Improved Activity and Modulated Action Pattern Obtained by Random Mutagenesis at the Fourth β-α Loop Involved in Substrate Binding to the Catalytic (β/α)8-Barrel Domain of Barley α-Amylase 1*

The functionality of the sequence Arg183-Gly184-Tyr185 of the substrate binding fourth β-α loop in the (β/α)8-barrel of barley α-amylase isozyme 1 (AMY1) was studied by random mutagenesis. A motif of polar Gly184hydrophobic residues was present in active mutants, selected by starch plate screening of yeast transformants. Gly184 was important, probably due to the carbonyl group binding to Ca2+ and the spatial proximity of Phe181. Mutation of both flanking residues as in Ser183-Gly184-Met185 (SGM-) and TGL-AMY1 decreased the Ca2+ affinity. SGM-AMY1 has 2-fold increased activity for amylose but reduced activity on maltooligosaccharides, whereas KGY-AMY1 has up to 3-fold elevated activity toward the oligosaccharides. TGL-AMY1 has modest activity on all substrates. Shifted action pattern on maltooligosaccharides for NGY-, SGM-, and TGL-AMY1 support that Arg183 in wild type is located at subsites +1 and +2, accommodating two sugar rings toward the reducing end from the site of cleavage. In the crystal structure of barley α-amylase 2 (AMY2), Lys182 (equivalent to AMY1 Arg183) is hydrogen-bonded with sugar OH-3 in subsite +2. Higher K i  app for acarbose inhibition of KGY-AMY1 and parent AMY1 compared with the other mutants suggests favorable substrate interactions for Arg/Lys183. KGY-AMY1 was not inhibited by the AMY2-specific proteinaceous barley α-amylase/subtilisin inhibitor, although Lys182 of AMY2 is salt-linked to the inhibitor.

* This work was supported by grants from the Scandinavia-Japan Sasakawa Foundation and the European Union Third Framework Program on Biotechnology (BIO2-CT94-3008). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ‡ Present address: Molecular Biology Division, National Inst. of Bioscience and Human-Technology, Tsukuba, Ibaraki 305, Japan.
Using local random mutagenesis we have been able to engineer barley ␣-amylase to have improved activity even toward amylose. This approach moreover provided a powerful method to both extract information on the tolerance of individual positions to substitutions and to define the role of the target residues. Thus replacements obtained in the sequence Arg 183 -Gly 184 -Tyr 185 of AMY1 allowed detailed structure/function relationship investigations and resulted in enzyme variants with significantly improved activity and altered substrate specificity. This is the first report on random mutagenesis applied to a specific substrate-binding ␤-␣ loop of the catalytic (␤/␣) 8 barrel of a member of glycosylase family 13.

EXPERIMENTAL PROCEDURES
Materials-Soluble starch and the 2-chloro-4-nitrophenyl ␤-D-maltoheptaoside kit (Granutest 3) were from Merck; G 1 -PNP and G 2 -PNP were purchased from Sigma; G 3 -PNP, G 4 -PNP, G 5 -PNP, and G 6 -PNP were from Calbiochem; and G 7 -PNP was from Boehringer Mannheim. Degree of polymerization (DP) 17 amylose EX-I was from Hayashibara Chemical Laboratories (Okayama), and insoluble Blue Starch was from Pharmacia Biotech Inc. Acarbose was a generous gift of Dr. E. Moller (Bayer AG, Wuppertal). Restriction enzymes were from Promega and were used according to the manufacturer's recommendation. AMY2 was purified from barley malt (cultivars Triumph or Menuet) (19,40) and BASI from barley seeds (cultivar Piggy) (41). Rabbit antibodies against barley ␣-amylase were raised using AMY2 as antigen (custom immunization by Dako A/S, Denmark).
