Molecular basis of the alcohol dehydrogenase-negative deer mouse. Evidence for deletion of the gene for class I enzyme and identification of a possible new enzyme class.

The molecular basis of the alcohol dehydrogenase (ADH)-negative deer mouse (Peromyscus maniculatus) has been investigated. Several classes of mammalian ADHs have been recognized based upon biochemical and structural properties. ADH cDNA clones identified by hybridization to a mouse class I ADH cDNA clone were obtained from a deer mouse ADH-positive liver cDNA library. This cDNA has been identified as being a class I sequence and represents the deer mouse Adh-1 gene. An additional cDNA sequence identified in both the ADH-positive and -negative deer mouse cDNA libraries was identified by weak cross-hybridization to the mouse cDNA. This cDNA encodes an amino acid sequence representing a new class of mammalian ADH, and the deer mouse gene for this ADH is named Adh-2. ADH-negative deer mice do not produce mRNA, that is detected by the Adh-1 cDNA probe. However, both stocks of deer mice produce high levels of Adh-2 mRNA in liver. Southern analysis using an essentially full-length Adh-1 cDNA probe has shown that the Adh-1 gene is deleted in the ADH-negative mice. Biochemical analysis of enzyme activity suggests at least three ADH polypeptides are expressed in different tissues and have somewhat different substrate specificities, as in the mouse.

The alcohol dehydrogenase (E.C.1.1.1.1) (ADH)-deficient' (ADH-negative) and ADH-positive deer mouse (Peromyscus maniculrrtus) stocks (1,2) have been widely used to study the contributions of various pathways of ethanol metabolism in the liver. ADH-negative animals metabolize ethanol in vivo at rates approximately one-half that found for ADH-positive animals. Both stocks contain an ethanol-inducible microsomal ethanol-oxidizing system, although the activity is elevated about %fold in ADH-negative deer mice compared with ADHpositive deer mice in both control and ethanol-fed animals (3,4).
The relative contributions of the non-ADH pathways (namely, catalase and cytochrome P-450-mediated microsomal ethanol-oxidizing activity) have been studied in this * This work was supported by National Institute on Alcohol Abuse and Alcoholism Grant AA 06608. The costs of publication of this article were defrayed in part by the payment of page charges, This article must therefore be hereby marked "aduertisernent" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in thispaper hm been submitted to the GenBankTM/EMBL Data Bank with accession number ($ L.15703 for Adh-1 and L.15704 for  $To whom correspondence should be addressed. Tel.: 803-777-5135; Fax: 803-777-4002. The abbreviations used are: ADH, alcohol dehydrogenase; kb, kilobase(s). animal model with differing interpretations. Using different aminotriazole treatment protocols to inhibit catalase activity in ADH-negative deer mice, investigators have concluded that insignificant ( 5 ) or'substantial (6) ethanol metabolism via the peroxidatic activity of catalase occurs in the liver. Reports relying upon isotope discrimination effects on the different pathways to assess the role of non-ADH-mediated ethanol metabolism have suggested a predominant cytochrome P-450 role (7) or, alternatively, a significant catalase contribution (8) in ADH-negative animals. Deuterium exchange experiments have suggested dehydrogenase contributions in the ADH-negative deer mice (8), whereas this was not noted in other studies (7).
Because this is a widely used model to study the role of various pathways in ethanol metabolism, a more detailed molecular and biochemical study of the ADHs in the deer mouse has been undertaken. Previous studies have shown that a highly basic ADH is detected in the deer mouse liver using physiological concentrations of substrate ethanol. This isozyme is encoded by a single gene with three identified alleles (1). Two alleles encode electrophoretic variants of the enzyme, and the third allele encodes the enzyme deficiency. ADH-negative mice also lack cross-reacting antigenic material (2).
