Involvement of Glutamate 399 and Lysine 192 in the Mechanism of Human Liver Mitochondrial Aldehyde Dehydrogenase*

Mutation to the conserved Glu399 or Lys192 caused the rate-limiting step of human liver mitochondrial aldehyde dehydrogenase (ALDH2) to change from deacylation to hydride transfer (Sheikh, S., Ni, L., Hurley, T. D., and Weiner, H. (1997) J. Biol. Chem.272, 18817–18822). Here we further investigated the role of these two NAD+-ribose-binding residues. The E399Q/K/H/D and K192Q mutants had lower dehydrogenase activity when compared with the native enzyme. No pre-steady state burst of NADH formation was found with the E399Q/K and K192Q enzymes when propionaldehyde was used as the substrate; furthermore, each mutant oxidized chloroacetaldehyde slower than propionaldehyde, and a primary isotope effect was observed for each mutant when [2H]acetaldehyde was used as a substrate. However, no isotope effect was observed for each mutant when α-[2H]benzaldehyde was the substrate. A pre-steady state burst of NADH formation was observed for the E399Q/K and K192Q mutants with benzaldehyde, and p-nitrobenzaldehyde was oxidized faster than benzaldehyde. Hence, when aromatic aldehydes were used as substrates, the rate-limiting step remained deacylation for all these mutants. The rate-limiting step remained deacylation for the E399H/D mutants when either aliphatic or aromatic aldehydes were used as substrates. The K192Q mutant displayed a change in substrate specificity, with aromatic aldehydes becoming better substrates than aliphatic aldehydes.

Two conserved residues of mitochondrial aldehyde dehydrogenase (ALDH) 1 were found to bind the NAD ϩ ribose rings. Lys 192 bound the adenosine ribose, while Glu 399 bound the nicotinamide ribose (1). When they were separately mutated to a glutamine residue, the mutants were found to have a small decreased specific activity and only the K192Q form showed a markedly enhanced K m for NAD ϩ (2). It was totally unexpected to find that the mutants caused a change in the rate-limiting step. In the native enzyme, the rate-limiting step is deacylation (k 7 ), while for the K192Q and E399Q, it appeared to have become hydride transfer (k 5 , based on Fig. 1 of Ref. 2). Mutations to the other conserved residues which possessed a reactive side chain, while affecting some kinetic parameters, did not change the rate-limiting step. The exceptions were with Cys 302 and Glu 268 , which have been shown previously to function as the essential nucleophile (3) and general base (4), respectively.
We extended the study to investigate additional mutations of the two ribose-binding residues. Primary isotope effects with both aromatic and aliphatic substrates were measured to verify that the rate-limiting step did indeed become hydride transfer. In addition, a more specific role for Glu 399 was sought because, although it binds to the nicotinamide ribose, the E399Q mutant did not affect coenzyme binding. In contrast, mutation of Lys 192 , which binds to the adenosine ribose, did affect NAD ϩ binding. Furthermore, the results of the structure determination on ALDH2 suggested that the nicotinamide mononucleotide portion of NAD ϩ was not well ordered in the absence of Mg 2ϩ ions. Thus, it appears that position 399 might play some role in catalysis other than binding to the coenzyme.
Oligonucleotide-directed Mutagenesis-The human ALDH2 E399Q/ K/H/D and K192Q/E mutants were constructed by site-directed mutagenesis with the Mutagene kit following the manufacturer's instructions, as described previously (2,4). The mutant colonies were selected by sequencing using the dideoxynucleotide chain-termination method (7).
Expression in E. coli and Purification of the Native and Mutant Enzymes-All mutated forms of the enzyme were expressed (2, 8) and purified as described previously (2,4). The purity of the enzymes was checked by SDS-polyacrylamide gel electrophoresis using the Coomassie Blue staining procedure. The protein concentration was determined as described previously (2,4).
Fluorescence Assay for the Dehydrogenase Activity-The dehydrogenase activity assays were performed as described previously (2,4,9). The K m and V max values for NAD ϩ were determined in the presence of 14 M propionaldehyde, whereas the K m and V max values for propionaldehyde were determined in the presence of 1 mM NAD ϩ for the native enzyme and the E399D mutant, 2 mM NAD ϩ for the E399Q and E399H mutants, 3 mM NAD ϩ for the E399K mutant, and 7 mM for the K192Q mutant.
