Heterotetramers of human liver mitochondrial (class 2) aldehyde dehydrogenase expressed in Escherichia coli. A model to study the heterotetramers expected to be found in Oriental people.

About 50% of the Oriental population have less liver mitochondrial aldehyde dehydrogenase (ALDH2) activity than do other people. It was found that they possessed an enzyme with a lysine at position 487 (E487K) instead of glutamate (Glu487). We previously found that the Km for NAD of recombinant human and rat E487K enzymes increased more than 150-fold (Farrés, J., Wang X., Takahashi, K., Cunningham, S. J., Wang, T.T., and Weiner, H (1994) J. Biol. Chem. 269, 13854-13860). Many aldehyde dehydrogenase-deficient people were found to be heterozygous when genotyped for ALDH2. In this study liver tissue from heterozygous people was analyzed and found to possess mRNAs for both the glutamate and the lysine subunits. Western blot analysis showed that the glutamate subunit was present. The cDNAs for Glu487 and E487K were coexpressed on one plasmid in Escherichia coli, and the enzyme forms were separated from each other by isoelectric focusing to show that heterotetramers were formed. Only one Km value for NAD could be measured with the purified heterotetrameric enzyme that possessed just 16-18% activity of the glutamate homotetrameric enzyme. The E487K homotetramers had 8% specific activity of the Glu487 enzyme. There was no pre-steady state burst of NADH formation with the heterotetramer, a property found with the glutamate enzyme. Similar results were found for the coexpressed rat liver enzyme, except that a higher specific activity, 48%, was obtained. Thus, we conclude that presence of the lysine subunit altered the activity of the glutamate subunit in the heterotetramer to make it function more like an E487K enzyme.

Liver mitochondrial (class 2) ALDH2 1 is the major enzyme responsible for the metabolism of acetaldehyde to acetate (1,2). It has been shown that many Oriental people were lacking active mitochondrial ALDH2 (3,4). Work from Yoshida's labo-ratory (5) showed that these people possessed a mitochondrial enzyme that cross-reacted with antibodies prepared against the active enzyme. Amino acid sequencing revealed that this variant differed from the active form by just one amino acid substitution. The Oriental variant possessed a lysine residue at position 487, while the active form contained a glutamate (6,7). The accumulation of toxic acetaldehyde occurs after people with this phenotype drink ethanol. This leads to alcohol-associated symptoms, such as facial flushing and nausea (8,9). It was unexpected to have found that the people possessing the inactive lysine-variant actually possessed the DNA coding for both the active glutamate enzyme and the inactive lysine enzyme (10,11).
Three genotypes of ALDH2 have been identified in Oriental populations: ALDH2*1/ALDH2*1 allele for homotetramer Glu 487 ; ALDH2*2/ALDH2*2 for homotetramer E487K; ALDH2*1/ALDH2*2 for heterotetramer (12,13). Virtually all Caucasians have the genotype ALDH2*1/ALDH2*1. The catalytic properties of the homotetramers of Glu 487 and E487K have been studied. It was found that the Glu 487 form had a low K m for NAD, 30 -70 M (14, 15). The K m value for NAD for the recombinantly expressed E487K increased over 150-fold (15). Although no detectable ALDH2 activity was found on electrophoretograms and in liver homogenates of ALDH2-deficient people (5,16), the recombinantly expressed E487K was found to be active. The k cat value was decreased 10-and 2.5-fold compared to that of the recombinant Glu 487 human and rat forms (15), respectively.
