Organization of six functional mouse alcohol dehydrogenase genes on two overlapping bacterial arti®cial chromosomes Gabor Szalai 1 , Gregg Duester 2 , Robert Friedman 1 , Honggui Jia 3 , ShaoPing Lin 3 , Bruce A. Roe 3 2 and Michael R. Felder 1 1 Department of Biological Sciences, University of South Carolina, Columbia, USA; 2 Gene Regulation Program, The Burnham Institute, La Jolla, CA, USA; 3 Department of Chemistry and Biochemistry, University of Oklahoma, Norman, OK, USA Mammalian alcohol dehydrogenases (ADH) 3 form a com- plex enzyme system based on amino-acid sequence, func- tional p roperties, and gene expression pattern. At least four mouse Adh genes are known to encode dierent enzyme classes that share less than 60% amino-acid seq uence iden- tity. Two ADH-containing and overlapping C57BL/6 bacterial arti®cial chromosome clones, RP23-393J8 and -463H24, were identi®ed in a library screen, physically mapped, and sequ enced. The gene order in the complex and two new mouse genes, Adh5a and Adh5b, and a pseudogene, Adh5ps, were obtained from the physical map and sequence. The mouse genes are all in the same transcriptional orien- tation in the order Adh4-Adh1-Adh5a-Adh5b-Adh5ps-Adh2- Adh3. A phylogenetic tree analysis shows that adjacent genes are most closely related suggesting a series of duplication events resulted in the gene complex. Although mouse and human ADH gene clusters contain at least one gene for ADH classes I±V, the human cluster contains 3 class I genes while the mouse cluster has two class V genes plus a class V pseudogene. Keywords: alcohol dehydrogenase; mouse; gene complex. The alcohol dehydrogenases (ADHs; EC 1.1.1.1) are zinc- containing, dimeric enzymes fo und in the cytosolic fraction of the cell that are capable of reversible oxidation of a spectrum of primary and secondary alcohols to the corre- sponding aldehydes and ketones. Mammalian ADHs currently known a re grouped into six distinct classes [1±3] with members of different classes sharing less than 70% amino-acid sequence identity within a species. Humans possess classes I, II, III, IV and V [2] whereas the mouse is known to have expressed genes encoding ADHs of class I , II, III and IV [4,5]. Humans have three class I genes [6] encoding proteins with greater than 90% positional identity whereas the mouse has a single class I g ene [7,8]. A human class V ADH with approximately 60% positional identity a t the amino-acid level with the other human classes has been revealed from cDNA encoded sequence but not yet associated with an enzyme [9]. Deermouse [10] and rat [11] ADH c DNAs have been isolated that would e ncode proteins most closely related to this human class V with the deermouse ADH cDNA encoding a protein with 67% positional identity to this class. The ADH family of enzymes perform important meta- bolic functions. In the mouse, class III functions as a glutathione-dependent formaldehyde dehydrogenase 4 [12] and is involved in S-nitrosoglutathione metabolism [13] and retinol metabolism (G. Duester, unpublished results). Class IV has a n important role in retinol metabolism leading to retinoic acid p roduction. Adh4 is expressed during embry- ogenesis [14,15] and Adh4 null mouse mutants have reduced fetal survival during vitamin A de®ciency [16]. This obser- vation coupled with reduced conversion of retinol to retinoic acid in tissues of Adh4 mutant mice [12] suggests a role for class IV enzyme i n retinoid signaling during embryogenesis. Expressed at high levels in liver, the role of class I in ethanol metabolism has been con®rmed by natural [17] and engineered [12] enzyme-de®cient animals. Adh1 null mutant mice also demonstrate a signi®cant decrease in metabolism of retinol to retinoic acid [12]. The in vivo physiological roles of class I, III, and IV ADHs in the mouse are being unveiled with the development of targeted d isruptions for e ach gene. The mouse Adh1 and Adh4 genes are lin ked [18] on chromosome 3 at 71.2 cM (Mouse Genome Database), and this complex is a candidate region for a quantitative trait locus (Alcp3) for alcohol preference [19]. The genes encoding different classes of mammalian ADH have different tissue expression patterns suggesting that Correspondence to M. R. Felder, Department of Biological Sciences, University of South Carolina, Columbia, SC 29208, USA. Fax: + 803 777 4002, Tel.: + 803 777 5135, E-mail: felder@mail.biol.sc.edu Abbreviations: BAC, bacterial arti®cial chromosome; NJ, neighbor-joining; PC, Poisson corrected; CHEF I , contour-clamped homogenous electric ®eld. De®nitions: The nomenclature for the alcohol dehydrogenase gene and enzyme family being followed is that of Duester, G ., Farre  s, J., Felder, M.R., Holmes, R.S., Ho È o È g, J O., Pare  s, X., Plapp, B.V., Yin, S J. & Jo È rnvall, H. (1999) Biochem. Pharmacol. 58, 389±395. The previously used human ADH1, ADH2, ADH3 genes encoding class I enzymes are renamed ADH1A, ADH1B,andADH1C indicating their greater than 90% similarity. The enzyme class names and g en e names now corre- spond. The class II human and mouse enzymes are encoded by ADH2 and Adh2, r espectively; class III enzymes by human ADH3 (old ADH5)andmouseAdh3 (old Ad h5); class IV enzymes by human ADH4 (old ADH7)andmouseAdh4 (old Adh3); class V enzyme by human ADH5 (old ADH6). The deermouse c lass VI ADH is closest to human ADH5 protein in amino -acid sequence, and based on phy- logeny and genome organization of t he mouse presented here is now referred to as class V with a gene symbol of Adh5. New mouse genes reported here are named Adh5a, Adh5b,andAdh5ps. (Received 2 0 September 2001, accepted 30 October 2001) Eur. J. Biochem. 