R. Pietruszko, Center of Alcohol Studies, Rutgers, The State University of New Jersey, 607 Allison Road, Piscataway, New Jersey 08854-8001, USA. Fax: + 1 732 4453500. Tel.: +1 732 4453643. E-mail: email@example.com
Low concentrations of citral (3,7-dimethyl-2,6-octadienal), an inhibitor of retinoic acid biosynthesis, inhibited E1, E2 and E3 isozymes of human aldehyde dehydrogenase (EC126.96.36.199). The inhibition was reversible on dilution and upon long incubation in the presence of NAD+; it occurred with simultaneous formation of NADH and of geranic acid. Thus, citral is an inhibitor and also a substrate. Km values for citral were 4 µm for E1, 1 µm for E2 and 0.1 µm for E3; Vmax values were highest for E1 (73 nmol·min−1·mg−1), intermediate for E2 (17 nmol·min−1·mg−1) and lowest (0.07 nmol·min−1·mg−1) for the E3 isozyme. Citral is a 1 : 2 mixture of isomers: cis isomer neral and trans isomer, geranial; the latter structurally resembles physiologically important retinoids. Both were utilized by all three isozymes; a preference for the trans isomer, geranial, was observed by HPLC and by enzyme kinetics. With the E1 isozyme, both geranial and neral, and with the E2 isozyme, only neral obeyed Michaelis–Menten kinetics. With the E2 isozyme and geranial sigmoidal saturation curves were observed with S0.5 of ≈ 50 nm; the n-values of 2–2.5 indicated positive cooperativity. Geranial was a better substrate and a better inhibitor than neral. The low Vmax, which appeared to be controlled by either the slow formation, or decomposition via the hydride transfer, of the thiohemiacetal reaction intermediate, makes citral an excellent inhibitor whose selectivity is enhanced by low Km values. The Vmax for citral with the E1 isozyme was higher than those of the E2 and E3 isozymes which explains its fast recovery following inhibition by citral and suggests that E1 may be the enzyme involved in vivo citral metabolism.
NAD+-linked aldehyde dehydrogenase (EC188.8.131.52) with broad substrate specificity and activity with short chain aliphatic aldehydes, catalyzes both the irreversible dehydrogenation of aldehydes and hydrolysis of esters. In both cases the reaction mechanism involves a covalent acyl-enzyme intermediate formed between substrates and the Cys302 residue of aldehyde dehydrogenase (Fig. 1) [1,2]. The role of Cys302 as the catalytic residue has been confirmed recently by X-ray crystallography [3,4] and by site-directed mutagenesis . It is generally accepted that the carbonyl carbon of the aldehyde is attacked by the sulfhydryl group of the cysteine to form a thiohemiacetal tetrahedral intermediate (II, Fig. 1). Upon hydride transfer, an acyl-enzyme intermediate, a thioester (III, Fig. 1) is formed which subsequently undergoes hydrolysis to form an acid product. The enzyme occurs in several molecular forms, all homotetramers of ≈ 220 kDa which exhibit differences in primary structure and substrate specificity. Human aldehyde dehydrogenase, which has high affinity for short chain aliphatic aldehydes, occurs as three known isozymes, E1, E2, and E3; E1 and E3 are cytoplasmic and E2 is mitochondrial. The human genes encoding the isozymes used during this investigation are known as ALDH1 (coding for E1), ALDH2 (coding for E2), and ALDH9 (coding for E3), respectively (Human Genbank Database). All three isozymes were purified to homogeneity in our laboratory [6,7] and were shown to have low activity with all-trans-and 13-cis-retinaldehyde .
Citral (3,7-dimethyl-2,6-octadienal), a monoterpene, is a reactive and volatile α,β-unsaturated aldehyde that occurs naturally in herbs, plants and citrus fruits. Due to its intense lemon aroma and flavor, citral is used widely as a food, cosmetics and detergent additive. Both natural and synthetic citral occurs as a mixture of the geometric isomers, geranial (trans-3,7-dimethyl-2,6-octadienal) and neral (cis-3,7-dimethyl-2,6-octadienal). The chemical structures of the citral isomers are shown in Fig. 2.