Determination of the Primary 2 H Isotope Effect-Aldehyde dehydrogenase activity was determined at 25°C in 100 mM sodium phosphate buffer (pH 7.4) in the presence of acetaldehyde and [ 2 H]acetaldehyde, benzaldehyde, and ␣-[ 2 H]benzaldehyde.
Pre-steady State Burst of NADH Formation-The pre-steady state burst magnitude of NADH formation was determined with propionaldehyde or benzaldehyde as the substrate, as described in the accompanying paper (2). Concentrations of NAD ϩ were 1-7 mM for the native and different mutant enzymes. The propionaldehyde concentration was * This work was supported in part by National Institutes of Health Grant AA05812. This is journal paper 15431 from the Purdue Agricultural Experiment Station. 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.
‡ To whom correspondence should be addressed:  Agarose Gel Isoelectric Focusing-Analytic isoelectric focusing was performed in a 1% agarose slab gel using Pharmalyte, pH 4 -6.5. Focused samples were detected by protein staining with Coomassie Blue (10).

Expression and Purification of Native, E399Q/K/H/D, and K192Q/E Mutants of ALDH2
Enzymes-Recombinantly expressed native, Glu 399 , and Lys 192 mutant forms of human liver mitochondrial (class 2) ALDH were purified to homogeneity by established methods, as judged by SDS-PAGE followed by Coomassie Blue staining (2,4,8). It was estimated that all the new mutants except for K192E were expressed at levels similar to native ALDH2 based on the results of SDS-PAGE and Western blotting. Non-denatured PAGE showed that all of the mutants were still tetrameric enzymes. The isoelectric points (pI) of E399K and E399Q/H were different from that of the native enzyme. The removal of the negative charge caused an increase in the pI values from 4.9 to about 5.1 for E399K and about 5.0 for E399Q and E399H. The pI of the E399D mutant was the same as that of the native enzyme. Replacement of Lys 192 by glutamine decreased the pI value to 4.8. The fact that pI changed when the two ribose-binding residues were mutated shows that they are exposed to the solvent in the absence of NAD ϩ .
Dehydrogenase Activities of the Glu 399 and Lys 192 Mutants-The accompanying study showed that the E399Q mutant had approximately 10% of the dehydrogenase activity while the K192Q mutant had about 20% activity of the native ALDH2 (2). Converting Glu 399 to aspartate produced an enzyme that had 50% activity of the native ALDH2, while the E399H mutant had about 5% of the dehydrogenase activity. The E399K mutant had a k cat for dehydrogenase activity less than 1% of the native ALDH2 enzyme (Table I). This is similar to what was found with the E487K Oriental variant of human ALDH2 (9), and verified that the negative charge was important for the amino acid residue at position 399. Similarly, the positive charge at position 192 was also important; the K192Q mutant had about 20% activity of the native ALDH2, whereas the K192E mutant had less than 2% activity.
The K m values for NAD ϩ and propionaldehyde were determined for these mutant enzymes. There was up to a 10-fold increase in the K m for NAD ϩ for the Glu 399 mutants, but the K m values for propionaldehyde were approximately the same as for the native enzyme. In contrast, the K m for both NAD ϩ and propionaldehyde were altered for the K192Q mutant; there was a 100-fold increase in the K m for NAD ϩ and a 7-fold increase for propionaldehyde. Unexpected, the K m values for NAD ϩ and propionaldehyde for the K192E mutant were more similar to that of the native enzyme, although the k cat was decreased.

Rate-limiting Step Differed When Aliphatic and Aromatic
Aldehydes Were Used as Substrates for the E399Q/K Mutants-It was shown that there was a pre-steady state burst of NADH formation with human liver mitochondrial ALDH; the burst magnitude was approximately 2 mol of NADH/mol of tetrameric enzyme (4). This implied that deacylation (k 7 ) or NADH dissociation (k 9 ) was the rate-limiting step for the native ALDH2 enzyme (see Fig. 1 in Ref. 2). Other data proved that deacylation (k 7 ) was the rate-limiting step for the native ALDH2 enzyme (11)(12)(13). Similar to what was found with the E399Q mutant (2), no pre-steady state burst of NADH formation was observed with the E399K enzyme when propionaldehyde was used as the substrate (Table II). Furthermore, the E399K mutant oxidized chloroacetaldehyde slower than propionaldehyde (Table III), indicating that the rate-limiting step changed. Since the chloro group makes oxidation more difficult to occur, it implied that the rate-limiting step for the E399K became hydride transfer (k 5 ), as with the E399Q mutant.