It can be predicted that heterozygous people (ALDH2*1/ ALDH2*2) should have 50% activity of the homozygous people (ALDH2*1/ALDH2*1) if inheritance of ALDH2 were co-dominant. Studies on the activity of ALDH2 in heterozygous individuals are conflicting. Activity staining of different tissue homogenates, after separation by IEF, showed the lack of ALDH2 activity in heterozygous people even though immunoreactions showed the presence of ALDH2 protein (11,(17)(18)(19). Other studies showed that heterozygous individuals had 13-15% of the Glu 487 activity (16,20). A lower percentage of people who were heterozygous for ALDH2 had alcohol-associated symptoms after drinking ethanol than did people who were homozygous for E487K (21). This implies that the heterotetrameric ALDH2 might be catalytically active. None of these studies, however, determined whether or not both the E-and the Ksubunits of ALDH2 were present in the heterozygous individuals. In a recent study it was reported that the insertion of the cDNA coding for the E487K subunit into a cell line expressing Glu 487 subunits caused a reduction of the ALDH2 activity (22). The best interpretation of the latter observation is that the Glu 487 subunits in the heterotetramers from the doubly transformed cell line had less activity than did the native enzyme.
In this study we investigate the reasons for finding less ALDH2 activity in the livers of people who were found to be * This work was supported in part by National Institutes of Health Grant AA05812 (to H. W.). 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. This is journal paper no. heterozygous with respect to the ALDH2 gene. We will demonstrate that mRNAs for both Glu 487 and E487K, as well as E-subunits of ALDH2, were present in heterozygous livers. The cDNAs for Glu 487 and E487K were coexpressed in E. coli to produce heterotetrameric forms of ALDH2 and the kinetic properties were determined. The data can best be interpreted as showing that the heterotetrameric form of ALDH2 had less activity than would be expected from the simple combination of subunits.

EXPERIMENTAL PROCEDURES
Materials-NAD and NADH were purchased from Sigma; Sequenase version 2.0 kit was obtained from U. S. Biochemical Corp.; propionaldehyde was from Aldrich; Magic Minipreps DNA purification system, random primers and T4 DNA ligase were from Promega Corp.; Eco57I, BamHI, and NdeI were from New England Biolabs; IEF standards and Affi-Gel 15 Bio-Gel agarose were from Bio-Rad; agarose IEF and Pharmalyte were from Pharmacia Biotech Inc.; superscript reverse transcriptase II was from Life Technologies, Inc.; Taq DNA polymerase was from Boehringer Mannheim.
RT-PCR-Total RNA (5 g) isolated from a heterozygous liver genotyped by David Crabb, M.D., Indiana University School of Medicine, was used for the synthesis of the first strand of cDNA by following the protocol from Life Technologies, Inc. Random primers were used for the reverse transcription which was performed at 42°C for 1 h. The first strand of cDNA was then used as the template for PCR. The two primers, 5ЈTTTGAATTCCATATGATGTGTTTGGAGCCCAGT-3Ј (containing NdeI site, underlined) and 5Ј-TTTGGATCCTTATGAGTTCT-TCTGAGGCACTT-3Ј (containing BamHI site, underlined) were used to amplify the last 45 amino acids of human liver ALDH2. PCR was carried out for 30 cycles on a GeneAmp PCR System 9600 (Perkin-Elmer) using Taq DNA polymerase at 94°C for 1 min, 49°C for 2 min, and 72°C for 2 min.
The RT-PCR products were digested with BamHI and NdeI and cloned into pT7-7 vector. The amplified cDNA was sequenced using the Sequenase version 2.0 kit. In addition, the RT-PCR products were digested with Eco57I for 2 h at 37°C and then separated on a 2% agarose gel.
Plasmid and Bacterial Strains-The full-length cDNAs for human native and E487K ALDH2 (15,23) were cloned on the pT7-7 vector. Each cDNA had its own ribosome binding site, but both were under the control of one T7 promoter. E. coli DH5␣ (Bio-Rad) was transformed with pT7-7 expression vector, and the constructions were confirmed by DNA sequencing. The construction of the rat coexpression vector was identical to that of the human coexpression vector, except the native and E487K were under control of separate T7 promoters. Restriction analysis using different enzymes was carried out to verify the presence of rat E-and K-cDNAs on the same plasmid. Human ALDH2 coexpression was performed in E. coli strain JM 109(DE3) (Promega), and rat coexpression was in E. coli strain BL 21(DE3) PLysS.