Site-directed Mutagenesis and Random Mutagenesis-Standard cloning techniques (45) were used throughout. Because two SacI and two SacII sites were present in the AMY1 cDNA (46), silent mutations were introduced into the SacI (at position 646 -651) and SacII (at 471-476) sites of the AMY1 C95A gene (43) to make the outer SacI (at 376 -381) and SacII (at 1388 -1393) sites unique. This involved using polymerase chain reaction (PCR) (Perkin-Elmer Instrument DNA Thermal Cycler) with Vent DNA polymerase (New England BioLabs), two outside primers (SAC01BAM, 5Ј-TTTTTGGATCCTTTTTGAGCTCA-AGTCGCTCATCGGC-3Ј, and SAC02ECO, 5Ј-TTTTTGAATTCTTTT-TCCGCGGCACCGTTCTTCTCCC-3Ј), and two partially complementary inside primers (SAC05DEL, 5Ј-GACCGGGTCCAGCGCGAACT-CAAGGAGTGGCTC-3Ј, and SAC06DEL, 5Ј-GAGCCACTCCTTGAG-TTCGCGCTGGACCCGGTC-3Ј) (nucleotides in bold type were replaced for mutation). The two outside primers have BamHI and EcoRI recognition sites, respectively, at the 5Ј end. The PCR product was purified with QIAquick Spin PCR Purification kit (QIAGEN) and was digested with EcoRI and BamHI. The purified fragment was ligated into pUC18, which was used to transform E. coli by electroporation (Gene Pulsar, Bio-Rad). Transformants were screened on ampicillin containing LB (LBϩAmp) plates (47) incubated at 37°C overnight and subsequently propagated in 5 ml of LBϩAmp medium at 37°C overnight, and pUCBAL C95A ⌬S I was purified after centrifugation using a Midi Plasmid kit (QIAGEN). Introduction of the silent mutation as well as the absence of additional mutations were verified by sequencing on Applied Biosystems 373A DNA sequencer (Taq DyeDeoxy Terminator Cycle Sequencing kit; Perkin-Elmer). The SacII silent mutation was subsequently introduced in pUCBAL C95A ⌬S I by PCR using the outside primers SAC01BAM and SAC02ECO (see above), with a different set of inside primers (SAC03DEL, 5Ј-GCCGACTACAAGGAT-AGCAGAGGCATCTACTGCATCTTC-3Ј, and SAC04DEL, 5Ј-GAA-GATGCAGTAGATGCCTCTGCTATCCTTGTAGTCGGC-3Ј). The PCR product was subcloned into pUC18, and the recombinant plasmid pUCBAL C95A ⌬S I, II was purified as described above.
Random mutants of AMY1 at Arg 183 -Gly 184 -Tyr 185 were made in pUCBAL C95A ⌬S I, II by PCR using the outside primers described  above and the inside primers: PL02, 5Ј-CGTGGCGCCTTGACTTCG-CTNNNNNNNNNTCGCCGGAGATGGCCAAGGT-3Ј, plus PL05RV,  5Ј-AGCGAAGTCAAGGCGCCACGCGTCGAAGCCGAGGT-3Ј; random  mutants at Gly 184 were made using the same outside primers and the  inside primers: PL11, 5Ј-CGTGGCGCCTTGACTTCGCTAGGNNNT-ACTCGCCGGAGATGGCCAAGGT-3Ј, plus PL12RV, 5Ј-CCTAGC-GAAGTCAAGGCGCCACGCGTCGAAGCCGA-3Ј. The PCR products encoding the random mutations were purified, subcloned into pUC18, and then used to transform E. coli. For the Arg 183 -Gly 184 -Tyr 185 and Gly 184 random mutations, 10% (100 l) of the transformation mixture was screened as above. The transformant colonies were counted, each of seven clones was propagated in 5 ml of LBϩAmp medium, and the plasmids were purified and sequenced. The size of the pool of random mutants was estimated from the total number of transformants and the mutation efficiency as determined by DNA sequencing of the isolated plasmids. Each of the remaining portions of the transformation mixtures (900 l) was propagated in 1 liter of fresh LBϩAmp medium at 37°C overnight. The plasmids were purified (Maxi Plasmid kit; QIA-GEN) and designated as pUCBAL C95A X(183-185) ⌬S I, II and pUCBAL C95A X(184) ⌬S I, II, respectively.
Site-directed mutagenesis of Arg 183 to Lys in pUCBAL C95A ⌬S I, II was done by PCR as above using two different inside primers (PL13, 5Ј-CGTGGCGCCTTGACTTCGCTAAAGGCTACTCGCCGGAGATGGC-CAAGGT-3Ј, and PL05RV). The BamHI-EcoRI fragment was subcloned into pUC18 to yield pUCBAL C95A R183K ⌬S I, II, and the DNA sequence was confirmed.