269, 224±232 (2002) Ó FEBS 2002 complex regulatory mechanisms control this gene family. The mouse Adh1, Adh2, Adh3,andAdh4 genes h ave some overlap in tissue expression pattern, but levels of expression in different adult tissues and hormonal responsiveness of different genes in the complex vary widely [4,20±23]. cis-Acting sequences controlling the expression pattern of the Adh genes in mouse are not fully understood. A minimal promoter directing expression in transfected hepatoma cells has been de®ned for Adh1 [24,25]. How- ever, as much as 10 kb of sequence upstream of an Adh1 minigene does not direct expression in liver of transgenic mice [26] although kidney a nd adrenal e xpression is promoted. An entire Adh1 gene containing 7 kb of 5¢ and21kbof3¢ ¯anking sequence in transgenic mice expresses properly in most t issues except live r and intestine [27]. More distal sequences must control expression of Adh1 in these tissues, and this has stimulated an effort to obtain mouse Adh1 bacterial arti®cial chromosome (BAC) clones to identify these cis-acting regions. Analysis of these BAC clones has provided a more detailed knowledge of the genomic organization of the mouse Adh gene complex. This is an important prerequisite to understanding the regulatory strategy of this gene family. In this report, we show that the previously known four mouse Adh genes a re clustered within 250 kb of the mouse genome. Furthermore, two additional Adh genes and one pseudogene are found within this DNA region, and the order of these genes on two overlapping BAC clones has been determined. All genes are transcribed in the same orientation, and s equence comparisons and location within the complex suggests that the two newly identi®ed Adh genes are most closely related to the human ADH5 gene. MATERIALS AND METHODS Materials Chemicals a nd reagents were obtained f rom Sigma Chem- ical Co., Fisher Chemicals, J. T. Baker (Phillipsburg, NJ, USA), and New England Biolabs unless otherwise indicated. RPCI-23 segment 2 BAC library was purchased from BACPAC Resources of Roswell Park Memorial Institute (Buffalo, NY, USA). Individual clones w ere also purchased from this resource. CDNA hybridization probes for mouse genes The ADH cDNA probes used in this study were inserts from pADHn1 f ordeermouse ADH5 andpCK2 for m ouse ADH1 [10]. The ADH 4 and ADH3 cDNAs have been previously described [4]. Mouse ADH2 cDNA was prepared using methods previously described [4] by performing RT-PCR on mouse liver RNA with primers overlapping the start and stop codons of the published mouse ADH2 cDNA sequence [5]. The ADH2 cDNA was con®rmed b y nucleotide sequence analysis. Isolated cDNA inserts were used to prepare all probes for labeling by random priming [28]. Screening the BAC library for mouse Adh1 -containing clones Five ®lters containing a total of over 90 000 clones were screened by hybridization with 32 P-labeled mouse ADH1 full-length insert cDNA in pCK2. Probe was used in the hybridization solution at 2.4 ´ 10 6 c.p.m.ámL )1 . Filters were prehybridized for ³ 2 h at 65 °Cin6´ NaCl/P i /EDTA (1 ´  0.15 M NaCl, 0.01 M sodium phosphate, 1 m M EDTA, pH 7 .4)/6 ´ Denhardt's solution, 0.5% SDS, and 50 lgámL )1 denatured salmon sperm DNA. After prehy- bridization, the solution was discarded a nd replaced with an identical hybridization solution except without Denhardt's containing the radioactive probe. After overnight hybrid- ization the ®lters were washed twice with 6 ´ NaCl/ P i / EDTA/0.5% SDS at room temperature for 15 min each, twice in 1 ´ NaCl/P i /EDTA/0.5% SDS at 37) 42 °Cfor 15 min each, and once in 0 .1 ´ SSPE 5 /0.5% SDS at 65 °C. The ®lters were blotted of exc ess moisture, wrapped in Saran wrap, and exposed to Kodak XAR ®lm at )80 °Cfor several hours to a day depending upon intensity of signal and level of background necessary to reveal outline of the ®elds on the ®lter. BAC DNA isolation and restriction enzyme digest analysis A s ingle isolated bacterial colony containing a BAC clone of interest was obtained from a freshly streaked plate and used to inoculate 300 mL of Luria±Bertani media containing 20 lgámL )1 of chloramphenicol. After overnight gr owth at 37 °C, cells were harvested by centrifugation. Isolation of BAC DNA was by a rapid alkaline lysis method using no organic extractions following the protocol provided by BACPAC Resources. P ipetting of the ®nal BAC DNA preparation was carried out with large ori®ce tips from USA Scienti®c. The BAC DNA was digested with various restriction endonucleases and the fragments were resolved o n 1 % agarose gels in 0.5 ´ Tris/borate/EDTA (1 ´ Tris/borate/ EDTA  0.089 M Tris, 0.089 M boric acid, 0 .002 M EDTA) using a contour-clamped homogenous electric ®eld (CMHF) apparatus. Electrophoresis was conducted at 6 Vácm )1 at 15 °C for 16 hrwith the switch time ramped from 1 to 12 s. After completion of electrophoresis, gels were stained in 0.5 lgámL )1 ethidium bromide in 0.5 ´ Tris/borate/EDTA and washed for several hours in distilled water. After photographing, the gels were treated with 0.25 M HCl for 15 min, rinsed in distilled water, denatured with 0.5 M NaOH/1. 0 M NaCl, and neutralized with 0.5 M Tris/HCl/ 1.5 M NaCl (pH 7.4). The gels w ere inverted and DNA was transferred overnight onto Hybond Nytran membranes in 10 ´ NaCl/P i /EDTA by capillary movement. After trans- fer, membranes were baked for 2 h at 80 °C. BAC isolation, shotgun sequencing, and custom-synthetic primer directed closure The detailed procedures for cloned, large insert genomic DNA isolation, random shot-gun cloning, ¯uorescent- based DNA sequencing a nd subsequent analysis were used as described previously [29±31]. Brie¯y, BAC DNA was isolated free from host genomic DNA via a cleared lysate- acetate p recipitation-based protocol [ 29]. Subsequently, 50 lg portions of puri®ed BAC DNA were randomly sheared and made blunt ended [30,31]. After kinase treatment and gel puri®cation, fragments in the 1- to 3- kb range were ligated into SmaI-cut, bacterial alkaline Ó FEBS 2002 Mouse alcohol dehydrogenase gene complex (Eur. J. Biochem. 