THE JOURNAL 0 1993 by The American Society for Biochemistry

OF BIOLOGICAL CHEMISTRY and Molecular Biology, Inc

Vol .268, No. 2, Issue of January 15, pp. 10Y5-1100,1993 Printed in MS.A.

Site-directed ~utagenesis Identifies Aspartate 33 as a Previously Unidentified Critical Residue in the Catalytic Mechanism of Rabbit Aldolase A*

(Received for publication, July 7, 1992)

Aaron J. Morris and Dean R. TolanS From the Biology Department, Boston University, Boston, Massachusetts 02215

The expression and purification of the rabbit muscle aldolase A (D-fructose 1,6-bisphosphate:~-glyceralde- hyde-3-phosphate lyase, EC 4.1.2.13) from an expres- sion plasmid in bacteria is described. The enzyme is produced in bacteria at a level of 300 mg/liter and is indistinguishable from the enzyme isolated from mus- cle in assays using fructose 1,fi-bisphosphate and fruc- tose l-phosphate. The recombinant enzyme has the same primary, secondary, and quaternary structure as the muscle enzyme. Aspartic acid 33, found near the active site lysine in the crystal structure, is changed to alanine, serine, and glutamic acid by site-directed mu- tagenesis, resulting in the mutant proteins, D33A, D33S, and D33E, respectively. The mutant enzymes are purified by substrate affinity elution from carbox- ylmethyl-Sepharose, the same method as that used for the wild-type enzyme. The secondary and quaternary structure of D33A is identical to wild-type aldolase when analyzed by light scattering, gel filtration, and circular dichroism. Moreover, the hexose substrate can be fixed in the active site by reduction of the Schiff base with sodium borohydride, indicating that the ac- tive site is not drastically altered, These single muta- tions in the active site have a serious effect on the activity of the enzyme. In addition, the rate of carb- anion oxidation for D33A is 17-29 times slower when the substrate is fructose l,6-bisphosphate versus dih- ydroxyacetone phosphate, whereas in the wild-type there is no significant difference in these rates. This evidence and the conservation of this residue in other class I aldolases indicate that aspartic acid 33 is an essential residue in the catalytic mechanism, possibly involved in abstraction of the carbon 4 hydroxyl pro- ton.

The glycolytic enzyme, fructose-l,6-bisphosphate aldolase (EC 4.1.2.13), has been extensively studied with respect to its isozyme forms, tissue distribution, developmental patterns of expression, structure, and mechanism of catalysis (1, 2 ) . The enzyme cleaves Fru-1,6-P; or Fru-1-P into two trioses in an

* This work was supported by Grants DK38821 and DK43521 (to D. R. T.) from the National Institutes of Health and Boston Univer- sity Biomedical SEED Grant 860-BI. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked ‘ ‘ ~ ~ e r ~ ~ ~ ~ n ~ ’ ’ in accord- ance with 18 U.S.C. Section 1734 soleIy to indicate this fact.

4 TO whom correspondence should be addressed. The abbreviations used are: Fru-1,6-P2, fructose-1,6-bisphos-

phate; Fru-l-P, fructose-l-phosphate; DHAP, dihydroxyacetone phosphate; PAGE, polyacrylamide gel electrophoresis; MOPS, 3-[N- m~~ho~ino]propanesulfanic acid; TEA, t~ethanolamine; TAPS, N - tris[hy~oxymethyl]methyl-3-aminopropane sulfonic acid; MES, 2- [N-morpholino]ethanesulfonic acid.

ordered uni-bi mechanism. Although vertebrates have three distinct isozymes, aldolases A, B, and C, each with distin- guishable catalytic properties (3), all are thought to act through an identical catalytic mechanism involving a Schiff base between the carbonyl of the ketose and the €-amino group of a lysine at the active site (4). All vertebrate aldolases are found as tetramers of 40-kDa subunits. Hybrid tetramers can be formed, although the substrate specificity of each isozyme in the hybrid is maintained (1). This specificity suits the particular metabolic role of the tissues in which the different isozymes are expressed (5). The loss of activity of the human liver-specific isozyme, aldolase B, leads to heredi- tary fructose intolerance, a potentially fatal metabolic disor- der (6). Furthermore, a thermally unstable mutant of human aldolase A causes nonspherocytic hemolytic anemia ( 7 ) . The precise mutations for each of these disorders have been eiu- cidated (8-11), a~though an understanding of how the muta- tions lead to a loss of activity wilI require more knowledge of the structure and mechanism of the enzyme.

The primary sequence for a number of aldolases from different species has been determined, all of which have a highly conserved sequence around the active site lysine 229 (12-14). The most extensively characterized isozyme is aldo- lase A from rabbit muscle. Studies of this enzyme using protein modifying reagents have identified several residues potentially involved in the catalytic mechanism (15, 16). Indeed, a mechanism based on these studies has been pro- posed (17) and is found in many textbooks (18). Besides the active site lysine, either of 2 cysteine residues (Cys-7Z/Cys- 338) and a histidine at position 361 have been implicated. However, this proposal is not consistent with current evi- dence; they residues are not in the active site, as indicated by the 2.7 A structure of the rabbit aldolase A (19), they are not conserved in evolution (14), and site-directed mutagenesis of these residues had no effect on activity (20 ) .

