Tetrameric NAD-dependent alcohol dehydrogenase

Tetrameric NAD-dependent alcohol dehydrogenase

Andreas Karlsson a, Mustapha El-Ahmad b, Kenth Johansson a,Jawed Shafqat b, Hans Jornvall b, Hans Eklund a,*, S. Ramaswamy c

a Department of Molecular Biology, Swedish University of Agricultural Sciences, S-751 24 Uppsala, Swedenb Department of Medical Biochemistry and Biophysics, Karolinska Institutet, S-171 77 Stockholm, Sweden

c Department of Biochemistry, University of Iowa, Iowa City, IA 52242, USA

Abstract

Three-dimensional structures of the ethanol-induced, tetrameric alcohol dehydrogenase from Escherichia coli have

recently been determined in the absence and presence of NAD. The structure of the E. coli enzyme is similar to those of

the dimeric mammalian alcohol dehydrogenases, but it has a deletion of 21 residues located at the surface of the

catalytic domain. The catalytic zinc ions have two different types of coordination, which are also observed in the class

III dimeric mammalian alcohol dehydrogenase. Comparison of the structures provide new insights into the relationship

between tetrameric and dimeric alcohol dehydrogenases and provide a link to the structure of the tetrameric yeast

alcohol dehydrogenase.

# 2002 Elsevier Science Ireland Ltd. All rights reserved.

Keywords: Zinc coordination; Quaternary structure; X-ray crystallography

1. Introduction

The two major types of alcohol dehydrogenases

(ADHs) that were initially the subject of extensive

investigations [1] were the tetrameric yeast enzyme

(YADH) that produces alcohol during fermenta-

tion and the dimeric mammalian enzymes that in

different forms participate in detoxifications. Se-

quence similarities between these two enzyme

types established that they were related [2], but

distantly so, and they are now recognized as

members of different enzyme families [3]. Their

relationships could clearly be interpreted in terms

of the three-dimensional structure [4] by analysis

of the then available structure of horse liver

alcohol dehydrogenase (LADH) [5]. The main

finding was that the two enzyme types had the

same ligands binding the active site zinc ion

(Cys46, His67 and Cys174 in the LADH sequence

numbers). Both were also found to have the four

cysteine residues that bind a structural zinc ion

and the characteristic glycine fingerprint at the

coenzyme binding site. A major difference was that

the YADH chain was shorter, and a long deletion

of 21 residues was suggested to represent the lack

of a loop in the catalytic domain.

Whereas the knowledge of three-dimensional

structures of the dimeric ADHs has been extended

enormously since then [6�/15], the advancement of

knowledge has been more limited for tetrameric* Corresponding author.

E-mail address: [email protected] (H. Eklund).

Chemico-Biological Interactions 143�/144 (2003) 239�/245

www.elsevier.com/locate/chembioint

0009-2797/02/$ - see front matter # 2002 Elsevier Science Ireland Ltd. All rights reserved.

PII: S 0 0 0 9 - 2 7 9 7 ( 0 2 ) 0 0 2 2 2 - 3

mailto:[email protected]

enzymes, and the three-dimensional structure of aYADH is still not available. The structures of

bacterial ADHs, which use NADPH [16�/18], as

well as a bacterial NADH-dependent ADH with a

different coordination of the structural zinc [19]

have been determined. Sorbitol dehydrogenase

structures, which belong to the same superfamily

but with a different coordination of the active site

zinc, have also been determined [20,21].We have recently solved the structure of bacter-

ial ADH (Escherichia coli ADH), which is so far

the closest relative to YADH. This gives us the

possibility to make a detailed comparison of the

tetrameric enzymes of the YADH type and the

classical dimeric ADHs. The structure reveals a

number of interesting aspects concerning the

conformational changes on coenzyme bindingand an unusual conformational change of the

active site zinc coordination which links it to other

bacterial ADHs and class III ADH [15].

2. Materials and methods

The crystals of the E. coli ADH belong to thespace group P212121 with cell dimensions of 87.8,

124.3 and 143.4 A. The structure was solved by

molecular replacement using a poly-alanine model

constructed from the model of ketose reductase

[20] (pdb entry 1E3J). The crystallographic refine-

ment used standard methods. The three-dimen-

sional structures were determined both with and

without NAD present [23]. The apoenzyme (with-out NAD) was determined at 2.0 A resolution

whereas the holoenzyme was determined at 2.5 A

resolution. Both structures have been refined, and

the final R -values are 20.1% (R -free 23.1%) and

24.4% (R -free 29.6%), respectively.

