To investigate the function of the linker region of RoGA, several constructs containing full-length, truncated or partially deleted GA fragments were generated (Figure 1). All constructs used in this study contained a leader signal sequence of the N-terminal 25 residues of RoGA. When the full-length GA was expressed in S. cerevisiae, secreted enzyme activity was detected in the culture medium following incubation at 30°C (Figure 2A, lane 2 and 2B, lower panel, lane 2). In contrast, removal of the interdomain 36-amino acid linker of RoGA (GAΔ132–167) led to generation of a mutant with no detectable secreted GA as determined by plate assay or Western blot analysis (Figure 2A, lane 3 and 2B, lower panel, lane 3), suggesting that the linker played an important role in the function of RoGA. To further examine whether the linker region was essential for the formation of active starch-binding or catalytic domains, four deletion mutants devoid of either the substrate binding or catalytic domain in the presence or absence of the linker region were engineered to generate plasmids pS1-GAΔ168–604 encoding the SBD with the C-terminal linker region; pS1-GAΔ132–604 encoding the SBD alone; pS1-GAΔ26–131 encoding the linker region preceding the catalytic domain; and pS1-GAΔ26–167 encoding the catalytic domain only. It was found that two SBD-containing clones, one, GAΔ168–604, containing the linker sequence and the other, GAΔ132–604, without the linker sequence were both successfully expressed and secreted (Figure 2B, upper panel, lanes 3 and 4). Since no effect was observed in the absence of the linker in both cases, the linker region appeared to be not essential for the formation and secretion of the SBD. As for the catalytic domain-containing variants, GAΔ26–131 was successfully expressed and secreted in yeast, but GAΔ26–167 lacking of the linker sequence was not expressed at all (Figure 2B, lower panel, lanes 4 and 7), indicating that the linker region was specifically required for formation and secretion of the catalytic domain. To further define which region in the linker was crucial for its role in facilitating formation and secretion of the catalytic domain, several constructs encoding partial deletions of the linker region fused to the catalytic domain were subsequently expressed and characterized. Interestingly, both GAΔ26–145 and GAΔ26–160, unlike GAΔ26–167, were successfully expressed in secretory forms (Figure 2B, lower panel, lanes 5 and 6), indicating that the linker region from residues 161–167 was essential for in vivo folding and secretion of the active catalytic domain. It is of interest that a recently published article 7 on Rhizopus oryzar found a second GA gene, amyB, lacks the N-terminal SBD. The amyB gene contains the conserved residues in the catalytic domain important for starch hydrolysis, and the gene also contains a stretch of linker sequence (33 bp) preceding the catalytic domain. This finding also indicates that the small region of linker plays an important role in function of RoGA.

In order to characterize RoGA and to study the function of the linker region, the recombinant proteins including full-length GA, GAΔ168–604, GAΔ132–604, GAΔ26–131, GAΔ26–145, and GAΔ26–160 were individually expressed in the budding yeast S. cerevisiae and were grown in SD media at 30°C. Extracellular (secreted) proteins were concentrated from culture supernatants and purified by cation-exchange chromatography. As shown in Figure 3A, purified full-length GA, GAΔ168–604, GAΔ132–604, GAΔ26–131, GAΔ26–145, and GAΔ26–160 exhibited single protein bands with purity higher than 90%. SDS-PAGE and/or mass spectrometric determination revealed that the respective molecular mass of full-length GA, GAΔ168–604, GAΔ132–604, GAΔ26–131, GAΔ26–145, and GAΔ26–160 was approximately 78, 21, 12, 69, 60, and 56 kDa (Table 2). The differences between the calculated and observed molecular weights ranged from 3% to 28%, indicating that post-translational modifications occurred in some of the recombinant proteins, presumably due to differential degree of glycosylation. There are two main types of protein glycosylation: N-glycosylation and O-glycosylation. The former refers to the attachment of oligosaccharides to a protein through the amide of asparagine residues, whereas the latter involves attachment of sugars to the hydroxyamino acids serine and threonine via their hydroxyl groups 31. Potential N-glycosylation sites in RoGA analyzed using the on-line prediction server NetNGlyc version 1.0 32 revealed that RoGA contained three putative N-linked glycosylation consensus sites, one (Asn¹⁶⁷-Ser-Thr) located in the linker region and two (Asn²³⁰-Thr-Thr and Asn²³⁶-Lys-Thr) in the catalytic domain (Figure 1). Although no specific consensus sequence for O-linked glycosylation has yet been reported 3334, O-glycosylation may occur on any of the Ser or Thr residues within a short peptide region 35. Interestingly, the program NetOglyc version 3.1 36 predicted that potential O-glycosylation sites in RoGA were clustered only in the linker region. Of the residues in the 36-amino-acid linker region, 61% were Ser and Thr, among which some might serve as potential sites for O-linked glycosylation. MALDI fingerprint mass spectrum from the tryptic digestion fragments of RoGA (data not shown) also showed that no peptide fragment was mapped to the linker region, presumably due to the presence of heterogeneous glycosylation within this region.