Expression of Heterotetramer ALDH2-E. coli strains harboring the human coexpression vector pT7-7 were grown in 2 ϫ YT medium at 37°C for 4 -5 h in the presence of 50 g/ml ampicillin. The cultures were then grown at 16°C overnight after induction with 0.4 mM isopropyl-1-thio-␤-D-galactopyranoside. The 2 ϫ YT medium containing both ampicillin (50 g/ml) and chloramphenicol (50 g/ml) were inoculated with E. coli cells harboring the rat coexpression plasmid. After growing the cells for 3-4 h, the expression of enzyme was induced by 0.4 mM isopropyl-1-thio-␤-D-galactopyranoside and grown for a additional 5-6 h at 37°C.
Purification of Coexpressed Aldehyde Dehydrogenases-The enzymes were purified through DEAE-cellulose and 4Ј-hydroxyacetophenoneaffinity chromatography (15,24). The purity of ALDH2 was determined by SDS-polyacrylamide gel electrophoresis followed by protein staining with Coomassie Blue. The protein concentration was determined using the Bio-Rad protein assay kit with bovine serum albumin as a standard.
Fluorescence Assay for the Dehydrogenase Activity-Dehydrogenase activity assays were performed with a Aminco filter fluorometer (25) in 100 mM sodium phosphate, pH 7.4, at 25°C. The formation of NADH was recorded as a function of time.
Slab-Gel IEF-Agarose (1%) IEF was performed on a Pharmacia flat-bed electrophoresis apparatus using Pharmalyte at 10°C. Pharmalytes, pH 4.5-5.4 and 4.0 -6.5, were used for the human and rat samples, respectively. The gel was stained either for activity or for protein with Coomassie Blue (26). The concentrations of NAD and propionaldehyde were 5 mM and 100 M, respectively, in the activity staining solution prepared in 100 mM sodium phosphate (pH 7.4).
IEF Fractionation of Human Coexpressed ALDH2-About 20 mg of purified coexpressed ALDH2 was fractionated on a Rotofor preparative isoelectric focusing cell (Bio-Rad). A mixture of Pharmalyte, pH 4.5-5.4, and glycerol was applied to the focusing chamber and was pre-focused for about 1 h. The coexpressed ALDH2 was then injected into the sample chamber. The final concentrations of the Pharmalyte and glycerol were 2 and 18%, respectively. After focusing, each fraction was run on a IEF gel and stained for activity. The fractions containing heterotetramers were combined and were fractionated again.
Pre-steady State Burst Analysis-The pre-steady state burst of NADH formation was performed in 100 mM sodium phosphate, pH 7.4, as described previously (15,27). The concentration of propionaldehyde was 140 M, and the concentrations of NAD were 1 mM, 10 mM and 5 mM for native, E487K, and heterotetramers, respectively.
Preparation and Purification of Antibodies Raised against Glu 487 and E487K-Antibodies were raised against a synthetic peptide attached to the multiple antigen peptide system as described by Tam (28), which possessed the last 16 residues (485-500) of ALDH2. One had a lysine at the position corresponding to 487, while the other had a glutamate. The E-and K-antibodies were purified by immunoaffinity purification scheme described by the manufacturer (Bio-Rad). The native and the E487K ALDH2 were coupled to the Affi-Gel 15 which was supported on Bio-Gel A-agarose. The coupling was allowed to take place overnight at 4°C. The next day, 0.1 M ethanolamine HCl (pH 8.0) was added to complete the blocking reaction. After 2 h the gel was transferred to a column and was washed with water until it was free of reactants. The column was then equilibrated with 10 mM sodium phosphate (pH 7.4). The anti-E antiserum was passed through a column of E487K enzyme and anti-K was passed through a column of Glu 487 enzyme several times. Preparation and purification of antibodies against ALDH2 was described previously (15).

Identification of the mRNA for Glu 487 in the Heterozygous
Liver-To determine if mRNA for the Glu 487 variant was present in the liver of an ALDH2 heterozygous person, total liver RNA was analyzed by RT-PCR. The Oriental variant of ALDH2 differed from the active Glu 487 only at that position, so the cDNA encoding the last 45 amino acids of ALDH2, including position 487 was amplified to produce a 158-base pair fragment.