The wild type ApaI-SacII fragment was replaced to give pUCBAL C95A R183K ⌬S I, II. The expression vector pBAL7 C95A R183K ⌬S I, II was constructed as above and used for yeast transformation. The transformants were grown on SD/ϪLeu ϩ 1% soluble starch plates at 30°C for 2 days. Clear halos in the starch were visualized by exposure to I 2 vapor for 15 min.
Enzyme Purification-S. cerevisiae DBY746 harboring mutant or wild type pBAL7 C95A ⌬S I, II was propagated in 10 liters of SD/ϪLeu medium at 25 or 30°C for 2 days (MBR BioReactor, AG Switzerland). The recombinant AMY1 was purified by affinity chromatography on ␤-cyclodextrin-Sepharose (49) and analyzed by isoelectric focusing (IEF), SDS-polyacrylamide gel electrophoresis (PAGE) (50), and Western blotting using PhastSystem (Pharmacia). For IEF (PhastGel, pI 4 -6.5) 30 ng of protein was applied (15 ng of RGC-AMY1) and visualized by silver staining according to the manufacturer's recommendation. A starch/I 2 -zymogram was prepared as earlier described (49). Native PAGE was carried out on PhastGel (10 -15%) using the same amount of protein for zymogram development as in IEF. For SDS-PAGE (PhastGel, 10 -15%) enzyme (6 l, 10 g/ml; RGC-AMY1, 5 g/ml) was mixed with sample buffer (6 l) and boiled for 5 min, marker dye (1 l) was added, and 3 l was applied to the gel; protein was detected by silver staining. Western blotting was performed with ProtoBlot Western blot AP system (Promega) using rabbit anti-AMY2 serum at 10 3 -fold dilution. The concentration of AMY1, parent and mutants, and BASI was determined by amino acid analysis (15,49). The secretion level of recombinant AMY1 in fermentor cultures was estimated from the amount of purified enzyme.

HPLC Analysis of the Hydrolysis Products of 4-Nitrophenyl ␣-D-
Maltooligosaccharides-Enzyme and G 5 -PNP, G 6 -PNP, or G 7 -PNP (120 l) in 50 mM sodium acetate, pH 5.5, 12.5 mM CaCl 2 were incubated at 37°C. Aliquots (40 l) were removed at appropriate time intervals, and the reaction was stopped by addition of glacial acetic acid (20 l). Substrate and products were separated by HPLC on a TSK-GEL oligo-PW column (Toso, 7.8 ϫ 300 mm) and eluted with distilled water at room temperature at a flow rate of 0.7 ml/min (2150 HPLC pump, LKB) (7, 51). 4-Nitrophenyl compounds were monitored at 313 nm (SPD-6A, Shimadzu ultraviolet detector). A mixture containing 1% of each of G 1 -PNP, G 3 -PNP, G 5 -PNP, G 6 -PNP, and G 7 -PNP in 50 mM acetate buffer, pH 5.5, was used for calibration. The relative activities were calculated using where S t and P t represent residual substrate and total amount of products at reaction time t. Enzyme Activity Assays-Hydrolysis of DP 17 amylose by mutant and wild type AMY1 (1-5 nM) at pH 5.5 in 20 mM sodium acetate, 5 mM CaCl 2 , and 0.5 mg/ml bovine serum albumin at 37°C was followed by reducing sugar analysis using copper bicinchoninate (52,53). Boiled samples were aliquoted into microtiter plates in triplicates of 300 l, and the absorbance at 540 nm was measured using a Ceres UV-900-HDi scanning autoreader (Bio-Tek Instruments, Inc). Maltose was used as a reference. Kinetic parameters were calculated by fitting the initial velocities at seven amylose concentrations in the range 0.06 -10.0 mg/ml to the Michaelis-Menten equation using Erithacus-software GraFit, version 3.0. Standard deviations were calculated from the data obtained by triplicate experiments performed independently. The activity toward G 7 -Cl-PNP was determined using the Granutest 3 ␣-amylase assay kit; ␣-amylase (15 nM) was added at nine concentrations of G 7 -Cl-PNP (0.26 -16 mM) in 20 mM MES, pH 6.8, 5 mM CaCl 2 (total volume, 100 l) at 30°C in microtiter plates. The progress of hydrolysis was followed at 405 nm, and kinetic parameters were calculated as above. ␣-Amylase (30 nM) was preincubated with the inhibitor acarbose (0.5-50 M) for 15 min at pH 5.5 (20 mM sodium acetate, 5 mM CaCl 2 , 0.5 mg/ml bovine serum albumin) at 25°C, the substrate DP 17 amylose (6.25 mg/ml) was added at the given [acarbose], and the activity was measured as above. K i app was calculated as ( where A max and A x are A 540 values at 0 and L x M acarbose, respectively. Influence of CaCl 2 on Activity-The activity of mutants of AMY1 and wild type AMY1 and AMY2 (2 nM) toward insoluble Blue Starch (6.25 mg/ml) was measured at varying [CaCl 2 ] (0.0025-100 mM) at pH 5.5 (20 mM sodium acetate, 0.5 mg/ml bovine serum albumin) at 37°C. Standard deviations were calculated from the data obtained in triplicate experiments performed independently.