269) 225 phosphatase-treated pUC18 (Pharmacia) and Escherichia coli, strain XL1BlueMRF¢ (Stratagene), was transformed by electroporation. A random library of approximately 1200 c olonies were picked fr om each transformation, grown in t erri®c broth [32] s upplemented with 100 lgámL 7 )1 of ampicillin for 14 h at 37 °C with shaking at 250 r.p.m., and the sequencing templates were isolated by a cleared lysate-based protocol [30]. Sequencing reactions were performed as previously described [31] using Taq DNA polymerase, the Perkin Elmer Cetus ¯uorescently labeled Big Dye Taq termina- tors. The reactions were incubated for 60 cycles in a PerkinElmer Cetus DNA Thermocycler 9600 and after removal of unincorporated dye terminators by ®ltration through Sephadex G-50, the ¯uorescently labeled nested fragment sets were resolved by electrophoresis on an ABI 3700 Capillary DNA Sequencer. After base calling with the ABI analysis software, the analyzed data was trans- ferred to a Sun workstation cluster, and assembled using the PHRED and PHRAP programs [33,34]. Overlapping sequences and contigs were analyzed using CONSED [35]. Gap closure and proofreading was performed using either custom primer walking or using PCR ampli®cation of the region corresponding to the gap in the sequence followed by subcloning into pUC18 and cycle sequencing with the universal pUC-primers via Taq terminator chemistry. In some instances, additional synthetic custom primers were necessary to obtain at least threefold coverage for each base. Draft and ®nished BAC s equences were analyzed on Sun workstations with the programs contained within the GCG package [36] as well as the BLAST [37], BEAUTY [38] and BLOCKS [39] programs. The sequences of BAC clones RPCI23-463H24 and RPCI23-463H24 have been deposited into GenBank and give n accession numbers AC079823 and AC079682, respectively, Computer analysis of DNA sequence CLUSTAL W (version 1.81) was used to align the amino-acid sequences [40] for phylogenetic treatments. Any site at which the alignment postulated a gap in any sequence was removed from the data set for all pairwise comparisons so that a s imilar data set was used for each comparison. Phylogenies were constructed u sing the following methods: (a) the neighbor-joining (NJ) method [41] based on the Poisson corrected (PC) amino-acid distance [42] and (b) th e NJ method based on the gamma-corrected (a  2.0) amino- acid distance [42]. Both methods produced es sentially identical results; therefore, only the results of the NJ tree based on PC are presented here. The NJ method is re liable a t reconstructing phylogenies even when evolutionary rates differ among branches of a phylogenetic tree [42]. The reliability of clustering patterns in NJ trees was tested by bootstrapping [43], which involved clustering of trees based on pseudos- amples of sites sampled in the data set (with rep lacement). Five-hundred bootstrap pseudosamples were used. GenBank and EST database searches were performed using BLAST programs [44] (available at http:// www.ncbi.nlm.nih.gov/BLAST website). Sequence analysis and manipulation 8 were carried out using the GCG software package. RESULTS Identi®cation of Adh1 -containing BAC clones DNA was i solated from all BAC clones hybridizing to the ADH1 cDNA probe. Puri®ed DNA was digested with EcoRI and analyzed by Southern blotting and hybridization to ADH1 cDNA. Finally, 13 positive clones were identi® ed and among these were found all EcoRI restriction fragments detectable in genomic DNA with an ADH1 cDNA probe. The Adh1-containing EcoRI fragments in C57BL/6 DNA are 6.8-, 3.8-, 2.5- and 1.1 kb. In addition, more faintly hybridizing 2.0- and 2.3 kb EcoRI fragments present in the genome were identi®ed among the BAC clones. Of these 13 clones, tw o were chosen for e xtensive analysis because they overlap within the Adh1 gene. BAC 463H24 contains the entire Adh1 gene while 393J8 contains only t he most 3¢ end, the 3.8- a nd 1.1-kb Eco RI [7] h ybridizing fragmen ts (Fig. 2 A). T herefore, 393J8 extends farther downstream than 463H24 relative to Adh1 orientation. 393J8, in contrast to 463H24, includes the 2.0 k b non-Adh1 [7] hybridizing fragment as does another BAC clone 461A12. As 461A12 contains no Adh1 sequence, this con®rms that the cross- hybridizing 2.0-kb EcoRI fragment is downstream of the Adh1 g ene. T wo additional B AC clones, 388D7 a nd 434F16, were identi®ed that contained the 2.3-kb EcoRI Adh1-crosshybridizing fragment [7], but n o Adh1 sequence, as the only species to hybridize to ADH1 cDNA. Adh genes on 463H24 BAC 463H24 restriction endonuclease digests were pro- duced, and the resulting fragments were resolved by 9 CHEF analysis. A s estimated by the sizes of the r estriction fragments, 463H24 contains an insert of over 165 kb (Fig. 1 A). The blots prepared from CHEF separation of restriction fragments were hybridized to several mouse ADH cDNA clones. Only ADH1 and ADH4 cDNAs hybridize to restriction fragments in 463H24 (Fig. 1A). The results of several restriction digests revealed that the positions of Adh1 and Adh4 are resolved on this clone. For example, Adh1 is locatedonthe60-kbRsrII fragment while Adh4 is found on the 100-kb fragment (Fig. 1 A, N/R). Either because of context or the sequence recognized by RsrII [CGG (A/T)CCG] t here was never complete digestion a t t his s ite as revealed by the remaining 160-kb undigested insert present in a nonstoichiometric amount. As expected, this undigested fragment hybridizes to both ADH1 and ADH4 cDNAs. Adh4 is found on the 135-kb Eag I fragment while Adh1 is found on the 135- and 25-kb fragments (Fig. 1A, N/E). Within t he Adh1 gene a single EagI site is f ound near the 3¢ end of exon 6 accounting for the two hybridizing fragments. This also localizes the 3 ¢ end of Adh1 approxi- mately 21 kb from one end of the clone. Adh1 and Adh4 are on the large SalIfragmentsuggestingthissiteisnearerthe opposite end of the clone th an EagI. As EagI, Rsr II, and SalI all cut the clone once and EagIandSalI are nearer the ends of the BAC, other double digests were performed to construct a more detailed map o f 463H24 (Fig. 2A). Two PmeI sites are present (Fig. 1A, N/P) and both Adh1 and Adh4 are found on the large 90-kb fragment. The location of the additional PmeI site was not determined from the digests performed. The RsrII/PmeI double digest 226 G. Szalai et al. (Eur. J. Biochem. 