Studies on the subunit structure and molecular size of the human alcohol dehydrogenase isozymes determined by the different loci, ADH1, ADH2, and ADH3.

It has been proposed (Smith, Hopkinson & Harris 1971, 1972) that human alcohol dehydro-gellase (ADH) is determined by three separate structural gene loci, designated ADH,, A DH, and ADH,, which code for structurally different polypeptides a, /3 and y respectively. Evidence has been obtained for the occurrence of two common alleles at each of the ADH, and ADH, loci, designated ADH; and ADH; and ADH; and ADH;, which code for structurally distinct forms of the corresponding polypeptides (pl, pz and yl, y 2 respectively). According to the hypothesis the isozymes of human ADH are dimers and any one particular isozyme may be homodimeric, consisting of two identical polypeptides (e.g. 0101, plpl, y'y', etc.) coded by a specific allele at one of the loci, or heterodimeric, consisting of two non-identical polypeptides (e.g. ~$1, ay', ay2, etc.) coded by alleles at separate loci or heterodimeric but coded by different alleles a t the same locus (e.g. y1y2). observed after tissue extracts depend the relative activities of the three loci and they vary from tissue to tissue and also with development. in the homodimeric aa isozyme polypeptides predominates early foetal life but later ADH, polypeptides appear and the heterodimeric ap the homodimeric pp isozymes can be detected. the first year of post-natal life ylyl and plpl, ply2, y2y2. Mixtures of foetal lung extracts (plpl) and foetal kidney or intestinal extracts of the ADH, 1 (y'y') or ADH, 2 (y2y2) phenotype subjected to the dissociation-recombination procedure were found to generate a new isozyme (Ply1 or ply2 respectively) with intermediate electrophoretic mobility. Also a preparation of partially purified ply2 isozyme was treated and shown to generate plpl and y2y2 isozymes. A clean preparation of the ply1 isozyme was not available for study but a mixture of the ap' and ply1 isozymes obtained from CM cellulose chromatography of an adult liver ADH, 1 sample was available.

of the corresponding polypeptides (pl, pz and yl, y 2 respectively). According to the hypothesis the isozymes of human ADH are dimers and any one particular isozyme may be homodimeric, consisting of two identical polypeptides (e.g. 0101, plpl, y'y', etc.) coded by a specific allele at one of the loci, or heterodimeric, consisting of two non-identical polypeptides (e.g. ~$1, ay', ay2, etc.) coded by alleles a t separate loci or heterodimeric but coded by different alleles a t the same locus (e.g. y1y2).
The isozymes actually observed after electrophoresis of tissue extracts appear to depend on the relative activities of the three loci and they vary from tissue to tissue and also with development. Thus, in liver, the homodimeric aa isozyme consisting of ADH, polypeptides predominates in early foetal life but later ADH, polypeptides appear and the heterodimeric ap and the homodimeric pp isozymes can be detected. During the first year of post-natal life ADH, polypeptides also appear and a complex mixture of homodimeric (0101, and y y ) and heterodimeric (ap, ay and By) isozymes can be detected in liver from children and adults.
I n contrast, the isozyme pattern of lung tissue is simple in both foetal and adult samples and consists only of / 3 polypeptides determined by the ADH, locus. In kidney and gastro-intestinal tract the ADH, polypeptides predominate in early foetal life and the homodimeric jsozymes y1y1 and y2y2 can be readily seen in the respective homozygotes, while in the heterozygotes (ADH, 2-1) they occur together with the hybrid heterodimeric y1y2 isozyme. I n adult gastrointestinal tract specimens the ADH isozymes determined by the A D H , locus also predominate but traces of the ADH, type isozymes are seen as well. In adult kidney it has now been found by careful homogenization of ice-cold post-mortem tissue that isozymes determined by the ADH, locus do occur, contrary to what was thought earlier (Smith et al. 1971) but most of the total ADH activity can be attributed to the A D H , locus. The present paper is principally concerned with the results of in vitro hybridization experiments involving the dissociation and recombination of the postulated subunits of the different ADH isozymes. The findings support the hypothesis outlined above. In addition, we describe experiments using gel filtration column chromatography to determine the molecular size of the various isozymea and the separation of the individual isozymes by ion exchange chromatography on carboxymethyl (CM) cellulose.