BASI Inhibition-AMY1 mutants and wild type AMY2 (30 nM) were preincubated with BASI (0.02-992 M) at pH 8.0 (40 mM Tris, 5 mM CaCl 2 , 0.5 mg/ml bovine serum albumin) for 15 min at 37°C. Residual activity was determined by adding an aliquot (25 l) of this mixture to an insoluble Blue Starch suspension (4 ml, 6.25 mg/ml) in the same buffer. After 15 min the reaction was stopped by the addition of 0.5 M NaOH (1 ml). The absorbance at 620 nm was measured after centrifugation, and the percentage of inhibition by BASI was calculated as described previously (15,41).

RESULTS
Mutant DNA Sequences at Positions 183-185-In the present study we focused on active mutants of barley AMY1 secreted from the S. cerevisiae host strain. Parent enzyme C95A-AMY1 was used to avoid inactivating glutathionylation of the free thiol group of Cys 95 in AMY1 during yeast growth and processing (43). C95A-AMY1 is fully active toward Blue Starch (43) and will here be referred to as wild type or parent AMY1. A pool a R183K-AMY1 was obtained by site-directed mutagenesis. Other mutants resulted from random mutagenesis. The full length of the AMY1 insert was sequenced for transformant clones listed. b 800-fold concentrated culture filtrates were used. Italic letters and bold letters are polar and hydrophobic residues, respectively, as classified according to Ref. 45. of 10 4 different random mutants was created on pUC18 and designated as pUCBAL C95A X(183-185) ⌬S I, II. 120 yeast transformants harboring pBAL7C95A X(183-185) ⌬S I, II were selected and subjected to DNA sequencing, and ␣-amylase activity was measured. Of these transformants, 83% had a mutant gene, but only 4% secreted active enzyme. DNA sequencing showed that Gly 184 as well as a polar residue at position 183 and a hydrophobic residue at position 185 were retained in active mutants (Table I) (Table I). Transformants producing Asn 183 -Gly 184 -Tyr 185 (NGY), KGY, and wild type (RGY) AMY1 gave large starch plate halos, whereas those secreting SGM-and TGL-AMY1 produced smaller halos, and RGC-AMY1 produced a very small halo. The halo size agreed well with the estimated amount of enzyme in fermentor cultures: wild type, 82 g/l; KGY, 120 g/l; NGY, 80 g/l; SGM, 20 g/l; TGL, 10 g/l; and RGC mutants, Ͻ 10 g/l.
Mutational Analysis of the Requirement for Glycine at Posi-tion 184 -After random mutation of only the codon for Gly 184 , 150 yeast transformants were isolated. Following repeated colony isolation, enzyme secreting transformants (11) were identified by a positive starch plate assay, and the target DNA region was sequenced and found to encode Gly 184 . Codons for Trp, Tyr, Phe, Val, Arg, Ser, or Cys 184 were identified in genes from transformants not secreting immunoreactive enzyme. Electrophoretic Patterns and Western Blot of Mutant and Wild Type AMY1-In the IEF zymogram only NGY-AMY1 resembled wild type with the expected modest decrease in pI (Fig.  1A, lanes 4 and 5). RGC-AMY1 exhibited much weaker bands (Fig. 1A, lane 1). TGL-and SGM-AMY1 lacked activity in the zymogram and gave faint diffuse bands by silver staining (Fig.  1, A and B, lanes 2 and 3). Clear activity and protein staining, however, were obtained for these mutants in native PAGE (Fig.  1, C and D). Essentially the same molecular size and antigenecity were found for the mutants and wild type AMY1 in SDS-PAGE and Western blotting (Fig. 1, E and F). KGY-AMY1 also showed activity in an IEF zymogram and had the same mobility in SDS-PAGE and native PAGE, and immunoreactivity as the wild type (data not shown).
