Fundamental molecular differences between alcohol dehydrogenase classes.
ABSTRACT Two types of alcohol dehydrogenase in separate protein families are the "medium-chain" zinc enzymes (including the classical liver and yeast forms) and the "short-chain" enzymes (including the insect form). Although the medium-chain family has been characterized in prokaryotes and many eukaryotes (fungi, plants, cephalopods, and vertebrates), insects have seemed to possess only the short-chain enzyme. We have now also characterized a medium-chain alcohol dehydrogenase in Drosophila. The enzyme is identical to insect octanol dehydrogenase. It is a typical class III alcohol dehydrogenase, similar to the corresponding human form (70% residue identity), with mostly the same residues involved in substrate and coenzyme interactions. Changes that do occur are conservative, but Phe-51 is of functional interest in relation to decreased coenzyme binding and increased overall activity. Extra residues versus the human enzyme near position 250 affect the coenzyme-binding domain. Enzymatic properties are similar--i.e., very low activity toward ethanol (Km beyond measurement) and high selectivity for formaldehyde/glutathione (S-hydroxymethylglutathione; kcat/Km = 160,000 min-1.mM-1). Between the present class III and the ethanol-active class I enzymes, however, patterns of variability differ greatly, highlighting fundamentally separate molecular properties of these two alcohol dehydrogenases, with class III resembling enzymes in general and class I showing high variation. The gene coding for the Drosophila class III enzyme produces an mRNA of about 1.36 kb that is present at all developmental stages of the fly, compatible with the constitutive nature of the vertebrate enzyme. Taken together, the results bridge a previously apparent gap in the distribution of medium-chain alcohol dehydrogenases and establish a strictly conserved class III enzyme, consistent with an important role for this enzyme in cellular metabolism.
- Chemico-biological Interactions - CHEM-BIOL INTER. 01/2000; 2000(10):337-355.
- Chemical Reviews 07/2013; · 41.30 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: Two tetrameric secondary alcohol dehydrogenases (ADHs), one from the mesophile Clostridium beijerinckii (CBADH) and the other from the extreme thermophile Thermoanaerobacter brockii (TBADH), share 75% sequence identity but differ by 26 °C in thermal stability. To explore the role of linear segments of these similar enzymes in maintaining the thermal stability of the thermostable TBADH, a series of 12 CBadh and TBadh chimeric genes and the two parental wild-type genes were expressed in Escherichia coli, and the enzymes were isolated, purified and characterized. The thermal stability of each chimeric enzyme was approximately exponentially proportional to the content of the amino acid sequence of the thermophilic enzyme, indicating that the amino acid residues contributing to the thermal stability of TBADH are distributed along the whole protein molecule. It is suggested that major structural elements of thermal stability may reside among the nine discrepant amino acid residues between the N-terminal 50-amino acid residues of TBADH and CBADH.Letters in Peptide Science 10/1998; 5(5-6).
Proc. Nati. Acad. Sci. USA
Vol. 91, pp. 4980-4984, May 1994
Fundamental molecular differences between alcohol
dehydrogenase/class m alcohol dehydrogenase/mo ur patterns/zinc enyme famy)
OLLE DANIELSSON*, SILVIA ATRIANt, TERESA LUQUEt, LARS HJELMQVIST*, ROSER GONZALEZ-DUARTEt,
AND HANS J6RNVALL*f
*Department of Medical Biochemistry and Biophysics, Karolinska Institutet, S-171 77 Stockholm, Sweden; tCenter for Biotechnology, Karolinska Institutet,
S-141 86 Huddinge, Sweden; and tDepartment of Genetics, University of Barcelona, E-08071 Barcelona, Spain
Communicated by Sune Bergstrom, January 18, 1994
rate protein families are the "medium-chain" zinc enzymes
(including the classical liver and yeast forms) and the "short-
chain" enzymes (including the insect form). Although the
medium-chain family has been characterized in prokaryotes
and many eukaryotes (fungi, plants, cephalopods, and verte-
brates), insects have seemed to possess only the short-chain
enzyme. We have now also characterized a medium-chain
alcohol dehydrogenase in Drosophila. The enzyme is identical
to insect octanol dehydrogenase. It Is a typical classm alcohol
dehydrogenase, similar to the correspondin
(70% residue identity), with mostly the same residues involved
in substrate and coenzyme interactions. Changes that do occur
are conservative, butPhe-Si is offunctional interest in relation
to decreased coenzyme binding and Increased overall activity.