In order to elucidate the roles of various residues in and around the active site, and more correctly describe the mech- anism of catalysis of this crucial metabolic enzyme, we have chosen an approach using site-directed mutagenesis of rabbit aldolase A. This paper describes the const~ction of an expres- sion vector for the production of rabbit aldolase A in bacteria. The bacterially derived enzyme is identical to the enzyme isolated from muscle as determined by both functional and structural measurements. Site-directed mutagenesis of the evo~utionari~y conserved Asp-33 drastically reduces the activ- ity of the enzyme, while apparently leaving the structure unchanged. The properties of this mutant indicate that the Asp-33 is critical for catalysis, possibly involved in proton abstraction at the carbon 4 hydroxyl.

EXPERIMENTAL PRO~EDURES

111, mung bean nuclease, and DNA polymerase I were from New Materiuls-Restriction endonucleases, T4 DNA ligase, exonuclease

1095

1096 Role of Asp-33 at the Active Site of Aldolase

England Biolabs. DNA polymerase I (Klenow fragment), calf intes- tine alkaline phosphatase, and glycerol-3-phosphate dehydrogenase/ triose phosphate isomerase were from Boehringer Mannheim. Deox- ynucleoside triphosphates, Cm-Sepharose" CL-GB Fast Flow, Seph- adexm G-150, and pKK233-2 were from Pharmacia LKB Biotechnol- ogy Inc. [a-32P]Deoxynucleoside triphosphates were from Amersham Corp. Radiolabeled [U-"C]Fru-1,6-P2 was from ICN. Nitrocellulose filters were from Sartorius. GF/C glass microfiber filters were from Whatman. Oligonucleotides for construction, sequencing, and site- directed mutagenesis were synthesized on Milligen/Biosearch DNA synthesizers using phosphoramidite chemistry and the manufactur- er's protocols. When necessary the larger oligonucleotides were pu- rified by urea-PAGE. SDS low M, standards, conjugate antibody, and immunostaining reagents were from Bio-Rad. pKK233-PK was a gift from P. Kelly (University of Nebraska). Rabbit anti-spinach aldolase antibody was a gift from H. Lebherz (San Diego State University). DHAP and trichloroacetic acid were from Fluka. Rabbit muscle aldolase, Fru-1,6-P2, Fru-1-P, and other chemicals were from Sigma.

Strains-Escherichia coli strains JM103 and TG1 (21) were used for M13 cloning and mutagenesis. JM83 (22) and DH5a (21) were used for expression.

Site-directed Mutagenesis of Expression Plasmids-Site-directed mutagenesis was performed (23) to change the Asp-33 (GAU) codon to Ala using the oligodeoxyribonucleotide, 5"CTGGCTGCA- GAGTCGACC; to Ser using the oligodeoxyribonucleotide, 5'- CTGGCTGCABAGTCGAC; or to Glu using the oligodeoxyri- bonucleotide, 5'-TGGCTGCAmGAGTCGACCG. The Lys-229

5'-CCTTGCTGaCCCAACAT. The potential mutants were (AAG) codon was changed to Ala using the oligodeoxyribonucleotide,

screened by DNA sequence determination using dideoxy termination (24). Restriction enzyme digestions, ligation reactions, and transfor- mations were performed as described in Sambrook et al. (21). Se- quence determination employed 7-deaza dGuo-triphosphate in place of dGTP which alleviated much of the G/C compression (25) present in the rabbit aldolase A sequence (26).

Growth and Purification of Recombinant Aldolase-A single colony of E. coli DH5a bearing the appropriate expression plasmid was used to inoculate 1 liter of 2xYT (21) medium plus 50 pg/ml ampicillin. Cultures were grown with vigorous shaking for 24-36 h at 37 "C. Cells were harvested and resuspended in 20 ml of lysis buffer (250 mM MOPS .KOH, pH 7.0, 10 mM EDTA, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 1% glycerol). DNase and RNase (2 mg each) were added and cells were lysed in a French press at 10,000 psi. The extract was centrifuged for 30 min at 10,000 X g, and twice for 60 min at 100,000 X g. A 35-65% ammonium sulfate cut of the supernatant fraction was dissolved in 5 ml of MG buffer (50 mM MOPS.glycine.KOH, pH 7.0, 1 mM dithiothreitol) and dialyzed against the same buffer (600-fold excess). The clarified dialysate was loaded onto a 10-ml column of Cm-Sepharose CL-GB Fast Flow equilibrated with MG buffer. The column was washed with MG buffer until the absorbance at 280 nm was 50.015 and repeated with 50 mM TGK buffer (TAPS.glycine.KOH, pH 8.3, 1 mM dithiothreitol). Aldolase was eluted with TGK buffer containing 2.5 mM Fru-l,6-P2 (1). Most enzymes were further purified by passage through a 90-cm X 3.5-cm column of Sephadex G-150. Aldolase containing fractions were pooled and precipitated in 65% saturated ammonium sulfate. Protein samples were stored at 4 "C as ammonium sulfate suspen- sions.

