ABSTRACT
A color-variant strain of Aureobasidium pullulans (NRRL Y-12974) produced α-l-arabinofuranosidase (α-l-AFase) when grown in liquid culture on oat spelt xylan. An extracellular α-l-AFase was purified 215-fold to homogeneity from the culture supernatant by ammonium sulfate treatment, DEAE Bio-Gel A agarose column chromatography, gel filtration on a Bio-Gel A-0.5m column, arabinan-Sepharose 6B affinity chromatography, and SP-Sephadex C-50 column chromatography. The purified enzyme had a native molecular weight of 210,000 and was composed of two equal subunits. It had a half-life of 8 h at 75°C, displayed optimal activity at 75°C and pH 4.0 to 4.5, and had a specific activity of 21.48 μmol · min−1· mg−1 of protein againstp-nitrophenyl-α-l-arabinofuranoside (pNPαAF). The purified α-l-AFase readily hydrolyzed arabinan and debranched arabinan and released arabinose from arabinoxylans but was inactive against arabinogalactan. TheKm values of the enzyme for the hydrolysis of pNPαAF, arabinan, and debranched arabinan at 75°C and pH 4.5 were 0.26 mM, 2.14 mg/ml, and 3.25 mg/ml, respectively. The α-l-AFase activity was not inhibited at all byl-arabinose (1.2 M). The enzyme did not require a metal ion for activity, and its activity was not affected byp-chloromercuribenzoate (0.2 mM), EDTA (10 mM), or dithiothreitol (10 mM).
More than one billion gallons of ethanol are produced annually in the United States, with approximately 95% of it being derived from the fermentation of corn starch (4). Various lignocellulosic agricultural residues such as corn fiber, corn stover, straw, and bagasse can also serve as low-value and abundant feedstocks for the production of fuel alcohol. Currently, the utilization of lignocellulosic biomass to produce fuel ethanol faces significant technical and economic challenges, and its success depends largely on the development of highly efficient and cost-effective enzymes for the conversion of pretreated biomass to fermentable sugars. Hemicelluloses, the second most common polysaccharides in nature, represent about 20 to 35% of lignocellulosic biomass (38). l-Arabinosyl residues are widely distributed in hemicelluloses as they constitute monomeric and/or oligomeric side chains on the β-(1→4)-linked xylose or galactose backbones in xylans, arabinoxylans, and arabinogalactans and are the core in arabinans forming α-(1→5) linkages (26, 36). These side chains restrict the enzymatic hydrolysis of hemicelluloses by xylanases (2). α-l-Arabinofuranosidases (α-l-arabinofuranoside arabinofuranohydrolase, EC3.2.1.55 ) (α-l-AFase) are exo-type enzymes which hydrolyze terminal nonreducing α-l-arabinofuranosyl groups from l-arabinose-containing polysaccharides. These enzymes can hydrolyze (1→3)- or (1→5)-α-l-arabinofuranosyl linkages of arabinan or both. The α-l-AFases are part of the microbial xylanolytic systems required for the complete breakdown of heteroxylans (2, 9, 22, 27).
In recent years, arabinofuranosidases have received much attention because of their practical applications in various agro-industrial processes such as the efficient conversion of hemicellulosic biomass to fuels and chemicals, delignification of pulp, efficient utilization of plant materials for animal feed, and hydrolysis of grape monoterpenyl glycosides during wine fermentation (3, 8, 10, 35). There is a need to develop a suitable α-l-AFase for use in the conversion of hemicellulose to fermentable sugars for the subsequent production of fuel ethanol and other value-added chemicals. α-l-AFases are produced by several bacteria and fungi, but only a few of these enzymes have been purified and characterized (8, 11, 13, 22, 33). The color-variant strains of the yeast-like fungus Aureobasidium pullulans have been recognized as excellent producers of amylases, xylanase, and β-glucosidase (20, 30, 31). These color-variant strains are differentiated from typically pigmented (off-white to black in appearance) strains of A. pullulans by their brilliant pigments of red, yellow, pink, or purple and their low DNA relatedness (21, 37). We have found that a color-variant strain ofA. pullulans produced an extracellular highly thermostable α-l-AFase which was able to hydrolyze both (1→3) and (1→5) linkages in arabinan. In this paper, we report on the purification and characteristics of this novel enzyme.