Extra residues versus the human enzyme near position 250
affect thecoenzyme-binding domain. Enzymatic properties are
similar-i.e., very low activity toward ethanol (K. beyond
measurement) and high selectivity for formaldehyde/glu-
tathione (S-hydroxymethylgutathione; kt/Km = 160,000
min'1 mM'1). Between the present class m1 and the ethanol-
active class I enzymes, however, patterns ofvariability differ
greatly, highlighting fundamentally separate molecular prop-
erties of these two alcohol dehydrogenases, with class Im
resembling enzymes in general and class I showing high
variation. The gene coding for the Drosophila class HI enzyme
produces an mRNA of about 1.36 kb that is present at all
developmental stages of the fly, compatible with the constitu-
tive nature of the vertebrate enzyme. Taken together, the
results bridge a previously apparent gap in the distribution of
medium-chain alcohol dehydrogenases and establish a strictly
conserved class m enzyme, consistent with an important role
for this enzyme in cellular metabolism.
Two types of alcohol dehydrogenase in sepa-
The "classical" alcohol dehydrogenase is part of a wide-
spread system ofzinc-containing enzymes (1). In mammalian
tissues, at least six classes ofthis enzyme occur. They differ
considerably and represent stages between separate enzymes
and ordinary isozymes. Class I is the well-known liver
enzyme with ethanol dehydrogenase activity (2), class III is
identical with glutathione-dependent formaldehyde dehydro-
genase (3), class IV is a form preferentially expressed in
stomach (4, 5), while classes II, V, and VI, although little
studied, are known also to exhibit distinct properties (6, 7,
44). The class origins have been traced to gene duplications
early in vertebrate evolution [the I/III duplication (8)] or
during that evolution [the IV/I duplication (5)], with emerg-
ing activities toward ethanol (9); class III corresponds to an
ancestral form. These properties and the different evolution-
ary patterns, with class III being "constant" and class I
"variable" (10), result in a consistent picture of the enzyme
system and place the classes of medium-chain alcohol dehy-
drogenases as separate enzymes in the cellular metabolism.
Similarly, another protein family, short-chain dehydroge-
nases, has also evolved into a family comprising many
different enzyme activities, including an alcohol dehydroge-
nase (11). This form operates by means of a completely
different catalytic mechanism and is related to mammalian
prostaglandin dehydrogenases/carbonyl reductase (12).
Thus far, this alcohol dehydrogenase has been found in
insects, the Drosophila enzyme being recognized early to
differ from the zinc-containing alcohol dehydrogenases (13,
14). Its properties in various Drosophila species are well
These two alcohol dehydrogenase types demonstrate that
ethanol dehydrogenase activity has evolved in different man-
ners, with many organisms now employing a medium-chain
enzyme, while others depend on a short-chain enzyme. The
medium-chain family has not been identified in insects,
although it is ofancient origin and has been characterized in
other eukaryotes and in prokaryotes. We now show that the
family is indeed present also in insects and that its major
representative is the typical class IIItype. Infact, theenzyme
turns out to be identical to Drosophila octanol dehydroge-
nase, long known (16-20) but little studied. We have char-
acterized the enzyme from Drosophila melanogaster enzy-
matically and structurally to prove its consistency with other
class III forms. We also have identified and sequenced the
correspondinggene§ and detected its transcription product at
all developmental stages. Thus, the family is now known to
be present essentially in all life forms, supporting the view
that medium-chain alcohol dehydrogenases are universal
factors in cellular defense mechanisms from prokaryotes to
humans. In addition, we find fundamental differences be-
tween the class I and III enzymes, defining separate prop-
erties of these related proteins.
MATERIALS AND METHODS
Protein. D. melanogaster whole flies were bred and har-
vested as described (21). After homogenization, centrifuga-
tion, isoelectric focusing, and activity staining with 1 mM
octanol or 33 mM ethanol at pH 10 and 1 mM glutathione/1
mMformaldehyde atpH 8, alcohol dehydrogenase activity of
the medium-chain type was detected and purified by utilizing
ion-exchange chromatography on DEAE-Sepharose, affinity
chromatography onAMP-Sepharose, and afastprotein liquid
chromatography (FPLC) steponMonoQ as described forthe
class III enzyme from other sources (9). Short-chain alcohol
dehydrogenase activity was also monitored (isopropanol at
§The sequence reported in this paper has been deposited in the
GenBank data base (accession no. U07641).