Electrophoresis and Zmmunoblotting-SDS-PAGE (12.5%) (27) was performed and gels were stained with Coomassie R-250 (0.1% in water/methanol/acetic acid (5:5:1)). Protein concentration was deter- mined by dye binding (28) using bovine serum albumin as a standard, or for pure aldolase by absorbance using EZW (1%) = 0.91 (29). Immunoblotting (30) was performed on unstained gels by electropho- retic transfer to nitrocellulose at 250 mA for 20 h in a 20 mM Tris. glycine, pH 8.5, buffer. Membranes were probed with rabbit anti- spinach aldolase antibody followed by goat anti-rabbit immunoglob- ulin-G antibody conjugated to horseradish peroxidase and stained with 0.5% ch1oronaphtho1/0.015% H202.

Kinetic Analysis-The substrate cleavage rate was determined by measuring the decrease in absorbance/minute at 340 nm in a coupled assay (31); aldolase was diluted in 50 mM TEA.HCI, pH 7.4, and added to a cuvette containing 50 mM TEA.HC1, pH 7.4, 10 mM EDTA, 0.16 mM NADH, 10 pg/ml glycerol-3-phosphate dehydrogen- ase/triose phosphate isomerase. Assays of 1 ml were performed in triplicate a t 30 "C following addition of substrate. The cleavage rate for Fru-l,6-P2 was measured over a substrate concentration range of 160 to 1.25 p ~ , with 0.3 pg of wild-type enzyme, or 100-350 pg of

D33A, D33S, or D33E. The cleavage rate for Fru-1-P was measured over a substrate concentration range of 3.0 to 0.5 mM, with 85 pg of wild-type enzyme. Kinetic values were determined from double-recip- rocal plots using a least squares method that explicitly included the errors in rate measurement as weights to avoid misweighting the data (32). Protein concentration was determined by absorbance using E,, (1%) = 0.91.

Structural Annlysis of Recombinant Proteins-Purified protein and a set of M, standards (ferritin, 440 kDa; catalase, 220 kDa; lactate dehydrogenase, 146 kDa; bovine serum albumin, 66 kDa; myoglobin, 17.3 kDa) were separated on a 70-cm X 1.6-cm column of Sephadex G-150 in 50 mM TEA.HCI, pH 7.5, with a flow rate of 0.15 ml/min at 4 "C. Activity assays detected fractions containing wild-type aldo- lase, while densitometric tracing of SDS-PAGE on a LKB Ultro-scan XL densitometer detected mutant aldolase fractions. The CD spectra were determined using a protein concentration of 3.8 mg/ml (29) in 1 mM Tris.HCI, pH 7.5, at 20 'C on an AVIV 60DS spectrometer using a 0.1-mm path length cuvette. CD spectra were taken from 180 to 260 nm with the reading averaged for 5 s at each nm. Light scattering analysis of 1 mg/ml protein samples was performed with a Oros Instruments Limited, model 801 Molecular Size Detector.

Analysis of Schiff Base Formation-Radioactive [U-"C]Fru-1,6-P2 was combined with purified protein and incubated at pH 7.3, at 25 "C for 1 h to reach equilibrium before treatment with sodium borohydride (33). The 50-p1 reaction contained 100 pg of enzyme and 208 p~ Fru- l,6-P2 (0.0096 Ci/mmol) in 20 mM Tris.HCI, pH 7.3. After 1 h, 50 pl of freshly made 80 mM sodium borohydride in 200 mM K.MES, pH 6.0, was added and incubated 10 min. Samples were precipitated with 400 pl of 20% trichloroacetic acid and collected on Whatman GF/C glass microfiber filters. Filters were washed with 5% trichloroacetic acid, then 95% ethanol and were counted in a Beckman LS 2800 scintillation counter ("C-counting efficiency was 82%). Partial pro- teolysis and analysis of peptides was performed with 100 ng of V8 protease and 7 pg of labeled aldolase as described by Cleveland et al. (34).

Carbanion Assay-The carbanion present at carbon 3 of the sub- strate was oxidized by hexacyanoferrate(II1) (35). The rate of hexa- cyanoferrate(II1) reduction, which is proportional to the level of carbanion, was monitored by the decrease in absorbance at 420 nm. The 0.7-ml assay contained 200 mM Tris. HCI, pH 7.4, 0.7 mM hexacyanoferrate(III), 10 mM Fru-1,6-P2 or DHAP, 0.1 mg/ml bovine serum albumin, and 0.02-1.0 mg of purified aldolase. After a 5-min preincubation, enzyme was added to start the reaction. The rate of the decrease in absorbance was measured within the first 60 s a t 25 "C.