To study the rate-limiting step for the mutants E399Q/K, the primary isotope effect on the aldehyde oxidation was determined (Table IV). No primary isotope effect was found for the native ALDH2 enzyme with either ␣-[ 2 H]benzaldehyde or [ 2 H]acetaldehyde, consistent with hydride transfer not being the rate-limiting step for the native enzyme, and also consistent with what we found for the horse liver mitochondrial ALDH (11,12). A V H /V D of 2.5, however, was observed for the E399Q and E399K mutant enzymes, when [ 2 H]acetaldehyde was used as the substrate, consistent with hydride transfer (k 5 ) being at least a partial rate-limiting step.
In contrast to what we found with [ 2 H]acetaldehyde, no primary isotope effect was observed when ␣-[ 2 H]benzaldehyde was used as the substrate. This indicated that the rate-limiting step was not hydride transfer (k 5 ) when benzaldehyde was used as the substrate for the E399Q/K mutants. This was verified by finding a pre-steady state burst when benzaldehyde was the substrate for the E399Q/K mutant enzymes (Table II). Furthermore, the k cat of aromatic aldehyde oxidation was related to the structure of the substrate. p-Nitrobenzaldehyde, an aromatic aldehyde having an electron-withdrawing group, was oxidized much faster than those lacking the group (Table V). This indicated that the rate-limiting step for aromatic substrates was still deacylation (k 7 ) for the E399Q/K mutant enzymes, as with the native ALDH2 enzyme.

Rate-limiting
Step Changed for the K192Q Mutant when Aliphatic and Aromatic Aldehydes Were Used as Substrates-No pre-steady state burst of NADH formation was found with the K192Q enzyme when propionaldehyde was used as the substrate (Table II). We showed that the K192Q mutant oxidized chloroacetaldehyde slower than propionaldehyde (2), indicating that the rate-limiting step changed. Since the chloro group makes oxidation more difficult to occur, it implied that  the rate-limiting step became hydride transfer (k 5 ), which was now verified by the pre-steady state burst data (Table II).
A V H /V D of 2.0 was observed for the K192Q mutant enzyme (Table IV), when [ 2 H]acetaldehyde was used as the substrate, consistent with hydride transfer (k 5 ) being involved in the rate-limiting step. In contrast, when ␣-[ 2 H]benzaldehyde was used as the substrate for the K192Q mutant, no primary isotope effect was observed. This indicated that the rate-limiting step was not hydride transfer (k 5 ) when benzaldehyde was used as the substrate. Only a small pre-steady state burst of NADH formation was observed during the oxidation of benzaldehyde by the K192Q mutant enzyme (Table II). The k cat of aromatic aldehyde oxidation was related to the structure of the substrate. p-Nitrobenzaldehyde was oxidized much faster than the others (Table V). This indicates that the rate-limiting step was not hydride transfer (k 5 ) for the K192Q mutant when aromatic aldehydes were used as substrates. Thus, the rate-limiting step differed for the K192Q mutant when aliphatic or aromatic aldehydes were oxidized.
The E399H/D Mutants of Human Mitochondrial ALDH2-The E399H/D mutants oxidized chloroacetaldehyde faster than propionaldehyde (Table III). No isotope effects were found with the E399H/D mutants when using either ␣-[ 2 H]benzaldehyde or [ 2 H]acetaldehyde as substrates, showing that hydride transfer was not the rate-limiting step for the E399H/D mutants when either aliphatic or aromatic aldehydes were used as substrates (Table IV). Furthermore, a pre-steady state burst of NADH formation was observed for the E399H/D mutants during the oxidation of both propionaldehyde and benzaldehyde (Table II), and the k cat was a function of substrates (Table V). These results indicated that the rate-limiting step for the E399H/D mutants was still deacylation (k 7 ).
