0735-0414/90 J3.00 + 0.00 Pergamon Press pic © 1990 Medical Council on Alcoholism
Alcohol & Alcoholism. Vol. 25. No. 2/3. pp. 105-116.1990 Printed in Great Britain
ALCOHOL AND ALDEHYDE DEHYDROGENASE* TORSTEN EHRIG, WILLIAM F. BOSRON and TING-KAI Lit Departments of Biochemistry and Medicine, Indiana University School of Medicine, Indianapolis, IN 46202, U.S.A. (Received 23 January 1990) Abstract — The enzymes mainly responsible for ethanol degradation in humans are liver alcohol dehydrogenases (ADH) and aldehyde dehydrogenases (ALDH). Polymorphisms occur in both enzymes, with marked differences in the steady-state kinetic constants. The /fm-values for ethanol of ADH isoenzymes relevant for alcohol degradation range from 49 uM to 36 jiM, and the Kma,-values from 0.6 to 10 U/mg. Expression of an inactive form of the ALDH2 isoenzyme, the so-called Oriental variant, results in impaired acetaldehyde metabolizing capacity. The differences in ethanol and acetaldehyde metabolizing activities of allelic enzyme forms may be responsible in part for the large variation in the alcohol metabolism rate in humans. Interindividual differences in the isoenzyme pattern may contribute to the genetically determined predisposition for excessive alcohol intake.
this notion, twin studies have shown that the alcohol elimination rate is, in part, genetically The conversion of ethanol to acetate in the controlled (Kopun and Propping 1977; Martin liver is catalyzed mainly by NAD+-dependent etal., 1985). However, it has not been possible alcohol dehydrogenases (ADH; EC 1.1.1.1) to relate the alcohol elimination rate directly to and aldehyde dehydrogenases (ALDH; EC the isoenzyme pattern in single individuals, 1.2.1.3), whereas other pathways, namely the since the pattern could in the past only be microsomal ethanol-oxidizing system (MEOS) reliably determined from liver autopsy or and catalase, play a minor role. These enzyme biopsy samples. More recently, the developsystems are mainly located in the liver and ment of the polymerase chain reaction (PCR) oxidize ethanol to acetate via acetaldehyde. has made the genotyping of ADH and ALDH The acetate is subsequently oxidized to CO2 in living individuals possible. In this technique, minimal amounts of DNA such as contained in and H2O in extrahepatic tissues. During the last two decades, it has become the leukocytes of an ordinary blood sample can clear that ADH and ALDH are polymorphic in be enzymatically amplified in vitro. The amhumans, and a great deal of progress has been plification generates enough DNA to allow made in the purification and kinetic character- characterization by restriction analysis or ization of the ADH and ALDH isoenzymes, as hybridization with specific oligonucleotides. well as in elucidating the organization of their Using this technique, the role of the ADH and structural genes. Since the isoenzymes differ ALDH isoenzymes in alcohol pharmacokinemarkedly in their steady-state kinetic con- tics can now be directly determined, and first stants, it has been hypothesized that, according results have already been obtained. We will to the heritability of the individual isoenzyme review here the catalytic properties as well as patterns, the ethanol elimination rate should recent advances in the molecular genetics of be genetically controlled. In agreement with ADH and ALDH isoenzymes, with special reference to the genetic control of the elimination rate of ethanol. We will also review some •Supported by NIAAA grants AA 02342, AA 07117, AA aspects of structure-function relationships in 07611. ADH and ALDH isoenzymes derived from tTo whom all correspondence should be addressed. INTRODUCTION
105
106
T. EHRIG et al.
comparison of naturally occurring isoenzymes and from chemical modification studies.