The 158-base pair RT-PCR product was treated with Eco57I to determine if mRNAs for both Glu 487 and E487K forms of ALDH2 were present in the heterozygous liver, as outlined by Tu and Israel (29). This enzyme recognizes the sequence 5Ј-CTGAAG(N) 16 -3Ј and 3Ј-GACTTC(N) 14 -5Ј, which is present in the cDNA for the Glu 487 enzyme flanking the codons for position 487. When glutamate (GAA, underlined) at the position of 487 is replaced by lysine (AAA) the cDNA cannot be recognized by Eco57I. If mRNA for both native and E487K were present in the heterozygous liver, some of the RT-PCR products should be digested by Eco57I, others should not. As a control, the PCR products amplified from two pT7-7 plasmids (15), which carried the cDNAs encoding Glu 487 and E487K, respectively, were treated with Eco57I. The PCR products amplified from the cDNA encoding native form of human liver ALDH2 were digested as expected, while the PCR products of E487K were not cleaved by Eco57I. About half of the RT-PCR products amplified from the heterozygous liver were cleaved by the endonuclease, as shown in Fig. 1. The results of the Eco57I digestion indicated that the mRNAs for both Glu 487 and E487K were present in the heterozygous liver.
To further confirm the presence of mRNAs for Glu 487 and E487K in the heterozygous liver, RT-PCR products were cloned into the NdeI and BamHI sites of pT7-7 vector. Six colonies were sequenced; three were found to possess Glu 487 form of ALDH2, having GAA at the position corresponding to 487. The other three possessed E487K, having AAA at this position. The sequencing results confirmed that the heterozygous liver had the mRNAs for both Glu 487 and E487K ALDH2.

Antibodies Established the Presence of Glu 487 Subunits in
Livers from Heterozygous People-The antibodies prepared against the two synthetic peptides were used in cross-reaction experiments. Both antisera were found to cross-react with pure Glu 487 and E487K. An attempt was made to purify the antibodies so they would become determinant specific. Purified anti-E antibody recognized the native enzyme but not the E487K form. Unfortunately, purified antibodies prepared against E487K recognized both forms of ALDH2. It was possible, though, to use the anti-E antibody to determine if the E-subunits were produced in the livers of heterozygous persons.
Human livers from people with different ALDH2 genotypes were analyzed by Western blotting as shown in Fig. 2. It was found that anti-ALDH2 (15) reacted equally well with liver homogenates from the homozygous E, the homozygous K, and heterozygous EK persons. Anti-E antibodies reacted with both homozygous E and heterozygous EK, but not with homozygous K liver homogenates. Neither anti-ALDH2 nor anti-E crossreacted with cytoplasmic aldehyde dehydrogenase. These results indicated that the E-subunits were present in the heterozygous people. Due to the age of the liver samples and the potential differential stability of the subunits, it was not possible to accurately quantify the concentration of E-and K-subunits.
Expression and Purification of Coexpressed Human ALDH2-It appears that livers from people genotyped to be heterozygous contained both Glu 487 and E487K subunits. Assuming that the two subunits were equally expressed, one could expect that the tissue would possess 50% of the ALDH2 activity of the homozygous Glu 487 liver. However, the results were found to be different (11, 16 -17, 20). It appears, supported by recent cell culture work (22), that the heterotetrameric enzyme had activities more like the E487K homotetrameric enzyme.