The publication costs ofthis article were defrayed in part by page charge
payment. This article must therefore be hereby marked "advertisement"
in accordance with 18 U.S.C. §1734 solely to indicate this fact.
Proc. NatL Acad. Sci. USA 91 (1994)
pH 8.6), pooled after the ion-exchange chromatography step,
and purified to homogeneity on Blue-Sepharose (22). The
strain used contained the adhF allele yielding the rapidly
migrating form of the short-chain enzyme.
Structural Analysis. The pure protein was carboxymethy-
lated by treatment with 14C-labeled iodoacetate and digested
in separate batches with proteolytic enzymes (8). Peptide
digests were fractionated by reverse-phase HPLC, and all
fragments obtained were submitted to structural analysis.
Amino acid compositions were determined with a Pharmacia
LKB Alpha plus analyzer after acid hydrolysis for 24 hr at
1100C with 6MHCl/0.5%phenol, and sequence degradations
were carried out with MilliGen Prosequencers 6600 and 6625
utilizing arylamine coupling for membrane attachments or
with an Applied Biosystems 470 sequencer with an on-line
PCR Amplfication and Northern Analysis. Peptide struc-
tures determined were utilized forconstruction oftwo 23-mer
degenerate oligonucleotide probes (corresponding to amino
acidresidues 92-99 and 261-268), which served as primers for
PCR amplification with genomic DNA of several Drosophila
species. Total cellular RNA was isolated by the guanidinium
isothiocyanate method (23) from larval, pupal, and adult D.
melanogaster; separated by 1.2% agarose/formaldehyde gel
electrophoresis; and transferred to a nylon membrane (Am-
ersham) for hybridization at high stringency (42TC in 50%
formamide) (24). A control rehybridization was performed
with a D. melanogaster actin gene probe. Autoradiographs
were measured with an UltroScan XL (Pharmacia LKB)
enhanced laser densitometer.
Enzymatic Characterization. Substrate specificities were
screened by activity staining with ethanol, isopropanol, oc-
tanol, and S-hydroxymethylglutathione (formed by sponta-
neous reaction offormaldehyde and glutathione) (9). Km and
kin values were determined with alcohols atpH 10.0 and with
S-hydroxymethylglutathione at pH 8.0 (5).
Structural Comparisons. The structure obtained was cor-
related with the three-dimensional model deduced for human
class III alcohol dehydrogenase (25) to evaluate all replace-
ments. The conformational representation in Fig. 4 was
prepared by using a program supplied by Protein Science (26)
and the coordinates (27) in the Protein Data Bank (28, 29) of
the related (25) class I human alcohol dehydrogenase. Align-
ments required only few insertions and utilized the class III
alcohol dehydrogenases from a prokaryote (30), yeasts (31,
32), a cephalopod (33), and vertebrates (34).
Presence of Class II Alcohol Dehydrogenase (Octanol De-
hydrogenase). Homogenates of D. melanogaster were sub-
mitted to isoelectric focusing and subsequent activity staining
with ethanol, isopropanol, octanol, and formaldehyde/
glutathione. Results (Fig. 1) clearly show the presence oftwo
enzyme types with partly overlapping substrate specificities.
One, active with ethanol, isopropanol, and octanol, repre-
sents the well-known Drosophila short-chain alcohol dehy-
drogenase (13-15), which is known in multiple electropho-
retic forms (35), and confirms that this is the only ethanol-
active alcohol dehydrogenase in Drosophila. The other is
active with octanol and formaldehyde/glutathione in a man-
ner typical of mammalian medium-chain class III alcohol
dehydrogenase. This suggests that the zinc-containing class
III enzyme is present in Drosophila and that this glutathione-
dependent formaldehyde dehydrogenase may represent the
little-studied Drosophila octanol dehydrogenase previously
reported (16-20) but not characterized structurally.