Geranial, the trans form of citral bears structural resemblance to all-trans retinaldehyde (Fig. 2). In mammals, all-trans-retinaldehyde is formed from β-carotene by central or excentric cleavage [9–11] and is subsequently oxidized to retinoic acid or reduced to retinol. Retinoic acid has been recognized only recently as a major hormone of differentiation and development . In rat conceptal homogenates, low concentrations of citral were found to effectively inhibit retinoic acid formation . Although citral has been reported to inhibit the mitochondrial isozyme of rat aldehyde dehydrogenase , neither its effect on the other isozymes nor its mechanism of inhibition have been investigated previously. During this investigation, citral was found to inhibit all three enzymes effectively. The results led to the determination of the mechanism of inhibition of aldehyde dehydrogenase by citral as described in this paper.
Materials and methods
Citral, geraniol, nerol, glycolaldehyde, horse liver alcohol dehydrogenase, and NADH were obtained from Sigma Chemical Co. Geranic acid was from Bedoukian Research, Inc. (Danbury, CT, USA). Propionaldehyde, γ-aminobutyraldehyde diethyl acetal, and trifluoroacetic acid were obtained from Aldrich. Propionaldehyde and acetaldehyde were redistilled before use. γ-Aminobutyraldehyde was prepared, immediately before use, by acid hydrolysis of the fractionally redistilled acetal . Acetonitrile (HPLC grade) was from Fisher Scientific. NAD+ (grade 1) was from Boehringer Mannheim. All other compounds were reagent grade.
The E1, E2, and E3 isozymes were purified from human liver as described previously [7,16], and stored in 30% glycerol at 4 °C, under nitrogen. Before use, the enzymes were dialyzed extensively against nitrogen saturated 30 mm sodium phosphate buffer, pH 7.0, 1 mm EDTA, to remove glycerol and 2-mercaptoethanol. Protein concentration was determined by a microbiuret method , using BSA as a standard. Protein was also assayed spectrophotometrically at 280 nm, using an extinction coefficient of 1.0 (mg·mL−1·cm−1) [6,7].
E1, E2, and E3 isozymes were assayed in 0.1 m sodium pyrophosphate buffer, pH 9.0, containing 1 mm EDTA, 500 µm NAD+ and 1 mm propionaldehyde. In these conditions 0.6, 1.6, and 0.6 µmol NADH·min−1·mg−1 are produced by fully active E1, E2, and E3 isozymes, respectively [6,7]. The E3 isozyme was also assayed in 0.1 m sodium phosphate buffer, pH 7.4, containing 1 mm EDTA, 500 µm NAD+, and 100 µmγ-aminobutyraldehyde. Reactions were started by the addition of enzyme and followed spectrophotometrically at 340 nm and 25 °C in cuvettes of 1 cm light path. An extinction coefficient of 6.22 mm−1·cm−1 for NADH was used for the calculation of reaction rates.
Steady state kinetics
Citral (10 mm) stock solutions were prepared in ethanol, as were the appropriate citral dilutions and added to a 3-mL cuvette in 10 µL of ethanol. Geranial and neral were added directly to buffer from HPLC separation. For use in kinetics the concentrations of HPLC separated geranial and neral were determined by dehydrogenation to completion with the E1 isozyme and determination of total NADH formed. The reaction was started by addition of varied substrate, following 4 min preincubation of the enzymes with inhibitors. Reaction rates were determined by tangents to steady state velocities. For determination of Michaelis constants reactions were initiated by the addition of enzyme. Both the single reaction curve method of Yun and Suelter  and the Lineweaver–Burk method  were used. All kinetic parameters were obtained using the HYPER Program of Cleland , using the nonlinear regression fit of rates vs. substrate concentration. Fluorescence measurements were made using a Perkin-Elmer MPF-2A Fluorescence Spectrophotometer, and the instrument was calibrated for each experiment with stock solutions of NADH.
A Varian recording spectrophotometer (model 635) with a double beam was used at 340 nm. The reactions were carried out in 1 mL total volume, in cuvettes of 1 cm light path at 25 °C and in 0.1 m sodium phosphate buffer (pH 7.4) containing 1 mm EDTA. The reactions were initiated by the addition of citral, geranial, or neral to both experimental and reference cuvettes. Steady state velocity was extrapolated to 0 time to determine the ‘burst’ amplitude.