MATERIALS AND METHODS
Substrates and chemicals.Arabinogalactan, all saccharides, all aryl-glycosides, and molecular weight markers for gel filtration were purchased from Sigma Chemical Co., St. Louis, Mo. Arabinan (beet sugar), debranched arabinan, wheat arabinoxylan, and rye arabinoxylan were purchased from MegaZyme, North Rocks, Australia. Molecular weight markers and precast gels for sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and DEAE Bio-Gel A agarose, Bio Gel A-0.5m, and Aminex HPX-87C columns for high-pressure liquid chromatography (HPLC) were obtained from Bio-Rad Laboratories, Hercules, Calif. SP-Sephadex C-50 and epoxy-activated Sepharose 6B were from Pharmacia LKB Biotechnology, Piscataway, N.J.
Organism, cultivation, and enzyme production.The color-variant strain NRRL Y-12974 of A. pullulans was obtained from the ARS culture collection (National Center for Agricultural Utilization Research, Peoria, Ill.). The medium (32) used for seed culture and enzyme production had the following composition (per liter): 10 ml of solution A, 10 ml of solution B, 100 ml of solution C, 10 g of yeast extract, and 10 g of oat spelt xylan. Solution A contained (per liter) 1.1 g of CaO, 0.4 g of ZnO, 5.4 g of FeCl3 · 6H2O, 0.25 g of CuSO4 · 5H2O, 0.24 g of CoCl2 · 6H2O, 0.06 g of H3BO3, and 13 ml of concentrated HCl. Solution B contained (per liter) 10.1 g of MgO and 45 ml of concentrated HCl. Solution C contained (per liter) 64 g of urea, 12 g of KH2PO4, and 1.8 g of Na2HPO4. Oat spelt xylan (5% suspension in water) was sterilized (121°C, 15 min) separately. The pH of the medium was adjusted to 5.0 with 1 M HCl before inoculation. A 125-ml Erlenmeyer flask containing 50 ml of medium with oat spelt xylan (1%, wt/vol) as a carbon source was inoculated with a loopful of cells taken from a stock slant and incubated at 28°C on a rotary shaker (200 rpm) for 2 days. The shake flasks (250-ml Erlenmeyer flasks containing 100 ml of medium with 1% oat spelt xylan) were inoculated with 2 ml of this culture and cultivated on a rotary shaker (200 rpm) at 28°C. After 3 days, the cells were removed from the culture broth by centrifugation (18,000 × g, 20 min). The resulting supernatant solution was used as the crude enzyme preparation.
Enzyme assay.α-l-AFase activity was assayed in a reaction mixture (1 ml) containing 1 mM p-nitrophenyl α-l-arabinofuranoside (pNPαAF), 50 mM acetate buffer (pH 4.5), and appropriately diluted enzyme solution. After incubation at 50°C for 30 min, the reaction was stopped by adding ice-cold 0.5 M Na2CO3 (1 ml), and the color that developed as a result of p-nitrophenol liberation was measured at 405 nm. One unit of α-l-AFase activity was defined as the amount of enzyme that liberates 1 μmol of p-nitrophenol (pNP) per min in the reaction mixture under these assay conditions.
Purification of α-l-AFase.All purification steps were performed at 4°C, unless otherwise stated.
(i) Ammonium sulfate treatment.The culture supernatant (1,500 ml) was treated with ammonium sulfate (80% saturation) and left overnight. The precipitate was collected by centrifugation at 48,000 × g for 30 min, dissolved in 50 mM acetate buffer (pH 5.0), and dialyzed overnight against the same buffer.
(ii) DEAE Bio-Gel A agarose column chromatography.The dialyzed enzyme solution (630 ml) was concentrated to ∼20 ml by ultrafiltration with a stirred cell (model 202; Amicon, Inc., Beverly, Mass.) equipped with a PM 10 membrane under nitrogen pressure of 20 lb/in2, diluted 10-fold with 50 mM imidazole buffer (pH 6.5), and applied to a DEAE Bio-Gel A agarose column (2.5 by 26 cm) preequilibrated with 50 mM imidazole buffer, pH 6.5. The column was washed extensively with the same buffer and eluted with a continuous gradient of 0 to 0.5 M NaCl in the same buffer (280 ml each). The α-l-AFase activity was eluted as a single peak. The highly active fractions (fractions 21 to 27; fraction volume, 9 ml) were pooled, concentrated by ultrafiltration with a PM 10 membrane, and dialyzed overnight against 50 mM acetate buffer, pH 5.0.