269) Ó FEBS 2002 (Fig. 1 A, N/R/P) delineates t he position of Adh1 and Adh4 (Fig. 2 A) on th e BAC. Adh genes on BAC 393J8 The p resence of the 3¢ end of Adh1 near one end of 393J8 provides a useful approach to determine the location of each restriction enzyme site located nearest this end by probing with ADH1 cDNA. Sizes of fragments produced by digestion of 393J8 with various restriction enzymes were determined by CHEF (Fig. 1B). The results of probing a blot of these CHEF separated restriction endonuclease fragments o f 393J8 with four ADH cDNA probes are shown in Fig. 1B. The single SalI site is located approxi- mately 30 kb from the Adh1-tagged end of the clone con®rmed by the strong hybridization signal of the 30-kb fragment when probed with ADH1 cDNA (Fig. 1B, N/S). The faint signal observed from t he large fragment must be due to cross-hybridization with other ADH genes. The only RsrII site is located a bout 100 kb from the Adh1 end of the clone (Fig. 1B, N/R). The single SgrAI site is locate d about 135 k b from the Adh1 end (Fig. 1B, N/Sg), and the EagIsite Fig. 1. CHEF analysis of restriction enzyme digests of BAC clone. The r estriction enzymes used to prepare BAC digests loaded into each lane ar e indicated at the top. The enzymes used were N, NotI; B, BssHII; E, EagI; P, PmeIR,RsrII; S, SalI; and Sg, SgrAI. The top panel in both (A) an d (B) shows the fragments detected by ethidium bromide staining. (A) Identi®cation of ADH-containing restric- tion fragments in BAC 463H24. Blots probed with ADH1 or ADH4 cDNAs are indicated. (B) Identi®cation of ADH-containing restric- tion fragments in BAC 393J8. Autoradiogram resulting from p robing identically prepared blots with ADH4, ADH2, ADH3 or deer- mouse ADH5 (mammalian class V) cDNAs are shown. Ó FEBS 2002 Mouse alcohol dehydrogenase gene complex (Eur. J. Biochem. 269) 227 located nearest Adh1 is over 60 kb (Fig. 1B, N/E) from the end (the largest EagI f ra gmen t). A DH1 c DNA a ls o hybridizes to a 75-kb PmeI fragment (Fig. 1B, N/P), but this band is actually composed of two nearly identical fragments. The smaller 12-kb PmeIfragmentisfromthe middle of this c lone. The observation t hat SalI, RsrII, and SgrAI digests not only have a strong signal from the fragment containing the 3¢ end of Adh1 but also a weak hybridization signal from the other fragment in t he digest suggests other Adh genes that weakly cross-hybridize to Adh1 are found on this clone. In EagI digests, ADH1 cDNA hybridizes to the 5-kb f ragment w hich contains the portion of the Adh1 gene located 5¢ of this site to the end of the clone, t o t he 60-kb fragment a lready mentioned, and t o the slightly smaller, weakly hybridizing fragment, which must contain other Adh genes. The ability to precisely locate the position of several restriction enzyme sites relative to the Adh1 end of the clone enabled a more detailed map to be constructed by analyzing additional double digests. The resulting blots were also hybridized with ADH2, ADH3 and deermouse ADH5 (Fig. 1B) cDNAs. ADH1 and deermouse ADH5 cDNAs both gave strong and weak signals suggesting that these probes cross-hybridize to other ADHs. For instance, ADH1 probe gives strong and w eak signals with the 30-kb and 140- kb NotI/ SalI fragments, respectively. However, deermouse ADH5 probe gives approximately equal signals in both fragments. The d eermouse ADH5 probe does c ross-hybrid- ize to Adh1 sequences, but the strong signal suggests there may be other cross-hybridizing sequences in the small fragment. Both mouse ADH2 and ADH3 cDNA probes hybridize only to the large fragment. Resolution of the positions of the genes on the BAC was obtained from single and double digests. The Adh2 gene is found on the large NotI/S grAI fragment whereas Adh3 is on the small fragment (Fig. 1 B, N/Sg). A map (Fig. 2A) of the two overlapping BACs was generated based upon estimated fragment sizes generated in single and double digests, hybridization of the different probes, and the ability to determine precisely the location of restriction sites relative to the Adh1 end of the BAC. The positions of Adh4, Adh2,andAdh3 are arbitrarily placed in the middle of the ¯anking restriction sites. Adh1 is anchored by the presence of the EagI site in the gene. T he deermouse ADH5 cross-hybridizing sequence is positioned encompass- ing restriction sites thought to reside in the hybridizing sequence. However, additional Adh sequences may reside between the position of Adh5 and Adh1 base d upon weak cross-hybridization signals observed with both cDNAs as probe and the fact that the EagI i s p laced w ithin Adh5, but could b e betw een relate d ge nes. However, the other nearby EagI site may be able to resolve t wo Adh5 cross-hybridizing species. The precise location of the mouse ortholog of deermouse Adh5 cannot be determined from mapping alone but was determined from sequence data (see below). Transcriptional orientation of Adh4 , Adh1 , Adh2 and Adh3 genes The draft sequence of 463H24 (AC 079682.14) consists of four unordered pieces and is s uggested to be approximately 182084 bp in length. BLAST analysis reveals that the Adh4 gene is on the 32 589±106 087 bp contig in that transcrip- tional orientation. The contig also contains three SalIsites, all located within 7-kb of each other, upstream from t he 5¢ end of the gene. These sites are too close to resolve on CHEF gels and must represent the single SalI site shown on the map (Fig. 2A). Based on the sequence, the PmeIsite on the m ap is loca ted upstream o f Adh4. T he sequence also identi®es the location of the add itional PmeI site which is near the 5¢ end of Adh4.TheRsrII site on the map is on a different contig (1 82 084±106 188) and is upstream of the 5¢ end of the Adh1 gene located on this contig. The single Eag I site located near the 3¢ end of exon 6 in the Adh1 gene localizes this gene on 463H24 and the draft sequence con®rms t he 5¢ to 3¢ orientation o f t his g ene. As one end of 393J8 begins in the 3¢ end of Adh1, this con®rms that Adh4 and Adh1 are transcribed in the same orientation . BLAST analysis of 393J8 complete sequence (AC 079832.16) with ADH1, ADH2, and ADH3 cDNAs also revealed that these genes