Identification of reaction products and of citral isomers
Citral (500 µm) was incubated for 12–21 h at 25 °C in 0.05 m sodium phosphate buffer (pH 7.4) containing 1 mm EDTA and 500 µm NAD+ with E1 (0.12 µm), E2 (0.38 µm), or E3 (0.57 µm) isozymes. Two controls were used: (a) as above with no enzyme; (b) as above with no NAD+. Before HPLC analysis, the enzymes were removed from the mixtures by centrifugation using 0.45 µm UltraFuge filters (Micron Separations, Inc., Westborough, MA, USA). Two hundred µL of each filtrate were then injected into an HPLC column for the identification of product.
For synthesis of citral isomers, the reactions were carried out in 0.1 m sodium phosphate buffer, pH 7.4, containing 1 mm EDTA, 500 µm NAD+ and 1.9 mm of either geraniol (trans-3,7-dimethyl-2,6-octadien-1-ol) or nerol (cis-3,7-dimethyl-2,6-octadien-1-ol). Each reaction was initiated by the addition of alcohol dehydrogenase, incubated overnight at 25 °C and analyzed by HPLC after filtration.
A 5 µ Supelco C18 reversed phase column (0.46 cm × 25 cm) was used with a Waters 501 HPLC system. A single solvent system consisting of solvent A [aqueous 0.1% trifluoroacetic acid (v/v)] and solvent B (100% acetonitrile) was used for all separations. A 10-min linear gradient from 0% B to 40%B was followed by a 30-min linear gradient to 41% B, at a flow rate of 1 mL min−1. The separation of alcohols, aldehydes, and their acid products was followed at 219 nm. For isolation of geranial and neral on a larger scale, an isocratic elution with 0.1% trifluoroacetic acid/acetonitrile (3 : 1) was used with the same column, and a flow rate of 2.5 mL min−1.
Inhibition of aldehyde dehydrogenase isozymes by citral
At Km concentrations of glycolaldehyde (325 µm for E1, 50 µm for E2 and 220 µm for E3 isozyme), citral (10 µm) totally abolished E3 isozyme activity and inhibited the E2 and E1 isozymes to 10% and 20% remaining activity, respectively. The reversibility of inhibition was tested by saturation kinetics . The E2 isozyme (1 mg·mL−1) was incubated with and without 10 µm citral at 25 °C in 0.1 m sodium phosphate buffer, pH 7.4, containing 1 mm EDTA and 500 µm NAD+. At various time intervals, 10-µL aliquots were withdrawn and assayed for enzyme activity as described: activity equal to that of the control was recovered each time. The time courses, however, exhibited a lag of several minutes duration, which was absent from controls which were not incubated with citral. These experiments showed that the recovery from inhibition was not instantaneous and occurred with lags, and that the inhibition was reversible and did not occur by formation of covalent bonds.
When the E1 and E2 isozymes inhibited by citral were incubated in the presence of NAD+, activity with glycolaldehyde was regained. The duration of incubation with NAD+ for activity regain was dependent on the citral concentration; at 10 µm citral, it took 3 h for E2 isozyme to regain activity, whereas only 12–15 min were required at 0.5 µm citral. Activity regain was also dependent on the amount of enzyme used: the more enzyme the faster the regain. The activity regain was also dependent on the isozyme used: the regain was fastest with the E1 isozyme, followed by E2 isozyme, an incubation of several days was required for E3 isozyme to regain activity.
Formation of NADH from citral in the presence of E1 isozyme
Using the E1 isozyme, which was not totally inhibited by citral, it was found that in the presence of NAD+, the E1 isozyme formed NADH from citral and that the amount of NADH formed was stoichiometrically equivalent to the amount of citral present in the incubation mixture.