(iii) Gel filtration on Bio-Gel A-0.5m column.The α-l-AFase was further purified by gel filtration on a Bio-Gel A-0.5m column (1.5 by 120 cm) preequilibrated with 50 mM acetate buffer, pH 5.0. The enzyme solution in 50 mM acetate buffer, pH 5.0, was applied to the column and eluted with the same buffer. The highly active α-l-AFase fractions (fractions 38 to 50; fraction volume, 2.5 ml) were pooled and concentrated by ultrafiltration (PM 10 membrane).
(iv) Arabinan-Sepharose 6B affinity chromatography.An arabinan-Sepharose affinity matrix was prepared by coupling arabinan to epoxy-activated Sepharose 6B according to the manufacturer’s recommendations (Pharmacia Fine Chemicals, Uppsala, Sweden). The α-l-AFase obtained after Bio-Gel A-0.5m gel filtration was applied to an arabinan-Sepharose 6B column (2.5 by 6 cm) preequilibrated with 50 mM acetate buffer, pH 5.0. The column was extensively washed with this buffer, and the enzyme was eluted with 1 M NaCl in the same buffer. The active enzyme fractions were pooled, concentrated by ultrafiltration (PM 10 membrane) and dialyzed against 50 mM citrate-phosphate buffer, pH 3.5.
(v) SP-Sephadex C-50 column chromatography.The dialyzed enzyme solution was applied to a SP-Sephadex column (2.5 by 10 cm) preequilibrated with 50 mM citrate-phosphate buffer, pH 3.5. The column was washed extensively with the same buffer and eluted with a continuous gradient of 0 to 0.5 M NaCl in the same buffer (150 ml each). The α-l-AFase activity was eluted as a single peak. The active enzyme fractions (fraction volume, 6.4 ml) were pooled, concentrated by ultrafiltration with a PM 10 membrane, and dialyzed overnight against 50 mM acetate buffer, pH 5.0. The dialyzed enzyme solution was used as purified α-l-AFase for subsequent studies.
Other methods.The protein concentration was estimated by the method of Lowry et al. (24) with bovine serum albumin as the standard. The protein concentration in the column effluents was monitored by measuring absorbance at 280 nm. SDS-PAGE was performed on a 12% gel according to the method of Laemmli (19). The apparent molecular weight (MW) of the native enzyme was determined by gel filtration on a Bio-Gel A-0.5m column, as described by Andrews (1), with apoferritin (MW 443,000), sweet potato β-amylase (MW 200,000), yeast alcohol dehydrogenase (MW 150,000), bovine serum albumin (MW 66,000), and ovalbumin (MW 45,000) as standard proteins. The half-life of the enzyme was estimated by incubating the enzyme at 75°C and determining the residual enzyme activity after certain intervals. The Km andVmax values were determined by the double-reciprocal plot method of Lineweaver-Burk (23) by using the KINET software program. Arabinose analysis was performed by HPLC (Spectra-Physics, San Jose, Calif.) with an ion-moderated partition chromatography column (Aminex HPX-87C). The column was maintained at 85°C, and the sugars were eluted with Milli-Q water at a flow rate of 0.6 ml/min. Peaks were detected by refractive index and were identified and quantified by comparison to retention times of authentic standards (l-arabinose, d-glucose, and d-xylose).
RESULTS
Production of α-l-AFase.The time course of α-l-AFase production by A. pullulans grown on oat spelt xylan was studied. The whole-broth α-l-AFase activity (assayed without breaking the cells) increased sharply up to 24 h, increased slowly from 24 up to 48 h, and then declined slowly. The extracellular enzyme was released into the culture fluid after obtaining almost maximum total enzyme production (in about 24 h), and its concentration increased sharply during the 24- to 48-h growth period after which it remained almost the same. About 50% of the enzyme activity was extracellular.