3. Results and discussion

3.1. Enzymatic properties of E. coli ADH

The ethanol-induced ADH isolated from E. coli

is distinct from the fermentative alcohol dehydro-

genase and the E. coli class III ADH [22]. The

ethanol-induced E. coli ADH has properties

similar to the class I enzymes and a Km for ethanolof 0.7 mM. However, it has a very high kcat (4050

min�1) and is inhibited by pyrazole (Ki�/0.2 mM)

and 4-methylpyrazole (Ki�/44 mM), in a ratio that

is the inverse of the inhibition of LADH. The

enzyme is even more efficient in the reverse

direction of acetaldehyde reduction (Km�/30 mM

and kcat�/9800 min�1), suggesting a physiological

function like that seen for the fermentative alcoholdehydrogenases. The present ADH is ethanol-

inducible with apparently maximal induction at

an ethanol concentration of about 17 mM.

3.2. Subunit structure of E. coli ADH

The E. coli ADH subunit structure has much in

common with the well-known class I ADH struc-tures. Each subunit is divided into a coenzyme

binding domain comprising residues 143�/283 and

a catalytic domain comprising residues 1�/142 and

284�/336. The catalytic domain is essentially built

up by a complex network of b-structure, but also

contains five helices. The coenzyme binding do-

main is one of the most symmetric Rossmann-

folds where the systematic ab-pattern in two (ab)3

motifs are only violated by the extra strand bS at

the subunit interface.

The long loop corresponding to residues 118�/

139 of horse LADH is replaced with a short turn

with 310 and a-hydrogen bonds that expose Tyr111

at the substrate binding site in the E. coli ADH

structure. Beside this, there are only small differ-

ences in chain length, one or two residues, mostlyin connections between secondary structure ele-

ments which usually occur as deletions in LADH,

the longer enzyme. Each domain can be super-

imposed with an r.m.s. deviation of about 1.5 A

for Ca-atoms.

3.3. Tetramer formation

The tetrameric E. coli ADH molecule is a dimerof dimers where each dimer is highly similar to the

dimeric ADHs (Fig. 1). In each dimer, the six-

stranded b-sheet of the coenzyme binding domain

is extended to a twelve-stranded sheet by a

molecular twofold axis. The hydrophobic interac-

tion area between the coenzyme binding domains

A. Karlsson et al. / Chemico-Biological Interactions 143�/144 (2003) 239�/245240

is covered on the opposite side by an antiparallel

strand from each subunit. Beside this area, there

are interactions of the structural zinc region with

the coenzyme binding domain.

The tetramer in E. coli ADH is formed by two

dimers interacting back to back (the back-side is

opposite the open active site clefts). The tetramer

interactions between the two dimers involve inter-

actions of one subunit A in one dimer with both

subunits C and B in the other. One of these

interactions, A�/C, involves interactions with the

loop containing the structural zinc ion (residues

89�/99), residues 24�/25, 69�/73 and 124�/127. The

only direct hydrogen bonds are from Arg287 in the

last part of the catalytic domain to the main chain

of residue 90 and Asn96 to the main chain of

residue 288. His24 from each subunit stacks

against each other, and Ser73 and His91 are in

van der Waals contact with each other.

The interactions of subunit A in one dimer and

subunit B in the other dimer involve residues 161�/

162, 184�/186 and 292�/300. Double hydrogen

bonds are formed between Arg160�/Glu292 and

Asn186�/Asn186 and a single hydrogen bond

between Glu299 and Lys188.

If both coenzyme binding domains of a dimer in

the holoform of horse liver ADH are superim-

posed on the holoform of E. coli ADH, the

catalytic domains of the dimer are differently

located. The catalytic domains have to rotate by

about 158 in order to superimpose closely. The

extra 21 residues that exist in the dimeric ADHs

Fig. 1. (a) The subunit and dimer structure of E. coli ADH. The central part is formed by two coenzyme binding domains and the

catalytic domains are at the ends. (b) The tetramer is formed by two such dimers. (c) The dimer of E. coli ADH (top) compared with a

corresponding dimer of horse LADH (bottom). The catalytic domains are rotated differently with respect of the coenzyme binding

domains which forms a suitable smooth surface for tetramer interaction.

A. Karlsson et al. / Chemico-Biological Interactions 143�/144 (2003) 239�/245 241

interact differently, and the loss of these residuesin E. coli ADH allows this rotation of the domain.

The main effect of the rotation of the catalytic

domains is that the backside of the dimer (the side

that forms the tetramer interaction) gains a

different topology, a much more flat surface that

is suitable for the tetrameric organization of dimer

against dimer.