HPLC analysis of incubation mixtures of E1, E2 and E3 isozymes with citral
In order to determine if citral was a substrate, HPLC analysis was used. A 1 : 1 mixture of the commercial citral and geranic acid is shown in Fig. 3A. Citral is a mixture of two isomers (the trans-isomer, geranial, and the cis-isomer, neral) which could be separated by HPLC. The retention times of the two resulting peaks were 22.5 min and 24.4 min (Fig. 3A, peaks 3 and 4). Commercial geranic acid also migrated through the C18 reversed phase column as two peaks with retention times of 19.7 and 20.7 min (Fig. 3A, peaks 1 and 2). It can be seen that the citral isomers were separable from each other and from geranic acid.
Reaction assay mixtures containing 500 µm NAD+ and 500 µm citral, which had been incubated in the presence or absence of E1, E2, and E3 isozymes at 25 °C, were then analyzed by HPLC. NAD+ and NADH were found to migrate well ahead of the citral and geranic acid peaks, both with a retention time of ≈ 7.2 min. As shown by Fig. 3, a new peak with a retention time of 20.7 min, which corresponded to the major peak of commercial geranic acid, appeared in the presence of the E1 isozyme after a 12-h incubation (peak 2, Fig. 3B); this peak was not seen in the control reaction which did not contain the E1 enzyme. It was also absent from the control containing the enzyme but no NAD+. The same new peak was also observed in the presence of the E2 and E3 isozymes, after 12- and 24-h incubations, respectively, with citral (peak 2, Fig. 3C and D). The minor peak (retention time 19.7 min) of the commercial geranic acid was never observed as a product of enzymic oxidation of citral and was probably an impurity in commercial geranic acid.
Stereospecificity of citral dehydrogenation
Concomitant with the appearance of the geranic acid peak, the disappearance of both citral isomer peaks was observed (compare peaks 3 and 4 in Fig. 3A, B, C and D). In all cases, the citral isomer with retention time of 24.4 min (peak 4, Fig. 3) disappeared faster than the isomer with retention time of 22.5 min (peak 3, Fig. 3). The isomers could not be identified directly by HPLC because their reference standards, geranial and neral, were not commercially available. The identification of the citral isomer peaks was attempted enzymatically using horse liver alcohol dehydrogenase. The alcohol precursors of the citral isomers, geraniol (trans) and nerol (cis), which are available commercially, were used. The cis-isomer, nerol (peak 1, Fig. 4A) had a retention time of 20.9 min. The trans isomer, geraniol (peak 2, Fig. 4B) had a retention time of 21.6 min. Incubation of nerol with alcohol dehydrogenase resulted in the appearance of a new peak with a retention time of 22.5 min (peak 3, Fig. 4C), whereas incubation of the trans-isomer, geraniol, with alcohol dehydrogenase resulted in formation of a new peak with retention time of 24.4 min (peak 4, Fig. 4D). The citral peak with a retention time of 22.5 min has been identified as neral and that with the retention time of 24.4 min as geranial. All three aldehyde dehydrogenase isozymes oxidized citral, with preferential utilization of the trans-isomer, geranial.
Dehydrogenation of citral, neral and geranial by the E1 isozyme
Because of measurable activity of the E1 isozyme with these substrates, the direct procedure of Yun and Suelter  could be used for determination of Km and Vmax values (Table 1). Vmax values were also determined from steady state velocities at saturating concentrations of the substrates (see Table 1) after establishing that there was no substrate inhibition. The Km values of geranial were lower than those for neral, whereas Vmax were similar.
Table 1. KmandVmaxvalues for citral, geranial, and neral with E1 isozyme.
Determined in sodium phosphate buffer (0.1 m), pH 7.4, containing 1 mm EDTA, and 2.3 mm NAD+, at 0–4 µm citral, 0–10 µm neral, and 0–0.14 µm geranial. a The mean Vmax value for all substrates was determined for a total of six experiments, three of which were from Km determinations by the procedure of Yun and Suelter (YS) and three of which were steady state velocity determinations at 150 µm citral, 118 µm geranial and 120 µm neral. U, UV/VIS spectrophotometric determination (with λ at 340 nm); F, fluorescence spectrophotometric determination (λexcitation at 340 nm and λemission at 450 nm).