Attempts to identify other ADHs in the deer mouse have not been systematically explored. The ADH isozymes in mammals, including humans, were originally grouped into class I, 11, or I11 based upon electrophoretic, kinetic, and immunological properties (9). Structural similarity of nearly 85% in amino acid sequence is found between members of a class even between distant species, while sequence similarity of about 60% exists between different classes of isozyme found within a species (10). More recent structural studies have suggested that the rat stomach ADH represents a new class IV isozyme ( l l ) , and a recently characterized genomic and cDNA sequence reported from human is a structurally distinct class (12). Thus, the mammalian ADHs are known to exist as a minimum of five structurally distinct classes. Here, the molecular nature of the deer mouse ADH-negative phenotype is examined. A cDNA clone obtained from an ADH-positive liver cDNA library was determined to encode a class I ADH. ADH-negative deer mice do not contain liver mRNA detectable by hybridization to this clone, and analysis of genomic DNA suggests that the gene for class I ADH is deleted in ADH-negative mice. However, a cDNA clone isolated and characterized from the ADH-negative liver cDNA library was found to encode a new structural class of mammalian ADH. Of the tissues examined, the mRNA for this class of ADH is expressed at high level only in liver and in both strains of deer mice. Biochemical analysis has suggested at least three forms of the ADH in the deer mouse with similar electropho-

Molecular Basis for Alcohol Dehydrogenase-negative Deer Mice
retic and substrate specificity properties as found in the mouse.

EXPERIMENTAL PROCEDURES
Animals-ADH-positive (AdhF/AdhF genotype) and ADH-negative (AdhN/AdhN) deer mice were obtained from a breeding colony in the Peromyscus Stock Center at the University of South Carolina. The animals were fed chow diet and water ad libitum while being housed under 16 h of light and 8 h of dark conditions. cDNA Library Construction-Poly(A)+ RNA was isolated as described (13) from total cellular RNA prepared from liver of ADHpositive and -negative deer mice using the guanidinium thiocyanate method (14). Two independent liver double-stranded cDNAs were synthesized using described methods (15) and were made blunt-ended with T4 polymerase. EcoRI sites were methylated, and EcoRI linkers were added and digested with EcoRI restriction endonuclease. Fractions eluting from Bio-Gel A-15 M columns in the 0.5-7-kb range were pooled and ligated to EcoRI-digested XgtlO. The DNA was packaged using a Gigapack extract (Stratagene). Starting with approximately 200 ng of cDNA, about 5 X 10' recombinant phage were obtained.
Selection and Characterization of Deer Mouse Adh-1 cDNA Clones-Approximately 1 X lo' recombinant X g t l O plaques from both the ADH-positive and -negative liver cDNA libraries were screened by hybridization as previously described (16) (20) of some clones to aid in sequencing. Other subclones were obtained using convenient restriction sites to further aid in sequencing. One oligonucleotide primer was made for sequencing one part of one strand on the Adh-2 cDNA clone. Double-stranded plasmids were prepared, and both strands were sequenced by the dideoxynucleotide chain termination method (21) using [cI-~'S]~ATP as the radiolabeled nucleotide. Sequenase 2.0 was used following the supplier's protocals for the reactions (U. S. Biochemical Corp.). The sequencing gels were 8 M urea and 8% acrylamide. DNA sequences were analyzed on a VAX computer using the University of Wisconsin Genetics Computer Group Program and compared with Genbank and EMBL sequence banks (22).
RNA and DNA Isolation-High molecular weight genomic DNA was isolated from liver tissue after mice were starved overnight. The procedure used was a modification of other methods. Liver from a single animal was homogenized in 10 ml of 0.1 M EDTA, 1% SDS, 10 mM Tris-HC1 (pH 8.0) by 2-3 short bursts of a polytron homogenizer.
The homogenate was treated overnight with 100 pg/ml Proteinase K at 56 "C with rotary shaking. Two phenol/chloroform, two chloroform, and four ether extractions followed. The aqueous sample was then incubated at 37 "C for 4 h in the presence of 100 pg/ml of RNAse A and 1000 units/ml of RNAse T1 followed by 100 pg/ml of pronase for 4-12 h. After one phenol/chloroform and one chloroform extraction, the DNA was precipitated twice with ethanol and dissolved in TE (10 mM Tris-HC1, pH 7.0, 1 mM EDTA).
