Doubling Power Output of Starch Biobattery Treated by the Most Thermostable Isoamylase from an Archaeon Sulfolobus tokodaii

Biobattery, a kind of enzymatic fuel cells, can convert organic compounds (e.g., glucose, starch) to electricity in a closed system without moving parts. Inspired by natural starch metabolism catalyzed by starch phosphorylase, isoamylase is essential to debranch alpha-1,6-glycosidic bonds of starch, yielding linear amylodextrin – the best fuel for sugar-powered biobattery. However, there is no thermostable isoamylase stable enough for simultaneous starch gelatinization and enzymatic hydrolysis, different from the case of thermostable alpha-amylase. A putative isoamylase gene was mined from megagenomic database. The open reading frame ST0928 from a hyperthermophilic archaeron Sulfolobus tokodaii was cloned and expressed in E. coli. The recombinant protein was easily purified by heat precipitation at 80 oC for 30 min. This enzyme was characterized and required Mg2+ as an activator. This enzyme was the most stable isoamylase reported with a half lifetime of 200 min at 90 oC in the presence of 0.5 mM MgCl2, suitable for simultaneous starch gelatinization and isoamylase hydrolysis. The cuvett-based air-breathing biobattery powered by isoamylase-treated starch exhibited nearly doubled power outputs than that powered by the same concentration starch solution, suggesting more glucose 1-phosphate generated.

Enzyme-based biocatalysis has become an attractive alternative to chemical catalysis because of its higher reaction selectivity and more modest reaction conditions 14,15 . But most enzymes are not suitable for industrial applications due to their relatively poor stability. Discovery and utilization of thermoenzymes from hyperthermophilic microorganisms and exploding megagenome database is of great interest for numerous industrial applications 15 . Sulfolobus tokodaii was originally discovered in an acidic spa in Beppu Hot Springs of Kyushu Island, Japan, in the early 1980s 16 . It is a hyperthermophilic archeaon with an optimal growth temperature of 80 o C and an optimal pH of 2.5-3.0. S. tokodaii strain 7 is the most investigated because it is the most abundant, can be easily isolated and cultivated in labs. Its genomic DNA sequence has been completed in 2001 16 . S. tokodaii may be an invaluable source of intrinsically thermostable enzymes.
In this study, the open reading frame (ORF) ST0928 which was hypothesized to encode a glycoside hydrolase -glycogen debranching enzyme (E.C.3.2.1.-) was cloned in E. coli. The recombinant enzyme was purified and characterized for the first time. Isoamylase-treated starch was tested to power biobattery compared to non-treated starch (Fig. 1b).

Results
Identification of a putative IA. Compared to thermostable alpha-amylase, isoamylase received less attention because its hydrolytic product -amylodextrin has limited applications. Approximately 10 isoamylases have been purified and characterized (Table 1). Among them, one from a hyperthemophilic archaeon S. solfataricus is thermostable 7 and the other from Bacillus lentus is thermotolerant 8,17 . But their lifetime at 90 o C, a temperature needed for starch gelatinization, is not long enough (e.g., several hours for alpha-amylase) for simultaneous starch gelatinization and enzymatic hydrolysis.