MATERIALS AND METHODS
The sources of the tissues used in this study have been previously described (Smith et (11. 1971(Smith et (11. , 1972.

Dissociafion and reconahination experiments.
Tlie method used in these experiments is simple and is based on the procedure devised by Hart (1971) for in vitro 'hybridization' experiments on wheat ADH. In this method dissociation and recombination of subunits is achieved by freezing and thawing samples in the presence of 1 M-NaCl, 0.4 M sucrose and 0.1 M 2-mcrcaptoethanol in 0.1 M sodium phosphate buffer, p H 7.0. The salt is then removed by clialysis. Expcxriments were carried out using crude tissue extracts prepared by homogenizing the tissue with an equal volume of the salt, sucrose, mercaptoethanol, phosphate mixture, p H 7.0, in a Silverson mixer emulsifier. Appropriate mixtures of the homogenates were prepared and were then left in a deep freeze a t -20" C for about 17 hr. After thawing, dialysis was carried out for 4 hr. apiiist the phosphate buffer, pH 7.0, containing sucrose (0.4 M) and mercaptoethanol (0. I nr) and for a further 4 hr. against 0.005 M tris-phosphate buffer, pH 7.0, containing 0.002 M NAD. The specimens were then examined by starch-gel electrophoresis immediately. A number of control samples, with and without salt, were prepared in parallel with each experiment, subjected to the same procedure and finally analysed by electrophoresis alongside the test material. Experiments were also done using the same technique with partially purified isozymes prepared by CM chromatography of liver extracts. For this purpose 50: 50 mixtures of the sample and the salt-sucrose-mercaptoethanol phosphate buffer solution were made and then subjected to the freeze-thaw dialysis procedure described above.
Gel Jiltration chromatography. Experiments were carried out using Biogel P 150 (Calbiochem) in a water-jacketed column, 100 x 2.5 cm. Tlie column and fraction collector were used in a cold rooni a t 4" C . ; the water circulating around the column was maintained at a constant 8" C. The Biogel P 150 was swollen in a large excess of sterile 0.005 M tris-phosphate buffer, pH 7.7, coiitainiiig 0.1 M-NaCI, for 48 hr. a t 4" (7. The column was poured and allowed to settle overnight; then buffer was pumped in a n upward direction through the column a t a rate of about 20 ml./hr. for approximately 8 hr. During this period the sample was prepared by homogenizing 30-40 g tissue in 15-20 ml. of the column buffer (0.005 M tris-phosphate, pH 7.7). In some experiments pooled material of the same phenotype was used. The homogenatc was centrifuged at 90,OOOg for 45 min. and the supernatant then coilcentrated using Aquacitle (Calbiochcm) to about 3 nil. This was then applied to the bottom of the column, together with 10 mg. myoglobin, 20 mg. ovalbumin and 5 mg. Blue Dextran (Pharmacia Ltd.) as markers. The buffer was pumped through a t about 20 ml./hr. and fractions were collected a t 10 min. intervals. The fractions were tested for ADH activity by the spectrophotometric assay procedure described by Voii Wartburg, Papenberg & Aebi (1965) as detailed by Smith et al. (1971) and active fractions were examined by starch-gel electrophoresis a t pH 8.6 (Smith et al. 1971). Thc positions of the marker proteins, ovalbumin and myoglobiii in the eluate were determined by spectrophotometric scanning at 280 mp. and by starch-gel electrophoresis using the discontinuous buffer system of Poulik (1957). Several enzymes, peptidases A, B, C, D and S (Lewis & Harris, 1967 ; Rapley, Lcwis & Harris, 1971), lactate dehydrogenase (Bergmcyer, Bernt & Hess, 1963) and nucleoside phosphorylase (Kalckar, 1947 ;Edwards, Hopkinson & Harris, 197 1) prcsent in the original extract, were used as markers and were located in the eluake by direct assay and electrophoresis.