Total cellular RNA was isolated (14) and analyzed for integrity by agarose gel electrophoresis in 1 X TBE (0.089 M Tris-borate, 0.089 M boric acid, 2 mM EDTA). RNA purity was determined by the A~w / Northern and Southern Analyses-Electrophoresis of RNA in formaldehyde denaturing agarose gels, blotting, and hybridization under stringent conditions were done as previously described (17). X DNA digested with Hind111 was used as molecular size markers.
Electrophoretic Analysis of Tissue Alcohol Dehydrogenases-Twenty percent tissue extracts were prepared in 50 mM Tris-HC1 (pH 7.0) containing 1 mM dithiothreitol at 4 "C using a polytron homogenizer on medium setting for 20 s. Supernatants were prepared by centrifugation for 20 min at 27,000 X g. The supernatant was added (90 pl) to fill the slots in 12% starch gels made with 8 mM Tris/3 mM citric acid (adjusted to pH 7.2 with NaOH). The gel was connected to electrode-containing buffer chambers containing the same buffer 27.5 times more concentrated, and electrophoresis was conducted for 14 h at 7 V/cm at 4 "C. The gels were 18 X 31 X approximately 0.8 cm. Enzyme activity in the horizontal gel slices was visualized using various alcohols in the histochemical staining method described by Holmes et al. (24).

RESULTS
Cloning and Sequencing of Adh-1 cDNA-About 1 X lo4 recombinant X g t l O phage from the ADH-positive liver cDNA library were screened with a mouse class I ADH cDNA probe. Nearly 100 positive plaques were initially picked from this library, and two were initially chosen for sequencing and further study after subcloning the cDNA inserts into pGEM plasmids. Initially, the complete sequence of the 0.9-kb insert from pADHF72 was obtained. Since the cDNA in pADHF72 was not full-length, the library was rescreened with a 5'fragment obtained from a full-length mouse cDNA clone called pCK1.' Additional clones were obtained using this probe, and hADH12-3 was further analyzed and found to contain a larger cDNA insert of about 1.4 kb. The insert was subcloned into pGEM and subsequently called pADH12-3 from which the additional 5'-end sequence was obtained. The complete sequence of the deer mouse class I ADH cDNA representing the Adh-l gene was determined from these two clones and is presented in Fig. 1. Both strands were sequenced, as were all overlaps. The encoded amino acid sequence is 374 amino acid residues and has a relatively short 3"untranslated region. A polyadenylation signal (AATAAA) is located upstream of the poly(A) tract. The 5"untranslated region is 47 nucleotides, but this cDNA may not represent the full transcript.
Cloning and Sequencing of Adh-2 cDNA-An additional clone, XADHnl, was identified by a faint hybridization signal in the ADH-negative cDNA liver library. Further comparison identified a related clone, XADHF65, which had been isolated from the ADH-positive library. Inserts from both of these XADH clones were subcloned into pGEM plasmids. The sequence of the cDNA insert in pADHnl is shown in Fig. 2. This clone was sequenced on both strands, and all overlaps were determined. An open reading frame encoding a 374amino acid sequence is found within this sequence. The sequence has 135 nucleotides of 5"untranslated sequence and a long 612-nucleotide 3"untranslated sequence. A polyadenylation signal (AATAAA) is located upstream of the polyadenylation site.
Comparison of Deduced Amino Acid Sequences-The deduced amino acid sequence of the deer mouse Adh-l cDNA is 94% identical to the mouse Adh-I (16, 25) deduced sequence; therefore, the deer mouse Adh-1 gene clearly encodes a class I enzyme. This is further confirmed by comparison with the deduced amino acid sequence encoded by the six human genes ( Table I). The human ADH1, ADH2, and ADH3 genes all encode structurally closely related class I enzymes, and the deer mouse Adh-1-encoded sequence exhibits greatest similarity to these sequences. The encoded amino acid sequence C. K. Boyle and M. R. Felder, unpublished results.