have the same transcriptional orientation. The o verall length of the sequence is 194 850 bp. Identi®cation of Adh5a , Adh5b , and Adh5ps between Adh1 and Adh2 Sequences in 393J8 hybridizing to the deermouse ADH5 cDNA maps between Adh1 and Adh2 but closer to Adh2 (Fig. 2 A). It is possible that other sequences in this region may cross hybridize to ADH1 or deermouse ADH5 cDNA probes. BLASTN analysis of the region revealed that the exon 2±exon 6 deermouse ADH5-like sequence is found at position 97 694±110 93 4 in the BAC. Exon 6 sequence containsa9-bpdeletionfollowedbya7-bpstretchandthen a single b p deletion 10 when compared to the deermouse ADH5 or human ADH5 cDNAs. After 22 codons in the altered reading frame a stop codon is encountered strongly suggesting this is a nonfunctioning pseudogene, Adh5ps. Before the deletion, the encoded sequence has a p ositional identity of 24/26 amino-acid residues with human ADH5 and there is an identity of 29/57 after the deletion by returning to the proper reading frame. No coding regions beyond exon 6 were found by BLASTN searches with other ADH cDNAs. However, a perfect match w as found between nucleotide positions 7±95 of an adult male liver cDNA clone (GI: 12836366) and nucleotide position 89 989±90 077 in the BAC. This cDNA has a start codon at position 7 6 and encodes s ix amino acids with 3/6 identity with human ADH5 exon 1 encoded sequence. Also, this cDNA at position 511±2407 has a 99% identity with position 101 875±103 771 in 393J8 that includes potential exons 4 and 5. Because so much of the cDNA extends beyond these potential exons, it is probable t hat this cDNA represents partially spliced mRNA. Nucleotides 2513±2795 of this same cDNA have a 99% sequence identity with position 103 877±104 159 in the BAC. TBLASTN searches in the 75 kb of sequence b etween Adh1 and Adh5ps for sequences homologous to deermouse ADH5 or human ADH5 protein sequence revealed two probable complete genes each with nine exons. These genes are de®ned as Adh5a and Adh5b based on location and phylogeny (see below). TBLASTN and BLASTN analyses de®ne the location of Adh5a as encompassing 29 317±47 164 in the 5¢ to 3¢ orientation. Adh5b is located from 57 150 to 74 874. A TBLASTN search of the GenBank database failed to locate the ®rst and last exons of the Adh5 gene. 228 G. Szalai et al. (Eur. J. Biochem. 269) Ó FEBS 2002 Adh5a and Adh5b have available complete cDNA or EST sequence represented in the database, respectively. T his h as allowed the last exon of Adh5b to be de®ned by an EST expressed in mouse skin (GI: 4404412). A full-length cDNA (GI: 12840922) obtained from 10-day-old male pancreas represents the expressed product of Adh5a and has been used to determine g ene structure. The complete gene structures of Adh5a and Adh5b are fully annotated in GenBank (GI: 15383846). Both g enes consist of nine exons and e ncode 374 amino acids excluding the initiation methionine which is the same as the mouse Adh4 and Adh1 protein products. Exon 1 of Adh5b is tentatively identi®ed as a sequence encoding the i ntiation met and ®ve additional amino acids before encountering a gt splice site. The nucleotide sequence of the Adh2 gene is fully annotated in GenBank (GI: 1538346). This gene has nine exons a nd encodes a p olypeptide of 376 am ino a cids, again exclusive of the inititation methionine. All three genes, Adh2, Adh5a and Adh5b have consensus gt and ag nucleotides ¯anking the intron sequences. The translation initiation codons for Adh5a and Adh2 have 8/10 and 6 /10 nucleotide identity with the consensus initiator sequence, respectively. The Adh5b initiator shares 5/10 identity with the consensus sequence and does not have, in contrast to Adh2 and Adh5a, the important G residue after the ATG or the A/G three nucleotides upstream from t he ATG. Although t he ®rst intron in the Adh5b gene begins with gt the match with the consensus is poor only having the AG at the end of the exon before the gt in the intron. The 5¢ end o f the ®rst intron matches the consensus much more in Adh5a and Adh2. The locations of the intron interruptions in the coding sequence of the mouse genes in the cluster are shown in Table 1 . The mo lecular map of the mouse Adh complex is shown in Fig. 2B along with a comparison with the human ADH complex. The s izes of introns in all genes including the pseudogene are shown in Fig. 2C. Phylogeny of mouse and human Adh genes In the phylogen y of the mouse and human genes, there were ®ve major clusters (Fig. 3), each separated by a bootstrap Fig. 2. Structural org anization of the genes in the mouse Adh complex. (A) Restriction map o f the area encompassed by the two BACs indicating positions of the mapped Adh genes. Restriction e nzyme symbols are the same as in Fig. 1. Genes known to lie between two restriction sites a re arbitrarily placed i n the middle. The location of the two BAC clones a re shown b elow t he physical map. (B) Molecular map of the mouse Adh complex (top) comp ared to th e human ADH complex (bottom, based on the sequence in NT 022863). Arrows indicate transcriptional orientation. The draft sequence of 463H24 (GI 15042854) consists of four unordered contigs, b ut molecular distances and orientation wer e derived based on BAC end sequence from the d atab ase, restriction sit es located w ithin contigs compared to the physical map , and paire d BLASTN analysis of the BAC against ADH4 and ADH1 cDNA. The order of contigs in 463H24 relative to the 5¢ to 3¢ tra nscriptional orientation of the Adh4 and Adh1 genes on the BAC is (32489/5054)-(32589/106087; Adh4)-(182084/106188; Adh1). One contig (1/4953) cannot be placed on the map, but is not at either e nd of the BAC. ( C) Organization of intron s an d exon s in the mouse g enes. Introns are lines and exons are rectangles. The genes are given in the ir order within t he complex. Ó FEBS 2002 Mouse alcohol dehydrogenase gene complex (Eur. J. Biochem. 