RESULTS

Rabbit Muscle Aldolase Produced in Bacteria-An expres- sion vector was constructed from a cDNA clone and synthetic DNA based upon the mRNA sequence (26) (Fig. 1). The synthetic DNA, which was designed with silent-site modifi- cations to create several unique restriction enzyme sites, was inserted into M13 vectors in three separate fragments, sub- sequently excised, and ligated together into one vector. This clone, RA, was used as the host vector for a fragment from the cDNA clone, pRM223 (26), which was ligated at an in- frame SstI site to create the vector, RbA. The modified synthetic sequence contained a unique NdeI site at the start codon. This site was fused to the NcoI site of the bacterial expression vector, pKK233-2 (36) to recreate the AUG codon with only a T for C substitution at the -1 position to the start of translation (Fig. 1, inset). The resulting plasmid, pDT14, expressed rabbit aldolase A and was used for the construction of AMI, for site-directed mutagenesis. The DNA sequence of RbA and AM1 was confirmed by using several sequence-specific primers throughout the aldolase coding re- gion. For better expression, the rabbit aldolase coding se- quence was subcloned into a high copy plasmid, pPB1, to create pPB14 (37).

The rabbit aldolase A was expressed in E. coli DH5a trans- formed with the expression plasmid pPB14. Cell extracts were assayed for activity in the presence of EDTA which inacti- vates the class I1 metal-dependent aldolase from the bacteria

Role of Asp-33 at the Active Site of Aldolase 1097

Synthetic DNA

x E RA-1 R A - 2 r H@

K H RA-3

W 1. Hind 111 2. BlunVKlenow

3. Partial Sst I 4. Isolate Insert

R bA

pKK-PK 1. EcoIHindlll 2. Ligation

AGGAACAGAC TZGCCT NCO Nde

AM1

pDT14

FIG. 1. Construction of rabbit aldolase A expression vector. The three synthetic DNA fragments were combined in RA after each was separately cloned into M13. Insertion of a partial cDNA from pRM223 (31) into RA produced RbA. The entire RbA aldolase coding sequence from the NdeI site to the PstI site at the 3' end of the cDNA clone was inserted into pKK233-PK, a vector derived from pKK233-2 which made digestion with both NcoI and PstI more efficient. The EcoRIIHindIII fragment of pDT14 containing the trc promoter and aldolase coding sequence was placed into AM2, a modified M13 mp19, to generate AM1 for sequence determination and mutagenesis. To produce AM2, an M13 vector without a lac promoter, M13 mp19 was digested with AuaII, blunted with Klenow, joined with HindIII linkers, digested with HindIII, and religated. The letters denote the enzymes used for construction; P, PstI; X, XbaI; K , KpnI; H , HindIII; E, EcoRI; A, AseI; Nde, NdeI; Nco, NcoI; S , SstI; MB, mung bean nuclease; T4-po1, T4-DNA polymerase; CIP, calf intestine alkaline phosphatase. The larger letters denote the restric- tion enzyme sites in the vector which were cut and the smaller letters denote either ligation sites from the previous step or sites not cut in a partial digestion. Inset, the sequence of pDT14 at the site of ligation between the NcoI and NdeI (underlined) sites of pKK233-2 and RbA, respectively. The shaded box denotes those bases which were cleaved prior to ligation. The dashed line denotes the ribosome-binding site. The first codon is depicted below the Pro.

(4, 38). An additional protein of about 40 kDa was produced in the cells transformed with pPB14 when compared to crude extracts of DH5a (Fig. 2, panel A, lanes 2 and 3) . Immuno- blotting confirmed that this 40-kDa protein was the eukary- otic aldolase (data not shown).

The rabbit aldolase A was purified by two methods, one which was common for isolation from muscle tissue (39), and one which involved only three steps: ammonium sulfate frac- tionation, affinity elution from Cm-Sepharose CL-GB, and gel filtration (Fig. 2, panel A ) . This latter procedure resulted in an overall yield of 35-40% and produced aldolase of the same purity and activity as the former preparation (data not shown). In addition, this procedure resulted in preparations of greater purity than another, which employed a dye-binding affinity column (Fig. 2, panel A, lane 8) . These procedures yielded substantial amounts of purified enzyme ranging from 40-110 mg/liter culture. The rabbit aldolase A was expressed at a level of 10-20% of the E. coli DH5a-soluble protein as estimated from specific activities. There was no difference in

A 1 2 3 4 5 6 7 8

97.4 kda +

66.2 kda +

42.7 kda +

31.0 kda +

1 2 3 4 5 6 7 8 9

97.4 kda +

66.2 kda -t

"

42.7 kda -t

I 31.0 kda -t

FIG. 2. Expression and purification of rabbit aldolase A. Coomassie Blue staining of 12.5% SDS-PAGE. Panel A, purification of wild-type aldolase A expressed from pPB14 in DH5a. Lane I, M, standards, sizes depicted by the arrows; lane 2, crude lysate of DH5a with no plasmid; lane 3, French press lysate supernatant of DH5a containing pPB14; lane 4 , dialyzed 35-65% ammonium sulfate cut of sample depicted in lane 3; /ane 5, Cm-Sepharose column flow-through of sample depicted in lane 4; lane 6, eluate from Cm-Sepharose column; lane 7, aldolase peak from gel filtration of sample depicted in lane 6; lane 8, purified rabbit muscle aldolase (Sigma). Lanes 2-5 each contain 30 pg of protein, and lanes 6-8 each contain 10 pg of protein. Panel B, bacterially expressed mutant aldolases, before and after purification. Lane I , M, standards as in panel A; lanes 2 and 3, pAM6 (D33A); lanes 4 and 5, PAM9 (D33S); lanes 6 and 7, pAMlO (D33E); lanes 8 and 9, pAM13 (K229A). Lanes 2,4,6, and 8 contain 20 pg each of crude lysate of DH5a bearing a mutant expression plasmid. Lanes 3, 5, 7, and 9 each contain 10 pg of mutant aldolase, purified as in panel A , except lane 9 was not passed through Sephadex G-150.