zygous ADH\ ADH\ (Smith et al., 1971; von Wartburg and Schurch, 1968). By contrast, in Oriental populations, approximately the inverse ratio is found with. frequencies in ALCOHOL DEHYDROGENASE Japanese of 32 and 68% for the ADH\I Nomenclature and genetic model of human ADH\ allele, respectively (Teng et al., 1979). ADH isoenzymes The p 3 chain is seen in about 25% of the Numerous isoenzymes are seen on electro- African-Americans, the frequency of the phoresis of human liver homogenates in starch ADH\ allele in the population being 15% gels (Smith et al., 1971, 1972). The finding of (Bosron et al., 19836). Frequency ratios for ontogenetic and tissue-specific differences in the ADH\ (YI subunit)A4£>/^ (y2 subunit) expression, the purification and characteriza- are 0.6/0.4 for Caucasian (Smith et al., 1972) tion of isoenzymes including dissociation- and 0.9/0.1 for Oriental and African-American recombination experiments to identify homo- populations (Teng et al., 1979; Bosron and Li, and heterodimers (Bosron et al., 1983a; Yin et 1981). No heterogeneity is known for classes II al., 1984a,b), and the determination of the and III ADHs. In humans, these classes primary sequences (Hempel et al. 1984a; comprise the nn (Hoog et al., 1987a) and Buhler et al., 1984a, b; Jornvall et al., 1984; von XX (Kaiser et al., 1988) isoenzymes. Bahr-Lindstrom et al., 1986; Burnell et al., 1987; Hoog et al., 1987a; Kaiser et al., 1988) Molecular genetics of human class I ADH have allowed unequivocal identification of all Knowledge of the amino acid sequences of known human ADH isoenzyme subunits. the human ADH isoenzymes has made possBased on similarities in the primary structure, ible the synthesis of specific oligonucleotides to the human ADH isoenzymes can be grouped detect complementary sequences in cDNA into three classes, I, II and III, with at least libraries. By this method, partial cDNAs cod85%, mostly over 90%, homology within a ing for Pi (Duester et al., 1984), p 2 (Ehrig class and about 60-70% homology between et al., 1988), Yi and y2 (Hoog et al., 1986) classes. Within class I, there is genetic poly- chains and full-length cDNA clones for a (von morphism. Three different structural gene loci, Bahr-Lindstrom et al., 1986) and px (Heden et ADHU ADH2, and ADH3, code for the three al., 1986; Edenberg et al., 1989) have been different types of human class I peptide chains, isolated. The deduced sequences are in agreea, p\ and y. This model was originally ment with the protein sequences, except that inferred from the differences in development Val 276 in the Yi protein is Met in the cDNA. and tissue-specific expression: in the liver, The isolated P^clones show untranslated 3' there is a sequential initiation of expression in ends of different lengths generated by alternathe order a (first fetal trimester), p (second tive use of different polyadenylation signals. fetal trimester), and y (after birth). In the Accordingly, several p! mRNA bands are lung, only the P subunits, and, in the upper seen on Northern blots. Two bands appear for gastrointestinal tract, only y chains, are ex- the a mRNA and one band for the Y mRNA pressed (Smith et al., 1971, 1972). At the (Ikuta and Yoshida, 1986). ADH2 and ADH3 loci, allelic polymorphisms cDNA clones were subsequently used to are found: the ADH2 locus codes for p 1; P2 detect complementary sequences in libraries of and p 3 chains, and the ADH3 locus for yi and human genomic DNA, leading to the isolation Y2 chains (Smith et al., 1971, 1972). There are of three different types of genomic clones marked differences in frequency of occurrence coding for a, p and Y chains (Duester et al., among different ethnic populations. About 80- 1986), and to the isolation of the p gene 3 95% of Caucasian populations have the (Carr et al., 1988). Restriction analyses perADH]jADH\ genotype (homozygous for formed on human genomic DNA generated PiPO. The ADH\ allele (coding for the only fragments expected from the restriction P2 subunit) is seen in 4-20% of the popu- maps of these three types of clones, suggesting lation; probably most of these are hetero- that there are probably only these three class I
