Tài liệu Báo cáo khoa học: Cosubstrate-induced dynamics of D-3-hydroxybutyrate dehydrogenase from Pseudomonas putida ppt

Cosubstrate-induced dynamics of D-3-hydroxybutyrate
dehydrogenase from Pseudomonas putida
Karthik S. Paithankar1, Claudia Feller2, E. Bartholomeus Kuettner1, Antje Keim1, Marlis Grunow2
and Norbert Strater1
ă
1 Center for Biotechnology and Biomedicine, Institute of Bioanalytical Chemistry, Faculty of Chemistry and Mineralogy,
University of Leipzig, Germany
2 Institute of Biochemistry, Faculty of Biosciences, Pharmacy, and Psychology, University of Leipzig, Germany

Keywords
crystal structure; loop closure; protein
dynamics; SDR
Correspondence
M. Grunow, Institute of Biochemistry,
Faculty of Biosciences, Pharmacy, and
Psychology, University of Leipzig,
Bruderstraòe 34, D-04103 Leipzig, Germany
ă
Fax: +49 341 9736998
Tel: +49 341 9736907
E-mail:
N. Stra
ăter, Center for Biotechnology and
Biomedicine, Institute of Bioanalytical
Chemistry, Faculty of Chemistry and
Mineralogy, University of Leipzig, Deutscher
Platz 5, D-04103 Leipzig, Germany
Fax: +49 341 9731319
Tel: +49 341 9731311
E-mail:
(Received 5 July 2007, revised 8 August

2007, accepted 10 September 2007)
doi:10.1111/j.1742-4658.2007.06102.x

D-3-Hydroxybutyrate dehydrogenase from Pseudomonas putida belongs to
the family of short-chain dehydrogenases ⁄ reductases. We have determined
X-ray structures of the D-3-hydroxybutyrate dehydrogenase from Pseudomonas putida, which was recombinantly expressed in Escherichia coli, in
˚
three different crystal forms to resolutions between 1.9 and 2.1 A. The socalled substrate-binding loop (residues 187–210) was partially disordered in
several subunits, in both the presence and absence of NAD+. However,
in two subunits, this loop was completely defined in an open conformation
in the apoenzyme and in a closed conformation in the complex structure
with NAD+. Structural comparisons indicated that the loop moves as a
rigid body by about 46°. However, the two small a-helices (aFG1 and
aFG2) of the loop also re-orientated slightly during the conformational
change. Probably, the interactions of Val185, Thr187 and Leu189 with the
cosubstrate induced the conformational change. A model of the binding
mode of the substrate D-3-hydroxybutyrate indicated that the loop in the
closed conformation, as a result of NAD+ binding, is positioned competent for catalysis. Gln193 is the only residue of the substrate-binding loop
that interacts directly with the substrate. A translation, libration and screw
(TLS) analysis of the rigid body movement of the loop in the crystal
showed significant librational displacements, describing the coordinated
movement of the substrate-binding loop in the crystal. NAD+ binding
increased the flexibility of the substrate-binding loop and shifted the equilibrium between the open and closed forms towards the closed form. The
finding that all NAD+-bound subunits are present in the closed form and
all NAD+-free subunits in the open form indicates that the loop closure is
induced by cosubstrate binding alone. This mechanism may contribute to
the sequential binding of cosubstrate followed by substrate.

Short-chain dehydrogenases ⁄ reductases (SDRs) constitute a large protein family that now includes more
than 1000 enzymes in humans, mammals, insects and

bacteria [1]. The dehydrogenases act on a wide variety
of substrates, including steroids, retinoids, prostaglandins, sugars and alcohols. The name SDR is based on

their smaller subunit size of about 250 residues compared with the medium-chain dehydrogenase ⁄ reductase
family that has a subunit size of about 350 residues.
The SDR enzymes exhibit a sequence identity of
15–30% and share two signature motifs: a GxxxGxG
motif involved in coenzyme binding; and a YxxxK

Abbreviations
PfHBDH, Pseudomonas fragi D-3-hydroxybutyrate dehydrogenase; PpHBDH, Pseudomonas putida D-3-hydroxybutyrate dehydrogenase;
SDR, short-chain dehydrogenase ⁄ reductase.