Ion-exchange chromatography. Chromatography was carried out in a cold room at 4' C. using carboxymethyl (CM) cellulose (Whatman CPVI 5 2 ) , in glass columns 40 x 2.5 cni., and the starting buffer was 0.005 M tris-phosphate, p H 7.7, containing 2-mercaptoethanol (2 mni) and NAD (1 mai). These additives were found to improve the recovery of ADH activity and the resolution of the isozymes. Thirty to forty g. of liver was homogenized in 30-40 ml. starting buffer and the debris was spun down at 45,OOOg for 2 hr. The clear supernatant was dialysed against an excess of the same buffer and centrifuged again a t 45,OOOg for 1 hr. This material was applied to the column, followed by 300-400 ml. of the starting buffer at a flow rate of about 30 ml./hr. Five ml. fractions were collected and assayed for ADH activity but no ADH was detected during this phase. Suitable conditions for elution were devised by experimentation. At first a linear saline gradient (0-0.5 M-NaCl in 600 ml. starting buffer) was used but in later experiments the conditions were modified until eventually it was found that a two-stage procedure gave the best separation of the ADH isozymes. The first stage of elution was carried out with 0.008 M-NaC1 aiitl'tlie second stage was done with 0.02 iv-NaC1 in the starting buffer. Five ml. fractions were collected during each phase and the ADH activity was assayed spectrophotometrically (Von Wartburg et al. 1965;Smith et al. 1971). The active fractions were also examined by starch-gel electrophoresis (Smith et al. 1971). Each step in the elution phase was continued until after the ADH activity in the fractions had subsided to zero. Application of a final higher salt concentration (up to 0.04 M) did not result in the elution of further activity.

Molecular-weight determinations
A representative selection of tissues was examined by gel filtration chromatography on Biogel P 150. This included infant and adult liver, infant kidney and adult gastric mucosa extracts.
In each experiment ADH activity was detected in the eluted fractions as a single symmetrical peak, with a molecular weight of about 80,000. Analysis of the individual fractions by starch gel electrophoresis showed that all the isozymes were eluted together and are therefore of similar molecular weight, irrespective of their subunit composition. The results are summarized in Table 1.
In some experiments with liver extracts it was noticed that the aa isozyme, though prominent in the original extract, was often very weak and sometimes not detected in the eluate from the column, though the 01 containing heteroclimeric isozymes, ap arid ay remained relatively strong in staining intensity. This suggests that the aa isozyme is relatively less stable under the conditions used for chromatography than most other ADH isozymes; a conclusion which is supported by direct studies on the relative thermal stabilities of the ADH isozymes (Smith et al. in preparation). Another interesting observation concerning the relative stabilities was made in an experiment using a liver extract of the atypical pH ratio phenotype. Here the total ADH activity recovered in the fractions was only 25 yo of the original, compared with recoveries of between 90 yo and 75 yo from experiments carried out with livers of the usual pH ratio phenotype. This suggests that the isozymes containing the polypeptide determined by the variant ADH; ('atypical') allele are less stable than the others. Inol. wt.

CM cellulose chromatography
All these experiments were carried out using liver extracts.
(i) Foetal and infant liver, usual p H ratio. Three isozyrnes (aa, aP1, PIP') characteristic of these livcr extracts were cleanly separated by CM cellulose chromatography. Using an infant livcr extract, for example, three individual peaks of activity were detected in the eluate (Fig. 1 a ) which corresponded with aa, ap' and PIP' respectively ( Fig. l h ) . The order of clution of the isozymes was that expected from starch-gel electrophoresis at pH 8.6 : aa followed by aP1 tlien plp1. Comparison of the elution profiles obtained in a series of experiments in which liver from different infants and foetuses was used showed that in early foetuses the first peak (aa) was relatively much larger than the second (apl) and third (/?'PI) peaks but in older foetuses and young infants the second arid third peaks were relatively more prominent.
