andlar et al.

19 Jorgensen H, Vibe-Pedersen J, Larsen J and Felby C, Liquefaction of lignocellulose at high-solids concentrations. Biotechnol Bioeng 96:862–870 (2007). 20 Zhang J, Chu D, Huang J, Yu Z, Dai G and Bao J, Simultaneous saccharification and ethanol fermentation at high corn stover solids loading in a helical stirring bioreactor. Biotechnol Bioeng 105:718–728 (2010). 21 Šantek B, Ivanˇci´c M, Horvat P, Novak S and Mari´c V, Horizontal tubular bioreactors in biotechnology. Chem Biochem Eng 20:389–399 (2006). 22 Mohagheghi A, Tucker M, Grohmann K and Wyman C, High solids simultaneous saccharification and fermentation of pretreated wheat straw to ethanol. Appl Biochem Biotechnol 33:67–81 (1992). 23 Du J, Zhang F, Li Y, Zhang H, Liang J, Zheng H et al., Enzymatic liquefaction and saccharification of pretreated corn stover at high-solids concentrations in a horizontal rotating bioreactor. Bioprocess Biosyst Eng 37:173–181 (2014). 24 Knutsen JS and Liberatore MW, Rheology of high-solids biomass slurries for biorefinery applications. J Rheol 53:877–892 (2009). 25 Hodge DB, Karim MN, Schell DJ and McMillan JD, Model based fed-batch for high-solids enzymatic cellulose hydrolysis. Appl Biochem Biotechnol 152:88–107 (2009). 26 Miller GL, Use of dinitrosalicylic acid reagent for determination of reducing sugars. Analyt Chem 31:426–428 (1959). 27 Kristensen JB, Enzymatic hydrolysis of lignocellulose: substrate interactions and high solids loadings. Forest and Landscape, University of Copenhagen. 28 Roche CM, Dibble CJ and Stickel JJ, Laboratory-scale method for enzymatic saccharification of lignocellulosic biomass at high-solids loadings. Biotechnol Biofuels 2:28 (2009). 29 Du J, Zhang F, Li Y, Zhang H, Liang J, Zheng H et al., Enzymatic liquefaction and saccharification of pretreated corn stover at high-solids concentrations in a horizontal rotating bioreactor. Bioprocess Biosyst Eng. 37:173–181 (2014). 30 Palmqvist B, Wiman M and Lidén G, Effect of mixing on enzymatic hydrolysis of steam-pretreated spruce: a quantitative analysis of conversion and power consumption. Biotechnol Biofuels 4:10 (2011). 31 Hodge DB, Karim MN, Schell DJ and McMillan JD, Soluble and insoluble solids contributions to high-solids enzymatic hydrolysis of lignocellulose. Bioresource Technol 99:8940–8948 (2008). 32 Pavleˇci´c M, Vrana I, Vibovec K, Ivanˇci´c Šantek M, Horvat P and Šantek B, Ethanol production on different intermediates of sugar beet processing. Food Technol Biotechnol 48:362–367 (2010). 33 Panagiotou G and Olsson L, Effect of compounds released during pretreatment of wheat straw on microbial growth and enzymatic hydrolysis rates. Biotechnol Bioeng 96:250–258 (2006). 34 Pan X, Role of functional groups in lignin inhibition of enzymatic hydrolysis of cellulose to glucose. J Biobased Mater Bio 2:25–32 (2008). 35 Sørensen HR, Pedersen S and Meyer AS, Optimization of reaction conditions for enzymatic viscosity reduction and hydrolysis of wheat arabinoxylan in an industrial ethanol fermentation residue. Biotechnol Prog 22:505–513 (2008).

© 2016 Society of Chemical Industry

J Chem Technol Biotechnol (2016)