TABLE I Kinetic properties of recombinant rabbit aldolase A

Substrates

Aldolase Fru-1,6-P2 FN-1-P

Vma. K , Vma. K, unitsfmg X M unitslmg X 1O-' M

Rabbit 20.8rt 1.7 16.4 rt 1.8 0.44 rt 0.05 7.4 f 0.9

Wild-type 20.8f 0.5 14.3 rt 0.7 0.51 rt 0.14 7.8 f 2.2

D33A 0.0036 & 80.0 f 0.074 -'

D33S 0.0056 & 36.5 If: 0.02 -

D33E 0.023 & 105.7 f 0.05 -

K229A ND - -

muscle

(pPR14)

(1.8 X lo-')

(1.3 X

-

-

(9.4 x 10-61 - -

-, not determined. * ND, not detected.

yields when comparing soluble crude extracts and whole cell extracts, which indicated that little if any of the aldolase was in an insoluble form (37).

The kinetic properties of the recombinant rabbit aldolase A were compared to that purified from muscle. The kinetic values for K,,, and V,,,,,, using Fru-1,6-P2 and Fru-1-P, were determined for the recombinant aldolase and the muscle aldolase. There were no significant differences between the two enzymes (Table I).

The structure of the recombinant rabbit muscle aldolase was compared to that purified from muscle. The apparent M , of the aldolase subunits from both sources were essentially

1098 Role of Asp-33 at the Active Site of Aldolase

180 2w m 240 2MI

Wavelength

FIG. 3. Circular dichroism spectra of wild-type and mutant aldolases. One spectrum is shown for each enzyme; rabbit muscle aldolase from Sigma (- - -), rabbit aldolase A from pPB14 (.- .-. -), D33A (--), K229A (. . . .).

TABLE I1 Carbanion oxidation rates for wild-type and

mutant enzyme-substrate complexes

Aldolase Substrates (k',.J

Fw-l.B-P, DHAP min"

Wild-type (pPB14) 7.89 & 1.35 D33A

7.17 2 1.07

D33E 0.06 & 0.01 1.37 & 0.26

K229A 0.24 2 0.11 0.43 & 0.01

0.021 & 0.005 0.020 & 0.007

the same (Fig. 2, panel A , lanes 7 and 8). Gel filtration showed that the recombinant aldolase associated as a tetramer with an apparent molecular mass of 158 kDa, which was the same as rabbit muscle aldolase (data not shown) (1). In addition, light scattering showed a molecular mass of 140 f 10 kDa for both recombinant and muscle enzymes. The amino-terminal sequence of the recombinant aldolase was determined by solid-phase amino acid sequencing and measurement of the phenylthiohydantoin derivatives (40). The sequence deter- mined was; NH2-Pro-His-Ser-His-Pro-Ala-Leu-Thr-Pro- Glu-Gln-Lys-Lys-Glu-Leu, which was identical to the muscle enzyme (26), reflecting the efficient removal of the amino- terminal methionine in bacteria. Moreover, the secondary structures were identical as measured by CD (Fig. 3).

Active Site Mutagenesis-The vector construction that pro- duced the native muscle enzyme in bacteria enabled subse- quent investigation of the active site of this enzyme using site-directed mutagenesis. Mutagenesis was performed on AM1, and the EcoRI-Hind111 fragments of the mutant vectors were cloned into pPB1 (37) to create pAM6 (D33A), PAM9 (D33S), pAMlO (D33E), and pAM13 (K229A). The DNA sequence of the aldolase coding region confirmed that only these substitutions were present. These plasmids were trans- formed into D H h , and the aldolase expressed was purified by the same method used for the wild-type enzyme (Fig. 2, panel B ) . The ability of these mutant enzymes to bind to the Cm-Sepharose column and elute with Fru-l,6-P2 suggested that the substrate-binding site was not drastically affected.

To confirm the structural integrity of these mutant en- zymes, CD, gel filtration, and light scattering experiments were performed. The secondary structure of the mutant en- zymes was analyzed using CD. The spectrum for K229A indicated that it had slightly greater a-helical content than the other enzymes. D33A, D33S, and D33E spectra overlapped the wild-type spectra (Fig. 3) (D33S and D33E spectra not shown). Furthermore, the quaternary structure of the D33A

enzyme, determined by gel filtration and light scattering, was indistinguishable from the wild-type enzyme (data not shown).