We searched potential thermostable isoamylase genes by following the below protocol. First, we collected all characterized isoamylase protein sequences. Second, we blasted the known isoamylase protein sequences against the whole gene database of the National Center for Biotechnology Information (NCBI) and especially against special hypthermophilic micro-organisms, whose optimal growth temperature is more than 80 o C. Third, we double checked possible thermostable isoamylase annotations in two other database-the Kyoto Encyclopedia of Genes and Genomes (KEGG) and the glycoside hydrolase family 13 of CAZy (http://www.cazy.org/). It was found that an ORF (ST0928) was annotated to encode a putative glycogen debranching enzyme 16 . Its deduced amino acid sequence contains 716 amino acids and has a calculated molecular weight of 83.1 kDa. This predicted mature enzyme has a family 48 carbohydrate-binding module (17-108 AA) and a catalytic domain of alpha-amylase (204-545 AA) followed by an unknown function polypeptide (546-716 AA). This putative IA shared 80% and 79% identities with a well-characterized IA from the archaeon S. solfataricus 7 and another putative IA from S. acidocaldarius, respectively, and much lower identities with reported bacterial IAs, such as E. coli (43%) 18 , Archorbacter sp. (53%) 19 , Flavobacterium odoratum 20 , Pseudomonas amyloderamosa (34%) 21 , Erwinia chrysanthemi (41%) 22  (43%) 24 . According to CAZy (http://www.cazy.org/), this putative IA belongs to glycoside hydrolase family 13, which includes more than 20 different kinds of hydrolases, such as alpha-amylase, pullulanase, cyclomaltodextrin glucanotransferase, isoamylase, trehaloe synthase, sucrose phosphorylase, and so on. Figure 2 shows the three highly conserved amino acid sequences located in the catalytic domains among archaeal, bacterial and plant isoamylases. The three essential amino acid sites of this enzyme were Asp in region I, Glu in region II, and Asp in region III, in an agreement with Asp375, Glu435, and Asp510 of the P. amyloderamosa isoamylase, all of which play a catalytic role in activities of the α -amylase family 21 . Expression and purification of isoamylase. The ST0928 was sub-cloned into the T7-promoter plasmid pET20b by restriction enzyme-free, ligase-free Simple Cloning technique 25 . Two E. coli strains BL21(DE3) and Rosetta (DE3) were tested to express the recombinant IA with a His tag on its C terminus. Apparently, E. coli Rosetta was a better host than BL21 to express the soluble targeted enzyme (Fig. 3A, the left gel) because this gene contained a lot of rare codons in E. coli, including one three-rare codon cluster and several two-rare codon clusters. Although the host Rosetta can co-express the tRNAs for rare codons, a clear band with a molecular weight of ~81 kDa was observed in the pellet fraction by SDS-PAGE (Fig. 3a, Lane P), suggesting a significant amount of inclusion body formed. The His-tag enzyme was purified by affinity adsorption on nickel-charged resins. Alternatively, the cell lysate containing this enzyme was treated at 80 o C for 30 min, to denature E. coli cellular proteins. After centrifugation, the targeted protein was the predominant band in the supernatant, being approximately 85% purity (Fig. 3a, Lane HT). The protein recovery efficiency for nickel resin adsorption and heat precipitation were 81% and 98%, respectively. Approximately 10 mg of the purified His-tagged enzyme was purified from 200 mL of the cell culture grown in the LB media. This His-tagged enzyme had a specific activity of 6.4 IU/mg on amylopectin at 80 o C based on the reducing ends generated. The specific activity of heat precipitated enzyme was approximately 89% of that purified from nickel resin adsorption, in consistent of SDS-PAGE data. Heat precipitation is the easiest approach for purifying relatively pure thermostable enzymes suitable for in vitro biocatalysis 26,27 . Basic enzyme properties. The optimal pH of this enzyme was tested in two buffers -acetate and phosphate on amylopectin (Fig. 3b). This enzyme had a narrow optimal pH 5.5 in the acetate buffers but a relatively broad pH range in the phosphate buffers. In 40 mM acetate buffer (pH 5.5), this enzyme exhibited the optimal temperature of 85 o C and remained approximately 50% activity at 50 o C (Fig. 3c), suggesting that this enzyme had a broad temperature range. The effects of the addition of 0.5 or 5 mM metal ions (i.e., CuCl 2 , FeCl 3 , ZnCl 2 , CaCl 2 , MgCl 2 , CoCl 2 , NiCl 2 , MnCl 2 ) and EDTA on enzyme activities were studied in the acetate buffer (pH 5.5) at 80 o C. The addition of EDTA regardless of its concentration caused protein aggregation and drastically decreased this