The glutathione-dependent formaldehyde dehydrogenase
was purified 1900-fold in a 52% yield, resulting in a homo-
geneous preparation (Fig. 1, lanes 5 and 8) after chromatog-
purified alcohol dehydrogenases (B and C) ofD. melanogaster. (A)
Isoelectric focusing under native conditions and activity staining
withformaldehyde/glutathione (lane 1), octanol (lane 2), isopropanol
(lane3),and ethanol (lane4). (B)IsoelectricfocusingasinA,but with
protein staining ofthe purified proteins, octanoldehydrogenase (lane
5), and short-chain dehydrogenase (lane 6). (C) Isoelectric focusing
afterSDS/polyacrylamide gel electrophoresis andprotein stainingof
standard proteins (lane 7; molecular masses 94, 67, 43, 30, 20.1, and
14.4 kDa from top to bottom) and pure octanol dehydrogenase (lane
8). In lanes 3, 4, and 6, a minor more-acidic form is visible,
representing one ofinterconvertible forms in flies homozygous at the
ADH locus (35).
Gel electrophoretic patterns of homogenates (A) and
raphy on DEAE-Sepharose, AMP-Sepharose, and Mono Q
FPLC column in a protocol similar to that for other class Ill
alcohol dehydrogenases (9). All octanol dehydrogenase ac-
tivity was monitored and coincided either with that of the
glutathione-dependent formaldehyde dehydrogenasethrough
all purification steps or with the short-chain alcohol dehy-
drogenase, which was also purified. The pure glutathione-
dependent formaldehyde dehydrogenase exhibited both ac-
tivities, with values (Table 1) typical of class Ill alcohol
dehydrogenase and patterns similar to those of the corre-
sponding human enzyme (36-38). The specific activity ob-
tained was 12 units/mg with glutathione/formaldehyde,
while the activity toward 1 mM octanol was 11 units/mg.
Chromatography on Superose-12 indicated a molecular mass
ofabout 80 kDa, compatible with a dimer of40-kDa subunits
and in agreement with the estimate from SDS/polyacryl-
amide gel electrophoresis (Fig. 1C). Combined with the
structural analysis (below), all results now show that Dro-
sophila octanol dehydrogenase is a class III medium-chain
alcohol dehydrogenase and establish the conserved catalytic
activity of this enzyme.
class III alcohol dehydrogenase compared to those of human class
III alcohol dehydrogenase
Values for the Drosophila enzyme experimentally determined at
pH 10 in 0.1 M glycine/NaOH for the alcohol substrates and at pH
8 in 0.1 M sodium pyrophosphate with 2.4 mM NAD+ for S-hy-
droxymethylglutathione. Values for the human enzyme from refs.
36-38 are given below, within parentheses for comparison. NS, not
saturable. NAD+(alc) indicates values in the alcohol dehydrogenase
reaction measured with 0.5 mM octanol (pH 10).
Enzymatic properties ofD. melanogaster medium-chain
Biochemistry:Danielsson et al.
4982Biochemistry: Danielsson et al.
D KTIL Y
VSR I V
KEA ECIQ FSK--P
SE MS KIK
class HI structure (upper sequence; residues cited only where different; in addition, the human enzyme lacks residues corresponding to the three
first residues of the Drosophila form). Peptides passed by Edman degradations are denoted by lines beneath the sequences and were obtained
by cleavages with a lysine-specific protease (upper lines) and with a Glu-specific protease (lower lines). Positional numbers (within parenthesis)
above the human sequence line refer to the human class I enzyme and are given to allow correlation with the functional residues, which are
generally known under their numbers in the class I enzymes, while numbers below the Drosophila sequence refer to the Drosophila class m
enzyme now determined. Initiator methionine is not included since the protein is N-terminally blocked and is concluded to be acetylated like
the human form.
Primary structure ofD. melanogaster class III alcohol dehydrogenase (lower, continuous sequence) and its relationship to the human
Structure, Presence in Different Species, and Expression at
AU Developmental Stages. The enzyme was carboxymethy-
lated and digested in different batches with glutamic acid- and
lysine-specific proteases, respectively. Each digest was frac-
tionated by reverse-phase HPLC, and peptides were submit-
ted to sequence analysis. These results, combined with DNA
data (below), gave the primary structure of the protein chain
(Fig. 2). Only the N-terminally blocked peptides were not
degraded, and positions 1-3 rely on the DNA data.