FIG. 1. Nucleotide sequence of the Adh-1 cDNA. Nucleotides 1-520 were obtained from the insert in pADH12-3, and nucleotides 440-1304 were obtained from pADHF72. The overlapping sequence was identical in the two clones, and the insert in pADH12-3 is essentially full-length based upon insert size. The asterisk indicates the stop codon. Nucleotides are numbered in the left column, and amino acids are numbered in the right column.

MGTTTMCCTAGATCCACTGATTACCCACACCCTGACTCTCGATMGGT~T~~TTCAGCTCATG K F N L D P L I T H T L T L D K V N E A I Q L M
362 ~CGGGCAATGTATCCGCTGTGTCCTGTTACCTTAGTTACAAGAGCTGCAGTATTTCATCGCT~CTTG K N G Q C I R C V L L P .

7 4
FIG. 2. The Adh-2 cDNA nucleotide sequence. The complete sequence of the insert in pADHnl is presented. Numbering is as in Fig. 1. The asterisk indicates the stop codon. The (0) denotes those 9 amino acid residues conserved in 47 members of the zinc-containing ADH family, and the (m) indicates those additional conserved residues when ("crystallin is excluded (33). ADH2, and ADHB (26,27); ADH4 (28); ADH5 (29); ADHG (12). The a Deduced amino acid sequences were obtained as follows: ADH1, ADH1, ADH2, and ADHB genes all encode class I ADHs. ADH4 and ADH5 encode classes I1 and 111, respectively. ADHG encodes the recently identified distinct class, which is equally similar to the other three classes.

FIG. 3. Northern analysis of the expression of the Adh-Z and Adh-2 genes in liver of ADH-negative and -positive mice.
Lanes were loaded with 10 pg of total liver RNA from ADH-positive mice ( 1 ) and ADH-negative mice (2). The blots were probed with the Adh-1 cDNA insert from pADHF72 ( A ) , the Adh-2 cDNA insert from pADHnl ( B ) , and an equal mixture of the two probes (C). The approximate sizes of the detected species are given on the right side of the figure.
of Adh-1 is about equally distant from the other three classes represented by the human ADH4, ADH5, and ADHG genes. The amino acid sequence encoded by deer mouse Adh-2 is about equally similar (5148%) to the human class I, 11, and I11 ADHs and is 67% identical to the additional class defined by the ADH6-deduced sequence, where the greatest similarity is found.
Expression of the Adh-1 and Adh-2 Genes in ADH-negative and -positive Deer Mice-The cDNA inserts of pADHF72 and pADHnl were used as probes to study the expression of the Adh-1 and Adh-2 genes, respectively, in ADH-positive andnegative deer mice. ADH-1 and ADH-2 mRNAs of different size are observed in the liver (Fig. 3). The ADH-1 mRNA is approximately 1.5 kb in size and is absent in ADH-negative deer mice (Fig. 3A). The ADH-2 mRNA is nearly 1.9 kb and is present in the liver of both ADH-positive and -negative animals (Fig. 3B). The simultaneous use of both the Adh-1 and Adh-2 probes (Fig. 3C) clearly demonstrates the absence of the 1.5-kb ADH-1 mRNA in the ADH-negative deer mice. ADH-positive deer mice were used to examine the tissuespecific expression of the two genes. The Adh-1 gene was expressed at the highest levels in liver, kidney, and adrenal gland and, to a lesser extent, in seminal vesicle tissue (Fig.   4A). The Adh-2 gene was expressed at a high level only in liver with a faintly detectable signal seen in kidney RNA. Expression was not detectable in the other tissues examined (Fig. 4B).
Southern Analysis of the Adh-1 and Adh-2 Genes in ADHpositive and -negative Deer Mice-Southern blot analysis was used principally to examine the nature of the Adh-1 gene in ADH-negative and -positive animals. When DNA from ADHpositive deer mice is digested with restriction enzymes and the separated fragments are probed with Adh-1-specific sequences from the nearly full-length insert in pADH12-3, a single copy sequence appears to be detected. However, faintly detectable bands are also seen in the autoradiograph (Fig. 5). The strongly hybridizing bands detected in ADH-positive DNA with the Adh-1 probe are not seen in DNA from ADHnegative animals. Only very faintly hybridizing DNA restriction fragments are seen in the ADH-negative DNA, and these correspond closely to those weak signals seen in the ADHpositive DNA. This suggests that the ADH-negative mice are due to a deletion of all or most of the Adh-1 gene.