FEBS Journal 274 (2007) 5767–5779 ª 2007 The Authors Journal compilation ª 2007 FEBS

5767


D-3-Hydroxybutyrate dehydrogenase

K. S. Paithankar et al.

motif in the active site. Crystal structures are now
known for about 60 SDR members and it is now clear
that SDRs are single-domain enzymes; by contrast,
medium-chain dehydrogenases ⁄ reductases consist of a
cosubstrate-binding domain and a substrate-binding
domain [2,3]. SDR enzymes exist as monomers, dimers
or tetramers. The tetrameric enzymes exhibit 222
point-group symmetry. Conventionally, the three

mutually perpendicular two-fold axes are named P, Q
and R [4].
One of the most variable parts of the different SDR
enzymes is a loop consisting of two a-helices that protrudes out of the compact tetramer. This so-called substrate-binding loop, which is located between b-strand
F and a-helix G, is involved in recognition of the
structurally different substrates. The substrate-binding
loop differs significantly in length and sequence and
also adopts different conformations when comparing
open or closed forms of different SDR enzymes. It has
also been shown in some crystal structures that the
loop undergoes a conformational change upon substrate binding [4–6]. Substrate-induced conformational
changes from an open to a closed conformation, or
from a disordered (conformationally flexible) to an
ordered structure, have also been observed. Nakamura
et al. recently demonstrated by X-ray crystallography
and CD spectroscopy that coenzyme binding to 3a-hydroxysteroid dehydrogenase alone induces a transition
of the loop from a disordered structure to a conformation consisting of two a helices [7]. Spectroscopic
studies on 17-b-hydroxysteroid dehydrogenase by fluorescence energy transfer also indicated that a conformational change might occur upon coenzyme binding
[8]. To the best of our knowledge, a coenzyme-induced
conformational change to a closed conformation of
the substrate-binding loop has, to date, not been analyzed by crystallographic means.
D-3-Hydroxybutyrate dehydrogenase from Pseudomonas putida (PpHBDH) (EC 1.1.1.30, GenBank
accession number AJ310211.2) catalyzes the reversible
and stereospecific oxidation of D-3-hydroxybutyrate
to acetoacetate using NAD+ as a coenzyme. One
subunit contains 256 amino acids with a calculated
molecular weight of  26.6 kDa [9]. X-ray structures
of the homologous enzyme from Pseudomonas fragi
(PfHBDH) in the presence of NAD+ and inhibitor
(Protein Data Bank accession code: 1X1T) and without cosubstrate (Protein Data Bank accession code:

1WMB) have been recently determined [10]. Interestingly, in these structures the substrate-binding loop
was ordered in the absence of NAD+ and disordered
in the complex structure with bound NAD+. In addition, a crystal structure of a human cytosolic HBDH
5768

(DHRS6) with bound NAD+ is available [11]. Based
on homology modelling, substrate and inhibitor docking studies, and site-directed mutagenesis, residues
Gln91, His141, Lys149, Tyr152 and Gln193 were
found to be involved in substrate binding in
PpHBDH [9]. With the exception of Gln193, these
residues are located in the deep active-site cleft. The
exact boundaries of the flexible loop differ among
enzymes. Based on the structure of PpHBDH, the
loop runs from residue 187 to residue 210. We will
refer to the rest of the one-domain protein as the catalytic subdomain.
In this study we determined the structure of
PpHBDH using three different crystal forms. The
X-ray structures showed a completely ordered conformation of the substrate-binding loop in at least one
subunit in the open and closed forms. A comparison
of these conformers, and an analysis of the mobility of
the loop in the crystals, allowed a detailed description
of the enzyme dynamic properties and conformational
change during HBDH catalysis.

Results and Discussion
Monomer and tetramer structure
Three different crystal forms of PpHBDH were
obtained in the presence or absence of NAD+
(Table 1). In crystal form I the asymmetric unit was
found to contain one tetramer and two dimers (the

tetrameric structure is generated by a crystallographic
two-fold axis). In crystal form II the asymmetric unit
was found to contain two dimers, and in crystal form III one tetramer was present in the asymmetric
unit. In crystal form I, two of the eight subunits
(designated as chains A and B) contained no bound
NAD+, as shown by the electron density maps. In
crystal form II, only subunit A was NAD+ free.
Therefore, crystal forms I and II contain tetramers
with all four binding sites occupied and tetramers
with two bound NAD+ molecules. In crystal form III
the enzyme was completely devoid of NAD+ cosubstrates. The comparison of several independent subunits, with and without bound NAD+, allowed an
analysis to be made of cosubstrate-induced movements within the enzyme, in particular of the substrate-binding loop.
Each subunit of HBDH has an a ⁄ b doubly wound
structure with the characteristic dinucleotide-binding
motif known as the Rossmann fold (Fig. 1). The subunit structure was made of a core b-sheet composed of
seven parallel b-strands (bA, bB, bC, bD, bE, bF and
bG) buried between three a-helices (aB, aC and aG,