To determine how much of the molecular weight discrepancy for each recombinant protein was derived from the addition of glycosyl groups, each protein was enzymatically deglycosylated by PNGase F or Jack bean α-mannosidase followed by SDS-PAGE analysis. PNGase F is an amidase that cleaves between the innermost GlcNAc and the asparagine residue of complex N-linked oligosaccharides from glycoproteins 37. The fully glycosylated RoGA possessed a molecular mass of 78 kDa. After removal of N-linked carbohydrates with PNGase F, its molecular mass was reduced to 65 kDa (Figure 3B), indicating that RoGA produced by S. cerevisiae was highly glycosylated and the N-linked carbohydrates contributed approximately 13 kDa to the molecular weight. Enzymatic deglycosylation of asparagine-linked glycans in the catalytic domain variants GAΔ26–131, GAΔ26–145, and GAΔ26–160 also showed that these proteins were efficiently glycosylated to various degrees (Figure 3B).

The linker sequences code for a 36 amino acid extension at the C-terminal end of GAΔ168–604, which correspond to a molecular mass of approximately 3.5 kDa. The difference in molecular mass between GAΔ168–604 and GAΔ132–604 was 8.7 kDa, implying that the linker region contributed to the hyperglycosylation in GAΔ168–604. However, treatment of GAΔ132–604 and GAΔ168–604 with the N-glycan-specific PNGase F did not reduce its apparent molecular weight; hence there seemed to be little or no N-glycosylation on either protein (data not shown). At present, no enzyme comparable to PNGase F is available for removing intact O-linked sugars 38. In general, O-glycans are short linear oligosaccharides consisting of one to five mannose residues in yeast 243439. Therefore, α-mannosidase, an exoglycosidase capable of removing terminal mannose, was used to treat the recombinant proteins. If O-linked carbohydrates were present, the overall mass of the protein would decrease after treatment with α-mannosidase. The SBD variant GAΔ132–604 did not respond to the treatment with α-mannosidase, such that no change was detected in the electrophoretic mobility (Figure 3C), suggesting that no major O-glycosylation occurred in this case. On the other hand, a clear shift in the mobility of GAΔ168–604 was observed after treatment with α-mannosidase, indicating that presence of the linker region led to a considerable degree of modification by high-mannose-type O-glycans (Figure 3C). Taken together, our data indicate that the linker region of RoGA is modified by both N- and O-linked glycosylation, but the exact positions and functions are not clearly understood and require further investigation.

The minimal active linker region Ala¹⁶¹ to Asn¹⁶⁷ was previously demonstrated to be extremely important for GA function. In addition, N-terminal sequencing analysis of the GAΔ26–160 yielded the following sequence: Ala¹⁶¹-Thr¹⁶²-Phe¹⁶³-Pro¹⁶⁴-Xaa^165-Gly¹⁶⁶-Xaa¹⁶⁷, where Xaa represented unidentified residues. Generally speaking, biochemically modified amino acid residues were not identifiable by regular Edman degradation; hence residues Thr¹⁶⁵ and Asn¹⁶⁷ were possibly modified after translation. To abolish the glycosylation and evaluate the consequence, Thr¹⁶⁵ and Asn¹⁶⁷ in the linker region were individually mutated to generate T165A and N167D, respectively, and their effects on protein secretion and enzymatic activity were examined at 30°C. T165A mutation showed no effect on the phenotype as compared with wild-type full-length GA (Figure 4A and 4B, upper panel, lanes 2 and 5), while a significant decrease in GA expression and function was observed in the N167D mutant (Figure 4A and 4B, upper panel, lane 6). Site-specific mutation at Asn¹⁶⁷ led to much less active GA secretion, presumably due to either misfolding of RoGA, or faster degradation of the nascent protein. In support of the misfolding hypothesis, several different methods were used to improve the production of the mutant GA, N167D. One evidence showed that N167D exhibited enhanced expression at 20°C (Figure 4B, lower panel, lane 6), and about 0.5 mg protein could be obtained from each liter of yeast culture at 20°C. The activity of the N167D mutant could also be detected directly on starch-agar plate with prolonged incubation at 4°C (Figure 4C, No. 6). Similar phenomena were also observed for the mutants GAΔ132–167 and GAΔ26–167 after prolonged incubation at low temperature (Figure 4C, No. 3 and No. 4).