3.4. Variations in zinc coordination

The coordination of the structural zinc in the

tetrameric E. coli ADH is similar to that in the

dimeric LADH. The zinc ions are coordinated in a

symmetric tetrahedral arrangement with four cy-steine ligands in the characteristic sequence motif

CX2CX2CX7C. In contrast, the active site zinc of

E. coli ADH displays different coordination types

in different subunits of the two structures. In one

type, the coordination is the classical tetrahedral

one with two cysteines, one histidine and a free

ligand site where water is coordinated. In the other

type, the active site zinc is coordinated in tetra-hedral arrangement with the two cysteines and the

histidine but with the free ligand substituted by a

glutamate residue corresponding to Glu68 in

LADH (Fig. 2).

The existence of two distinct zinc coordinations

in E. coli ADH is puzzling. It is not a symmetric

distribution in either the apo- or holo-enzymes

structures. In the apoenzyme structure one subunit

per tetramer has the classic coordination while the

others have Glu-coordinated zinc. This is the

reverse for the holoenzyme where only one subunit

has Glu coordinating the zinc, and this corre-

sponds to the one that does not have this type of

coordination in the apoenzyme. The occurrence of

one or the other type is obviously not related to a

‘half of the sites’ reactivity. If the catalytic

domains of the two types of subunits are super-

imposed, the position of the zinc atom is displaced

in the Glu-coordinated subunits by about 2 A

away from the active site. This change affects the

ligating residues such that the Glu turns around

and becomes a ligand. The Cys and His ligands

adapt to the new coordination by altering side

chain conformation.What is characteristic of the subunits with Glu

coordinating the zinc is that they do not contain

any bound water molecule. Instead, water is

bound further out in the substrate cleft, at about

5 A distance from the active zinc ion but still

hydrogen bonded to the active site Thr (corre-

sponding to Ser48 in LADH). From this, it

Fig. 2. (a) Normal coordination of the active zinc ion with two cysteines, one histidine and one water molecule. (b) Unproductive zinc

coordination with two cysteines, one histidine and one glutamate. Water is not coordinated in this case.


appears that the Glu-coordinating subunit exists

such that binding of a free ligand is impaired.

There is also a clear correlation in the

holo-enzyme that those subunits with the classic

zinc coordination bind coenzyme, whereas the

subunit with Glu coordinated zinc has no trace

of coenzyme bound. How can zinc coordination be

related to coenzyme binding? It appears that when

Glu moves in to coordinate zinc, it loses its

interactions with an arginine, which becomes

mobile and moves away from its normal position.

This impairs its possibilities for coenzyme binding

since the arginine binds the phosphates of the

coenzyme.

Active site zinc coordinated by Glu has been

observed also in other structures. Dimeric human

class III ADH in a ternary complex with NAD and

inhibitor has the active zinc coordinated to Glu,

but the apo-enzyme and a binary complex have the

normal zinc-coordination [15]. Two types of zinc

coordination also occur in the tetrameric ADH

from Clostridium bejerinckii where an additional

Glu is bound to zinc in the apo-form but not in the

holo-form [16,17]. The tetrameric ADH from

Sulfolobus solfataricus has a zinc-bound Glu in

the absence of coenzyme [19].

It is obvious that both types of coordination can

occur for both dimeric and tetrameric enzymes.

The type of coordination is correlated with bind-

ing of coenzyme, and the other tetrameric enzymes

have normal coordination in the presence of

coenzyme. However, there seems to be no general

rule. The unusual coordination occurs for the

apoenzymes and for a holoenzyme complex for

the class III ADH. In any case, the coordination of

the active zinc with the Glu cannot be an active

form of the enzyme. The coordinating Glu is a

conserved residue in ADHs and the role has been

debated. The experimental evidence of the impor-

tance of this residue is that the catalytic efficiency

(V /Km) for ethanol oxidation and aldehyde reduc-

tion were decreased by a factor of 100 when this

Glu was changed to Gln in YADH [24]. Theore-

tical calculations also indicated that Glu could be

coordinating during the reaction cycle as a means

to displace the bound water molecule and/or the

product [25,26].