The mutants and wild type AMY1 migrated as two bands in IEF and native PAGE. Electrospray ionization mass spectrometry of SGM-AMY1, KGY-AMY1, and wild type AMY1 (data not shown) showed a full-length and a truncated form lacking the C-terminal Arg/Ser 414 dipeptide. The truncation occurs by the known processing of recombinant AMY1 in S. cerevisiae by Kex1p, a carboxypeptidase specific for basic side chains (43,49). Mass spectrometry also indicated Thr 410 in a significant fraction of recombinant AMY1 to carry two hexose residues, as identified earlier for recombinant wild type AMY1. Further glycosylation was not observed (data not shown). These modifications did not affect activity (43,54,55).
Action Pattern on 4-Nitrophenyl Maltooligosaccharides-To investigate the role of Arg 183 in substrate interaction in detail, the product distribution in hydrolysis of G 5 -, G 6 -, and G 7 -PNP was determined. Drastic changes were seen in the action pattern on G 6 -PNP, in particular for NGY-, SGM-, and TGL-AMY1, whereas small changes occurred with G 5 -PNP (Table  II). The dominant productive binding mode for hydrolysis of G 6 -PNP with wild type, KGY-AMY1, and RGC-AMY1 resulted in release of G-PNP in 32-50% yield, whereas NGY-, SGM-, and TGL-AMY1 released PNP in 48 -84% yield. Although wild type (RGY) released some PNP (24%), KGY-and RGC-AMY1 formed no or very little PNP (Table II). A similar shift of the preferred bond to be cleaved in G 7 -PNP was seen for NGY-AMY1 and, to a lesser degree, SGM-AMY1, as a significantly increased hydrolysis of the first bond to release PNP. Wild type, KGY-AMY1, TGL-AMY1, and RGC-AMY1, in contrast, hydrolyzed the second bond in G 7 -PNP in 63-81% yield (Table II). For the shorter G 5 -PNP, NGY-, SGM-, and TGL-AMY1 having small side chains at position 183 showed 3-26% of the wild type activity, whereas KGY-AMY1 retained 72% (Table III). In no case, however, did the action pattern shift on G 5 -PNP, although SGM-AMY1 generated an unusually large amount of glucose (Table II). The results reflect the subsite map of AMY1 (19) with its low binding affinity at subsites ϩ3 and Ϫ3 and the high affinity at subsite Ϫ6 that plays an important role in binding near the nonreducing end of longer maltooligosaccharides.
Kinetic Parameters for Hydrolysis of DP 17 Amylose, G 7 -Cl-PNP, and the Inhibition by Acarbose-The activity after mutation was either increased or essentially retained for amylose, which is a substrate that spans the entire active site binding area. k cat thus increased nearly 2-fold for SGM-AMY1; K m slightly increased, except for TGL-AMY1 (Table III). k cat /K m after mutation varied as 64 -171% of the wild type value. For the oligosaccharide G 7 -Cl-PNP, k cat was highest for KGY-AMY1 (160%) and lowest for TGL-AMY1 (35%). On this substrate NGY-and SGM-AMY1 gave intermediary k cat values, and K m increased or decreased by a factor of two dependent on the mutant (Table III). G 7 -Cl-PNP contains a ␤-glucosidic bond at the reducing end, which may interfere adversely with Lys 183 , the equivalent of Lys 182 in AMY2 that interacts with the sugar OH-3 at subsite ϩ2 in the crystal structure of the acarbose complex (71). In comparison, the shorter Asn, Ser, or Thr 183 gave decreased K m of around 4 M, resulting in k cat /K m above the wild type value for NGY-and SGM-AMY1 (Table III). K i app of acarbose inhibition of the hydrolysis of amylose was 2-4 M for the NGY-, SGM-, and TGL-AMY1. Because K i, app of the parent and KGY-AMY1 was around 16 M (Table III), the pseudotetrasaccharide acarbose more effectively inhibited the Asn, Ser, or Thr 183 containing mutants.