E399H had many properties that mimic those of the native enzyme. The K m for NAD ϩ increased 8-fold in the E399H mutant, and the k cat decreased to 5% of the native enzyme. Although the k cat of E399H was lower than the k cat of E399Q, the rate-limiting step remained deacylation (k 7 ) for E399H.
Aspartate could replace glutamate at position 399. Both K m for NAD ϩ and k cat were native-like. This indicated that a negative charge was important for the proper function of the nicotinamide ribose-binding residue at position 399.
Effect of Magnesium Ions on the k cat of the Native, Glu 399 , and Lys 192 Mutant Forms of ALDH2-It has been reported that the mammalian liver mitochondrial ALDHs can be activated by magnesium ions (9, 14 -17). Native ALDH2 and the E399H/D enzymes could be activated approximately 2-fold by Mg 2ϩ ions when using either propionaldehyde or benzaldehyde as the substrate. In contrast, the E399Q/K and K192Q mutants were not activated by Mg 2ϩ ions when propionaldehyde was used as the substrate. They could be activated 2-5-fold by Mg 2ϩ ions when benzaldehyde was used as the substrate. DISCUSSION The structure of mitochondrial ALDH reveals that Lys 192 binds to the adenosine ribose and that Glu 399 binds to the nicotinamide ribose. When the Mg 2ϩ ions were present at low concentration in the crystals, a disordering of the nicotinamide portion of the coenzyme was observed (1). ALDH is active in the absence of Mg 2ϩ ions, but the presence of Mg 2ϩ ions causes the enzyme to have an increased specific activity and the presteady state burst magnitude of NADH formation, at least for the enzyme from horse (14) and rat (9), increases from 2 mol of NADH/mol of enzyme to 4. Here we found that only the E399H or E399D mutants were affected by the addition of Mg 2ϩ ions, analogous to what is found with the native human enzyme. These two mutants had the most native-like behavior in that the rate-limiting step remained deacylation (k 7 ).
A glutamate residue can replace the function of the lysine at position 192; the K m for NAD ϩ and the rate-limiting step for the K192E were native-like. The specific activity of K192E was depressed, compared with the native enzyme or even the glutamine mutation. Since residue 192 does not appear to be in direct contact with any residue in the substrate binding domain or near other components of the active site, as shown in Fig. 1 of the accompanying paper (2), we can only assume that the affect on k cat was due to a transitional state stabilization such that hydride transfer is not functioning efficiently when the adenosine ribose is not bound by the lysine residue. For the esterase reaction of the enzyme, where hydride transfer is not involved, the K192Q mutant was shown to have 50% the activity of the native enzyme, compared with 19% of the dehydrogenase activity (2). Finding that the esterase activity was diminished shows that residue 192 does indeed influence the catalytic site even in the absence of NAD ϩ .
In the accompanying paper (2), we proposed that mutations to the two ribose-binding residues, Lys 192 and Glu 399 , caused the rate-limiting step to change to hydride transfer. Here we show that when [ 2 H]acetaldehyde was the substrate, a primary isotope effect was found with K192Q and E399Q (Table IV). Unlike native enzyme, where a pre-steady state burst of NADH was found, neither of the two mutants produced a burst, consistent with the other data presented, which indicated that the rate-limiting step did change. A most unexpected result was found when [ 2 H]␣-benzaldehyde was employed as the substrate. No primary isotope effect was found for the 192 and 399 mutants. This implies that the rate-limiting step with aromatic aldehydes remained deacylation. Finding that there was a presteady state burst and that p-nitrobenzaldehyde was oxidized more rapidly than was benzaldehyde is consistent with this conclusion.
It is not possible to unequivocally explain why for the two mutant enzymes did the rate-limiting step become hydride transfer with aliphatic aldehydes but remain deacylation with aromatic aldehydes. Cys 302 is located 7.2 Å from Glu 399 and 17 Å from Lys 192 . Thus, it is not possible for these two residues to interact directly with the substrate being oxidized, but their presence is clearly transmitted to the region surrounding the substrate pocket. Hydride transfer from aliphatic substrates appears to be more sensitive to the precise positioning of NAD ϩ , which is held in place by Lys 192 and Glu 399 . The transition state for hydride transfer from aromatic aldehydes could be partially stabilized by the presence of the ⌸ electron cloud. If so, it could explain why the rate-limiting step differs.