ALCOHOL AND ALDEHYDE DEHYDROGENASE ADH genes in the human genome, in agreement with the conclusions derived from the ontogenetic and tissue-specific characteristics of ADH expression (Duester et al., 1986). All three human class I genes are situated on the long arm of chromosome 4. This was established by studying human-rodent somatic cell hybrids where DNA restriction fragments of the expected sizes, detected with a human cDNA probe, were present in cells carrying human chromosome 4, and also in cells carrying fragments of this chromosome (Smith et al., 1984, 1985). As stated above, the development of the polymerase chain reaction (PCR) has greatly facilitated the genotyping of human individuals with respect to the ADH alleles. Comparison of the genotype with the isoenzyme pattern of single autopsy livers has shown that individuals expressing only one type of P or Y chains are homozygous at the corresponding locus (Gennari et al., 1988; Xu et al., 1988). Physical and catalytic properties of human ADH isoenzymes All human ADH isoenzymes are dimers of peptide chains with a molecular weight of 40,000. The class I subunits (a, P and y) associate randomly to form homo- and heterodimers in all possible combinations. Association across the three classes (e.g. p with jr.) has not been observed (Bosron et al., 1983a; Yin et al., 19846). The steady-state kinetic constants of the
107
isoenzymes differ markedly, as can be seen from Table 1, which lists kinetic constants for the homodimeric isoenzymes. All isoenzymes follow Michaelis-Menten kinetics, except for the Yy-isoenzymes which exhibit nonlinearity in double-reciprocal plots. For the latter ones, So 5-values for half-saturation instead of ATmvalues are given in Table 1 (Bosron et al., 1983a). Heterodimers behave as a mixture of the parent homodimers, i.e. their subunits seem to function independently of each other (Yin et al., 19846; Fong and Keung, 1987). The blood ethanol concentration of alcoholintoxicated individuals is usually around 10-20 mM. The Km- and 50.5-values in Table 1 indicate that, during ethanol oxidation, the PiPx and YY isoenzymes are saturated over most of the time, and the aa, P2P2, P3P3 and JIJI isoenzymes are partially saturated. Since the XX isoenzyme is not saturated with ethanol concentrations below 1 M, it does not contribute significantly to ethanol elimination. A comparison of the Vmax values in Table 1 shows that the P2P2 and p3p3 isoenzymes can be expected to make a high contribution to the ethanol eliminating capacity, whereas the contribution by the p t p x isoenzyme will be relatively low. The capacity of the p3p3 isoenzyme, however, will, in part, be offset by its high .Km values towards ethanol and NAD + . Structure—function relationships in human ADH The structure of the EE isoenzyme of horse
Table 1. Steady state kinetic constants of homodimeric isoenzymes of human liver ADH
I
Class Isoenzyme
aa
P.Pi
P2P2
P3P3
KmaoH ^NAD
4.2 13
0.049 7.4 90 0.23
0.94 180 . 340 10
36 710 2300 7.9
1.0* 7.9
0.49* 8.7
2.2
0.87
7.0
10.5
(mM) (uM)
Vnax forwd (U/mg) pH-optimum
0.6 10.5
10.5
8.5
Y1Y1
Y2Y2
10.5
II
III
rot
XX
34 14 86 0.5
Alcohol & Alcoholism. Vol. 25. No. 2/3. pp. 105-116.1990 Printed in Great Britain
ALCOHOL AND ALDEHYDE DEHYDROGENASE* TORSTEN EHRIG, WILLIAM F. BOSRON and TING-KAI Lit Departments of Biochemistry and Medicine, Indiana University School of Medicine, Indianapolis, IN 46202, U.S.A. (Received 23 January 1990) Abstract — The enzymes mainly responsible for ethanol degradation in humans are liver alcohol dehydrogenases (ADH) and aldehyde dehydrogenases (ALDH). Polymorphisms occur in both enzymes, with marked differences in the steady-state kinetic constants. The /fm-values for ethanol of ADH isoenzymes relevant for alcohol degradation range from 49 uM to 36 jiM, and the Kma,-values from 0.6 to 10 U/mg. Expression of an inactive form of the ALDH2 isoenzyme, the so-called Oriental variant, results in impaired acetaldehyde metabolizing capacity. The differences in ethanol and acetaldehyde metabolizing activities of allelic enzyme forms may be responsible in part for the large variation in the alcohol metabolism rate in humans. Interindividual differences in the isoenzyme pattern may contribute to the genetically determined predisposition for excessive alcohol intake.