Influence of CaCl 2 on Activity-The isozymes AMY1 and AMY2 show distinct variations in activity for insoluble Blue Starch as a function of the Ca 2ϩ concentration (Fig. 2). Thus, whereas the activity of AMY1 changed little from 0.0025 to 10 mM CaCl 2 , that of AMY2 increased to reach a maximum around 10 mM CaCl 2 (15). Both isozymes are inhibited at [CaCl 2 ] Ͼ 20 mM. Whereas SGM-and TGL-AMY1 were clearly less active than the parent enzyme at [Ca 2ϩ ] Ͻ 0.01 mM (Fig. 2), resulting in a profile that resembled that of AMY2, the behavior of KGY-AMY1 (data not shown) very much resembled that of the AMY1 parent.
Inhibition by BASI-KGY-and NGY-AMY1 were tested for inhibition by up to 3 ϫ 10 4 -fold molar excess of BASI. The natural target AMY2, under the conditions used, is inhibited even by stoichiometric amounts of BASI. However, although Lys 182 of AMY2 forms a salt bridge with a glutamate residue in BASI, 2 replacement of the equivalent Arg 183 in AMY1 by Lys, or Asn 183 (in KGY-and NGY-AMY1), did not confer these AMY1 variants detectable sensitivity for BASI.

DISCUSSION
The Sequence Motif at Positions 183-185-Very few and short sequences are well conserved among ␣-amylases and related amylolytic enzymes (28,29); only seven residues are invariant (25,27). Based on crystal structures (1-7) and mutational analyses, these residues are implicated either in catalysis and transition state stabilization or in the structural integrity of the active site (25,27,53). The molecular model of a barley ␣-amylase-inhibitor complex (71) confirmed that conserved regions are important in substrate specificity and action pattern. An alignment focused on the fourth ␤-␣ loop of the catalytic (␤/␣) 8 -barrel revealed remarkable sequence variation corresponding to Arg 183 -Gly 184 -Tyr 185 in AMY1 (boxed in Fig.  3), the present target of random mutagenesis. Family members with different specificity thus contain (i) (Arg/Lys)-Gly-Ar as in plant ␣-amylases, (ii) Lys-His-Z predominant in animal and microbial ␣-amylases, or (iii) a completely different sequence where Z and Ar signify hydrophobic and aromatic residues,   The relative activities were calculated as described under "Experimental Procedures" using the data shown in Table II. The values in parenthesis present the relative activities (%) in the hydrolysis of the substrate at the penultimate glycosidic bond from the reducing end.
b The values in parenthesis represent activities expressed as percentage of the values of wild type AMY1. c ND, not determined.
respectively (9,34,35,46,56). The carbonyl oxygen of the aligned residues, AMY2 Gly 183 (4,9), TAA His 210 (63), and PPA His 201 (3), is a ligand to the Ca 2ϩ conserved in ␣-amylases, i.e. Ca 2ϩ (Ca 500 in Ref. 4) in AMY2 (4). Lys 182 of AMY2 binds to sugar OH-3 at subsite ϩ2, and Lys 200 of PPA shows the same interaction (6). Furthermore, NE2 of His 201 from the Lys-His-Z motif in PPA forms a hydrogen bond with sugar OH-2 in subsite ϩ1, i.e. on the reducing end side of the cleavage (6). Sitedirected mutagenesis of His 201 in the 95% identical human pancreatic ␣-amylase revealed its multifunctional role in substrate affinity, pH activity dependence, Cl Ϫ activation, and binding of a proteinaceous inhibitor (61,62). Site-directed mutagenesis earlier indicated the roles in catalysis and binding of invariant carboxylic acid and histidine residues in barley AMY1 (53). The present goal is to alter substrate specificity and action pattern by random mutagenesis of Arg 183 -Gly 184 -Tyr 185 involved in substrate binding. All active mutant AMY1 found had conserved the motif: Gly 184 , a polar residue at position 183, and a hydrophobic residue at position 185. Gly 184 occurs in all known plant ␣-amylase sequences (27,64). We searched for possible variants by subjecting the corresponding codon to random mutagenesis, but no active mutants were obtained. Attempts to construct G184A failed repeatedly (not described). In accordance with the crystal structure of AMY2 (4), glycine is suggested therefore to be required at that position in plant ␣-amylases.