Both ADH-positive and -negative deer mice contain the Adh-2 gene (Fig. 5). This provides a good control, since the same DNA restriction digests were analyzed separately with the Adh-1 and Adh-2 probes. Some restriction fragments that faintly hybridize with the Adh-1-specific probe hybridize strongly with the Adh-2-specific probe. As examples, these include the 4.0-kb PstI fragments found in both genetic stocks, the approximately 4.9-kb EcoRI fragments, and the 6.6-kb PvuII fragments. One clear example is the HindIII polymorphic restriction fragment, which is 1 kb in ADH-positive mice and 0.8 kb in ADH-negative mice. This is easily detectable with the Adh-2 gene probe but is only faintly detectable with the Adh-1 gene probe. Another HindIII restriction site polymorphism is detected only with the Adh-2 probe. This polymorphism is the approximately 2-kb fragment in the ADHpositive sample and the nearly 5-kb fragment in the ADHnegative sample. It is not surprising to find such DNA polymorphisms either between or within these two stocks, since the animals are not inbred lines or congenic with one another.
Biochemical Detection of Tissue Alcohol Dehydrogenuses- The enzyme activities in various deer mouse tissues were analyzed by starch gel electrophoresis and histochemical detection. Electrophoretic mobility, tissue specificity, and substrate utilization of the ADHs detected in the deer mouse are similar to ADHs in the mouse, and the dimeric subunit structure of the isozymes is designated A2, Bz, and C2 as in the mouse system (24,30,31). With 20 mM ethanol as substrate, the cathodally migrating class I ADH-A2 is apparent in liver and kidney of ADH-positive mice (Fig. 6A, lanes 1 and 2) but is deficient in ADH-negative mice (lanes 7 and 8). This isozyme is the product of the Adh-1 gene. Anodally migrating enzyme activity, designated B2, is present mostly in liver of both stocks of animals, and a small amount of activity is also observed in the lung. The B2 designation on the gel is actually resolved into two bands of activity. At higher ethanol concentrations (250 mM), the stomach C2 enzyme is detectable (Fig. 6B), and this form is also detectable using benzyl alcohol and trans-2-hexene-1-01 as substrates (Fig. 6, C and D). The B2 isozyme seems to have broad substrate specificity using benzyl alcohol and trans-2hexene-1-01 efficiently as substrates. The A2 form is detectable with these substrates but stains less intensely than with ethanol. The liver has the highest level of expression of the activity designated ADH-B2, and the tissue specificity is similar to the expression of the Adh-2 gene as measured by Northern analy- The blots were probed with the Adh-I cDNA insert in pADHF72 ( A ) and the Adh-2 cDNA insert in pADHnl ( E ) . The RNA sample from each tissue was denatured, divided, and analyzed separately with the two probes. -negative ( N lanes) deer mice were digested with various restriction enzymes, and equal amounts of the digestion products were loaded onto each of the two gels, resolved by electrophoresis, and blotted. The restriction endonucleases used were as follows: H, HindIII; P, PstI; B, BamHI; E, EcoRI; and Pu, PuuII. The two blots were probed with Adh-I-specific (insert from pADH12-3) and Adh-2-specific (insert from pADHnl) cDNA sequences as indicated on the figure. The positions of X DNA HindIII fragments are indicated in kb. sis (Fig. 4). In the deer mouse, the B2 region does have slower and faster moving components, and this region may contain a previously unrecognized form of alcohol dehydrogenase. The lung appears to possess only the slower moving component in the B2 region, but this may be lack of detection due to the low amount of overall activity in this tissue.