Purification of α-l-AFase.An extracellular α-l-AFase was purified to apparent homogeneity from the culture filtrates of A. pullulans grown on oat spelt xylan. A summary of the purification procedures is presented in Table1. Only the first few active fractions with high specific activity were pooled in the case of Bio Gel A-0.5m gel filtration. The α-l-AFase was well adsorbed on the arabinan-Sepharose affinity matrix but was easily eluted with 1 M NaCl. The enzyme was also well adsorbed onto the SP-Sephadex C-50 cation-exchange matrix at pH 3.5. Upon SDS-PAGE of the purified α-l-AFase, a single band was visualized when stained with Coomassie brilliant blue (Fig. 1). The final purification resulted in a yield of 11% of the activity, 0.05% retention of total protein, and a 215-fold increase in specific activity (Table 1). The purified α-l-AFase had a specific activity of 8.58 U/mg of protein (assayed at pH 4.5 and 50°C with pNPαAF as substrate). The specific activity of the purified enzyme under optimal conditions (pH 4.5 and 75°C) was 21.48 U/mg of protein.
Purification of α-l-AFase from A. pullulans Y-12974
SDS-PAGE gel of purified α-l-AFase fromA. pullulans Y-12974. The enzyme (∼20 μg of protein) was electrophoresed at pH 8.3 on a 12% acrylamide gel and stained with Coomassie brilliant blue R-250. Lane 1, MW standards; lane 2, purified α-l-AFase. The standards were myosin (200,000), β-galactosidase (116,250), phosphorylase B (97,400), bovine serum albumin (66,200), and ovalbumin (45,000). MWs (in thousands) are shown at left.
Characterization of α-l-AFase. (i) MW.The apparent MW of the native α-l-AFase estimated by gel filtration on a Bio-Gel A-0.5m column was 210,000. By SDS-PAGE analysis, the MW of the enzyme was about 105,000 (Fig. 1), suggesting that the α-l-AFase was composed of two subunits of equal MW.
(ii) pH and temperature dependence.The thermostability and thermoactivity of the purified α-l-AFase from A. pullulans are shown in Fig. 2. The purified enzyme in 50 mM acetate buffer, pH 5.0 (18 mU/ml; 2.11 μg of protein/ml), was fairly stable up to 75°C for 30 min with a half-life of about 8 h at 75°C. It exhibited maximum activity at 75°C under the assay conditions used (Fig. 2). The enzyme was stable at pH 4.0 to 5.5 (1 h at 75°C) (Fig. 3). It displayed an optimum activity at pH 4.0 to 4.5 with 47 and 40% relative activities at pH 3.0 and 6.0, respectively (Fig. 3).
Effect of temperature on stability (○) and activity (•) of purified α-l-AFase from A. pullulansY-12974. For stability, the enzyme solution in acetate buffer (50 mM, pH 5.0) was incubated for 30 min at various temperatures, and then the residual enzyme activities were assayed. For activity, the enzyme activity was assayed at various temperatures by the standard assay method. The concentration of enzyme was 18 mU/ml.
Effect of pH on stability (○) and activity (•) of purified α-l-AFase from A. pullulans Y-12974. For stability, the enzyme solutions in 50 mM citrate-phosphate buffer at various pH values were incubated for 1 h at 75°C. After adjustment of pH, the residual activity was assayed by the standard method. The enzyme activity was assayed by the standard assay method by changing the buffer to obtain the desired pH. The range of pH values for the buffer was 3.0 to 8.0. The concentration of the enzyme was 18 mU/ml.
(iii) Substrate specificity and kinetic analysis.Relative initial rates of hydrolysis ofp-nitrophenyl-α-l-arabinopyranoside,p-nitrophenyl-α-d-glucoside,p-nitrophenyl-β-d-glucoside,p-nitrophenyl-β-d-xyloside,p-nitrophenyl-β-d-cellobioside, andp-nitrophenyl-β-d-galactoside (4 mM each) by the purified α-l-AFase (28 mU/ml) were 3.4, 2.3, 1.7, 3.1, 0.0, and 0.1%, respectively (expressed as a percentage of that obtained with pNPαAF at pH 4.5 and 75°C after a 15-min reaction). The rate dependence of the enzymatic reaction on the pNPαAF concentration at pH 4.5 and 75°C followed Michaelis-Menten kinetics. Reciprocal plot analysis showed an apparent Kmvalue of 0.26 mM and a Vmax value of 6.99 μmol of pNP · min−1 · mg−1 of protein for the hydrolysis of pNPαAF. Substrate inhibition was not observed with pNPαAF tested up to an 8 mM concentration.