To actually study the properties of human ALDH2 heterotetramers composed of both E-and K-subunits, the cDNAs coding for Glu 487 and E487K were incorporated on one plasmid (Fig.  3). The recombinant enzymes were expressed in E. coli and were purified to apparent homogeneity using techniques employed to purify the individual homotetrameric forms (15). A single band of protein was found on SDS-polyacrylamide gel electrophoresis. The expressed protein was analyzed by IEF gel electrophoresis to determine if heterotetrameric forms of the enzyme might exist. A smear of bands was found in the region between those for the two homotetrameric forms of the human enzyme whose pI differed by only 0.2 unit (Fig. 4). This implied that heterotetrameric forms of the enzyme were present. We found that when pure Glu 487 and pure E487K proteins were mixed and subjected to isoelectric focusing, two bands of protein, corresponding to each pure homotetramer could be found, lending further support to the notion that the smear of unresolved bands represented heterotetrameric forms of the en-zyme. The possibility of the presence of homotetramer E-and K-forms could not be ruled out. Therefore, in order to characterize the properties of pure heterotetramers, the coexpressed enzyme was further resolved by an IEF fractionating cell.
Preparative IEF of the Coexpressed Human Enzyme-The 4Ј-hydroxyacetophenone-affinity purified coexpressed enzyme was subjected to preparative IEF fractionation to make certain that the preparations were free of homotetramers. Each fraction of active enzyme was analyzed by IEF gel electrophoresis followed by activity staining. The fractions containing the potential heterotetramers were pooled and fractionated again. The second fractionation resulted in a ladder of ALDH2 enzymes indicating that some fractions contained the heterotetramers, free of homotetramers (Fig. 5).
Enzymatic Properties of the Human Heterotetramers-The specific activity of the enzymes in four fractions which lie between the pIs of homotetrameric Glu 487 and E487K was determined. They averaged 13% of the activity of the Glu 487 homotetramer when assayed at 2.5 mM NAD, a concentration which would saturate the low K m enzyme form. Under these assay conditions E487K homotetramer had just 2.5% the activity of the glutamate enzyme.
The K m for propionaldehyde for E487K homotetramer was essentially identical to that of the native enzyme (15) and was found to be the same for the heterotetrameric mixture. Although the double reciprocal plots appeared to be linear (Fig.  6), both a linear and a nonlinear regression analysis of the primary data were performed. The statistical fit of the data revealed that the r 2 values were all greater than 0.995. Essentially same kinetic constants were obtained by either treatment of the data. Only one K m for NAD was found for the mixture of heterotetramers even when the concentration of NAD was increased to 8 mM. The K m for NAD was between 91-117 M, which was approximately 3-4 times higher than that of homotetrameric Glu 487 , but was very different from the 6500 -7300 M value found for E487K (Table I).
Since the K m for NAD with the pure E487K homotetramer was so high and its specific activity so low, the contribution of the K-subunit to the overall activity would be small. To determine if it would even be possible to find a high K m component in the heterotetrameric mixture, equal concentrations of Glu 487 and E487K homotetramers were mixed and assayed. At NAD concentrations greater than 5 mM the Glu 487 enzyme was inhibited. If the E-subunits in the heterotetramer were also subjected to NAD inhibition, it would not be possible to measure the activity of a high K m component. In fact, when the heterotetrameric enzyme was assayed above 5 mM NAD the velocity started to decrease. Thus, although we could not determine FIG. 1. Eco57I digestion of the RT-PCR products from the heterozygous ALDH2 liver. The RT-PCR product amplified from the total RNA isolated from the liver of a heterozygous person was treated with Eco57I. The PCR products amplified from plasmids encoding Glu 487 and E487K forms, respectively, were also digested with Eco57I and served as controls. ϩ and Ϫ indicate with and without Eco57I, respectively. that a high K m component was present, we cannot unequivocally state that one did not exist.
To show that it would be possible to assay for a high K m component, the rat enzymes were used. This was done because rat E487K possess a higher specific activity and a lower K m for NAD than human E487K (15). Both rat Glu 487 and E487K were mixed together in a ratio such that they possessed equal activity when assayed separately. A biphasic Lineweaver-Burke double reciprocal plot was obtained and the K m values were similar to ones found for the rat Glu 487 and E487K enzymes.