4.2 ± 0.4
73.2 ± 2.6
0.07 ± 0.02
63.4 ± 9.1
3.2 ± 1.0
65.5 ± 5.5
Dehydrogenation of citral, neral and geranial by the E2 isozyme
With the E2 isozyme, citral, geranial, or neral dehydrogenation was not instantaneously apparent from spectrophotometric traces obtained within a relatively short time period. However, when enzyme concentration was increased and tracing was continued longer at higher instrument sensitivity, the complete curves of substrate utilization could be obtained, allowing determination of Km and Vmax values. The highest enzyme concentrations used were ≈ 30% of Km values, at which the Michaelis–Menten kinetics are still valid . The Km values were comparable to those with E1 isozyme (compare Table 1 with Table 2) but maximal velocities were ca. four times lower. The Km for geranial appeared to be low (≈ 0.14 µm), and problems were encountered in determination of Vmax values for geranial; the Vmax values for geranial were three times higher when determined by extrapolation of Lineweaver–Burk plots than when determined directly. Hill plots of data for geranial (Fig. 5A) indicated a strong positive cooperativity, with an n-value of 2.1–2.5, precluding determination of Km values. Vmax values could be obtained by measurement of steady state velocity. There was no positive cooperativity observed with neral. Hill plots of ln v/[Vmax–v] vs. ln[S] of kinetic data for geranial and neral for the E1 isozyme (Fig. 5B) showed n-values of ≈1.0, suggesting normal kinetics.
Table 2. Km and Vmax values for citral, geranial, and neral with E2 isozyme.
Determined in sodium phosphate buffer (0.1 m), pH 7.4, containing 1 mm EDTA, and 2.3 mm NAD+, at 0–7 µm citral, 0–4 µm neral, and 0–0.10 µm geranial. a The mean Vmax value for citral was determined for a total of six experiments, three of which were from Km determinations by the procedure of Yun and Suelter (YS) and three of which were steady state velocity determinations at 200 µm citral, 76 µm neral and 110 µm geranial. b With geranial as substrate, only steady state velocity measurements were used for Vmax determination. c The mean Vmax for neral was determined from four Km measurements and four steady state velocity determinations. U, UV/VIS spectrophotometric determination (with λ at 340 nm); F, fluorescence spectrophotometric determination (λexcitation at 340 nm and λemission at 450 nm).
Direct determination of kinetic constants was impossible with the E3 isozyme because of its extremely low velocity. Determination of Ki values (which for alternate substrates is equal to Km), was therefore attempted. Citral reacted slowly with the E3 isozyme and its inhibition was delayed, necessitating starting the reaction with the varied substrate. The results showed that in addition to the expected slope effect there was also an unexpected intercept effect (data not shown, but results given in Table 3). At high enzyme concentrations (E3 at 1.3 µm), the maximal velocity of citral dehydrogenation by the E3 isozyme was determined. For this experiment, citral was used at 580 µm and NAD+ at 2 mm (in 0.1 m sodium phosphate buffer, pH 7.4). The Km value for citral, obtained from slope replots, was 80 nm and Vmax was 0.07 nmol·min−1·mg−1.
Table 3. Inhibition constants of the E3 isozyme for citral and steady state velocity at saturating citral.
Substrate (range, µm)
Type of inhibition
Vmax of 0.07 µmol·min−1·mg−1 was determined by direct measurement at 580 µm citral and 1.3 µm E3 isozyme. Inhibition constants were obtained from three separate experiments in sodium phosphate buffer (0.1 m, pH 7.4) containing 1 mm EDTA, 500 µm NAD and 0.03–0.04 µm E3 isozyme. In these conditions the Km values for E3 isozyme for glycolaldehyde and γ-aminobutyraldehyde were 220 µm and 4 µm, respectively.