Moreover, elimination of the N-glycosylation site at position Asn¹⁶⁷ (N167D) resulted in a faster relative electrophoretic mobility and lower apparent molecular weight (Figure 4D), providing additional line of evidence that the only putative N-glycosylation site in the linker region, Asn¹⁶⁷-Ser-Thr, was indeed modified in S. cerevisiae. The biological function of carbohydrates in GA is not, however, completely understood, although it directs protein folding, facilitates secretion, and enhances stability of Aspergillus GAs 24. In this study, we have demonstrated that N-glycosylation at Asn¹⁶⁷ in the linker is required for functional expression of RoGA, whereas mutation abolishing N-glycosylation in the linker, N167D, decreased the efficiency of protein expression. Furthermore, similar result was obtained when Asn¹⁶⁷ was converted to Gln (data not shown). N-linked carbohydrates play important roles for a variety of structural and functional activities of glycoproteins 40. Here, we have characterized the N-glycans attached to the linker region of RoGA to be important for folding process, especially for its catalytic domain.

The SBD, the non-catalytic module of GA that binds raw starch 21, is found at the N terminus in RoGA and shares only 13.5% identity with that of AnGA 22. Characterization of the key functional groups of this domain has been accomplished using sequence-based structure alignment and NMR spectroscopy 2223. The effects of the linker region on SBD adsorption on to insoluble starch were thus investigated. The binding assay was conducted at pH 4.5 using insoluble corn-starch as the affinity matrix. Figure 5 showed the binding isotherms for the interaction between corn-starch and the SBD variants (GAΔ132–604 and GAΔ168–604). The binding isotherms were used to calculate binding parameters as described in the Materials and Methods section. The K_d values for GAΔ132–604 and GAΔ168–604 were determined to be 3.98 μM and 5.99 μM, respectively; and the B_max values of GAΔ132–604 and GAΔ168–604 were measured as 35.12 μmol/g and 17.24 μmol/g, respectively. It was apparent that the 106-residue SBD, GAΔ132–604, possessed stronger ligand affinity and higher capacity than those of the 142-residue SBD, GAΔ168–604, indicating that the presence of the linker region in the SBD construct did not increase but instead slightly interfered with the raw starch-binding affinity. The difference in starch binding affinity between GAΔ132–604 and GAΔ168–604 might be due to higher sugar content in the linker region or steric hindrance caused by the linker tail. Binding of the starch-degrading enzyme to its substrate is a critical step in starch hydrolysis because it involves the phase transfer of a soluble enzyme to the insoluble substrate 15. In our case the linker sequence influences the binding of the SBD to starch, possibly by affecting the ligand transfer process.

GA activity was assayed by measuring the reducing sugar released from the reaction on starch. The standard assay was performed as previously described with minor modification 17. The specific activities of the wild-type RoGA and the truncated mutants are listed in Table 3. Full-length GA and GAΔ26–131 exhibited similar activities, 4.58 × 10³ U/μmol and 4.88 × 10³ U/μmol, respectively, indicating that the absence of the N-terminal SBD had little or no effect on the ability to digest soluble starch. Moreover, comparison of the activities of the three catalytic domain variants GAΔ26–131, GAΔ26–145 and GAΔ26–160 revealed that longer size of the linker region was correlated with enhanced catalytic activity.