269) 229 value of 87 or greater: ( a) h ADH1A, hADH1B, hADH1C, and mADH1 cluster signi®cantly with a bootstrap value of 100; (b) hADH4 and mADH4 cluster (100); (c) hADH5, mADH5A and mADH5B cluster (87); (d) hADH2 and mADH2 cluster (100); and (e) hADH3 and mADH3 cluster (100). The proteins encoded by the newly identi®ed mouse Adh5a and Adh5b genes are most closely related to the human ADH5 andtoeachother,whichisthebasisfortheir nomenclature. DISCUSSION In this report, we have isolated a series of mouse BAC clones using a mouse ADH1 cDNA as a probe. As two of these BACs overlap at the Adh1 gene, they could be orientated relative to the 5¢ to 3¢ orientation of this gene. Using cDNAs for three other mouse ADH genes and a deermouse gene to probe restriction digests of these clones, we were able to determine the order of the four mouse genes and the sequences cross-hybridizing to the deermouse ADH5 cDNA probe. Sequence of the two overlapping BAC clones combined with the physical map allowed u s t o determine t hat the previously known mouse genes are found in the order Adh4-Adh1-Adh2-Adh3 in less than 250 kb of DNA, a nd all genes are in t he same transcriptional orientation being 5¢ to 3¢ from the Adh4 to Adh1 orientation. This is the same order and transcriptional orientation as found for the human genes except the human has three ADH1 genes (ADH1C, ADH1B and ADH1A) instead of the single mouse Adh1 ortholog. The mouse sequence cross- hybridizing to deermouse ADH5 cDNA was mapped physically to a region b etween Adh1 and Adh2 correspond- ing to the location of the human ADH5 gene. The amino- acid similarity of 67% between the deermouse encoded protein and the human ADH5 protein [10] previously provided the rationale for suggesting that the deermouse sequence represents a separate class VI. A similar s equence has been identi®ed in the rat [11]. However, the identity of 67% is an intermediate value between members of d ifferent classes and members of the same class between species. The orthologous physical location and this intermediate value suggests this represents sequence most related to human ADH5. Pairwise BLASTN comparisons between the draft sequence of the BAC in this area and the deermouse and human ADH5 cDNAs allowed the identi®cation of poten- tial exons 2±6 in this physically mapped r egion. However, exon 6 was found to contain deletions that altered the reading frame leading to a termination codon. Although available cDNA indicates this region is expressed with spliced exons 1±3, the d eletion in exon 6 strongly suggests that no functio nal protein is made. A partial sequence from exon 6 of an unidenti®ed mouse Adh gene from a C 57BL/6 library contains an identical deleted sequence in the exon 6 region [46] (D. Dolney 11 , University of South Carolina, Columbia, SC, USA, unpublished results). Because the location of this sequence is similar r elative to other genes i n human, it is de®ned as Adh5ps. TBLASTN pairwise searches between deermouse and human ADH5 cDNA translation products and the BAC sequence located between this mouse Adh5ps and Adh1 identi®ed two additional genes. The newly de®ned Adh5a and Adh5b genes reside between Adh1 and Adh2 ind icating an overall order of Adh4-Adh1 -Adh5a-Adh5b-Adh5ps-Adh2- Adh3. The mouse Adh5a and Adh5b reside in the same relative position as the human ADH5 gen e. The structures of bo th genes a re very similar to the human ADH5 gene [9] except an additional amino-acid residue is encoded by exon 4 in the human sequence compared to the mouse. The Table1. Amino-acidpositionsencodedbyeachexoninsixmouseAdh genes. Codons split between exons are shown at the end and beginning of adjacent exons. Blanks indicate exon coding is the same as in Adh4.ThehumanADH2 exon co ding regions are shown as this gene is most closely relatedtothemouseAdh2. Exons for hum an ADH2 are from [48], and human ADH5 organization is from [9,47]. Sources for organization of other mouse genes are: Adh4 [45], Adh1 [7] and Adh3 [49]. Gene Exon 123456789 Adh4 Met1±5 6±39 40±86 86±115 115±188 189±275 276±321 321±368 368±374 Adh1 Adh5a Adh5b Adh3 Met3±5 Adh2 115±192 193±279 280±323 323±369 369±376 ADH2 40±87 87±116 116±193 194±280 281±326 326±372 372±379 ADH5 40±87 87±116 116±188 189±275 276±321 321±368 368±375 hADH1C hADH1B hADH1 A mADH1 hADH7 mADH7 hADH6 mADH6A mADH6B hADH4 mADH4 hADH5 mADH5 100 100 100 100 70 100 100 96 87 72 0.1 PC Fig. 3. Phylogeny of mouse and human Adh genes constructed by the neighbor-joining meth od. Numbers on the branch es are perce ntages of 500 bootstrap samples t hat support the branch. 230 G. Szalai et al. (Eur. J. Biochem. 269) Ó FEBS 2002 positions where intron sequence disrupts coding sequence in Adh5a and Adh5b genes is identical to Adh4 [45] and Adh1 [7,46] However, Adh5a, Adh5b,andAdh2 are remarkably similar to human ADH5 at the exon 8/exon 9 boundary. I n all cases there is a potential stop codon just downstream o f exon 8 that would produce a truncated protein if splicing between exons 8 and 9 fails to oc cur. Recently, alternative splicing of exons 8 and 9 has been reported for human ADH5 [47]. All t he mouse genes except Adh4 [45] contain a potential translational stop codon in frame just downstream of exon 8, but it is unknown with what frequency a transcription product lacking exon 9 is produced for any of these genes. All genes in the complex are very similar in location of intron positions within the coding r egion of t he gene with the exception of the Adh2 gene that encodes four additional amino acids in exon 5, but two less in exon 7. This ADH2 protein contains 376 amino acids [5], but is not as large as the human ortholog that contains 379 amino acids [48]. An overview of the gene structure within the complex suggests that genes in the middle of the complex are larger than the ones at the ends. The Adh5ps even with only six exons is the largest in the complex. The genes (Adh4, Adh1, Adh5 a, Adh5b, Adh5ps)atthe5¢ end of the complex relative to transcriptional orientation characteristically have small introns between exons 4 a nd 5, but genes at t he 3¢ end (Adh2 and Adh3) have a larger intron 4. This report presents a detailed map of the mouse Adh gene complex a nd ®nds that six genes in the same transcriptional