Oligonucleotides (corresponding to positions 92-99 and
261-268) were synthesized and used for PCR amplifications
with DNA from five Drosophila species belonging to three
different subgenera. Data obtained (Fig. 3A) show that the
gene coding for the class III enzyme is present in all of the
species tested. However, slightly different sizes of the am-
plified fragments suggest that there is some variability in the
genomic region flanked by the two PCR primers.
dehydrogenase gene in several Drosophila species (A) and expres-
sion in different developmental stages (B). (A) PCR amplification
with DNA of five Drosophila species: D. immigrans (lane Di); D.
hydei (lane Dh); D. lebanonensis (lane Dl); D. subobscura (lane Ds);
and D. melanogaster (lane Dm). Lane M shows Hae III-digested
pBR322 as molecular weight markers. Amplification was performed
in 100 Ad containing 1 pgof genomic DNA, 40 pmols of each
degenerate primer, 1.5 units ofTaqpolymerase (Promega) in 2.5mM
MgCl2, and 200 pM each dNTP. After 2 min at 94CC and 45 cycles
of 940C for 60s, 50TC for 90 s, and 72TC for 90 s, 15Alwas resolved
on 1.2% agarose gels in Tris borate/EDTA buffer and stained with
ethidium bromide. Units are in kb. (B) Northern analysis with larval
(lane L), pupal (lane P), and adult (lane A) total cellular RNA from
D. melanogaster. Units are in kb.
Presence of a class Ill alcohol dehydrogenase/octanol
Using the PCR-amplified products, we cloned and se-
quenced the D. melanogaster gene, confirming the protein
structure. Total RNA from larval, pupal, and adult D. mel-
anogaster was also used for Northern analysis with the
PCR-amplified product (Fig. 3B; an actingene probewas also
used). A single band ofcomparable intensity (as measured by
laser densitrometry with the actin control) and a1.36 kb was
detected in each sample. This indicates that expressionofthe
class III gene is not confined to a particular developmental
stage but appears to be abundant during the life span of the
organism. Together, all data support a wide distribution of
the enzyme in the Drosophila genus. The constant presence
of the transcript at all developmental stages is compatible
with a constitutive pattern of expression.
Identification of Medium-Chain Alohol Dehydrogese as
Octanol Dehydrogenase in Drosophila. The demonstration of
a medium-chain alcohol dehydrogenase in Drosophila estab-
lishes that this protein family occurs throughout the living
world. It has previously been characterized structurally in
prokaryotes (30) and inmany highly differenteukaryotic lines
(31-34) but thus far not in Drosophila, where the short-chain
alcohol dehydrogenase appeared to constitute a peculiar
feature of alcohol oxidation. Nevertheless, the short-chain
enzyme is the only one with appreciable ethanol dehydroge-
nase activity inDrosophila (Fig. 1), even though the medium-
chain family is present, as is now evident. Its class III form
coexists with the short-chain enzyme and has both unaltered
enzymatic properties and a structure conserved in relation to
the human enzyme. The absence of a class I medium-chain
alcohol dehydrogenase in Drosophila apparently reflects the
later enzymogenesis of that enzyme type (9) but does not
imply that the protein family as such or the class III ancestral
form is absent.
Theextremelywideoccurrence ofthis protein highlights its
general importance and suggests that it has a role in basic
cellular metabolism. Moreover, the present results identify
octanoldehydrogenase, whichhas been discussedpreviously
(16-20) butwith somewhat contradictory estimatesregarding
molecularweightand quaternary structure, as being identical
to the class III medium-chain alcohol dehydrogenase. Thus,
the octanol dehydrogenase is involved also in glutathione-
dependent elimination of formaldehyde. However, negative
mutants are viable (32, 39), suggesting further multiplicity
and the existence of a back-up system.
Proc. Natl. Acad Sci. USA 91(1994)
Proc. NatL Acad. Sci. USA 91 (1994)
Correlation with Functional Properties. The identification
and analysis of Drosophila class III alcohol dehydrogenase
means that five divergent forms (conserved residues, 46%6) of
class III alcohol dehydrogenase have nowbeen characterized
structurally-mammalian forms including the human enzyme
(34), a cephalopod enzyme (33), a Drosophila enzyme (this
work), and two yeast enzymes (31, 32). Their properties can
be compared with those of the classical liver alcohol dehy-
drogenase ofclass I, whose structure, within the same family,
has been analyzed recently to a level showing approximately
the same residue divergence (conserved residues, 42%) by
analysis of five major vertebrate lines (40).