Figure 6 showed the effects of temperature and pH on enzyme activity. Regarding the thermal stability, the full-length RoGA remained stable at 40°C for 30 min at pH 4.5 with almost 100% of its activity remained, whereas at the same pH only approximately 35% and 5% residual activity was respectively detected at 50°C and 60°C (Figure 6, solid curve, open square). The three catalytic domain derivatives possessed inactivation profiles similar to that of the full-length GA (Figure 6, solid curves), indicating that the linker region did not contribute as much as expected to the thermal stability. In addition, the pH stability of the full-length RoGA at 25°C was found to be quite high over the pH range between 4.0 and 6.0, and more than 70% and 50% of the activity remained after 2 h incubation (Tm, 25°C) at pH 3.0 and pH 8.0, respectively (Figure 6, dotted curves). The truncated enzymes showed similar trends in stability over the pH range tested, indicating that the linker region was not crucial for the pH stability of GA either. Our data imply that the catalytic domain can function independently in terms of digesting soluble starch and the addition of linker sequences does not affect the thermal and pH stabilities of the enzyme.

The three-dimensional structure of RoGA is currently unavailable; circular dichroism spectroscopy has been widely used to study the secondary structure, with the goal of understanding the role of the linker in protein structure. In this study, the far-UV spectrum of full-length RoGA displayed two broad minima at 208 and 222 nm, characteristic of the presence of a mainly α-helical structure (Figure 7A, open square). The circular dichroism spectra of catalytic domain variants with different linker lengths were also indicative of a high content of helical conformation. In addition, the spectra of GAΔ26–131, GAΔ26–145, and GAΔ26–160 were very similar to each other, mainly differing in absorbance intensity (Figure 7A, closed symbols). GAΔ26–131 possessed a higher fraction of ordered secondary structure than did the other mutants, suggesting that the linker sequence might increase the stability of the secondary structural motifs. It was thus concluded that the linker sequences or their high degree of glycosylation facilitated stabilization of conformation of the catalytic domain of RoGA.

As for the SBD, Figure 7A also showed the calculated circular dichroism spectra of GAΔ132–604 and GAΔ168–604 (open triangle and circle, respectively). The negative ellipticity peak centered at 215 nm indicated that the secondary structure of the protein was predominantly β-sheet, but the latter showed a higher β-sheet conformation content. Similar to the case for the catalytic domain variants, the presence of the linker peptide in the recombinant SBD induced a substantial conformational change to a more ordered state. In addition, the thermal denaturation of GAΔ132–604 and GAΔ168–604 was monitored by circular dichroism at 215 nm. Figure 7B revealed that the denatured state of GAΔ168–604 possessed a greater negative ellipticity at 215 nm than did GAΔ132–604, suggesting that the presence of linker sequence might induce the SBD to form a more compact structure at high temperature. Taken together, although the 36-amino-acid linker alone of RoGA is predicted to be disordered, it strongly influences the structural ordering of the independent functional domains of RoGA.

Escherichia coli TOP10F' (Invitrogen) was used for plasmid manipulations, and the S. cerevisiae strain MNN10 (MATa, mnn10, leu2Δ0, his3Δ1, met15Δ0, ura3Δ0) was used for protein expression. E. coli cells were grown in Luria-Bertani medium (1% tryptone, 0.5% yeast extract, and 0.5% sodium chloride) at 37°C. Yeast cultures were grown in YPD (1% yeast extract, 2% peptone, and 2% glucose) or synthetic minimal (SD) medium (0.67% yeast nitrogen base supplemented with the appropriate amino acids and 2% fructose) at 20 or 30°C. The expression vector used in this work was pYEX-S1 (pS1, Clontech), a yeast-E. coli shuttle vector containing the phosphoglycerate kinase (PGK) promoter, the E.coli Amp^r gene, and the yeast selectable marker URA3.

The full-length cDNA encoding RoGA was kindly provided by Simpson Biotech, and the set of primers F-RoGA and R-RoGA was used to clone the RoGA gene. Both primers included an additional EcoRI site at their 5'-ends. The sequences of all the primers used are listed in Table 4. PCR amplification was carried out with KlenTaq-1 DNA polymerase (Clontech). The 1.8-kb PCR product was gel-purified and subcloned into yT&A TA cloning vector (Yeastern Biotech) to generate a pT-RoGA plasmid, which was further digested with EcoRI and then subcloned into pS1 at the same enzyme site to generate a pS1-RoGA plasmid.