o rientation are found within 250 kb o f DNA s equence. A pseudogene located in a transcribed region is also detected in this locus. The ®rst gene in the complex at the 5 ¢ end relative to transcriptional orientation, Adh4, is expressed at high level in adult stomach, esophagus and skin with lower levels produced in ovary, uterus, seminal vesicle [4,23]. The next gene, Adh1, is expressed at highest levels in liver, adrenal, and small intestine [4,21] with somewhat smaller amounts being found in kidney and still smaller amounts detectable in several tissues including ovary, uterus, seminal vesicle. Expression of Adh2 occurs in liver with lesser expression in kidney [5] although an extensive expression pattern at t he RNA level has not been established, while the Adh3 gene is widely expressed in mouse tissues [4]. The expression pattern of Adh5a and Adh5b is still to be de®ned. There is some order in the liver expression pattern of the different genes as related to their position on t he chromosome progressing from the 5¢ end of the complex in which Adh4 expression is totally absent from liver, to the middle and 3¢ end of the complex where Adh1, Adh2,andAdh3 are h ighly expressed in liver. At this time, it is unknown if individual regulatory elements between the genes c ontrol e xpression o f each gene during development and differentiation, or if a locus control region in combi- nation with local elements control expression of the complex. The knowledge of the organization of this locus will be useful in addressing these questions in transgenic mouse expression studies. ACKNOWLEDGEMENTS This work was supported by NIH grants AA 11823 ( M. R. F), AA 09731 (G. D), and HG 00313 (B. A. R.). The contribution of R. Friedman in constructing the phylogenetic tree was supported through NIH grant GM 43940 to A. L. Hughes. We are grateful to the NIH Mouse BAC Sequencing Program for generating the seque nce of RP23-393J8 and RP23-463H24 BAC clones. REFERENCES 1. Duester, G., Farre  s, J., Felder, M.R., Holmes, R.S., Ho È o È g, J O., Pare  s, X., Plapp, B.V., Yin, S J. & Jo È rnvall, H. (1999) Recom- mended nomenclature f or the v ertebrate alcohol dehydrogenase gene family. Biochem. Pharmacol. 58 , 389±395. 2. Jo È rnvall, H. & Ho È o È g, J O. (1995) Nomenclature of alcohol dehydrogenases. Alcohol Alcohol. 30 , 153±161. 3. Jo È rnvall, H., S hafqat, J., el-Ahmad, M ., Hjelmquist, L., Pe rsson, B. & Danielsson, O. (1997) Alcohol dehydrogenase variability. Evolutionary an d functional conclusio ns from characterization o f further variants. Adv. Exp. Med. Biol. 414, 281±289. 4. Zgombic-Knight, M., Ang, H.L., Foglio, M.H. & Duester, G. (1995) Cloning of the mouse class IV alcohol dehydrogenase (retinol dehydrogenas e) cDNA and tissue-spec i®c expression patterns of the murine ADH gene family. J. Biol. Chem. 270, 10868±10877. 5. Svensson, S., Stro È mberg, P. & Ho È o È g, J O. (1999) A novel subtype of class II alc ohol dehydrogenase in rodents. J. Biol. Chem. 274, 29712±29719. 6. Duester, G., S mith, M., Bilanchone, V. & H at®eld, G.W. (1986) Molecular analysis of the human class I alcohol deh ydrogenase gene family and nuc leotide seque nce of t he gene encod ing the b subunit. J. Biol. Chem. 261, 2027±2033. 7. Ceci, J.D., Zheng, Y W. & Felde r, M.R. ( 1987) Molecular anal- ysis of mouse alcohol dehydrogenase: nucleotide sequence o f t he Adh-1 gene and genetic mapping of a related n ucleotide sequence to chromosome 3. Gene 59, 1 71±182. 8. Edenberg, H.J., Zhang, K., F ong, K., Bosron, W.F. & Li, T. -K. (1985) Cloning a nd sequencing of cDNA encoding t he com plete mouse liver alcohol dehydrog enase . Proc. Natl Acad. Sci. USA 82, 2262±2266. 9.Yasunami,M.,Chen,C S.&Yoshida,A.(1991)Ahuman alcohol dehydrogenase gene (ADH6) encoding an additional c lass of isozyme. Proc.NatlAcad.Sci.USA88, 7610±7614. 10. Zheng, Y W., Bey, M., Liu, H. & Felder, M.R. (1993) M olecular basis of the alcohol dehydrogenase-negative deer mouse. Evidence fordeletionofthegeneforclassIenzymeandidenti®cationofa possible n ew enzyme class. J. Biol. Chem. 268, 24933±24939. 11. Ho È o È g, J O. & Brandt, M . (1995) Mammalian class VI alcohol dehydrogenase: nove l t yp es o f t he rodent en zyme s. Adv . Exp. Med. Biol. 372, 3 55±365. 12. Deltour, L., Foglio, M.H. & Duester, G. (1999b) Metabolic de®ciences in alcohol dehydrogenase Adh1, Adh3,andAdh4 null mutant mice. J. Biol. Chem. 274, 16796±16801. 13. Liu, L., H ausladen, A., Zeng, M., Que, L., Heitman, J. & Stamler, J.S. (2 001) A m etabol ic enzyme fo r S-nitrosothiol conserved from bacteria to humans. Nature 410 , 490±494. 14. Ang, H.L., Deltour, L., Hayamizu, T.F., Zgombic-Knight, M. & Duester, G. (1996) Retinoic acid synthesis in mouse embryos during gastrulation and craniofacial development linked to class IV alcohol dehydrogenase gen e expression. J. Biol. Chem. 271, 9526±9534. 15. Ang,H.L.,Deltour,L.,Zgombic-Knight,M.,Wagner,M.A.& Duester, G. (1996) Expression patterns of class I and class IV alcohol dehydrogenase genes in de veloping epithelia suggest a r ole for alcohol dehydrogenase in local retinoic synthesis. Alcohol Clin. Exp. Res. 20, 1050±1064. 16. Deltour, L., Foglio, M.H. & Duester, G. (1999a) Impaired retinol utilization in Adh4 alcohol dehydrogenase mutant mice. Dev. Genet. 25, 1±10. 17. Burnett, K.G. & Felder, M.R. (1980) Ethanol metabolism in Peromyscus genetically d e®cient i n a lcohol de hydroge nase. Biochem. Pharmacol. 29, 125±130. Ó FEBS 2002 Mouse alcohol dehydrogenase gene complex (Eur. J. Biochem. 269) 231 18. Holmes, R.S., Albanese, R., Whitehead, F.D. & Duley, J. (1981) Mouse alcoho l d ehydrogenase isoz ymes: products of closely localized duplicated genes exhibiting divergent kinetic properties. J. Exp. Zool. 217, 151±157. 19. Peirce, J.L., Derr, R., Shendure, J., Kolata, T. & Silver, L.M. (1998) A major in¯uence of sex-speci®c loci on alcohol p referenc e in C57Bl/6 and DBA/2 inbred mice. Mamm. Geno me 9, 942±948. 20. Balak, K.J., Keith, R.H. & Felder, M.R. (1982) Genetic and developmental regulatio n of m ouse liver alc ohol d ehyd rogenase . J. Biol. C hem. 257, 15000±15007. 