(ii) Adult liver, usud p H ratio, ADH, 2. A more complex elution profile was obtained on CM cellulose chromatography of adult liver samples of the ADH, 2 phenotype. Up to six separate peaks of activity were obtained which corresponded with maximum elution of the isozymes 7 2~2 , ay2, P'y2, aa, aP1 and PIP1 respectively. Results from a representative experiment arc shown in Fig. 2(a) and ( h ) . It will be seen that in addition to the isozymes mentionrtl above, several other minor zones were detected in the fractions. These minor bands were most prominent in the fractions exhibiting ay2, ply2 and apl activity and are believed t o be 'secontlnry' isozymes. They correspond in electrophoretic mobilities with similar minor isozymes tlemonstrablc after electrophoresis of crucle liver extracts, particularly in samples which have been stored for 48 hr. or longer at 4' C. Similar minor bands will be noted in the other photographs shown in this paper and it seems that secondary isozymes can be formed from almost all of the main ADH components. One easily recognized minor zone is seen in crude liver homogenates cathodal to, and believed to be derived from, the Pp zone. However, this particular component does not seem to occur in samples containing 2-mercaptoethanol and this reagent was therefore added routinely t o the buffers used for chromatography. Another confusing extra isozyme often scen in crude livcr extracts (e.g. in Fig. 2 b ) can be attributed t o the LDH 5 isozyme, since this is active in the presence of NAD even when the stain mixture does not contain the lactate substrate.   (Fig. la). Electrophoresis at pH 7.7.

Fraction number
Pig. ? ( a ) . Elution of ADH activity from a CM ccllulosc column. Source of enzyme: acliilt liver of the ADH, 2 phcnotypc, ut;iinl pH ratio.
(iii) Adult liver, usual p N ratio, d D H , 1. Only three peaks of ADH activity were cletcctetl in the eluate from experiments using samples of the ADH, 1 phenotype (Fig. 3a). Electrophorcsis (Fig. 3 b ) showed that the first peak contained a mixture of the y'y', ay' arid aa isozymes and the second peak contained both the ply' and the up1 isozymes. However, the third peak was homogeneous and displayed only BIP1 isozyme activity. The relative sizes of the three peaks varied considerably in diffcrent experiments according to the relative contributions of the various isozymes to the total activity of the liver sample used.
(iv) Adult liver, atypical pH ratio, ADH, 1. Three peaks of ADH activity were also clctectctl in the eluate obtained from chromatography of atypical pH ratio phenotype liver samples of the ADH, 1 type but the shape of the elution curve was quite distinctive (Fig. 4). Thc third peak (fractions 140-170 in Fig. 4) was relatively much larger than the corresponding peak of activity obtained from chromatography of the usual p H ratio phenotype liver samples of thct ADH, 1 type. Electrophoretic analysis revealed that the first peak (fractions 10-40) exhibited the ay' isozyme only, no aa or y1y1 isozymes were detected; the second peak (fractions 40-90) showctl the two isozymes thought to rcpresent ap2 and P2y1 but the ap' and ply' isozymes usually semi in electrophoresis of whole liver extracts of this phenotype could not be detected ; the third peak exhibited a single isozyme with the same mobility as the most cathodal component of the atypical ADH, 1 livers. The subunit structure of this component, which is characteristic of the atypical p H ratio phenotype, is still uncertain but is possibly p1p2.

Dissociation and recombination experiments
The aim of these experiments was to cause dissociation of the isozymes into their respective subunits and then allow the subunits to recombine at random to see whether new isozymes were formed. Thus mixtures of postulated homodimeric isozymesfor example, act and treated appropriately would be expected to generate an additional heterodimeric isozyme, in this example ap'. Vice versa, a single heterodimeric isozyme should be capable of generating the two corresponding homodimeric isozymes.