A series of deletion mutants was generated by a PCR-based technique using the pS1-RoGA plasmid as the template. For the C-terminal truncation clone GAΔ168–604, which lacks the catalytic domain, the template was amplified using the primer pair F-RoGA and R-GAΔ168–604. For GAΔ132–604, which lacks both the linker region and the catalytic domain, the template was amplified using primers F-RoGA and R-GAΔ132–604. The resulting PCR products were separated on a 1% agarose gel and subcloned into pS1 as described above. The internal deletion clone GAΔ132–167, which lacks the entire linker sequence, was generated by fusing an N-terminal fragment to a C-terminal fragment with a two-step PCR approach. In the first step, F-PS1 paired with R-GAΔ132–167 and F-GAΔ132–167 paired with R-PS1 were used to generate PCR products of 516 bp and 1517 bp, respectively. The PCR products thus obtained shared overlapping sequences that could anneal in a secondary PCR. In the second step, the two first-stage PCR products were purified, combined and reamplified together with the external primers F-PS1 and R-PS1. The generated PCR product was subcloned into pS1 and sequenced for confirmation. Similar technologies were employed to obtain the other internal deletion clones with the following primer pairs: GAΔ26–131, F-GAΔ26–131 and R-GAΔ26–131; GAΔ26–145, F-GAΔ26–145 and R-GAΔ26–145; GAΔ26–160, F-GAΔ26–160 and R-GAΔ26–160; GAΔ26–167, F-GAΔ26–167 and R-GAΔ26–167. All primer sequences are listed in Table 4.

S. cerevisiae transformants were patched on YPD agar plate containing 0.5% (w/v) soluble starch. Secreted GA activity was detected by observing halo formation on agar plates containing iodine solution (0.01%).

Yeast strains were transformed using the one-step transformation method as described 46. The transformed yeast was cultivated in SD medium for 3 days at 30°C, and the supernatant of the culture medium was concentrated and dialyzed against 10 mM NaOAc (pH 4.5) using an Amicon stirred-cell concentrator (Millipore) equipped with a PM-10 membrane (10-kDa cut-off). For purification, the concentrated sample was loaded onto a 5-mL HiTrap SP cation-exchange column (Amersham Pharmacia Biotech) that was pre-equilibrated with the same buffer used for dialysis. The column was washed with 5 column volumes of 10 mM NaOAc buffer and then eluted with 10 volumes of a linear gradient of NaCl from 0 to 1 M in the same buffer at a flow rate of 2 mL/min. Fractions in 1-mL volumes were collected, and protein peaks were monitored by UV absorption at 280 nm. The fractions containing the desired proteins were pooled, dialyzed, and concentrated. The protein concentration was determined using the bicinchoninic acid (BCA) protein assay reagent kit (Pierce), with BSA as the reference standard.

SDS-PAGE was performed according to the method of Laemmli 47 using 10 or 15% (w/v) polyacrylamide gels. The protein bands in the gel were revealed by staining with Coomassie Brilliant Blue R-250. In Western blot analysis, extra- and intra-cellular samples were separated by SDS-PAGE and transferred electrophoretically onto PVDF membranes. The resulting Western blots were incubated with 3% BSA in TBS as the blocking solution and probed with polyclonal anti-RoGA antibody, used at a 1:5000 dilution. The secondary antibody was the horseradish peroxidase-conjugated anti-rabbit IgG, diluted to 1:5000. The bound complexes were detected with ECL reagents (ECL kit, Pierce) and exposure to X-ray film.

Molecular mass determination of the recombinant proteins was performed by Liquid Chromatography/Mass Spectrometer (LC/MS). The intact proteins (100 pmol) were acidified with 0.1% formic acid in 50% (v/v) acetonitrile, and the data were acquired over the 800–1800 m/z range under normal scan resolution. The original electrospray mass spectrum with multiply charged ion series was deconvoluted to give a mass spectrum.

For N-terminal amino acid sequencing, purified protein was separated by SDS-PAGE, and electrophoretically transferred to a PVDF membrane. The blotted proteins were visualized using Coomassie Brilliant Blue R-250. After destaining, the blots were rinsed in deionized water and air-dried. N-terminal sequencing was performed by automated Edman degradation using an Applied Biosystems model 494 Procise sequencer at the National Taiwan University.