Inhibition of the E2 isozyme with citral neral and geranial
Citral also reacted slowly with the E2 isozyme in the presence of propionaldehyde or glycolaldehyde necessitating starting the reaction with varied substrates. Citral inhibited the E2 isozyme in the same way as that observed with the E3 isozyme (see above) producing both slope and intercept effects which could be easily visualized with either glycolaldehyde (Km = 50 µm) or propionaldehyde (Km = 0.8 µm) as the varied substrates (Fig. 6). Noncompetitive inhibition patterns with pronounced intercept effects were also easily demonstrable with neral as inhibitor (data not shown as a figure but are given in Table 4). Fig. 7 shows Dixon plots of the inhibition of the E2 isozyme with neral and geranial. As the slope replot for the E2 isozyme with geranial was hyperbolic (see inset, Fig. 7), the same data were also plotted vs. varied propionaldehyde (not shown). With geranial as the inhibitor, both the slope and intercept effects were produced; this was, however, not totally reproducible in repeated experiments where patterns intersecting in the first quadrant were also obtained. The slope replot for geranial was also hyperbolic, thus, the Ki(slope) for geranial (Table 4) is an approximate value, taken from the Dixon plot in Fig. 7. This Ki(slope) for geranial was lower than that of neral. Geranial inhibited the E2 isozyme o a greater extent than did neral; at 17 µm propionaldehyde, neral (20 µm) inhibited the control activity by 26%, whereas geranial (20 µm) inhibited it by 89%.
Table 4. Inhibition constants of the E2 isozyme for geranial and neral.
Type of inhibition
Ki (slope) (µm)
Ki (int.) (µm)
All experiments were carried out using sodium phosphate buffer (0.1 m, pH 7.4) containing 1 mm EDTA and 2.3 mm NAD+, with the E2 isozyme at concentrations of 9.8–11.7 nm. Nonlinear regression was used to analyze the primary data. The Km value for the E2 isozyme for propanal is 0.8 µM.
a An approximate estimate from Fig. 7. ND, Not determined.
No substrate inhibition was observed up to 150 µm citral with the E1 isozyme or up to 200 µm citral with the E2 isozyme. With 0.5 mm geranic acid, however, the activity of the E2 isozyme with glycolaldehyde (100 µm) was inhibited by 28%, and with 2.5 mm geranic acid, it was inhibited by 74%.
Pre-steady state ‘burst’
When geranial or neral were used at concentrations of 0.11–0.14 mm in two separate experiments, at E2 isozyme concentrations of 4.1–4.6 µm and with NAD+ at 1.0 mm, using a spectrophotometer scale on which one turnover could be easily seen, no ‘burst’ was observed with either substrate. Steady state velocity, similar to Vmax obtained in other kinetic experiments, was easily measured and extrapolated to determine zero time burst amplitude.
The rates of turnover of citral by E1, E2 and E3 isozymes
Maximal velocities of citral dehydrogenation for E1 and E2 and E3 isozymes (from Tables 1, 2 and 3) are listed in Table 5 and are used for calculation of the turnover numbers for citral. As shown in Table 5, turnover numbers expressed per min decreased rapidly from E1 to E3 isozyme. The reciprocals of turnover numbers show the time necessary for one molecule of citral to be dehydrogenated by one active site. This time varied from 0.125 min for E1 isozyme to 0.53 min for E2 isozyme to 100 min for E3 isozyme.
Table 5. Turnover numbers of citral dehydrogenation by the E1, E2, and E3 isozymes.
a Data from Table 1: equivalent weight of the E1 isozyme is 109 420 kDa. b Data from Table 2: equivalent weight of the E2 isozyme is 108 310 kDa. c Data from Table 3: equivalent weight of the E2 isozyme was also used for the calculation of the turnover number of the E3 isozyme.
All three isozymes of human aldehyde dehydrogenase were inhibited by low concentrations of citral. α,β-Unsaturated aldehydes such as acrolein, or α,β unsaturated ketones such as (Z)-hexadeca-1,11-dien-3-one, which have a double or triple bond adjacent to the carbonyl group, are highly reactive and have been found to react irreversibly with aldehyde dehydrogenase by forming covalent bonds with enzyme sulfhydryl groups [2,21,25]. Saturation kinetics with citral demonstrated that the inhibition was reversible and that total activity was regained on dilution after incubation of E2 isozyme with citral .
Regain of enzyme activity upon long incubation in the presence of NAD+ (see Results), and NADH formation from citral suggested that citral was a substrate. Identification of the reaction products following incubation of all three isozymes with citral, showing formation of geranic acid (Fig. 3), established that citral was a substrate. Of the two citral isomers, geranial appeared to be metabolized preferentially by all three isozymes. In order to better characterize citral as a substrate, determination of Michaelis constants was attempted (Tables 1 and 2). This was relatively easy with the E1 isozyme but considerably more difficult with the E2 isozyme because of its much lower velocity with citral. With the E3 isozyme the reaction velocity was extremely low and almost impossible to measure. Because of this, an attempt was made to characterize citral as substrate for the E3 isozyme by determination of inhibition constants (Table 3). The Ki(slope) of an alternate substrate used as an inhibitor, equals Km of an alternate substrate .