(i) Isozymes aa, apl, plP1. A number of experiments were carried out with crude tissue extracts, using early foetal liver as a source of the aa isozyme and foetal lung as a source of pl,bl. Treatment of a mixture of these two tissues in the manner described in the methods section led to the Fraction number Fig. 2 (b). Photographs of starch gels showing ADH isozymes present in the fractions obtained on CM chromatography of an adult liver of the ADH, 2 phenotype, usual pH ratio (Fig. 2a). Electrophoresis at pH 8.6.  (Fig. 3 a ) . Electrophoresis at pH 7.7. A noticeable feature in all these experiments, however, in keeping with the observations made during the gel filtration of foetal liver samples and in direct heat stability studies (Smith et al. in preparation), was the apparent lability and ready inactivation of the aol isozyme. Thus, to be certain of detecting the aa isozyme in the hybridization experiments, a n excess of aa was used in the starting material whenever possible. I n the reciprocal experiments, preparations of the up1 isozyme were shown t o generate the two homodimeric isozymes aa and plpl (Fig. 5 b ) .
(ii) Isozymes aa, ayl, ylyl and aa, ay2, y2y2. Experiments were carried out using mixtures of foetal liver (aa) and infant kidney of ADH, type 1 (ylyl) or ADH, type 2 (y2y2). I n both cases heterodimeric (ayl or ay2) isozymes were generated from the homodimeric forms by the salt freeze-thaw-dialysed treatment. The reciprocal experiment using ayl or ay2 as the starting material to generate aa and y1y1 or act and y2y2 respectively was not attempted due t o lack of suitable purified material with high specific activity.
(iii) Isoxymes plpl, ply1, ylyl and plpl, ply2, y2y2. Mixtures of foetal lung extracts (plpl) and foetal kidney or intestinal extracts of the ADH, 1 (y'y') or ADH, 2 (y2y2) phenotype subjected to the dissociation-recombination procedure were found t o generate a new isozyme (Ply1 or ply2 respectively) with intermediate electrophoretic mobility. Also a preparation of partially purified ply2 isozyme was treated and shown to generate plpl and y2y2 isozymes. A clean preparation of the ply1 isozyme was not available for study but a mixture of the ap' and ply1 isozymes obtained from CM cellulose chromatography of an adult liver ADH, 1 sample was available. showing (i)-(v). This difference is due to the fact that the portion of the gel which contained (vi) and (vii) waa incubated for a longer period in order t o detect all isozymes present in these smples. PSM = sodium phosphate, sucrose, mercaptoethanol; FTD = frozen, thawed and then didysed.
The dissociation treatment was seen to lead to the formation of three out of the four additional isozymes expected from such a mixture viz., 0161, /31,B1, and ccyl (Fig. 6). The ylyl isozyme was not detected.
(iv) Isozymes yvl, y v 2 , rv2. A number of attempts were made to produce the heterodimeric isozyme from a mixture of the homodimeric ylyl and rv2 isozymes prepared with crude extracts of foetal intestine or foetal kidney of ADH, 1 and ADH, 2 phenotypes. Quite often no activity was detected in the final product of the experiment, probably because of the low concentration of ADH relative to other proteins in these tissues. However, in some experiments successful hybridization was accomplished and the rv2 isozyme was detected. The reciprocal experiment ( y v 2 + y v l and y2y2) was not attempted.  Jornvall & Pietruszko, 1972) have shown that human liver alcohol dehydrogenase is a dimeric enzyme with a molecular weight of about 80,000, indicating that the molecular weight of the constituent polypeptides is in the region of 40,000. Our experiments support these conclusions since the ADH in newborn liver, adult liver, gastric mucosa and foetal kidney extracts ahowed a similar elution profile on Biogel P 150 and hence gave a similar estimate of molecular weight (c. 80,000) in each experiment. Furthermore, electrophoretic analysis of the individual fractions selected from different areas revealed no asymmetry in the elution pro%e, thus indicating that all the isozymes, irrespective of their constitution are of similar molecular weight.