FEBS Journal 274 (2007) 5767–5779 ª 2007 The Authors Journal compilation ª 2007 FEBS


K. S. Paithankar et al.

D-3-Hydroxybutyrate dehydrogenase

Table 1. Crystal and refinement data of crystal forms I, II and III.
au, Asymmetric unit.
I
Space group
Unit-cell

˚
dimensions (A)
b (°)
˚
Resolution (A)a
Completeness (%)a
Rsym (%)a
I ⁄ dIa
Redundancya
Mosaicity (°)
˚
Wilson B factor (A2)
Monomers ⁄ a.u.
Solvent content (%)
R ⁄ Rfree (%)
˚
rmsd bonds (A)
rmsd angles (°)
No. of water
molecules
˚
< Bprotein > (A2)
˚
< Bwaters > (A2)
˚
< BNAD > (A2)

II

III

C2
261.5
59.9
116.5
113.7
30–2.0
(2.09–2.02)
96.6 (74.4)
8.2 (59.2)
7.8 (1.0)
7.5 (4.8)
0.69
36.1
8
36.9
20.2 ⁄ 27.4
0.028
2.28
473

C2
115.3
58.2
119.4
92.3
30–1.9
(1.97–1.9)
99.5 (98.3)
4.8 (15.1)

13.1 (4.9)
11.3 (11.0)
0.71
18.6
4
34.1
16.8 ⁄ 21.8
0.016
1.66
447

C2
117.6
58.8
119.4
93.7
30–2.1
(2.2-2.12)
99.1 (94.4)
6.7 (36.9)
8.3 (2.1)
9.9 (7.8)
1.4
34.4
4
36.1
18.1 ⁄ 24.6
0.028
2.04
320

42.7
43.5
45.7

18.6
20.7
17.6

A

33.0
33.8
–

B

a

The values given in parentheses refer to the highest resolution
shell.

or aD, aE and aF) located on both sides of the
b-sheet. The substrate-binding loop consisted of two
helices, designated aFG1 and aFG2.
With the exception of the substrate-binding loop,
the catalytic subdomains formed a compact, flat tetra˚
meric structure of dimensions 70 · 80 · 40 A along
the Q, P and R axes, respectively. Only the substratebinding loop protruded from the main body of the
tetramer along the R axis (Fig. 1). This loop was

partially disordered in most subunits (Table 2). However, the loop was completely defined in subunit A of
crystal form II and in subunit B of crystal form III,
both in the absence of NAD+, as well as in subunit D
.
of crystal form I in the presence of NAD+

C

Fig. 1. Crystal structure of P. putida D-3-hydroxybutyrate dehydrogenase (PpHBDH). (A) Fold of one subunit in the closed conformation. The substrate-binding loop is colored red and the bound NAD+
is colored yellow. (B) View of the tetramer structure along the R
axis (blue). The P and Q axes are marked in red and green, respectively. Shown are subunits A, B, C and D of crystal form I. Only in
subunits C and D (green) is an NAD+ molecule (yellow) bound. In
subunits A and B (blue) the coenzyme-binding site is not occupied.
The substrate-binding loops are depicted in red. (C) View along the
P axis.