Pre-steady state burst analyses were performed for we previously showed that a burst of NADH formation existed with the human (26) and rat (15) native enzyme but not with the rat lysine variant (15). Due to the low yield of the human enzyme after IEF fractionation, it was not possible to determine the burst magnitude of the fractionated enzyme. Therefore, presteady state burst analysis was performed with enzymes only purified through 4Ј-hydroxyacetophenone-affinity chromatography. A burst magnitude of 2 was found with the Glu 487 homotetramers and no burst was found for E487K. Only a burst magnitude of 0.12 was found with the coexpressed heterotetramer enzyme. The fact that there was essentially no pre-steady state burst of NADH formation showed that the E-subunit in the heterotetramers was not functioning as it did in the homotetramers. Furthermore, the fact that the specific activity was just 16 -18% compared to the homotetrameric Glu 487 form, allows us to suggest that the presence of E487K subunits alters the activity of the Glu 487 subunits, as was proposed to occur in the cell culture experiment (22).
Expression and Purification of Coexpressed Rat Liver Aldehyde Dehydrogenases-In a manner analogous to what was done with the human liver enzymes, cDNAs coding for the rat liver native and the E487K mutant were placed on one plasmid (Fig. 3). All three forms, homotetramer E, homotetramer K, and heterotetramer EK, were found to be expressed to the same extent. The enzymes were purified to homogeneity using the procedures previously described (15). IEF analysis showed, as was found with the human system, that the enzyme obtained from the coexpressed system had a pI value between those of the E-and K-homotetrameric enzymes (Fig. 4). The determinant-specific antibody, specific for the glutamate subunit, showed that the E-subunit was present in the coexpressed enzyme (data not shown).
Kinetic Properties of the Coexpressed Rat Aldehyde Dehydrogenases-We previously reported the kinetic constants for the recombinantly expressed rat liver enzyme forms (15). These were redetermined along with the one for the coexpressed heterotetrameric mixture and are presented in Table II. If the three enzyme forms were assayed with concentrations of NAD which would saturate only the low K m native enzyme, then the K-homotetrameric form had 12% the activity of the E-homotetramer, while the heterotetramer had 34% of the activity. When assayed with high concentrations of NAD, the K-homotetramer had 37% while the heterotetramer had 48% the activity of the native enzyme. DISCUSSION It has tacitly been assumed that Oriental people who possessed the genes coding for both the lysine and the glutamate form of ALDH2 expressed both variants of the enzyme. Assuming this would be found in the livers of these people, one could not explain why heterozygous persons were lacking ALDH2 activity in their livers. Even if the E487K enzyme had less than 10% the activity of the active glutamate enzyme, one would have expected that the heterozygous persons have had approximately 50% of the activity of a person who was homozygous with respect to the ALDH2*1 gene. One explanation for not finding the ALDH2 activity in these livers is that Glu 487 subunits were not expressed due to transcriptional or translational alterations. An alternative explanation is that the heterotetrameric enzyme behaves more like a homotetramer of the Ksubunits. Evidence recently has been presented to show that, in cells grown in culture, the addition of the lysine-coding ALDH2 cDNA caused a suppression of the activity of the glutamate enzyme (22). This was the first evidence to suggest that the E-subunit in the heterotetrameric enzyme was less active.
To test for these possibilities we examined liver samples from people who were previously genotyped for ALDH2. RT-PCR analysis showed that two cDNA sequences were obtained from the liver of the heterozygous individual, one with the lysine codon (AAA), other with the glutamate codon (GAA). This demonstrated that the mRNA for each form of the enzyme was transcribed in the liver of the heterozygous person. However, the existence of the mRNA does not prove that protein is produced. Instead of sequencing the entire cDNA to determine if there were a premature stop codon or an alteration which could affect translation, an antibody approach was employed. Antibodies were prepared against the C-terminal 16 amino acids of the Glu 487 and E487K enzymes. Liver homogenates from the two heterozygous people analyzed were found to possess the glutamate subunits, which could not be detected in two liver homogenates from people who were genotyped to be homozygous for the ALDH2*2. This supports the notion that the liver of a heterozygous person had the glutamate subunit of ALDH2.