this notion, twin studies have shown that the alcohol elimination rate is, in part, genetically The conversion of ethanol to acetate in the controlled (Kopun and Propping 1977; Martin liver is catalyzed mainly by NAD+-dependent etal., 1985). However, it has not been possible alcohol dehydrogenases (ADH; EC 1.1.1.1) to relate the alcohol elimination rate directly to and aldehyde dehydrogenases (ALDH; EC the isoenzyme pattern in single individuals, 1.2.1.3), whereas other pathways, namely the since the pattern could in the past only be microsomal ethanol-oxidizing system (MEOS) reliably determined from liver autopsy or and catalase, play a minor role. These enzyme biopsy samples. More recently, the developsystems are mainly located in the liver and ment of the polymerase chain reaction (PCR) oxidize ethanol to acetate via acetaldehyde. has made the genotyping of ADH and ALDH The acetate is subsequently oxidized to CO2 in living individuals possible. In this technique, minimal amounts of DNA such as contained in and H2O in extrahepatic tissues. During the last two decades, it has become the leukocytes of an ordinary blood sample can clear that ADH and ALDH are polymorphic in be enzymatically amplified in vitro. The amhumans, and a great deal of progress has been plification generates enough DNA to allow made in the purification and kinetic character- characterization by restriction analysis or ization of the ADH and ALDH isoenzymes, as hybridization with specific oligonucleotides. well as in elucidating the organization of their Using this technique, the role of the ADH and structural genes. Since the isoenzymes differ ALDH isoenzymes in alcohol pharmacokinemarkedly in their steady-state kinetic con- tics can now be directly determined, and first stants, it has been hypothesized that, according results have already been obtained. We will to the heritability of the individual isoenzyme review here the catalytic properties as well as patterns, the ethanol elimination rate should recent advances in the molecular genetics of be genetically controlled. In agreement with ADH and ALDH isoenzymes, with special reference to the genetic control of the elimination rate of ethanol. We will also review some •Supported by NIAAA grants AA 02342, AA 07117, AA aspects of structure-function relationships in 07611. ADH and ALDH isoenzymes derived from tTo whom all correspondence should be addressed. INTRODUCTION
105
106
T. EHRIG et al.
comparison of naturally occurring isoenzymes and from chemical modification studies.
zygous ADH\ ADH\ (Smith et al., 1971; von Wartburg and Schurch, 1968). By contrast, in Oriental populations, approximately the inverse ratio is found with. frequencies in ALCOHOL DEHYDROGENASE Japanese of 32 and 68% for the ADH\I Nomenclature and genetic model of human ADH\ allele, respectively (Teng et al., 1979). ADH isoenzymes The p 3 chain is seen in about 25% of the Numerous isoenzymes are seen on electro- African-Americans, the frequency of the phoresis of human liver homogenates in starch ADH\ allele in the population being 15% gels (Smith et al., 1971, 1972). The finding of (Bosron et al., 19836). Frequency ratios for ontogenetic and tissue-specific differences in the ADH\ (YI subunit)A4£>/^ (y2 subunit) expression, the purification and characteriza- are 0.6/0.4 for Caucasian (Smith et al., 1972) tion of isoenzymes including dissociation- and 0.9/0.1 for Oriental and African-American recombination experiments to identify homo- populations (Teng et al., 1979; Bosron and Li, and heterodimers (Bosron et al., 1983a; Yin et 1981). No heterogeneity is known for classes II al., 1984a,b), and the determination of the and III ADHs. In humans, these classes primary sequences (Hempel et al. 1984a; comprise the nn (Hoog et al., 1987a) and Buhler et al., 1984a, b; Jornvall et al., 1984; von XX (Kaiser et al., 1988) isoenzymes. Bahr-Lindstrom et al., 1986; Burnell et al., 1987; Hoog et al., 1987a; Kaiser et al., 1988) Molecular genetics of human class I ADH have allowed unequivocal identification of all Knowledge of the amino acid sequences of known human ADH isoenzyme subunits. the human ADH isoenzymes has made possBased on similarities in the primary structure, ible the synthesis of specific oligonucleotides to the human ADH isoenzymes can be grouped detect complementary sequences in cDNA into three classes, I, II and III, with at least libraries. By this method, partial cDNAs cod85%, mostly over 90%, homology within a ing for Pi (Duester et al., 1984), p 2 (Ehrig class and about 60-70% homology between et al., 1988), Yi and y2 (Hoog et al., 1986) classes. Within class I, there is genetic poly- chains and full-length cDNA clones for a (von morphism. Three different structural gene loci, Bahr-Lindstrom et al., 1986) and px (Heden et ADHU ADH2, and ADH3, code for the three al., 1986; Edenberg et al., 1989) have been different types of human class I peptide