The amount of mutant protein produced was influenced by the side chains at position 183 or 185. Although KGY-and NGY-AMY1 were produced at the same level as wild type AMY1 (RGY), yields of SGM-, TGL-, and RGC-AMY1 were around 10% of that amount. The segment centered on Gly 184 may thus be critical for stability and/or proper folding. In the crystal structure of AMY2, Lys 182 (equivalent to Arg 183 of AMY1) is located at a turn in the fourth ␤-␣ loop, in which the side chain of Phe 180 (Phe 181 in AMY1) and the C␣ proton of Gly 183 (Gly 184 in AMY1) come close (4) (Fig. 4). It is remarkable that the ␣-amylases from higher plants, amylomaltases, and maltotetraohydrolases (27,64) have a large residue (Phe or His) together with glycine matching Phe 181 and Gly 184 in AMY1 (Fig. 4), whereas ␣-amylases from other sources, maltohexaohydrolase, and cyclodextrin glucanotransferase contain a small residue (Ala, Ser, or Thr) and histidine corresponding to Phe 181 and Gly 184 , respectively. This complementarity in the different enzymes (Fig. 4) supports that AMY1 Gly 184 is structurally important.
Role of Ca 2ϩ -AMY2 has three Ca 2ϩ , one of which is conserved in the different ␣-amylase structures (2)(3)(4)63). The carbonyl oxygen of Gly 183 is the only Ca 2ϩ ligand from domain A (4). Replacement of both residues flanking AMY1 Gly 184 may affect the position of this carbonyl oxygen, because SGM-and TGL-AMY1 both lost activity at [Ca 2ϩ ] Ͻ 10 M and showed no zymogram activity after IEF. In fact, the behavior of AMY1-AMY2 isozyme hybrids suggested that some part of the AMY1 structure after residue 161 was responsible for the high Ca 2ϩ affinity (65); this may be the Gly 184 region. The other Ca 2ϩ ligands belong to the long third ␤-␣ segment (amino acids 89 -152), called domain B, that is grafted onto the (␤/␣) 8 -fold (4). Although the AMY2 Ca 2ϩ ligands are all conserved in AMY1 (4,9), AMY1 compared with AMY2 has higher affinity for Ca 2ϩ (14). It remains to be seen if AMY1 contains three Ca 2ϩ , and it remians to be understood why the activity of the isozymes depends differently on Ca 2ϩ .
Enzymatic Properties of AMY1 Mutants-Arg 183 in AMY1 most probably binds to substrate. In AMY2 acarbose (an inhibitory pseudotetrasaccharide) (66), NZ of Lys 182 thus hydrogen bonds with OH-3 of the third acarbose ring (71) and is considered a critical determinant of subsite 2. The shifted action pattern on the substrate G 6 -PNP is compatible with a weakening of this interaction in NGY-, SGM-, and TGL-AMY1 releas-ing PNP, whereas wild type and mutants having (Arg/Lys) 183 release G 1 -PNP as major product.
KGY-AMY1 shows a 3-fold increase in activity and maintains wild type action pattern for hydrolysis of G 7 -PNP but has only 72% activity on G 5 -PNP. NGY-AMY1 shows 91 and 8% activity, respectively, on these substrates. The three-dimensional structures around Lys 182 of AMY2 and Lys 209 of TAA are similar (4). In case of Sfamy, which is highly homologous to TAA (7,23,37,38,67), K210R (38) had 20 and 200% (k cat /K m ) of wild type activity toward maltopentaose and maltotriose due to decreased affinity at subsite ϩ2 and increased affinity of subsite ϩ1, respectively. The k cat /K m values of K210N Sfamy for G 5 -and G 3 -PNP were only 15 and 1% of K210R Sfamy due to reduced affinity at both subsites (38). This behavior is reminiscent of the present AMY1 mutants. The mutant action patterns are consistent with Arg 183 being an important component of subsites ϩ1 and ϩ2, and the strong affinity at subsite ϩ1 required for efficient hydrolysis of short substrates, such as G 5 -PNP.