The conservation of functionally important residues in
class III is extensive. Of 35 positions participating in coen-
zyme or substrate interactions in the class I enzyme (41) and
considered to do so also in the class III enzyme (25, 42), no
less than 28 are strictly conserved between the human and
Drosophila class III enzymes, and all but 1 of the 7 ex-
changed constitutejust minor variations encountered also in
other species. Position 51 is the single exception. There, the
class III Drosophila enzyme has a phenylalanine residue, as
does the Escherichia coli enzyme (30), while human and the
other class III forms have a tyrosine residue [and the whole
family has a histidine, tyrosine, or serine residue (5, 33, 40)].
This residue usually participates in hydrogen bonding with
the ribose moiety of the NMN part of the coenzyme (25).
Although the hydrogen bond may have little effect (43),
Phe-51 cannot donate such a bond to the coenzyme. There-
fore, this exchange may contribute an explanation to the
weakened coenzyme binding ofthe Drosophila enzyme and,
hence, to its increased activity relative to the human enzyme
Disregarding the functionally important residues, the Dro-
sophila enzyme has two extra residues in relation to the
human class III form. These extra residues, close to position
250 (Lys-251 and Gly-252 in Fig. 2, or adjacent residues,
depending on the alignment chosen), are also present in the
yeast enzyme (31, 32) but not in the octopus enzyme (33).
This region corresponds to a surface helix in the coenzyme
binding domain (Fig. 4 Upper Left), and it appears possible
to extend the helix without adverse consequences.
The conservation of class III stands in marked contrast to
the spread within the class I enzyme. Just among vertebrates,
class I enzyme residue conservation [55% for the human/fish
pair (42)] is lower than that for class III across separate
eukaryotes (63% for the human/yeast pair). Among the 35
functionally important positions, no fewer than 20 vary,
many to a great extent (40, 42). Glycolytic enzymes-e.g.,
glyceraldehyde-3-phosphate dehydrogenase (human/yeast,
63-65% residue identity, depending on the types compared)
and enolase (human/yeast, 61-64% identity)-demonstrate
that the class III alcohol dehydrogenase variation (human/
yeast, 63%) is "normal" for an enzyme ofthis type in afixed
metabolic pathway. It is the faster evolving class I thathas an
"atypical" variation. In fact, the human/yeast values for the
two glycolytic enzymes and the class HI alcohol dehydroge-
nase are surprisingly close, just within ±2%, suggesting
similar functional restrictions for these three metabolically
..' .-. t -iI
give a schematic representation ofthe residue conservations along the polypeptide chains (positions numbered), withblack vertical linesdenoting
positions with strictly conserved residues among the species shown. Unfilled bar areas correspond to variable segments, V1-V3I in class I and
V1i1-WV2iii in class III. Their spatial positionsare shown with thick lines inside the models of the class I and III enzymes (25, 27). The
conformation is that from the Protein Data Bank coordinates (28, 29) ofthe human class I enzyme (27) when using a program supplied by Protein
Differences in patterns of variability for classes I (bar and conformation) (Lower) and Ill (bar and conformation) (Upper). The bars
. - ...
.m6 ohmm mb A
i mm. '. its
Biochemistry:Danielsson et al.
Biochemistry: Danielsson et al.
fixed enzymes. This agreement adds strength to the conclu-
sion of significantly different natures of the class I and III
Fundamental Differences in Molecular Architecture of Re-
lated Proteins. Apart from the differences in overall conser-
vation, the actual distributions of the constant and variable
segments of the class I and III alcohol dehydrogenases are
fundamentally different. In class I, three segments(V1-V30)
stand out as variable (empty spaces in the bottom bar in Fig.
4) and have been noted to make class I peculiar among
proteins in general by suggesting hypervariability in impor-
tant segments (40). As shown in the conformation (Fig. 4
Lower), they correspond to a segment adjacent to the active
site (V1i), a part of the loop around the second zinc atom
(V20),and a part of the subunit-subunit interacting segment
(V30).These three segments all are concentrated on one side
of the molecule (toward the observer in Fig. 4). Therefore,
apart from affecting the functional areas, their covariability
may have even further implications, suggesting variability of
a particular side of the class I molecule.