TO further demonstrate that the active site was not drasti- cally perturbed in the mutant enzymes, assays for Schiff base formation with Fru-l,6-P2 were performed (33). The amount of radioactive substrate incorporated for each enzyme was: wild-type, 320 pmols; D33A, 431 pmols; D33S, 410 pmols; D33E, 247 pmols; K229A, 5 pmols. Substrate was bound to the same lysine in both the normal and D33A enzymes, as indicated by an identical pattern of labeled peptides following partial V8 proteolysis (data not shown). Unlike a mutation of Lys-229, which removes the Schiff base forming amino group, mutations of Asp-33 do not drastically affect the ability of the enzyme to catalyze Schiff base formation.

Although the enzymes mutated at Asp-33 had greatly re- duced activity, it was possible to measure their v,,, and K, toward Fru-1,6-P2 by using large amounts of enzyme in the assay (Table I). The significant difference in the K,,, of all proteins examined clearly showed that this activity was not due to a low level contaminant in the enzyme preparation. Moreover, the K229A protein, which was prepared in an identical fashion had no detectable activity. The most severe effect on VmaX was seen with enzymes in which the mutant residue was the most dissimilar to wild-type. The V,,, of D33A, from which the carboxyl group was completely re- moved, was 5600 times slower than wild-type. The D33S enzyme had a Vmsx 3600 times slower than wild-type, whereas a decrease of 900 times was seen in D33E.

To gain insight into the catalytic role of Asp-33, the level of carbanion/enamine enzyme-substrate intermediate was de- termined from the rate of its oxidation by hexacyanofer- rate(II1). The steady state and equilibrium levels of carbanion were determined using Fru-l,6-P2 and DHAP, respectively (Table 11). The mutant enzymes had a decreased carbanion oxidation rate with both substrates but there was a more drastic effect with Fru-l,6-P2. Surprisingly, with D33A there was a 17-29-fold difference between Fru-l,6-P2 and DHAP, whereas with wild-type there was no difference, and with D33E there was a 1-3-fold difference. The lower steady state level of the intermediate compared with the equilibrium level indicated a role for Asp-33 in steps leading to formation of the carbanion from Fru-l,6-P2.

DISCUSSION

Overexpression of rabbit aldolase A in E. coli produced an enzyme that was functionally and structurally the same as the classic muscle enzyme. Although production from the trc promoter can be controlled, expression following induction or expression under constitutive conditions made no difference in the yield obtained (37). Using pDT14, the rabbit aldolase A was expressed at levels similar to those reported for the expression of maize aldolase and the human aldolase A and B, i.e. 2.4-6.3% of the total soluble protein (35, 41). By using the high copy number plasmid, pPB14, expression was greatly increased and the enzyme remained soluble (37). Clearly the eukaryotic aldolase is not detrimental to the bacterial cell.

The drastic loss of activity in Asp-33 mutant enzymes, which had apparent retention of secondary and quaternary structure and an active site structure sufficient for Schiff base formation, strongly suggested a critical role for this residue in the catalytic mechanism. Additional evidence in support of the crucial role of this Asp-33 was its invariance during evolution. Alignment of aldolase sequences from maize, Dro- sophila, Trypanosome, Plasmodium, and numerous verte-

Role of Asp-33 at the Active Site of Aldolase 1099

brates revealed that all have an aspartate at this position? Insight into the role of this residue was obtained from

measurement of the rate of carbanion oxidation, which is directly proportional to the steady state or equilibrium con- centration of the carbanionfenamine intermediate. The re- active intermediate is formed upon either carbon-carbon bond cleavage of the hexose substrate or deprotonation at carbon 3 of the triose substrate. With both substrates the level of carbanion in Asp-33 mutants is decreased, yet a more extreme effect is seen with Fru-1,6-Pz. The low level of carbanion when using Fru-1,6-Pz could be caused by a reduced rate of formation of the intermediate, an increased rate of protona- tion to the nonreactive ketimine, slow release of glyceralde- hyde 3-phosphate which sterically hinders the oxidant, and/ or greater susceptibility of mutant enzymes to inactivation by hexacyanoferrate(II1). The data presented here best supports the first possibility, although each will be discussed below.

The carbanion levels and drastically reduced V,,, of D33A are consistent with a defect in carbon-carbon bond cleavage, the essential step in formation of the carbanion intermediate from Fru-1,6-Pz (42). The D33A mutant generated 17-29-fold lower carbanion levels from Fru-1,6-P2 than from DHAP, in contrast to wild-type and carboxyl-terminal mutants (43), which have relatively equal levels with either substrate. This indicates a role for this residue in carbon-carbon bond cleav- age prior to formation of the carbanion. Consistent with a role of the carboxyl group in cleavage, changing the Ala to Glu s i ~ i ~ c a n t l y reduced the disparity in the carbanion levels detected with different substrates.

Another explanation for the lower levels of carbanion could be an increased rate of protonation. However, this would not explain the lower V,,, observed. Mutations that affect the rate of protonation of the carbanion have been observed. Studies of the aldolase carboxyl terminus have shown that mutations or modi~cations cause an increase in the rate of carbanion oxidation, consistent with a build-up of this inter- mediate in these mutant enzymes (35, 42).