The Km values for all three isozymes for citral and its component isomers were low (Tables 1, 2 and 3), similar in magnitude for all three enzymes, and resembled those of the best substrates of aldehyde dehydrogenase. The Vmax value was highest for the E1 isozyme and represented ≈ 20% of Vmax with glycolaldehyde in the same conditions. The Vmax for the E2 isozyme was ≈ 4 times lower and represented only 0.08% of its Vmax with glycolaldehyde; that for the E3 isozyme was very small and represented only ≈ 0.0007% of its Vmax with glycolaldehyde. Because of its low Vmax citral is a poor substrate. The low Vmax combined with the low Km makes it an excellent and a selective inhibitor because it is recognized by the isozymes at low concentrations at which it is able to compete with good substrates. The duration of inhibition was dependent on the rate of metabolism. Although only 0.125 min was required to process one molecule of citral via the E1 isozyme, 100 min were required via the E3 isozyme (Table 5). Due to its highest Vmax, the E1 isozyme recovered the fastest of all three human isozymes from inhibition by citral.
Kinetics with geranial and neral established no differences in Vmax between the isomers (Tables 1 and 2). Lower Km values of E1 isozyme for geranial than those for neral and the resultant high Vmax/Km ratio explain why geranial is a preferred substrate, in agreement with results from HPLC (compare Vmax/Km values in Table 1 with Fig. 3B). With neral, the E2 isozyme appeared to follow Michaelis–Menten kinetics (Table 2); with geranial, however, it exhibited positive cooperativity with an n-value of 2–2.5 (Fig. 5). The low S0.5 (≈ 50 nm) value also explains preferential utilization of geranial by the E2 isozyme. Positive cooperativity has not been observed previously with any aldehyde dehydrogenase substrates, but mammalian retinaldehyde dehydrogenases with positive cooperativity with retinaldehyde as substrate were recently described . Geranial, which bears structural resemblance to all-trans-retinaldehyde (Fig. 2) has been also found during this investigation to be a better inhibitor of aldehyde dehydrogenase than neral (see Results).
Because of low velocity of citral and its isomers, relative to other substrates, inhibition studies could be performed with either E2 or E3 isozymes, by measuring NADH formation. The E1 isozyme was not used in these experiments because its relatively high velocity necessitated determination of geranic acid – the specific product. Because citral is an alternate substrate, competitive inhibition patterns were expected when Km values were determined for the E3 isozyme . Unexpectedly, both slope and intercept effects were observed. When kinetic studies were repeated with the E2 isozyme, intercept effects were confirmed (Fig. 6). Because an alternate substrate is always competitive with the substrate it replaces (aldehyde in this case) and is noncompetitive with other substrates (NAD+ in this case), the presence of intercepts indicated that citral inhibited in a manner that could not be completely reversed by saturation with varied aldehyde substrate and that part of the enzyme that leads to product was removed from the reaction sequence. It was at first considered probable that intercepts were produced because citral also combined with the enzyme at another point to form a dead end complex. However, complete absence of substrate inhibition when either citral or geranial or neral were used, and product inhibition occurring only at high geranic acid concentrations, argued against a dead-end complex.
Citral and its isomers constitute a novel class of dehydrogenase inhibitors, which are themselves substrates. Such inhibitors are well known in the serine protease field  where they inhibit by formation of acyl-enzyme intermediates which decompose slowly . The reaction catalyzed by aldehyde dehydrogenase also involves a covalent  acyl-enzyme intermediate (III, Fig. 1), which could deflect the enzyme from the main reaction pathway in a manner comparable to that of irreversible inhibitors. The formation of acyl-enzyme intermediate occurs during hydride transfer. If the intermediate decomposed slowly or if NADH dissociation were slow, a ‘burst’ should be observed upon mixing the enzyme with aldehyde and NAD. However, as no ‘burst’ of NADH formation occurred, neither decomposition of acyl-enzyme intermediate nor the dissociation of NADH could be the catalytic steps accounting for the slow rate of citral dehydrogenation; the slow step had to occur before or during the hydride transfer.