The AMY1 mutants have efficient k cat /K m for amylose hydrolysis varying at 64 -171% of the wild type value. Noticeably the relative activity of especially SGM-AMY1 increased with substrate length. According to the subsite theory (68,69) k cat /K m ϭ k int ⅐ K p , where k int and K p are the intrinsic rate constant for bond hydrolysis and the binding constant of substrate in productive mode, respectively; k int ϭ k cat /K m in hydrolysis of long substrates that cover the entire binding site. The strong dependence of SGM-AMY1 on substrate length might be explained by the mutation affecting both parameters resulting in increased k int and decreased affinity at one or more subsites near the catalytic site. For hydrolysis of short substrates, K p decreases are probably more important than k int increases. This agrees with affinities at subsites ϩ1 and ϩ2 being reduced by the replacement of Arg 183 by Ser, hence k cat /K m is expected to decrease for the mutant compared with the wild type enzyme. Except for an elevated k cat /K m toward G 7 -Cl-PNP containing a ␤-anomeric bond between the nitrophenyl group and glucose at the reducing end, the relative activities against ␣-anomeric oligosaccharide derivatives, G 7 -PNP, G 6 -PNP, and G 5 -PNP, showed drastic decreases from 13 to 3%. For amylose, however, SGM-AMY1 has a high k cat /K m value, which simply reflects an increase in k int .
The parent (C95A) AMY1 has K m of 9.8 mM for G 7 -Cl-PNP, whereas normal wild type AMY1, both from malt and a recombinant (11,15) produced in Pichia pastoris, gave K m around 1 mM (55). Similarly an approximate 5-fold increase in K m to 2.2 mg/ml was obtained for parent AMY1 acting on the large substrate amylose, as compared with K m ϭ 0.4 mg/ml for AMY1 from malt and Pichia. This poorer affinity (Table III) may stem from the C95A mutation, which is located near His 93 known to be critical for transition state stabilization (53).
Finally, BASI specifically inhibits AMY2 (17,18) in a 1:1 complex with K d ϭ 2 ϫ 10 Ϫ10 M (41). The AMY2-BASI complex was crystallized (36), and in the molecular model Lys 182 of AMY2 forms a salt bridge to a glutamate in BASI. 2 Whereas Lys 182 probably contributes important stabilization of AMY2-BASI, the lack of inhibition of the mutant R183K even by a large excess of BASI showed that Lys 183 alone does not confer AMY1 sensitivity for BASI. This finding agrees with the fact that AMY1-AMY2 isozyme hybrids inhibited by BASI possess AMY2 sequence from residue 116 (15,65).
Conclusion-In a schematic of the Arg 183 -Gly 184 -Tyr 185 region in the fourth ␤-␣ loop, based on the crystal structure of AMY2 (4) complexed with acarbose (71), Ca 2ϩ stabilizes the contact between domains A and B (Fig. 4). Gly 184 (domain A) binds to this Ca 2ϩ together with Asn 92 , Asp 139 , Ala 142 , and Asp 149 (domain B). Gly 184 indeed may play an important conformational role involving the steric complementarity with the side chain of Phe 181 , which most probably is obstructed by introduction of a side chain at Gly 184 . Because the activity for amylose was improved up to 171% for engineered AMY1, the enhanced k cat of the ␤-␣ loop 4 mutants might reflect increased flexibility in this segment, which has the catalytic nucleophile Asp 180 located at the C terminus of ␤-strand 4 (4). Finally, k cat /K m of SGM-AMY1 is nearly 2-fold that of wild type on amylose, which occupies all ten subsites in the substrate binding site, but activity on oligosaccharides is below 9% of wild type. In contrast, KGY-AMY1 is up to three times more active than wild type on G 6 -PNP and G 7 -PNP. The results suggest that the long side chain of Arg 183 contributes to binding at subsites ϩ1 and ϩ2 (Fig. 4) but also confers constraints on substrate processing.
The present results give very promising prospects for activity improvement and modulation of action pattern in plant ␣-amylases by protein engineering in the distinct sequence (Lys/Arg)-Gly-Tyr at the active site ␤-␣ 4 segment. This is the first report on an engineered ␣-amylase with enhanced activity toward the natural substrate. Ultimately ␣-amylase variants might be useful in transgenic barley for enhancement of malt quality by improving the enzyme activity during germination and mashing (70).