chains, isolated. The deduced sequences are in agreea, p\ and y. This model was originally ment with the protein sequences, except that inferred from the differences in development Val 276 in the Yi protein is Met in the cDNA. and tissue-specific expression: in the liver, The isolated P^clones show untranslated 3' there is a sequential initiation of expression in ends of different lengths generated by alternathe order a (first fetal trimester), p (second tive use of different polyadenylation signals. fetal trimester), and y (after birth). In the Accordingly, several p! mRNA bands are lung, only the P subunits, and, in the upper seen on Northern blots. Two bands appear for gastrointestinal tract, only y chains, are ex- the a mRNA and one band for the Y mRNA pressed (Smith et al., 1971, 1972). At the (Ikuta and Yoshida, 1986). ADH2 and ADH3 loci, allelic polymorphisms cDNA clones were subsequently used to are found: the ADH2 locus codes for p 1; P2 detect complementary sequences in libraries of and p 3 chains, and the ADH3 locus for yi and human genomic DNA, leading to the isolation Y2 chains (Smith et al., 1971, 1972). There are of three different types of genomic clones marked differences in frequency of occurrence coding for a, p and Y chains (Duester et al., among different ethnic populations. About 80- 1986), and to the isolation of the p gene 3 95% of Caucasian populations have the (Carr et al., 1988). Restriction analyses perADH]jADH\ genotype (homozygous for formed on human genomic DNA generated PiPO. The ADH\ allele (coding for the only fragments expected from the restriction P2 subunit) is seen in 4-20% of the popu- maps of these three types of clones, suggesting lation; probably most of these are hetero- that there are probably only these three class I
ALCOHOL AND ALDEHYDE DEHYDROGENASE ADH genes in the human genome, in agreement with the conclusions derived from the ontogenetic and tissue-specific characteristics of ADH expression (Duester et al., 1986). All three human class I genes are situated on the long arm of chromosome 4. This was established by studying human-rodent somatic cell hybrids where DNA restriction fragments of the expected sizes, detected with a human cDNA probe, were present in cells carrying human chromosome 4, and also in cells carrying fragments of this chromosome (Smith et al., 1984, 1985). As stated above, the development of the polymerase chain reaction (PCR) has greatly facilitated the genotyping of human individuals with respect to the ADH alleles. Comparison of the genotype with the isoenzyme pattern of single autopsy livers has shown that individuals expressing only one type of P or Y chains are homozygous at the corresponding locus (Gennari et al., 1988; Xu et al., 1988). Physical and catalytic properties of human ADH isoenzymes All human ADH isoenzymes are dimers of peptide chains with a molecular weight of 40,000. The class I subunits (a, P and y) associate randomly to form homo- and heterodimers in all possible combinations. Association across the three classes (e.g. p with jr.) has not been observed (Bosron et al., 1983a; Yin et al., 19846). The steady-state kinetic constants of the
107
isoenzymes differ markedly, as can be seen from Table 1, which lists kinetic constants for the homodimeric isoenzymes. All isoenzymes follow Michaelis-Menten kinetics, except for the Yy-isoenzymes which exhibit nonlinearity in double-reciprocal plots. For the latter ones, So 5-values for half-saturation instead of ATmvalues are given in Table 1 (Bosron et al., 1983a). Heterodimers behave as a mixture of the parent homodimers, i.e. their subunits seem to function independently of each other (Yin et al., 19846; Fong and Keung, 1987). The blood ethanol concentration of alcoholintoxicated individuals is usually around 10-20 mM. The Km- and 50.5-values in Table 1 indicate that, during ethanol oxidation, the PiPx and YY isoenzymes are saturated over most of the time, and the aa, P2P2, P3P3 and JIJI isoenzymes are partially saturated. Since the XX isoenzyme is not saturated with ethanol concentrations below 1 M, it does not contribute significantly to ethanol elimination. A comparison of the Vmax values in Table 1 shows that the P2P2 and p3p3 isoenzymes can be expected to make a high contribution to the ethanol eliminating capacity, whereas the contribution by the p t p x isoenzyme will be relatively low. The capacity of the p3p3 isoenzyme, however, will, in part, be offset by its high .Km values towards ethanol and NAD + . Structure—function relationships in human ADH The structure of the EE isoenzyme of horse
Table 1. Steady state kinetic constants of homodimeric isoenzymes of human liver ADH
I
Class Isoenzyme
aa
P.Pi
P2P2
P3P3
KmaoH ^NAD
4.2 13
0.049 7.4 90 0.23
0.94 180 . 340 10
36 710 2300 7.9
1.0* 7.9
0.49* 8.7
2.2
0.87
7.0
10.5
(mM) (uM)
Vnax forwd (U/mg) pH-optimum
0.6 10.5
10.5
8.5
Y1Y1
Y2Y2
10.5
II
III
rot
XX
34 14 86 0.5