enzyme activities, suggesting that some metal ions were important. Both MgCl 2 and CaCl 2 (0.5 or 5 mM) increased this enzyme activity, while 5 mM CoCl 2 , NiCl 2 , MnCl 2 significantly decreased the enzyme activity; CuCl 2 , ZnCl 2 and FeCl 3 completely inhibited this enzyme activity. Amylopectin was hydrolyzed by this enzyme under its optimal condition (e.g., acetate buffer (pH 5.5) containing 5 mM MgCl 2 and 80 o C) (Fig. 4). The branched amylopectin shows a typical brown-blue color after the iodine dying (Fig. 4a) because branched amylopectin cannot form coils and thus it does not form a complex with iodine. After this enzyme treatment, the solution turned a purple color (Fig. 4a), suggesting that linear amylodextrin forms a representative starch/iodine color -purple/deep blue. Figure 4b shows the changes in absorption spectra of the iodine-staining solution for the amylopectin before and after the treatment of this enzyme. The absorbance increased and the maximum wavelength of absorption shifted to a longer wavelength from 530 to 560 nm. These results suggest that the enzyme hydrolyzed the 1,6-alpha-glycosidic linkage of branched amylopectin. This enzyme exhibited a very low activity on amylose (~5%) relative to that on amylopectin, indicating that this enzyme preferred hydrolyzing alpha-1,6-glycosidic bonds. This very low activity on amylose could be due to the high-sensitivity reducing end assay based on the BCA assay instead of the commonly-used Somogyi assay and/or some  suggesting that it cannot hydrolyze alpha-1,4,6-D-glucose branch-points for short maltodextrin, different from pullulanase. The above results seemed appropriate to refer to this enzyme as an isoamylase but its weak alpha-1,4-hydrolytic activity was not eliminated completely. This enzyme had a specific activity of 6.4 IU/mg on amylopectin at 80 o C based on the reducing ends generated.
This isoamylase in the acetate buffer (pH 5.5) were very stable at temperatures of 60-80 o C, less than 1% activity losses for 1 h, and remained 87% activity after 1 h incubation of 90 o C. Surprisingly, this enzyme was more stable in the presence of 5 MgCl 2 than the absence of bivalent ions (Fig. 5). The addition of MgCl 2 resulted in a half lifetime of 200 min at 90 o C. In contrast, CaCl 2 decreased this enzyme stability greatly, resulting in a half lifetime of 35 min.
A de novo synthetic enzymatic pathway was designed to generate electricity from starch (Fig. 1b). In it, alpha-glucan phosphorylase (α GP) cleaves alpha-1,4-glycosidic bonds from nonreducing ends of starch, maltodextrin or amylodextrin in the presence of phosphate, yielding glucose 1-phosphate; phosphoglucomutase (PGM) converts glucose 6-phosphate; glucose 6-phosphate dehydrogenase generates NADH from glucose 6-phosphate and release 6-phosphogluconate; diaphorase transfers hydrides from NADH via a mediator AQDS to anode. This pathway was slightly different from the previous pathway used 28 : (i) amylopectin instead of maltodextrin as the substrate, and (ii) non-immobilized AQDS instead of immobilized VK3 as the mediator. The entire sugar biobattery based on a typical plastic cuvette without mobile parts is shown in Fig. 6d. Figure 6 shows the results of electrochemical tests of sugar batteries powered by starch and isoamylase-treated starch. Figure 6a,b display the polarization curves using isoamylase-treated or nontreated starch as the sugar biobattery's substrate, respectively. When nontreated starch was used, the polarization curve shows that the open circuit potential (OCV) was 0.23 V with short connection current of 0.029 mA. At 0.14 V, the power density reached to a peak of 2.2 μ W/cm 2 . In contrast, feeding the biobattery with isoamylase-treated starch, the maximum power density was almost doubled from 2.2 to 4.1 μ W/cm 2 . In the meantime, short connection current increased to 0.042 mA, and OCV increased to 0.31 V. To eliminate the different cathode performance, individual potentials were recorded (data was not shown). Both of the cathode potentials were 0.53 V with different substrates. Only the anode leaded to varied whole cell performance, suggesting more glucose 1-phosphate generated from isoamylase-treated starch. To further confirm this testing result, cyclic voltammetry were recorded in two types of anolyte solutions. As shown in Fig. 6c, both of the starches showed very slight oxidation peaks which may result from low concentration starch (0.012% wt/v). Both oxidation peaks of isoamylase-treated and nontreated starch were approximately −300 mV relative to Ag/AgCl, but isoamylase-treated starch had higher current indicating isoamylase-treated starch was better than nontreated starch in the anode reaction.