It may be concluded therefore thati the ADH polypeptides a, / 3 and y determined by ADH,, ADH, and ADH, respectively have about the same molecular weight (c. 40,000). Also the alternative polypeptides (p1 and pa from ADH, and y1 and ye from ADH,) appear to be of the same size. The similarity of the ADH isozymes in molecular size indicates that their separation by starchgel electrophoresis is mainly dependent on charge differences between the isozymes. The results obtained from ion-exchange chromatography of human liver reported here and in previous studies (Von Wartburg et al. 1964 with CM cellulose were those carried out on newborn liver samples, which exhibit the simplest iaozyme patterns. In such cases it wm possible to obtain a clear separation of the three principal iaozymes aa, up1 and plpl and the order of their elution from the ion exchanger was as expected from their relative electrophoretic mobilities on starch gels. Furthermore, the preparations of the individual iaaymes thus obtained were found to b e suitable for the investigation of the subunit structure by in vitro hybridization and also for studying enzymic properties of the individual isozymes. More complicated elution profiles but less clear-cut separations of the individual isozymes were obtained from ion exchange chromatography of adult liver extracts. Different results were obtained according to the ADH, and ADH, phenotypes of the samples being analysed. For example, using liver extracts of the usual p H ratio phenotype (ADHiADHi) and the ADH, 2 phenotype (ABH,2ADH,) six peaks of ADH activity were detected in the eluate which corresponded with the principal isozymes characteristic of this type of sample : the y2 containing isozymes yy2, ay2 and /3y2 were eluted first and in that order, followed by act, up1 and ,91/31 although absolute separation of the individual isozymes was not obtained under these conditions.
Using adult liver of the usual p H ratio (ADHi AD&) and the ADH, 1 phenotype (ADHiADH;) less good separations of the individual isozymes were obtained, however, since only three peaks of ADH activity were usually detected in the eluate. The ylyl isozyme was eluted first but together with ayl and ua as the first peak, the ply1 and up1 isozymes came off together as the second peak and the PIP' isozyme was eluted as the third peak of ADH activity.
Ion-exchange experiments with adult livers of the atypical p H ratio phenotype showed elution profiles quite different from those obtained using livers of the usual pH ratio phenotype. The outstanding feature was the very large peak of ADH activity containing the characteristic ' atypical ' p isozyme.
These results obtained by ion-exchange chromatography of human liver of the usual and of the atypical pH ratio ADH phenotype are comparable with the observations made by Von Wartburg and his colleagues using similar techniques (1964, 1968Schenker et al. 1971). However, exact comparison with this previous work is difficult since the methods used were not exactly the same for either the chromatographic or the electrophoretic separation of the isozymes. Furthermore, the phenotype of their starting material had not been assessed in the same way as ours by starch-gel electrophoresis. However, there appear to be no major inconsistencies between the present work and that previously reported, and the ion-exchange technique is clearly very suitable for obtaining partially purified preparations of some of the individual isozymes of human ADH which occur in crude liver homogenates.
The in vitro hybridization experiments carried out using crude tissue extracts and partially purified material from CM cellulose chromatography were very successful and provide convincing evidence in support of the postulated subunit composition of the various ADH isozymes.
Experiments were designed to show the formation of 'hybrid ' heterodimeric isozymes starting with mixtures of two different homodimeric isozymes and vice versa and the results obtained were exactly as expected. Also the results are consistent with the in vitro hybridization experiments reported on purified preparations of different human liver ADH isozymes separated by ion exchange chromatography (Schenker et al. 1971).
isozyme characteristic of the atypical pH ratio phenotype still remains uncertain. I n one experiment in which hybridization of this isozyme, in an extract of atypical lung, was carried out with a foetal liver sample, containing principally the ua isozyme, two additional isozymes were generated having electrophoretic mobilities which would correspond to aP1 and up2. This suggests that the structure of the 'atypical' , 4 isozyme may be p'p2. However, more work will be required on this problem.
The subunit structure of the most cathodal