Formation of the enzyme–substrate complex was considered at this point. Slow reaction of citral and its component isomers with both E2 and E3 isozymes was observed during this investigation, necessitating starting the reaction with varied aldehyde substrates during kinetics. There are reasons for this that can be better understood by investigating chemistry of hemiacetal formation. Equilibrium constants for the addition of carbonyl compounds to alcohols show the same response to structural features as do hydration reactions ; the more hydrated the aldehyde the greater the ease of hemiacetal formation. 2,3-Nonsaturated aldehydes, like geranial or neral would be largely anhydrous  when compared with saturated aldehydes of similar structure and would therefore form hemiacetals more slowly than the propionaldehyde or glycolaldehyde used here as the varied substrates. Thus, varied propionaldehyde or glycolaldehyde can compete with citral at the point of citral binding leading to the ternary complex, I (Fig. 1) where substrates are not yet covalently bound. As a result of the above considerations, steps leading to formation of ternary complex, I (Fig. 1) can be excluded as accounting for intercepts in the steady state kinetics. Thus, it appears most likely that either the conversion of ternary complex I to thiohemiacetal II (Fig. 1) or the decomposition of the thiohemiacetal transition state via hydride transfer constitutes the rate limiting step of citral dehydrogenation and also the point in the reaction sequence at which another substrate cannot compete. This postulate is consistent with the observed intercept effect, complete lack of ‘burst’ and with the slow recovery from inhibition observed during saturation kinetics. Lags observed during activity regain most probably occur because the thiohemiacetal form of citral remains firmly bound to the enzyme and is slowly removed by dehydrogenation in the presence of NAD. These properties could make citral useful in X-ray crystallography to obtain more information about the transition state.
Citral is used routinely in studies aimed at inhibition of the retinoic acid formation pathway. Geranial bears a structural resemblance to trans-retinoic acid, which functions as a signaling molecule in cell growth and differentiation both during embryogenesis and in adult life : through retinoid receptors, it can alter the expression of a wide variety of genes, especially homeobox genes, which control organ topography. An excess of retinoic acid can cause cranial malformations and teratogenesis  and so the levels of retinoic acid for normal cell function have to be precisely regulated. For this reason, understanding the enzymes that control levels of retinoic acid is of prime importance. In rat conceptal homogenates , low concentrations of citral almost completely inhibited retinoic acid formation, suggesting that the enzyme(s) responsible were sensitive to low concentrations of citral. This sensitivity is demonstrated by all three isozymes of human aldehyde dehydrogenase; only the duration of inhibition is different, with E1 being inhibited for the shortest time. It was also reported previously  that enzymes involved in retinoic acid biosynthesis showed positive cooperativity with retinaldehyde as substrate. Positive cooperativity with geranial as substrate has also been observed during this investigation. On the basis of the above criteria, all three isozymes could be candidates for physiological retinaldehyde metabolism; it is possible that all three are involved in retinoic acid synthesis. Although involvement of ALDH1 in retinaldehyde metabolism has been generally accepted, there have been reports of retinaldehyde activity of ALDH2 [31,32].
Although citral has been shown to be metabolized to its monocarboxylic acid derivatives in vivo, this investigation provides the first direct evidence that citral and its isomers are substrates for aldehyde dehydrogenase. The E1 isozyme of human aldehyde dehydrogenase has a turnover number for citral that is larger than those of the E2 and E3 isozymes; it is also present in the liver at a level of ≈1 g·kg−1. For a liver of 1.4 kg and at a reaction velocity of 70 nmol·min−1·mg−1 (Table 1), this would have a capacity to metabolize citral at a rate of ≈100 µmol·min−1 or 6 mmol·h−1. These estimates are based on the assumption that the enzymes are not simultaneously dehydrogenating other xenobiotics with Km values comparable with those of citral.
Financial support of USPHS Grant 1R01 AA00186 from NIAAA is gratefully acknowledged.