Discussion
Starch is the most widely used energy storage compound in nature. The catabolism of starch mediated by starch phosphorylase lead to a slow and nearly constant release of chemical energy (i.e., glucose 1-phosphate) in living cells that is different from that of the monomer glucose 29 . Amylodextrin made by isoamylase is much better than maltodextrin, a partially hydrolyzed starch fragment by alpha-amylase, because maltodextrin contains some 1,4,6 branched points, resulting in low glucose utilization efficiency. On the other side, amylopectin is a superior fuel to glucose because it has 11% higher energy density than glucose. An equivalent weight of amylopectin has a much lower osmotic pressure than glucose. Moreover, slowly-metabolized glucose 1-phosphate can provide more stable electricity generation in closed biobattery 2 .
This enzyme has the highest lifetime among all characterized isoamylases (Table 1). Due to its highest stability (i.e., a half lifetime of 200 min at 90 o C), this hyper-thermophilic enzyme can be used in simultaneous starch gelatinization and enzymatic hydrolysis at ~90 o C for several hours, like the case of alpha-amylase. This enzyme exhibited much better stability than the reported S. solfataricus IA, where the His-tag enzyme nearly lost all its activity after 120 min at 90 o C 7 .
This enzyme has different metal ion requirement from other reported isoamylases. Compared its closest IA from S. solfataricus, which did not require any metal ions 7 , this enzyme required Mg 2+ or Ca 2+ for its maximum activity. Furthermore, its thermostability was improved greatly in the presence of Mg 2+ . This metal preference of this enzyme was a little different from those of B. lentus IA that preferred Ca 2+ but Mg 2+ was an inhibitor 8 .
In conclusion, this enzyme was the most stable isoamylase reported and had a half lifetime of 200 min at 90 o C. Different from the closest IA from S. solfataricus, this required Mg 2+ as an activator while EDTA impaired its activity greatly. Due to its highest stability, this enzyme can be used for simultaneous starch gelatinization and isoamylase hydrolysis, producing linear amylodextrin. Isoamylase-treated starch produced nearly doubled power outputs in a sugar biobattery relative to that powered by the same concentration starch.  gene expression. E. coli strains were cultivated in the Luria-Bertani (LB) medium at 37 °C. Ampicillin at 100 μ g/mL was added in the E. coli medium.
Optimization of isoamylase reaction conditions. To determine optimal pH, the reaction solution was mixed by 350 μ l of 0.5% amylopectin solution, 100 μ l of 0.2 M buffer (e.g., acetate buffers (pH 4.0-6.0) or phosphate buffers (pH 5.0-8.0)), and 50 μ l of the enzyme solution (75 μ g/ml). The mixture was incubated at 80 °C for 30 min. The reaction was stopped by using an ice bath. Ten μ l of the enzymatic reaction mixture was mixed with 490 μ l of distilled water and 500 μ l of the bicinchonic acid (BCA) solution 30 . The tubes containing the reaction solutions were incubated at 75 o C for 30 min. Concentrations of reducing ends were measured by the modified BCA assay with glucose as a reference 30 . To determine optimal temperature, the reaction mixtures in 40 mM acetate buffer (pH 5.5) were incubated at a temperature from 40 to 90 °C for 30 min. The reducing ends generated by IA were measured by the BCA assay. To determine optimal metal concentration, the reaction mixtures in 40 mM acetate buffer (pH 5.5) supplemented with 0.5 mM or 5 mM of CuCl 2 , FeCl 3 , ZnCl 2 , CaCl 2 , MgCl 2 , CoCl 2 , NiCl 2 , MnCl 2 or EDTA were incubated at 80 °C for 30 min. The reducing ends generated by IA were measured by the BCA assay.