(iv) Effect of l-arabinose and other sugars.Up to 1.2 M (21.6%) l-arabinose was not at all inhibitory to the enzyme. Xylose, glucose, mannose, galactose, xylitol, andl-arabitol (each at 0.11 M) did not inhibit the α-l-AFase activity.
(v) Effect of metal ions and reagents.The influence of certain inhibitors or activators on α-l-AFase activity was studied. The enzyme did not require Ca2+, Mg2+, Mn2+ (each at 5 mM), or Co2+ (0.5 mM) for activity. Enzyme activity was not affected by EDTA (10 mM), dithiothreitol (10 mM), orp-chloromercuribenzoic acid (0.2 mM).
(vi) Hydrolysis of arabinan, debranched arabinan, and arabinoxylans.The degradation of arabinan by α-l-AFase was monitored by analyzing the reaction product by HPLC. Only arabinose was detected in the reaction product. Arabinan (1%, wt/vol; l-arabinose: galactose: rhamnose: galacturonic acid, 88: 3:2:7) was completely hydrolyzed by the purified enzyme (0.445 U/ml) (Fig. 4). Debranched arabinan (1%, wt/vol;l-arabinose:galactose:rhamnose:galacturonic acid, 88:4:2:6) was also readily hydrolyzed by the α-l-AFase. TheKm values for the degradation of arabinan and debranched arabinan were 2.14 and 3.25 mg/ml, and theVmax values were 14.7 and 11.5 μmol · min−1 · mg−1 of protein, respectively. The arabinose concentrations released from arabinan and debranched arabinan (each at 1% [wt/vol]) by the enzyme were 0.145 and 0.13 mg/ml, respectively, after 18 h at pH 4.5 and 50°C with 0.3 U of enzyme per ml of reaction mixture. Arabinose was detected by HPLC analysis as a product from wheat arabinoxylan, rye arabinoxylan, oat spelt xylan, and birch wood xylan. The concentrations of arabinose released from wheat arabinoxylan, rye arabinoxylan, oat spelt xylan, and birch wood xylan (each at 1% [wt/vol]) by the enzyme were 0.8, 1.4, 0.25 and 0.56 mg/ml, respectively after 60 h at pH 4.5 and 50°C with 0.3 U of enzyme per ml of reaction mixture. However, no arabinose was detected in arabinogalactan hydrolysis by the α-l-AFase.
Time course of arabinan (1%, wt/vol;l-arabinose:galactose:rhamnose:galacturonic acid, 88:3:2:7) hydrolysis by purified α-l-AFase (0.445 U/ml) fromA. pullulans Y-12974 at pH 4.5 and 50°C.
DISCUSSION
This is the first report to our knowledge on the purification and characterization of α-l-AFase from a color-variant strain of A. pullulans. Also, this is the first α-l-AFase reported to have such a high thermostability. The purification results suggest that the enzyme preparation fromA. pullulans contains only one form of α-l-AFase as no other active peak was detected during each purification step. Multiple forms of α-l-AFase have been detected in the culture broth of Aspergillus nidulans(28), Aspergillus niger (29),Aspergillus terreus (25) and Penicillium capsulatum (7). The specific activity of α-l-AFase from A. pullulans was 21.48 U/mg of protein at pH 4.5 and 75°C.
The α-l-AFase from A. pullulans is a homodimer with an apparent native MW of 210,000 and a subunit MW of 105,000 (Fig.1). Work by Komae et al. (17) showed that the MW of α-l-AFase from Streptomyces purpurascens IFO 3389 was about 495,000 and the enzyme contained eight equal subunits with a MW of 65,000. The MW of α-l-AFase ofButyrivibrio fibrisolvens GS113 was 240,000 and the enzyme consisted of eight subunits of MW 31,000 (11). The α-l-AFase from Clostridium acetobutylicum ATCC 824 was a single polypeptide with a MW of 94,000 (22). The intracellular α-l-AFase from Aspergillus nigerwas a monomer with a MW of 67,000 (15). Thus, there is considerable diversity in enzyme structure for different microbial α-l-AFases. The maximal activity of the purified α-l-AFase from A. pullulans was observed at 75°C and pH 4.0 to 4.5 (Fig. 2 and 3). Other microbial α-l-AFases have a broad range of pH and temperature dependence, with optimum activity occurring between pH 3.0 and 6.9 and from 40 to 70°C (3, 6, 7, 12, 22). The purified enzyme from Rhodotorula flava is highly acid stable, retaining 82% of its activity after being maintained for 24 h at pH 1.5 and at 30°C, and has an optimum activity at pH 2.0 (34). Comparative properties of some microbial α-l-AFases including thermoactivity are summarized in Table2. To our knowledge, the purified α-l-AFase from A. pullulans is the most thermostable enzyme reported to date, with a half-life of 8 h at 75°C and an optimum temperature of 75°C. Therefore, the thermostability characteristic of the α-l-AFase from the mesophilic yeast-like fungus A. pullulansmakes the enzyme suitable for use in commercial hemicellulose saccharification processes operating at elevated temperatures.