3.5. Comparison with other tetrameric ADHs and

relevance to tetrameric yeast ADH

The crystal structure of the alcohol dehydro-

genase in the apo form from the hyperthermophi-

lic archaeon S. solfataricus is very similar in the

overall structure, domain relations and the tetra-

hedral coordination of the active site zinc ion but

has significant differences in the substrate cleft and

the coordination of the structural zinc. The ketose

reductases also have a very similar tetramer

organization, but in this case one of the cysteine

ligands is substituted with an Asp. The tetrameric

ADH from Clostridium bejerinckii also has an Asp

instead of one Cys.The E. coli ADH and LADH sequences are 22%

identical. The main difference is a deletion of 21

residues in the catalytic domain. E. coli ADH is

29% identical to human ethanol-active class I

ADHs, 30% to E. coli class III enzyme, 79% to

bacterial Zymomonas mobilis ADH, and 36%

identical to YADH from Saccharomyces cer-

evisiae. In contrast to comparisons of YADH

with dimeric ADHs, comparing YADH with E.

coli ADH shows only one two-residue gap and five

one-residue gaps, indicating a very similar subunit

structure (Fig. 3). The formation of tetramers is in

all probability also similar in the two enzymes. The

structural zinc loop, which is involved in dimer */

dimer contacts, should also be very similar and

form one of the key interaction surfaces of the

tetramer.

Regions around the active site zinc ligands are

highly conserved and all residues implicated in the

function (Thr and His in the proton relay system)

are conserved in E. coli ADH and YADH. The

substrate pocket has several similarities close to

the active site zinc ion dominated by Trp and Thr.

Farther out in the substrate pocket, there should

be some differences since YADH has two more

residues in the loop following the first zinc ligand.

Determination of the structures of the various

ADHs provides a better understanding of the

interactions that lead to formation of tetrameric

enzymes, but does not solve the problem of the

functional significance of oligomerization. Obser-

vation of the different types of zinc coordination


raises new possibilities for the mechanisms of

catalysis.

References

[1] C.-I. Branden, H. Eklund, B. Furugren, Alcohol dehy-

drogenases, in: P. Boyer (Ed.), The Enzymes, vol. XI,

Academic Press, New York, 1975, pp. 103�/190.

[2] H. Jornvall, The primary structure of yeast alcohol

dehydrogenase, Eur. J. Biochem. 72 (1977) 425�/442.

[3] E. Nordling, H. Jornvall, B. Persson, Medium-chain

dehydrogenases/reductases (MDR): family characteriza-

tions including genome comparisons and active site

modelling, Eur. J. Biochem. 269 (2002) 4267�/4276.

[4] H. Jornvall, H. Eklund, C.-I. Branden, Subunit conforma-

tion of yeast alcohol dehydrogenase, J. Biol. Chem. 253

(1978) 8414�/8419.

[5] H. Eklund, B. Nordstrom, E. Zeppezauer, G. Soderlund, I.

Ohlsson, T. Boiwe, B.O. Soderberg, O. Tapia, C.-I.

Branden, A. .Akeson, Three-dimensional structure of horse

liver alcohol dehydrogenase at 2.4 A resolution, J. Mol.

Biol. 102 (1976) 27�/59.

[6] T.D. Hurley, W.F. Bosron, J.A. Hamilton, L.M. Amzel,

Structure of human b1b1 alcohol dehydrogenase: catalytic

effects of non-active-site substitutions, Proc. Natl. Acad.

Sci. USA 88 (1991) 8149�/8153.

[7] G.J. Davis, W.F. Bosron, C.L. Stone, K. Owusu-Dekyi,

T.D. Hurley, X-ray structure of human b3b3 alcohol

dehydrogenase. The contribution of ionic interactions to

coenzyme binding, J. Biol. Chem. 271 (1996) 17057�/17061.

[8] T.D. Hurley, W.F. Bosron, C.L. Stone, L.M. Amzel,

Structures of three human beta alcohol dehydrogenase

variants. Correlations with their functional differences, J.

Mol. Biol. 239 (1994) 415�/429.

[9] H. Eklund, C.-I. Branden, Alcohol dehydrogenase, in: F.

Jurnak, A. McPherson (Eds.), Biological Molecules and

Assemblies: Enzymes, vol. 3, Wiley, New York, 1987, pp.

73�/142.

[10] S. Ramaswamy, M. El Ahmad, O. Danielsson, H. Jornvall,

H. Eklund, Crystal structure of cod liver class I alcohol

dehydrogenase: substrate pocket and structurally variable

segments, Protein Sci. 5 (1996) 663�/671.

[11] Z.N. Yang, W.F. Bosron, T.D. Hurley, Structure of

human xx alcohol dehydrogenase: a glutathione-depen-

dent formaldehyde dehydrogenase, J. Mol. Biol. 265 (1997)

330�/343.