Carbon bond cleavage is not necessary to form a carbanion when using DNAP, yet a lower level of carbanion was seen with this substrate. This indicates that there is an equilibrium destabilization of the carbanion in these mutants. This could be explained by changes in charge distribution and interac- tions of Lys-146 and Lys-107 that flank Asp-33 in the active site structure (19). Lys-107 has been implicated in C6-phos- phate binding (44) and Lys-146 has been implicated in catal- ysis by modification with N-bromoacetylethanolamine phos- phate (45). Mutations at Asp-33 may shift the pK, of groups involved in carbanion protonation.

A slower release of glyceraldehyde 3-phosphate could ex- plain the decreased V,,, of the Asp-33 enzymes. However, slow release of glyceraldehyde 3-phosphate would interfere with carbanion protonation (42), leading to an increased level of carbanion, which is not consistent with the oxidation rates observed. It is possible that the hexacyanoferrate(II1) is ster- ically hindered by the glyceraldehyde 3-phosphate in the active site, thus masking an increased level of carbanion. Although, in the wild-type enzyme the rate of carbanion oxidation is the same with either substrate, indicating that glyceraldehyde 3-phosphate does not hinder the hexacyano- ferrate(II1). The spatial constraints may be different in the mutant enzymes, but structural measurements and Schiff base trapping indicate that the mutant and wild-type enzymes are very similar. Moreover, regardless of the change in size of the position 33 substituent there is a consistent decrease in carb- anion levels.

* D. Tolan, unpublished observations.

-NU- 1 \

SCHEME I SCHEME I I

FIG. 4. Potential role for Asp-33 in the catalytic mecha- nism. Scheme I , Asp-33 as a proton acceptor from the C4-hydroxyl group. The subsequent electronic rearrangements lead to C344 cleav- age and formation of a planar carbanion. Scheme I I , Asp-33 polarizing a base; in this depiction Lys-146 is the proton accepkor. The 5 active site residues shown are aligned based upon the 2.7-A structure. Both schemes I and I1 depict the substrate, Fru-l,6-P2 as a Schiff base between the carbonyl of the sugar and Lys-229. Lys-107 and Arg-148 are depicted as the residues binding the C-6 and C-1 phosphates, respectively.

It is known that hexacyanoferrate(II1) inactivates aldolase (46). There is a possibility that the Asp-33 mutant enzymes were more susceptible to inactivation by this oxidizing re- agent. This seems unlikely because it would indicate that the extent of inactivation is dependent upon t,he particular sub- strate. In summary, the most likely problem seems to be formation of the carbanion intermediate, consistent with the decreased Fru-1,6-Pa cleavage activity of these m u ~ n t s .

The published mechanism of catalysis for this enzyme has never indicated the involvement of an acidic residue such as Asp-33 (17). The postulated mechanisms have the proton acceptor as one of either Cys-72 or Cys-338 and the proton donor to the carbanion as His-361. As discussed, these resi- dues are clearly not at the active site, based upon primary and tertiary structure, and site-directed mutagenesi~ (14, 19, 20), and probably do not play the predicted role. We propose that Asp-33 may be involved in proton abstraction from the C-4 hydroxyl group, either directly as the base or indirectly by polarizing another residue which acts as the base (Fig. 4). A similar role has been proposed for a glutamate at the active site for another class I aldolase, the 2-keto-4-hydroxyglutarate aldolase (47). However, a three-dimensional structure with the hexose substrate bound should clarify the role of this residue in the catalytic cycle.

Acknowledgments-We thank Dr. Robert Davenport for many helpful discussions and critical reading of the manuscript, Dr. Richard Laursen and Dr. J. D. Dixon for the amino acid sequence aad analysis, Dr. Jurgen Sygusch for making the coordinates for a 2.3-A structure available prior to publication. The technical assistance in the early phases of this work from Luc Berthiaume, Sarah Cosgriff, Joanne Woodward, Joyce Soprano, and Dr. Christopher Green is greatly appreciated. We thank Dr. Stephen Burley for use of the light scattering device, Dr. Herbert Lebherz for the gift of antibody, and Sharon Doyle for critical reading of the manuscript.

REFERENCES

1. Penhoet, E. E., Koehman, M., Valentine, R., and Rutter, W. J. (1967)

2. Horecker, B. L., Tsolas, O., and Lai C . Y. (1975) in The Enzymes (Boyer,

3. Penhoet, E. E., Kochman, M., and Rutter, W. 3. (1969) Biachemistw 8.

Biochemistry 6,2940-2949

P. D., edf V61.7, pp. 213-258, Acidemic Press, New York

4391-4402 4. Rutter, W. J. (1964) Fed. Proc. 23, 1248-1257

6. Hers, H. G., and Joassin, G., (1961) Enzymol. Biol. Clin. 1 4-14 5. Rlostein, R., and Rutter, W. J. (1963) J. Biol. Chem. 238,3280-3285

7. Kishi, H., Tsunehiro, M., Hirono, A, Fuji, H., Miwa, d., and Hori, K. 8. Cross, N. C. P., Tolan, D. R., and Cox, T. M. (1988) Cell 53,881-885 9. Cross, N. C. P., de Franchis, R., Sebastio, G., Dazzo, C., Tolan, D. R.,

Gregori, C., Odihvre, N., Vidalhet, M., Romano, V., Mascali, G., Romano, C., Musumeci, S., Steinmann, B., Gitzelmann, R., and Cox, T. M. (1990) I . ~ , U W ~ RRK m-.?nR

" .