Isoamylase activity assay. Isoamylase activity was measured in 500 μ l of the reaction mixture containing 350 μ l of 0.5% (wt/v) amylopectin solution, 100 μ l of 0.2 M acetate buffer (pH 5.5) containing 2.5 mM of MgCl 2 , and 50 μ l of the enzyme solution (75 μ g/ml). The reaction mixtures were incubated at 80 °C for 30 min. The reducing ends generated by IA were measured by the modified BCA assay with glucose as a reference 30 . One international unit (IU) of isoamylase activity was defined as one micromole of reducing ends generated one min.
To determine the substrate specificity, the reaction mixture containing 350 μ l of 0.5% (wt/v) solution containing amylopectin, amylose, pullulan, and maltodextrins, 100 μ l of 0.2 M acetate buffer (pH 5.5) supplemented with 2.5 mM MgCl 2 , and 50 μ l of the enzyme solution (75 μ g/ml) was incubated at 80 °C for 30 min. The reducing ends generated by IA were measured by the BCA assay.
Alternatively, IA assay was measured by the increased blue value of glucan-iodine complexes as described elsewhere 7 . The reaction mixture contained 350 μ l of 0.5% amylopectin solution, 100 μ l of 0.2 M acetate buffer (pH5.5), 50 μ l of the enzyme solution. The mixture was incubated at 80 °C for 30 min 300 rpm. A half ml of 0.005 M I 2 -0.1 M KI solution was added, followed by the addition of 10 ml of distilled water, and the mixed well. The increase in absorbance at 610 nm was measured.
Scientific RepoRts | 5:13184 | DOi: 10.1038/srep13184 Thermostability. Fifty μ l of 75 μ g/ml IA solution was diluted in 100 μ l of 0.2 M acetate buffer (pH 5.5) containing 2.5 mM MgCl 2 or CaCl 2 or no divalent ions. The enzyme solutions were incubated at 70, 80 and 90 °C for different times. The remaining IA activities were measured as described previously.
EFC preparation and measurement. A cuvette enzymatic fuel cell was set up for testing as described previously 2 with some modifications. Membrane electrode assembly including Nafion and cathode (1.8 × 2 cm; from Fuel Cell Earth Woburn, MA, USA) was adhered by epoxy glue to cover up the open window (0.5 × 1.5 cm) in a cuvette (1 × 1 × 4.5 cm). Oxygen in air acted as an electron acceptor. 1 × 1 cm carbon paper (Fuel Cell Earth Woburn, MA, USA) was anode. To test the effect of substrate on the performance of EFC, two types of anolyte solution were made by adding 0.012 (wt/v)% isoamylase-treated starch or 0.012% starch. The other enzymes in anolyte per ml were 7 U of α GP, 3 U of PGM, 1.5 U of G6PDH, 5.4 U of DI in a 50 mM HEPES buffer (pH 7.2) containing 0.3 M NaCl, 4 mM NAD + , 5 mM Mg 2+ , 0.25 mM Mn 2+ , and 1.7 mM, analogue antraquinone-2,6-disulfonate (AQDS) as an electron shuttle. All the electrochemical tests were performed on a 1000B Multi-Potentiostat (CH Instruments Inc., Austin, TX, USA) interfaced to a personal computer at room temperature (~20 °C). Each test was repeated twice to ensure the reliability of data. For the linear sweep voltammetry (LSV), scanning was carried out at the rate of 5 mV s −1 after 10 min wait to measure EFC's open circuit potential. For the cyclic voltammetry (CV) tests, the anolyte solution was aerated 20 min with nitrogen gas before testing to eliminate dissolved oxygen. Ag/AgCl electrode was used as a reference; platinum wire was applied as a counter electrode. The scanning rate was at 2 mV s −1 .