FEBS Journal 274 (2007) 5767–5779 ª 2007 The Authors Journal compilation ª 2007 FEBS

5769


5770

˚
The number given as ‘crystal contacts’ describes all interatomic distances smaller than 4.5 A between an atom of the substrate-binding loop and an atom of a symmetry related molecule.
a

–
39.1
–

35.8
–
39.5
–
46.5
19.7
19.5
20.5
17.7
–
29
–
57.3
–
50.9

63.2
45.2

36.2
41.2

39.6
42.1

51.0
48.5

42.9
43.1

42.2
42.1

12.8
15.8

32.8
33.0
32.3
32.8
18.5
19.5
18.4
43.2
42.8
42.6
42.7
42.3
42.3

B (substrate-binding
˚
loop 186–212) (A2)
B (catalytic subdomain
residues 2–185
˚
213–256) (A2)
˚
B (NAD+) (A2)

B (residues
˚
88–89) (A2)

43.2

42.8

43.8

E
+
6
2–198
205–256
41
D
+
12
2–256

C
+
5
2–194
207–256
47.2
B
–
31

2–198
202–256
42.4
A
–
38
2–197
203–256
42.8
Chain
NAD+
Crystal contactsa
Residues

I

Crystal form

Table 2. Overview of the subunit structures of the three crystal forms.

F
+
5
2–189
207–256
43.2

G
+
11

2–199
204–256
41.6

II

17.6

18.4

33.4

A
–
70
2–197
200–256
34

K. S. Paithankar et al.

H
+
5
2–188
204–256
44.5

A
–

89
2–256

B
+
21
2–199
202–256
18.8

C
+
14
2–197
205–256
19.2

D
+
21
2–197
206–256
19.3

III

B
–
196
2–256

C
–
150
2–196
201–256
34.3

D
–
110
2–195
203–256
32.13

D-3-Hydroxybutyrate dehydrogenase

Comparison of the subunit structures
Crystal form I
Figure 2A shows a superposition of the eight subunits
in crystal form I. It demonstrates that subunits with
and without NAD+ adopted different conformations
for regions in the substrate-binding loop, helix aC and
the residues Ala88 and Gly89 after b-strand D. Those
regions that showed conformational variability in the
superposition also displayed significantly higher B factors (Fig. 2B). The average B factor for the atoms of
˚
helix C was 47.1 A2 (average over all molecules) and
˚
its residues superimposed with an rmsd of 0.6 A, indicating an increased mobility of this helix. Whereas the

subunits without NAD+ had their substrate-binding
loop in an open conformation, all subunits with
NAD+ exhibited a closed conformation. Interestingly,
in subunit H (with bound NAD+), residues 186–189
corresponded to the open conformation of the substrate-binding loop, whereas residues 204–208 (at the
end of the loop) were in a position that is similar to
the position of this region in the closed conformation.
Residues 190–203 were disordered. This finding indicated that the loop can also change to the open conformation in the presence of NAD+.
Crystal form II
A superposition of the subunits from crystal form II
showed that the substrate-binding loop of subunit A
(NAD+ free) is in an open conformation and completely ordered whereas the corresponding loops in the
NAD+-complexed subunits B to D are in a closed
conformation and partially disordered (Fig. 2C).
Besides this, the largest variability was seen again in
helix aC and in the region after b-strand D. Compared
˚
with the average B factor of 19.7 A2 for all four protein molecules, helix aC had a somewhat higher B fac˚
tor, of about 29 A2, also in this crystal form. In the
NAD+-free subunit A, residues 88 and 89 of the central b-strand had a conformation different from the
NAD+-containing subunits, similar to the situation in
˚
subunit IB. Both residues shifted up to 2 A upon
NAD+ binding.
Crystal form III
In crystal form III (all subunits are NAD+ free), subunit B possessed a completely ordered substrate-binding loop, whereas this loop was partially disordered in
the other subunits. All four subunits in crystal form III
were in an open conformation with respect to the substrate-binding loop (Fig. 2D). In contrast to crystal

FEBS Journal 274 (2007) 5767–5779 ª 2007 The Authors Journal compilation ª 2007 FEBS

K. S. Paithankar et al.

forms I and II, residues 88 and 89 adopted a conformation similar to that observed in the subunit structures with bound NAD+. As in the other crystal
forms, helix C showed a higher variability, with an
˚
˚
rmsd of 0.4 A, and a higher B factor, of 46 A2, com˚ 2.
pared with the average B factor of 40.4 A

D-3-Hydroxybutyrate dehydrogenase

A

Crystal packing interactions
As in other crystal structures of HBDH enzymes, the
exposed substrate-binding loop, which forms two faces
on opposite sides of the HBDH tetramer (Fig. 1), was
involved in crystal contacts in almost all subunits
(Table 2). Interestingly, only in the NAD+-bound subunits IIC and IID, where the loop is in a closed conformation, it was not involved in crystal contacts.
Furthermore, the loop was in a closed conformation in
all subunits with bound NAD+ and in an open conformation in all cosubstrate-free subunits. These findings suggest that the closed conformation of the loop
is a result of cosubstrate binding.