Peptide N-glycosidase F (PNGase F, New England Biolabs) and Jack bean α-mannosidase (Sigma-Aldrich) were used in enzymatic deglycosylation of the recombinant proteins according to the manufacturers' instructions. After enzymatic treatment, the samples were separated by SDS-PAGE and stained with Coomassie Brilliant Blue R-250 to detect any differences in protein migration.

Site-directed mutagenesis was performed with a PCR-based technique to change Thr¹⁶⁵ to Ala and Asn¹⁶⁷ to Asp. The primers used were as follows: universal primers, F-PS1 and R-PS1; mutagenic primers F-T165A and R-T165A for the Thr¹⁶⁵→Ala substitution; and F-N167D and R-N167D for the Asn¹⁶⁷→Asp substitution. All primer sequences are listed in Table 4.

The adsorption of the SBD to granular corn-starch (Sigma-Aldrich) was performed as follows. The purified SBD-containing protein at a concentration ranging from 1.0 to 7.5 μM was mixed with 1 mg of prewashed corn-starch in a final volume of 1 mL in 10 mM NaOAc, pH 4.5. After incubation at 25°C for 16 h, the starch was removed by centrifugation at 16,000 g for 3 min at 4°C, and the amount of unbound protein remaining in the supernatant was determined by the BCA assay. The amount of adsorbed protein was calculated from the difference between the initial and unbound protein concentrations. The maximal amount of bound protein (B_max) and the dissociation constant (K_d) were determined by fitting to the non-linear regression of the binding isotherms, and the following equation was used for saturation binding with one binding site: B = B_maxF/(K_d+F), where B (μmol) represents the bound protein, B_max (μmol) is the maximal amount of bound protein, F (μmol) is the free protein in the system, and K_d (μmol) is the equilibrium dissociation constant. The units of calculated B_max and K_d were converted to micromoles per gram and micromolar, respectively.

Substrate was prepared by boiling 1% soluble corn-starch in 10 mM NaOAc, pH 4.5. The amount of glucose liberated from the starch by enzymatic activity was determined using the glucose oxidase/peroxidase kit (Sigma-Aldrich). For each reaction, 50 μL of substrate solution was equilibrated in a 37°C water bath for 5 min, and the assay was initiated by adding 50 μL of appropriately diluted enzyme solution. After incubation at 37°C for 5 min, 200 μL of kit solution was added to each sample. The reaction mixtures were incubated for 5 min at 37°C and stopped by the addition of 200 μL of 6 M H₂SO₄. The absorbance was recorded at 540 nm using a U-3310 spectrophotometer (Hitachi). One unit of enzyme activity was defined as the amount of enzyme that produces 1 μmol glucose/min under the assay conditions described above.

Thermal stability of the enzyme was determined by pre-incubating the purified protein for 30 min at various temperatures ranging from 20 to 65°C. At time intervals, the residual activity was measured with the glucose assay kit as described above. Stability of the recombinant protein at different pH values was studied by incubating the enzyme in various buffers with pH values ranging from 2.0 to 11.0 for 2 h and then measuring the residual GA activity. The buffers used were 0.1 M glycine-HCl (pH 2.0, 3.0), 0.1 M sodium acetate (pH 4.0, 5.0), 0.1 M potassium phosphate (pH 6.0, 7.0), 0.1 M Tris-HCl (pH 8.0), and 0.1 M glycine-NaOH (pH 9.0, 10.0, 11.0).

Circular dichroism measurements were carried out with an AVIV model 202 spectropolarimeter (Aviv Associates, Lakewood, NJ). Far-UV wavelength scans were recorded from 200 to 260 nm with a bandwidth of 1.0 nm, using a 0.1-cm path length cuvette at 25°C. Each spectrum was an average of three consecutive scans and was corrected by subtracting the buffer spectrum. Thermal denaturation experiments were performed by increasing the temperature from 25 to 96°C, allowing temperature equilibration for 1.5 min before recording each spectrum. All of the data are expressed in terms of mean residue ellipticity, [θ]_m.r.w., calculated by the equation: [θ]_m.r.w. = (100θ_obs)/(nlc), where θ_obs is the observed ellipticity in degrees, n is the number of amino acids, l is the length of the light path in centimeters, and c is the molar concentration of the protein.