Comparative properties of some thermostable microbial α-l-AFases
The purified α-l-AFase from A. pullulanshydrolyzed pNPαAF, arabinan, and debranched arabinan and released arabinose from wheat and rye arabinoxylans and oat spelt and birch wood xylans. In contrast to the α-l-AFases fromAspergillus niger (13) and S. purpurascens IFO 3389 (17), which hydrolyze either (1→5)- or (1→3)-arabinosyl linkages, the purified α-l-AFase from A. pullulans released only arabinose from α-(1→3)-arabinoxylans, arabinan, and debranched α-(1→5)-arabinan. From these results, it is concluded that the α-l-AFase from A. pullulans has hydrolytic activity for both α-(1→3)- and α-(1→5)-linked, nonreducing, terminal l-arabinofuranose residues and does not act on internal α-l-arabinosyl linkages. These properties are similar to the substrate specificity of other α-l-AFases from Streptomyces sp. strain 17-1 (14),Streptomyces diastaticus (33), and Bacillus subtilis 3-6 (16). The enzyme purified from S. purpurascens was active on pNPαAF but inactive against arabinans and arabinogalactans (17). Kormelink et al. (18) described another type of α-l-AFase that was active only on arabinoxylans. The α-l-AFase from Streptomyces lividans exhibited no activity against oat spelt xylan or arabinogalactan (26). It slowly acted on arabinan and arabinoxylans by releasing arabinose after prolonged incubation (overnight). The limit of hydrolysis of arabinan by the α-l-AFase from B. subtilis 3-6 was only 15%, even when the enzyme was present in excess (16).
It is interesting that the α-l-AFase from A. pullulans had very little activity (3.4%) onp-nitrophenyl-α-l-arabinopyranoside. It was essentially free from α-glucosidase, β-galactosidase, β-glucosidase, β-xylosidase, or cellobiosidase activity. The enzyme was not inhibited by 8 mM pNPαAF or 1.2 M (21.6%)l-arabinose in the reaction mixture. In contrast, one α-l-AFase from P. capsulatum was competitively inhibited by l-arabinose with a Kiof 16.4 mM (7), while l-arabinose at an 80 mM concentration caused a 40% reduction of the hydrolytic activity of α-l-AFase from Aspergillus nidulans(6). None of the metal ions tested stimulated or inhibited α-l-AFase activity and no enzyme inhibition was observed in the presence of EDTA, suggesting that no metals are needed for the enzymatic reaction. The α-l-AFase activity was not inhibited by dithiothreitol or the thiol-specific inhibitor (p-chloromercuribenzoic acid), indicating that disulfide bonds are not critical for enzyme activity. The α-l-AFases from Aspergillus nidulans(6), Ruminococcus albus (9), B. fibrisolvens (11), Streptomyces sp. strain 17-1 (14), and S. purpurascens (15) were all sensitive to sulfhydryl reagents.
The high activity of the α-l-AFase from A. pullulans on both arabinan and debranched arabinan, its ability to release l-arabinose from arabinoxylans, and its high thermostability and activity make the enzyme a promising candidate for use in the production of fermentable sugars from hemicellulosic biomass such as corn fiber and to improve animal feed digestibility by hydrolyzing arabinoxylans, a major component of animal feed (5).
FOOTNOTES
- Received 5 August 1997.
- Accepted 29 October 1997.
- Copyright © 1998 American Society for Microbiology