DISCUSSION
In this report, the molecular basis of the ADHs in the deer mouse has been investigated. The ADH-positive and -negative stocks have been widely used in studies on ethanol metabolism, and the molecular basis of the ADH-negative variant is now better understood. Two different cDNAs for ADH have been obtained and sequenced from the deer mouse. One complete cDNA sequence was obtained from two overlapping clones isolated from the ADH-positive liver cDNA library. This cDNA sequence contained a 374-amino acid open reading frame, which shared 94% sequence identity with the mouse Adh-1-encoded class I ADH amino acid sequence (16,25). This deer mouse class I ADH sequence is designated as being encoded by the Adh-1 gene. Further support that the deer mouse Adh-1 gene encodes a class I ADH is the greater than 1.9Kb 1.5Kb 80% sequence identity at the amino acid level with the three class I human ADHs (26,27). Most importantly, the ADHnegative mice do not produce an mRNA in liver, which is detectable by hybridization to the Adh-1 cDNA probe. Furthermore, Southern analysis has shown that the Adh-l gene is substantially or entirely deleted in the ADH-negative deer mice.
The other cDNA sequence was detected by faint hybridization signals found when the ADH-negative cDNA library was screened with a mouse Adh-1 cDNA. The cDNA sequence in the clone obtained from the ADH-negative library was also found to encode a 374-amino acid polypeptide. This amino acid sequence was found to be only 57% identical to the amino acid sequence encoded by the deer mouse Adh-1 gene. This cDNA is designated as representing the mRNA of the deer mouse Adh-2 gene. When the Adh-2-encoded amino acid sequence is compared with the sequences encoded by the six known human alcohol dehydrogenase genes, the sequence was only 51-58% identical to five of the sequences representing class I, 11, and I11 ADHs. The deer mouse Adh-2-encoded sequence is 67% identical to the sequence encoded by the human ADHG gene, which represents the recently identified additional human class of ADH (12). Furthermore, an identity of only 50% was found between the deer mouse Adh-2-encoded sequence and the partial rat stomach ADH sequence (173 available amino acid residues), which is the only additional known mammalian class of ADH (11). This suggests that the deer mouse Adh-2 gene encodes a new enzyme class not represented by any of the six known human genes representing four classes and the rat stomach ADH representing a fifth class of mammalian ADH. The Adh-2-encoded protein has the highest sequence identity with the protein product of the ADHG gene, but the human ADHG gene is expressed in liver and stomach, whereas the deer mouse Adh-2 gene is not expressed in stomach.
The protein encoded by the deer mouse Adh-2 gene contains all 13 residues (Fig. 2) conserved in 47 members of the zinccontaining ADH family excluding {"crystallin (33). N' me residues are conserved when {"crystallin is included, and 8 of these are glycine, suggesting a side chain in these locations would disrupt a structure required for enzymatic function. One cluster of strictly conserved residues located in the substrate-binding domain has glycine at positions 66, 71, 77, and  192,201, 204, and 236. The four additional residues that are conserved if {"crystallin is excluded are all related to zinc binding. Ligands to the catalytic zinc are Cys-46 and His-67; Asp-49 and Glu-68 have been shown to affect the electrostatic environment near the yeast ADH catalytic zinc. Sequence alignment of the Adh-&encoded sequence with other animal ADHs indicates an insertion of Cys-62 and a deletion after the Thr-123. The deer mouse ADH-2 sequence also has Asp-223, which has been suggested to determine coenzyme specificity, and Thr-48. Either Ser-48 or Thr-48 is conserved in all the ADHs except {"crystallin and is thought to be hydrogen-bonded to the alcohol hydroxyl group bound to the catalytic zinc. Cysteines 97, 100, 103, and 111, which are responsible for binding the noncatalytic zinc, are conserved in the ADH-2 protein and in all the ADHs except for the enzymes from two bacterial sources.
Among 18 animal ADHs, 116 residues are conserved (33). The ADH-2 amino acid sequence is identical at 103 of these positions. Seven of the substitutions are conservative, and 6 of the substitutions involve amino acid replacements with partial similarity.