21. Felder, M.R., Watson, G., Hu, M.O. & Ceci, J.D. (1988) Mechanism of induction of m ouse kidney alcohol dehydrogenase by androgen. J. Biol. Chem. 263, 1 4531±14537. 22. Tussey, L. & Felder, M.R. (1989) Tissue-speci®c genetic variation in th e level of m ouse alcohol dehydrogenase is controlled tran- scriptionally in kidney and posttranscriptionally in liver. Proc. Natl Acad. Sci. USA 86, 5 903±5907. 23. Dolney, D.E.A., Szalai, G., Duester, G. & Felder, M.R. (2001) Molecular analysis of genetic dierences among inbred mouse strains controlling tissue expression pattern of alcohol dehydro- genase 4. Gene 267, 145±156. 24. Lin, Z., Edenberg, H.J. & Carr, L. (1993) A novel negative element in the promoter of m ouse liver alcohol dehydrogenase. J. Biol. Chem. 268, 10260±10267. 25. Carr, L.G., Zhang, K. & Edenberg, H.J. (1989) Protein±DNA interaction in the 5¢ region of the mouse alcohol dehydrogenase gene Adh-1. Gene 78, 277±285. 26. Xie, D., Narasimhan, P., Zheng, Y W., Dewey, J.J. & Felder, M.R. (1996) Ten kilobases of 5¢-¯anking region confers proper regulation of the mouse alcohol dehydrogenase-1 (Adh-1) gene in kidney and adrenal of transgenic mice. Gene 181, 173±178. 27. Szalai, G., Ceci, J., D ewey, M. & Felder, M.R. (2001) Identi®ca- tion and expression of cosmids w ith an allelic variant of class I alcohol dehydrogenase in transgenic mice. Chemico-Biol. Intera ct. 130±131, 481±490. 28. Feinberg, A.P. & Vogelstein, B. (1983) A technique for radiola- belling DNA restriction endonuclease fragments to high speci®c activity. Anal. Biochem. 132, 6±32. 29. Pan, H.Q., Wang, Y.P., Chissoe, S.L., Bodenteich, A., W ang, Z., Iyer,K.,Clifton,S.W.,Crabtree,J.S.&Roe,B.A.(1994)The complete nucleotide sequence of the SacBII domain of the P1 pAD10-SacBII cloning vector and three cosmid cloning vectors: pTCF, svPHEP, a nd LAWR IST16 . Genet. Anal. Tech. Appl. 12 11, 181±186. 30. Bodenteich, A., Chissoe, S., Wang, Y.F. & Roe, B.A. (1993) Shotgun cloning as the strate gy of choice to gener ate template s for high-throughput dideoxynucleotide sequencing. In Automated DNA Sequencing and Analysis Techniques (Venter, J.C., ed.), pp. 42±50. Academic Press, London, UK. 31. Chissoe, S.L., Bodenteich, A., Wang, Y.F., Wang, Y.P., Burian, D.,Clifton,S.W.,Crabtree,J.,Freeman,A.,Iyer,K.,Jian,L., Ma, Y., McLaury, H.J., Pan, H .Q., Sharan, O., Toth, S., Wong, Z., Zhang, G., Heisterkamp, N., G roen, J. & Roe, B.A. ( 1995) Sequence and analysis of the human ABL gene, t h e BCR gene, and regions invo lved i n t he Philadelphia ch romoso mal t ranslo cation. Genomics 27, 6 7±82. 32. Sambrook, J., Fritsch, E.F. & Maniatis, T. (1989) Molecular Cloning: a Laboratory Manual, Cold Spring H arbor Laboratory Press, Cold Spring Harbor, New York. 33. Ewing, B., Hillier, L., Wendl, M. & Green, P. (1998) B asecalling of automated s equencer trac es using phred. I. A ccuracy assessment. Genome Res. 8, 175±185. 34. Ewing, B. & Green, P. (1998) Basecalling of automated sequencer traces using p hred. I I. Erro r probabilities. Genome Res. 8, 186±194. 35. Gordon, D., Abajian, C. & Green, P. (1998) Consed: A. graphical tool for sequence ®nishing. Genome Res. 8, 195±202. 36. Genetics Computer Group (1994) Program Manual for the Wisconsin Package, Version 8, Genetics Computer Group, Madison, WI, USA 37. Altschul, S.F., Gish, W., Myers, E.W. & Lipman, D.J. (1990) Basic local alignment search tool. J. Mol. Biol. 215, 403±410. 38. Worley, K.C., W iese, B.A. & Smith, R.F. (1995) B EAUTY: an enhanced BLAST-based search tool that integrates multiple bio- logical information resources into sequence similarity searches. Genome Res. 5, 173±184. 39. Heniko, S. & Heniko, J .G. (1994) P rotein family classi®cation based on searching a database of blocks. Genomics 19, 97±107. 40. Thompson, J.D., Higgins, D.G. & Gibson, T.J. (1994) CLU- STALW: improving the sensitivity o f progressive multiple sequence alignment through sequence weighting, position-speci®c gap penalties and weight matrix choice. Nucleic Acids Res. 22, 4673±4680. 41. Saitou, N. & Nei, M. (1987) The neighbor-joining method: a new method for r econstruc ting phyloge netic t rees. Mol. Biol. Evol. 4, 406±425. 42. Nei, M. & Kumar, S. (2000) Molecular Evolution and Phylogenies. Oxford University Press, New York. 43. Felsenstein, J. (1985) Con®dence limits on phylogenies: an approach using the bootstrap. Evolution 39, 783±791. 44. Altschul, S.F., Madden, T.L., Scha È er, A.A., Zhang, J., Zhang, Z., Miller, W . & Lipman, D.J. (1997) Gapped B LAST and PSI- BLAST: a n ew generation of protein database search programs. Nucleic Acids Res. 25, 3389±3402. 45. Zgombic-Knight, M., Deltour, L., Haselbeck, R.J., Foglio, M.H. & Duester, G. (1997) Gene structure and promoter for Adh3 encoding mouse class IV alcohol dehydrogenase (retinol dehy- drogenase). Genomics 41, 105±109. 46. Zhang, K., Bosron, W.F. & Edenberg, H.J. (1987) Structure of the mouse Adh-1 gene and identi®cation of a deletion in a long alternating purine-pyrimidine sequence in the ®rst intron of strains expressing low alcohol dehydrogenase activity. Gene 57, 27±36. 47. Stro È mberg, P. & Ho È o È g, J O. (2000) Human class V alcohol dehy- drogenase (ADH5): a complex transcription unit generates C-t e r - minal multiplicity. Bioc hem. Biophys. R es. Comm. 278, 544±549. 48. von Bahr-Landstro È m, Jo È rnvall, H. & Ho È o È g, J O. (1991) Cloning and characterization of the human ADH4 gene. Gene 103, 269±274. 49. Foglio, M.H. & Duester, G . (1996) Characterization of the func- tional gene encoding mouse class III alcohol dehydrogenase (glutathione-dependent formaldehyde dehydrogenase) and an unexpressed processed pseudogene with an intact open reading frame. Eur. J. Biochem. 277, 496±504. 232 G. Szalai et al. (Eur. J. Biochem. 269) Ó FEBS 2002 . Organization of six functional mouse alcohol dehydrogenase genes on two overlapping bacterial arti®cial chromosomes Gabor Szalai 1 ,. oxidation of a spectrum of primary and secondary alcohols to the corre- sponding aldehydes and ketones. Mammalian ADHs currently known a re grouped into six