(1987) Proc. Natl. Acad. Sei. U. 9. A. 84,8623-8627

10. D a m , C., and Tolan, D. R. (1990) Am. J . Human Genet. 46,1194-1199 "_, _" _""

1100 Role of Asp-33 at the Active Site of Aldolase 11.

13. 12.

14.

15.

16.

17. 18.

19.

20.

21.

22. 23.

24.

25.

26.

Brooks, C. C., Buist, N., Tuerck, J., and Tolan, D. R. (1991)Am. J. Human

Kelly, P. M., and Tolan, D. R. (1987) Plant Physiol. 82, 1076-1080 Kukita, A,, Mukai, T., Miyata, T., and Hori, K. (1988) Eur. J. Biochem.

Rottmann, W. H., Tolan, D. R., and Penhoet, E. E., (1984) Proc. Natl.

Kobashi, K., and Horecker, B. L. (1967) Arch. Biochem. Eiophys. 121,

Hartman, F. C., and Welch, M. H. (1974) Eiochem. Eiophys. Res. Commun.

Lai, C. Y., Nakai, N., and Chang, D. (1974) Science 183,1204-1206 Voet, D., and Voet, J. G. (1990) Biochemistry, pp. 433-435, John Wiley and

Sons, New York, NY Sygusch, J., Beaudry, D., and Allaire, M. (1987) Proc. I'latl. Acad. Sci.

U. S. A. 84, 7846-7850 Takahashi, I., Takasaki, Y., and Hori, K. (1989) J. Eiochem. (Tokyo) 105,

281-286 Sambrook, J. , Fritsch, E. F., and Maniatis, T. (1989) Molecular Cloning: A

Laboratory Manual, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY

Genet. 49, 1075-1081

88,471-478

Acad. Sci. U. S. A . 81,2738-2742

178-186

57,85-92

Vieira, J., and Messing, J. (1982).Gene (Amst . ) 19,259-268 Taylor, J. W., Ott, J., and Eckstem, F. (1985) Nuclecc Aclds Res. 13,8765-

Saneer. F.. Nicklen. S.. and Coulson. A. R. (1977) Proc. Natl. Acad. Sci. 8785

UT S.'A. '74, 546315487

Tolan, R. (1986) BioTechnipues 4, 428-432

E. (1984) J. B id . Chem. 259, 1127-1131

Barr, P. J . , Thayer, R. M., Laybourn, P., Najarian, R. C., Seela, F., and

Tolan, D. R., Amsden, A. B., Putney, S. D., Urdea, M. S., and Penhoet, E.

27. Laemmli, U. K. (1970) Nature 227,680-685 28. Bradford, M. (1976) Anal. Eiochem. 72,248-254 29. Taylor, J. F. (1955) Methods Enzymol. 1, 310-320 30. Towbin, H., Staehelin, T., and Gordon, J . (1979) Proc. Natl. Acad. Sci.

31. Racker, E. (1952) J. Eiol. Chem. 196,347-351 32. Cleland, W. W. (1990) Enz mes 19,99-158 33. Tromhetta, G., Balhoni, 2, di Iasio, A,, and Grazi, E. (1977) Biochem.

34. Cleveland, D. W., Fischer, S. G., Kirschner, M. W., and Laemmli, U. K.

35. Healy, M. J., and Christen, P. (1973) Biochemistry 12, 35-41 36. Amann, E., and Brosius, J. (1985) Gene (Amst . ) 40, 183-190 37. Beernink, P. T., and Tolan, D. R. (1992) Protein Expression Purification

38. Baldwin, S. A., and Perham, R. N. (1978) Biochem. J. 169,643-652 39. Berthiaume, L., Beaudry, D., Lazure, C., Tolan, D. R., and Sygusch, J.

40. Pappin, D. J. C., Coull, J. M., and Koster, H. (1990) Anal. Eiochem. 187,

41. Sakakibara, M., Takahashi, I., Takasaki, Y., Mukai, T., and Hori, K. (1989)

42. Rose, I. A,, O'Connell, E. L., and Mehler, A. H. (1965) J. Biol. Chem. 240,

43. Berthiaume, L., Loisel, T. P., and Sygusch, J. (1991) J. Eiol. Chem. 266,

44. Anai, M., Lai, C. Y., and Horecker, B. L. (1973) Arch. Biochem. Eiophys.

45. Hartman, F. C., and Brown, J. P. (1976) J. Eiol. Chem. 251,3057-3062 46. Birkenhager, J. C. (1960) Eiochim. Eiophys. Acta 40 182-183 47. Vlahos, C. J., and Dekker, E. E. (1990) J. Bid. Ched. 265, 20384-20389

U. S. A. 76,4350-4354

Eiophys. Res. Commun. 74, 1297-1301

(1977) J. Eiol. Chem. 252, 1102-1106

3,332-336

(1989) Arch. Eiochem. Eiophys. 272, 281-289

10-19

Biochcm. Ecophys. Acta 1007, 334-342

1758-1765

17099-17105

156, 712-719