The E399K mutant was prepared since a lysine functions at position 192. Unlike with E399Q, the K m for NAD ϩ increased 12-fold and the k cat decreased to 0.5% of the native enzyme. Thus, the ligand to the nicotinamide ribose can not be lysine. Aspartate could function well at that position; the E399D mu- tant had native-like properties. The corresponding histidine mutant had properties more like the glutamine mutant except that the rate-limiting step did not change to hydride transfer. The burst magnitude was lower than that of the native enzyme, at 1.1 mol of NADH/mol of enzyme compared with 1.6 -2.0 for the native enzyme.
Based on the mechanism of ALDH, the burst magnitude ⌸ can be related to the ratio of the deacylation step (k 7 ) to one of the steps leading to it (k 5 or k 3 ).
⌸ ϭ E T / ͓1 ϩ ͑k 7 /k f ͔͒ 2 (Eq. 1) k f is either k 5 or k 3 , and E T is the concentration of active sites. The ratio of k 7 /k f would have to be between less than 0.05-0.1 to obtain a burst magnitude of 1.6 -2.0. To obtained a burst magnitude of 1.1, as found with the E399H enzyme, the ratio would have to be around 0.4. Although the individual rate constants were not measured, it was possible to determine whether k f was governed by a change in k 5 or k 3 . For the E399H mutant, no primary isotope effect was observed, showing that k 7 was still the rate-limiting step (k cat in Table I). The ratio k cat /K m(propionaldehyde) is k 3 (3). From the data in Table I, it can be estimated that k 3 decreased from 380 to 16 M Ϫ1 min Ϫ1 in the E399H mutant, only a factor of 20. From the ratios of rate constants (k 7 /k f ) and the k cat values from Table I, it can be estimated that k f would be between 1900 and 3800 min Ϫ1 for the native and just 25 min Ϫ1 for the E399H mutant, a factor of 75-150. Since the factor of 20 in k 3 is far less than the factor of 75-150 in k f , it appears that k f is not governed by k 3 , but must be related to a change in k 5 . Thus it seems that any mutation to Glu 399 causes a change to occur in the hydride transfer step (k 5 ). For the Gln and Lys mutants, the decrease was sufficient to make k 5 become the rate-limiting step, while for the His mutant, it was not.
It was found that the aliphatic aldehydes were better substrates than were aromatic aldehydes for the native ALDH2 enzyme. However, K192Q oxidized aromatic aldehydes faster than aliphatic aldehydes. In fact, the specific activity when aromatic aldehydes were used as substrates was greater than that of the native enzyme. p-Nitrobenzaldehyde was oxidized approximately 4 times faster by the K192Q mutant than by the native enzyme. It appears that an aromatic-specific ALDH should have a glutamine binding to the adenosine ribose, but no ALDH was reported to have one.
Addition of Mg 2ϩ ions increases the activity of the mammalian ALDH2 by approximately 2-fold (9, 14 -17). The stoichiometry of NADH binding remained 2 for the human ALDH2 in the presence of Mg 2ϩ ions (data not shown), but it was increased to 4 for horse, beef, and rat ALDH2 (9,17,18). This suggested that for the human ALDH2 Mg 2ϩ ions affected the rate-limiting step (k 7 ), while it increased the number of active site for the others. Consistent with the statement is the fact that Mg 2ϩ ions only increased the activity of the E399Q/K and K192Q mutants when aromatic aldehydes were oxidized, where the rate-limiting step was still k 7 .
Between the data presented in this and the accompanying paper (2), we show that, although many residues were completed conserved in the ALDH family of enzymes, only those at position 302 and 268 appear completely essential while those at 399 and 192 are important for hydride transfer. The first two are involved chemically, where the later most likely are involved in stabilizing the transition state, resulting from their binding to the ribose ring in NAD ϩ .

TABLE V
The dehydrogenase activities for the native and mutant forms of human liver mitochondrial aldehyde dehydrogenase (ALDH2) when different aldehydes were used as substrates The unit of activity is n mol/min/mg protein.