In contrast, these segments are not at all variable in class
III but are exactly those much conserved (bar in Fig. 4
Upper), supporting the view that class III represents a
"normal" enzyme, with maximal conservation at the active
site and other important segments. Nevertheless, class Ill
also has segments of variability, but its two such segments
(V1m andV2E1)affect nonfunctional, superficial regions,
completely differently positioned regarding both functional
representation and molecular surfaces (Fig. 4 Upper). Inter-
estingly, though, V2I
corresponds to the helix constituting
one of the two domain-interconnections of the subunit (Fig.
4). Since domain movements are known to be associated with
coenzyme binding in the class I enzyme (41), this segment
may indicate further functional differences in catalytically
active states between the classes.
The remaining classes are not yet known in similar detail.
Nevertheless, their inclusion in bar comparisons of the type
in Fig. 4 (as given in ref. 40) suggests that the class I pattern
is typical of variable classes of medium-chain alcohol dehy-
In conclusion, we notice that fundamentally different in-
ternal variability patterns affect two related proteins of a
protein family. These patterns add to the overall differences
between the classes and show that differences apply to
building elements in the molecular architecture of the pro-
teins. Class III behaves as enzymes in general, constant in
function, enzymology, overall structure, and important seg-
ments, while class I is emerging in function, exhibiting
enzymatic differences, rapid evolutionary changes, and van-
ability atimportant regions. The two variability patterns (Fig.
4)illustrate fundamental differences between an evolving and
a constant protein within a single family, and correlate the
differences with the molecular architecture and enzymatic
We are grateful to Dr. Bengt Persson (Department of Medical
Biochemistry and Biophysics) for supplying the program utilized for
the bar constructions in Fig. 4, and to Heldne Olsson (Center for
Biotechnology) for assistance with structural analysis. This study
was supported by the Swedish Medical Research Council (Project
03X-3542), the Swedish Alcohol Research Fund (Project 88/12:4),
and a Spanish Comisi6n Interministerial de Ciencia y Tecnologfa
Jdmvall, H., Danielsson, 0., Eklund, H., Hjelmqvist, L., H66g, J.-O.,
Par6s, X. &Shafqat, J. (1993) in Enzymology and Molecular Biology of
Carbonyl Compounds 4, eds. Weiner, H., Crabb, D. W. & Flynn, T. G.
(Plenum, New York), pp. 533-544.
Vallee, B. L. & Bazzone, T. J. (1983) Curr. Top. Biol. Med. Res. 8,
Koivusalo, M., Baumann, M. & Uotila, L. (1989) FEBS Lett. 257,
Moreno, A. & Pares, X. (1991) J. Biol. Chem. 266, 1128-1133.
Pares, X., Cederlund, E., Moreno, A., Hjelmqvist, L., Farr6s, J. &
JOrnvall, H. (1994) Proc. Nat!. Acad. Sci. USA 91, 1893-1897.
Jmrnvall, H., H66g, J.-O., von Bahr-Lindstr6m, H. & Vallee, B. L.
(1987) Proc. Nat!. Acad. Sci. USA 84, 2580-2584.
Yasunami, M., Chen, C.-S. & Yoshida, A. (1991) Proc. Natl. Acad. Sci.
USA 88, 7610-7614.
Cederlund, E., Peralba, J. M., Pares, X. & J6rnvall, H. (1991) Biochem-
istry 31, 2811-2816.
Danielsson, 0. & J6rnvall, H. (1992) Proc. Natl. Acad. Sci. USA 89,
Yin, S.-J., Vagelopoulos, N., Wang, S.-L. & J6rnvall, H. (1991) FEBS
Lett. 283, 100-103.
Persson, B., Krook, M. & J6rnvall, H. (1991) Eur. J. Biochem. 261,
Krook, M., Ghosh, D., Str6mberg, R., Carlquist, M. & J6mvall, H.
(1993) Proc. Natl. Acad. Sci. USA 9W, 502-506.
Schwartz, M. F. & J6rnvall, H. (1976) Eur. J. Beochem. 68, 159-168.
Thatcher, D. R. & Sawyer, L. (1980) Biochem. J. 187, 884-886.
Atrian, S., Marfany, G., Albalat, R. & GonzIez-Duarte, R. (1992)
Biochem. Genet. 11, 19-29.
Urspnrng, H. & Leone, J. (1965) J. Exp. Zool. 161, 147-154.
Sieber, F., Fox, D. J. & Ursprung, H. (1972) FEBS Lett. 26, 274-276.