Expression of the Adh-1 and Adh-2 genes in tissues of the deer mouse is substantially different. The ADH-1 mRNA is found at high levels in liver, kidney, and adrenal gland in ADH-positive mice with a much lower level being found in seminal vesicle tissue. In contrast, ADH-2 mRNA is detected only at high level in liver and at very low level in kidney. While ADH-1 mRNA is not detectable in liver of ADHnegative mice, ADH-2 mRNA is found in liver of both stocks of animals. The ADH-2 mRNA is about 400 nucleotides longer than the ADH-1 mRNA, as measured by Northern analysis, and this is seemingly due to the substantially longer 3'untranslated region in the ADH-2 mRNA as determined from the cDNA sequence.
An effort was made to correlate molecular expression of the Adh-1 and Adh-2 genes with ADH enzyme activities in various tissues. ADH-positive liver and kidney supernatants possess a basic ADH protein (ADH-A2) (Fig. 6) with activity at low ethanol substrate concentrations. This activity is deficient in the ADH-negative deer mice correlating with the expression of the Adh-1 gene. Both stocks of mice have an acidic liver ADH activity (ADH-B2) (Fig. 6), which is detectable with ethanol at low substrate concentrations and with other alcohols. This ADH isozyme seems to actually be resolved into two forms in the gel electrophoresis system used. The liverspecific expression of ADH-B2 correlates with the molecular expression of the Adh-2 gene. However, the mouse ADH-BZ cDNA has recently been sequenced and is 87% identical at the nucleotide level to the human class I11 ADH (34). The mouse ADH-B2 cDNA nucleotide sequence is only 62.9% identical to the deer mouse Adh-2 cDNA sequence, suggesting these are not orthologous genes. That the gel system employed here resolves the ADH-B2 region into two zones of activity makes it a possibility that this region of activity may be the result of expression of two genes, one of which could be Adh-2. An effort was made to identify an electrophoretic variant for the ADH-B2 region between Peromyscus polionotus and P. maniculatus, but none was found. This could have determined whether more than one gene encodes the two bands of activity in this region. The multiple bands in the ADH-A:! form are due to differential binding of NAD cofactor (1) and are the product of a single gene. The role the ADH-B2 activity and the product of the Adh-2 gene may play in ethanol metabolism remains unclear. Part of the residual ethanol metabolism found in the ADH-negative mice may be due to these ADHs, and some may be due to the cytochrome P450mediated microsomal ethanol-oxidizing system.
Not only do both ADH-positive and -negative deer mice express ADH-B2 in liver tissue, but both stocks express ADH-C2 in stomach tissue, as does the mouse (24, 30, 31). The isozyme is detectable with high ethanol substrate concentrations and with benzyl alcohol and trans-2-hexene-1-01 as is the mouse isozyme (24,31).
Not only are the ADH-negative deer mice reduced in their ability to metabolize ethanol, but they are also greatly reduced in their ability to synthesize retinoic acid from retinol (35), although the mice appear to possess the ability to produce retinoic acid a t physiologically required levels. In the liver and kidney, ADH-negative deer mice retain about one-eighth the ability to metabolize retinol as found in Adh-positive deer mice. These results would suggest that the protein product of the Adh-2 gene is not involved in this residual metabolism, since the Adh-2 gene is expressed at near negligible level in kidney.
The ADH-negative deer mice are here clearly shown to be due to a deletion of the Adh-1 gene, which encodes a class I ADH. A number of protein deficiencies are known in mammalian systems, but not a large number are due to substantial or entire gene deletions. A multiexon deletion of about 9 kb in the procollagen I11 gene in humans is known to cause a mild disorder (36). The entire gene for arylamine N-acetyltransferase is deleted in a genetically identified group of rabbits deficient in the ability to metabolize certain drugs (37). The deletion of the p-hexosaminidase a-chain in some French Canadians with Tay-Sachs disease appears to be associated with Alu sequences flanking the deleted sequences (38). A deletion of a large portion of the gene for the low density lipoprotein receptor is the molecular basis of familial hypercholesterolemia. This deletion is also flanked by Alu sequences (39). The possible molecular mechanism by which the Adh-1 gene became deleted in the ADH-negative deer mouse is currently unknown.