It was necessary to measure the activity of the heterotetrameric forms of human liver ALDH2 composed of both E-and K-subunits to determine why the activity in the heterozygous person was lower than what was expected if the two forms of ALDH2 were co-dominantly inherited. Since we did not have access to large quantities of fresh human liver tissue from genotyped people, we reverted to using an E. coli coexpression system. The cDNA for both the lysine and the glutamate subunits were cloned on one plasmid. The coexpressed enzymes after IEF separation were found between the pI values for the homotetrameric Glu 487 and E487K enzymes. Similar results were found with the rat coexpressed enzymes.
The purified human enzyme was fractionated by preparative IEF. The fractions of active enzyme were selected which had pI values falling between those of the two pure homotetramers. Thus, we could be assured that the activities being measured was not due to the presence of homotetramers of the E-or the K-subunits. From the pure heterotetramer only one low K m for NAD was found and the specific activity was 16 -18% of Glu 487 homotetramer. When the specific activity was measured at 2.5 mM NAD, a concentration which would saturate the E subunit, the heterotetramer possessed only 13% activity. Under these assay conditions, E487K homotetramer had a specific activity of just 2.5%. Even though the precise subunit compositions of these heterotetramers were unknown, we conclude that heterotetramers were less active and possessed kinetic properties different from the parent homotetramers.
Assuming that there were an equal number of E-and Ksubunits expressed, then independently of how they were assembled, one would have expected the heterotetramers to have 50% the specific activity of the Glu 487 homotetramer when assayed at the low concentration of NAD which would saturate only the E-subunits. Under these conditions, though, only 13% the activity was found, indicating that the presence of the K-subunit did indeed suppress the activity of the E-subunit, supporting the conclusions from the cell culture work (22).
We previously showed that the human Glu 487 enzyme had a pre-steady state burst of approximately 2 moles of NADH per mole of enzyme (26), indicative of the half-of-the-site reactivity. No burst was found with the human E487K enzyme as we previously showed for the rat E487K enzyme (15). The human coexpressed enzyme had a burst magnitude of only 0.12, not a value of 1 or 2 as would be expected if the E-subunits were acting independent of the K-subunits. The 0.12 burst magnitude might be due to the presence of a small concentration of FIG. 6. Double reciprocal plot for NAD as the variable substrate to the Glu 487 and E487K homotetramer and heterotetramer enzymes. The primary data was analyzed by both linear and nonlinear regression analysis and is presented in Table I. For each fit the r 2 value was greater than 0.995. homotetrameric Glu 487 in the coexpressed system because the coexpressed ALDH2 used for the burst assay was not fractionated by IEF. The rate-limiting step changed for the E-subunit in the heterotetramer. It appears, then, that E-subunit was not acting independently of the K-subunit and was not functioning as it would in the homotetrameric glutamate enzyme. It is difficult to predict exactly what properties could be expected to be found for a heterotetramer composed of both high active and low active subunits. First, as discussed, the K-subunit decreases the activity of the E-subunit, but the exact amount is not known. The second complexity is that ALDH2 functions with half-of-the-site reactivity (26,27). That is, in the tetramer the subunits are not acting independently of each other but are functioning as two pairs and not four independent subunits. Assuming that an equal concentration of the E-and the K-subunits were expressed and that assembly were truly random, then the distribution of species should have been as shown in Scheme 1. Since the homotetramers were removed, the percentage of the heterotetramers would have been 28.7, 42.5, and 28.7 for the E 3 K, E 2 K 2 , and EK 3 , respectively. In any tetramer, six pairs of dimers could theoretically exist in the total population, as is illustrated in Table III. If we define the total activity of the Glu 487 homotetramer as 100, then each of six E-E dimer pairs in the half-of-the-site model would have an activity of 16.6. When assayed at 2.5 mM NAD the activity of the E487K homotetramer was just 2.5% that of the Glu 487 homotetramer. Thus, each of the six K-K dimer pairs would have an activity of 0.4. The expected activity for each heterotetrameric species is tabulated in Table III, assuming that the EK-dimeric pair had the activity of the KK-homodimer. In E 3 K, where there are three possible pairs of E-E interactions in the half-of-the-site model, the enzyme would have 51.2% the activity of the E 4 -homotetramer. In contrast, if the subunits were functioning independent of each other, the species would have had 75.6% the activity of the E 4 homotetramer. Similarly, the E 2 K 2 form would have just 18.7% the activity in the half-of-the-site model compared to 51.2% in the full-site model. Lastly, EK 3 would have had essential no activity in the half-of-the-site model compared to 26.8% in the full-site model. On average, the heterotetramer mixture would have had approximately 51.1% the activity of the E 4 -homotetramer if the subunits were truly independent of each other, but only 23.3% if the subunits were functioning as a pair of dimers and the EK pair had properties similar to a KK pair. We found that the heterotetramer had just 13% the activity of the high active homotetramer. This shows that the K-subunit decreases the activity of the E-subunit, as was suggested by the cell culture experiments (22).