Dickinson, W. J. & Sullivan, D. T. (1975) Gene-Enzyme Systems in
Drosophila (Springer, Heidelberg), pp. 78-80.
Madhavan, K., Conscience-Egli, M., Sieber, F. & Ursprung, H. (1973)
J. Insect Physiol. 19, 235-241.
Ogonji, G. 0. (1971) J. Exp. Zool. 178, 513-522.
Ribas de Pouplana, L., Atrian, S., Gonzilez-Duarte, R., Fothergill-
Gilmore, L. A., Kelly, S. M. & Price, N. C. (1991) Biochem. J. 276,
Juan, E. & GonzAlez-Duarte, R. (1980) Biochem. J. 189, 105-110.
Chomczynski, P. & Sacchi, N. (1987) Anal. Biochem. 162, 156-159.
Maniatis, T., Fritsch, E. F. & Sambrook, J. (1982) MolecularCloning:A
Laboratory Manual(Cold SpringHarborLab. Press, Plainview, NY), 1st
Eklund, H., Miller-Wille, P., Horales, E., Futer, O., Holmquist, B.,
Vallee, B. L., H66g, J.-O., Kaiser, R. & J6rnvall, H. (1990) Eur. J.
Biochem. 193, 303-310.
Richardson, D. C. & Richardson, J. S. (1992) Protein Sci. 1, 3-9.
Hurley, T. D., Bosron, W. F., Hamilton, J. A. & Amzel, L. M. (1991)
Proc. Natl. Acad. Sci. USA 88, 8149-8153.
Bernstein, F. C., Koetzle, T. F., Williams, G. J. B., Meyer, E. F., Jr.,
Brice, M. D., Rodgers, J. R., Kennard, O., Shimanouchi, T. & Tasumi,
M. (1977) J. Mol. Biol. 112, 535-542.
Abola, E. E., Bernstein, F. C., Bryant, S. H., Koetzle, T. F. & Weng,
J. (1987) in Crystallographic Databases-Information Content, Software
Systems, Scientific Applications, eds. Allen, F. H., Bergerhoff, G. &
Sievers, R. (Data Commission Intl. Union Crystallogr., Bonn), pp.
Gutheil, W. G., Holmquist, B. & Vallee, B. L. (1992) Biochemistry 31,
Sasnauskas, K., Jomantiene, R., Janu~ka, A., Lebediene, E., Lebedys,
J. & Janulaitis, A. (1992) Gene 122, 207-211.
Wehner, E. P., Rao, E. & Brendel, M. (1993) Mol. Gen. Genet. 237,
Kaiser, R., Fernindez, M. R., Pards, X. &J6rnvall, H. (1993)Proc. Natl.
Acad. Sci. USA ", 11222-11226.
Kaiser, R.,Holmquist, B., Vallee, B. L. & Jmrnvall, H. (1989) Biochem-
istry 28, 8432-8438.
Jacobson, K. B., Murphy, J. B., Knopp, J. A. &Ortiz, J. R. (1972)Arch.
Biochem. Biophys. 149, 22-35.
Wagner, F. W., Pares, X., Holmquist, B. & Vallee, B. L. (1984) Bio-
chemistry 23, 2193-2199.
Moulis, J.-M., Holmquist, B. & Vallee, B. L. (1991) Biochemistry 30,
Uotila, L. & Koivusalo, M. (1974)J. Biol. Chem. 249, 7653-7663.
Voelker, R. A., Langley, C. H., Leigh-Brown, A. J., Ohnishi, S., Dick-
son, B., Montgomery, E. & Smith, S. C. (1980) Proc. Nat!. Acad. Sci.
USA 77, 1091-1095.
Persson, B., Bergman, T., Keung, W. M., Waldenstr6m, U., Holmquist,
B., Vallee, B. L. & J6rmvall, H. (1993) Eur. J. Biochem. 216,49-56.
Eklund, H., Samama, J.-P. & Jones, T. A. (1984) Biochemistry 23,
Danielsson, O., Eklund, H. & JMrnvall, H. (1992) Biochemistry 31,
Schmid, F., Hinz, H.-J. & Jaenicke, R. (1978) FEBS Lett. 87, 80-82.
Zheng, Y.-W., Bey, M., Liu, H. & Felder, M. R. (1993)J. Biol. Chem.
Proc. Nad. Acad Sci. USA 91(1994)