[12] P. Xie, S.H. Parsons, D.C. Speckhard, W.F. Bosron, T.D.

Hurley, X-ray structure of human class IV ss alcohol

dehydrogenase. Structural basis for substrate specificity, J.

Biol. Chem. 272 (1997) 18558�/18563.

[13] S. Svensson, J.O. Hoog, G. Schneider, T. Sandalova,

Crystal structures of mouse class II alcohol dehydrogenase

reveal determinants of substrate specificity and catalytic

efficiency, J. Mol. Biol. 302 (2000) 441�/453.

[14] M.S. Niederhut, B.J. Gibbons, S. Perez-Miller, T.D.

Hurley, Three-dimensional structures of the three human

class I alcohol dehydrogenases, Protein Sci. 10 (2001) 697�/

706.

[15] P.C. Sanghani, H. Robinson, W.F. Bosron, T.D. Hurley,

Human glutathione-dependent formaldehyde dehydrogen-

ase. Structures of apo, binary, and inhibitory ternary

complexes, Biochemistry 41 (2002) 10778�/10786.

[16] Y. Korkhin, A.J. Kalb, M. Peretz, O. Bogin, Y. Burstein,

F. Frolow, NADP-dependent bacterial alcohol dehydro-

genases: crystal structure, cofactor-binding and cofactor

specificity of the ADHs of Clostridium beijerinckii and

Thermoanaerobacter brockii , J. Mol. Biol. 278 (1998) 967�/

981.

[17] Y. Korkhin, A.J. Kalb, M. Peretz, O. Bogin, Y. Burstein,

F. Frolow, Oligomeric integrity*/the structural key to

Fig. 3. Structural alignment of E. coli ADH (at the top) and

horse LADH (at the bottom). The yeast ADH sequence is also

shown in between. Identical residues in E. coli and yeast ADHs

are marked with a vertical bar.


thermal stability in bacterial alcohol dehydrogenases,

Protein Sci. 8 (1999) 1241�/1249.

[18] C. Li, J. Heatwole, S. Soelaiman, M. Shoham, Crystal

structure of a thermophilic alcohol dehydrogenase sub-

strate complex suggests determinants of substrate specifi-

city and thermostability, Proteins 37 (1999) 619�/627.

[19] L. Esposito, F. Sica, C.A. Raia, A. Giordano, M. Rossi, L.

Mazzarella, A. Zagari, Crystal structure of the alcohol

dehydrogenase from the hyperthermophilic archaeon Sul-

folobus solfataricus at 1.85 A resolution, J. Mol. Biol. 318

(2002) 463�/477.

[20] M.J. Banfield, M.E. Salvucci, E.N. Baker, C.A. Smith,

Crystal structure of the NADP(H)-dependent ketose

reductase from Bemisia argentifolii at 2.3 A resolution, J.

Mol. Biol. 306 (2001) 239�/250.

[21] K. Johansson, M. El-Ahmad, C. Kaiser, H. Jornvall, H.

Eklund, J. Hoog, S. Ramaswamy, Crystal structure of

sorbitol dehydrogenase, Chem. Biol. Interact. 130�/132

(2001) 351�/358.

[22] J. Shafqat, J.O. Hoog, L. Hjelmqvist, U.C. Oppermann, C.

Ibanez, H. Jornvall, An ethanol-inducible MDR ethanol

dehydrogenase/acetaldehyde reductase in Escherichia coli :

structural and enzymatic relationships to the eukaryotic

protein forms, Eur. J. Biochem. 263 (1999) 305�/311.

[23] A. Karlsson, M. El-Ahmad, K. Johansson, J. Shafqat,

Jornvall, H. Eklund, S. Ramaswamy, Domain rotation,

variable zinc-coordination in tetrameric NAD-dependent

alcohol dehydrogenase and their relation to coenzyme

binding, (2002) in preparation.

[24] A.J. Ganzhorn, B.V. Plapp, Carboxyl groups near the

active site zinc contribute to catalysis in yeast alcohol

dehydrogenase, J. Biol. Chem. 263 (1988) 5446�/5454.

[25] U. Ryde, Molecular dynamics simulations of alcohol

dehydrogenase with a four- or five-coordinate catalytic

zinc ion, Proteins 21 (1995) 40�/56.

[26] U. Ryde, The coordination of the catalytic zinc in alcohol

dehydrogenase studied by combined quantum-chemical

and molecular mechanics calculations, J. Comput. Aided

Mol. Des. 10 (1996) 153�/164.


Documents

Tetrameric NAD-dependent alcohol dehydrogenase