We cannot offer an explanation for how the K-subunits alters the properties of the E-subunit. It appears to be a result of a change in rate-limiting step as the E-subunit no longer had a pre-steady state burst. Other investigators have reported that hetero-systems did not function as the sum of the properties of the independent subunits. This has been shown to occur with lactate dehydrogenase (31) and with aspartate transcarbamylase (32). The three dimensional structure of ALDH2 is not know so it is not possible to discuss subunit interactions. It is possible, though, that a glutamate residue in one subunit is interacting with a lysine in another in the heterotetrameric form. This new salt bond could be responsible for the altered properties of the heterodimer. Furthermore, if there were an attraction for heterotetrameric formation, the distribution of dimer pairs would not be the statistical average as illustrated in Table III but, would be skewed toward more E 2 K 2 heterotetramers. This condition could exist because the activity of the mixture was lower than the value estimated from the statistical model.
We reported previously that the recombinantly expressed rat E487K homotetramer had a high K m for NAD but, the specific activity was about 40% of the rat Glu 487 enzyme (15). This was reconfirmed in this study. Analysis of the kinetic data at a low concentration of NAD showed that E487K homotetramer had 12% and the heterotetramers had 34% the activity of the native enzyme. In contrast, the human E487K homotetramer had just 2.5% and heterotetramers had 13% the native human enzyme activity. Thus, it appears that the K-subunit in the human enzyme more dramatically affects the activity of the E-subunit than it does in the rat liver enzyme.
The amino acid sequences of the rat and human mitochondrial ALDH2 are 95% identical (30). Hence, it is difficult to rationalize why the change of a glutamate to a lysine at position 487 would produce such different affects on the specific activity of the rat and human enzymes. Investigators have discussed inserting the K-cDNA into animals to determine if this will affect their ethanol drinking behavior. Based on the data obtained in this study, we can conclude that in a liver of a transgenic rodent, the heterotetrameric ALDH2 might be much more active than it is in human heterozygous liver. The aldehyde oxidizing capacity in the rat liver will be larger than what would be expected to be found in heterozygous human. How this would affect acetaldehyde metabolism is not known. It is the increased blood acetaldehyde concentration that has been postulated to be the deterrent for alcohol drinking in Oriental people who are heterozygous with respect to their ALDH2 (16,21). Thus, the transgenic animal may not be a good model to study ethanol toxicity found in heterozygous Oriental people.   full-site and half-of-the-site reactivity In the half-of-the-site model there are six potential pairs of dimeric interactions. If on average all are occurring, then each E-E dimer pair would contribute 16.7% toward the total activity of the homotetramer, defined as 100 units. In the full-site model each subunit would contribute 25% toward the total activity. Each K-K subunit pair in the half of the site model would contribute 0.4 unit of activity, as the K-heterotetramer has 2.5% the specific activity of the native enzyme. In the full-site model each subunit would contribute 0.6 unit of activity. If one assumes that E-K has the same specific activity as does the K-K pair and there were a statistical distribution of subunits, then the full-site model predicts that the heterotetrameric mixture would have 51.1% activity, while the half of the site model would predict 23.3%.