B

Comparison of PpHBDH with PfHBDH and DHRS6
The major differences between the PfHBDH and
PpHBDH structures are in the substrate-binding loop,

in residues of the central b-strand D and in helix aC
(Fig. 3). The substrate-binding loop in the cosubstratefree PfHBDH structure has an intermediate position
between the open and closed forms of this loop in
PpHBDH. The loop is completely disordered in
the structure of PfHBDH in complex with NAD+.
The conformational change of residues 88 and 89 in
the central b-strand of PpHBDH upon NAD+ binding
was not observed in PfHBDH. In both structures of
the latter enzyme, the b strand was in exactly the same
conformation as in the NAD+-bound form of
PpHBDH.
In addition to these conformational differences, helix
aC is four residues shorter in PpHBDH because of a
deletion of residues. Moreover, the substrate-binding
loop of PpHBDH is one residue shorter that that of
PfHBDH, because of a deletion at the end of helix
aFG1.
In human DHRS6, a cytosolic type 2 human HBDH
enzyme [11], the substrate-binding loop is present in a

C

D

Fig. 2. Superposition of subunit structures. (A) Crystal form I: A,
green; B, black; C, yellow; D, red; E, magenta; F, blue; G, cyan; H,
brown. (B) Ca traces of the eight monomers of crystal form I col˚
˚
ored by the B factor (from blue at B < 30 A2 to red at B > 60 A2).
(C) Crystal form II: A, green; B, red; C, blue; D, black. (D) Crystal

form III: A, black; B, red; C, blue; D, green.

FEBS Journal 274 (2007) 5767–5779 ª 2007 The Authors Journal compilation ª 2007 FEBS

5771


D-3-Hydroxybutyrate dehydrogenase

K. S. Paithankar et al.

Asn87

Ala88
Fig. 3. Comparison of P. putida D-3-hydroxybutyrate dehydrogenase (PpHBDH) (open form green, closed form red), Pseudomonas
fragi D-3-hydroxybutyrate dehydrogenase (PfHBDH) (open form
blue) and DHRS6 (black). The structure of the NAD+-bound form of
PfHBDH (not shown) with the disordered substrate-binding loop
superimposes closely with the open form.

closed conformation (Fig. 3). A sulfate molecule from
the crystallization solution is bound to the active site
at the presumed binding site for the substrate. The
loop closure may be caused by the sulfate ion, as discussed by the authors, and also by interactions with
NAD+, as described below. The substrate-binding
loop of DHRS6 is six residues shorter, as in the Pseudomonas HBDH enzymes. Consequently, there is a
shortening of helix aFG1 and a slight relocation of
helix aFG2.
NAD+ binding
The conformation of NAD+ and the structure of the

cosubstrate-binding pocket of PpHBDH are very similar to those of PfHBDH, with a distance of about
˚
14 A between the C2 of nicotinamide and the C6 of
the adenine ring. As indicated in the superpositions of
Figs 2 and 3, the main difference is the displacement
˚
of Ala88 and Gly89 by about 2 A in the NAD+-free
subunits IB (chain B of crystal form I) and IIA. The B
˚
value for these residues is around 10 A2 higher than
the average B factor of the subunit. In the subunits
with bound NAD+, the B factor of the two residues is
comparable with the average B value of the subunit.
Nevertheless, the conformation of residues 88 and 89
is clearly defined in all electron density maps. The
main chain torsion angles of residues 87–90 all correspond to allowed regions of the Ramachandran plot.
In crystal form III, residues 88 and 89 adopted a conformation similar to that of the NAD+-bound subunits in crystal forms I and II. In this crystal form, the
˚
B factor of residues 88 and 89 was 5–10 A2 higher
than the average B value of the corresponding subunit.
5772

Gly89
Gln91
lle90

Fig. 4. Conformational change of residues 87–91 upon NAD+ binding. The carbon atoms are colored grey in the conformation in the
presence of NAD+ (yellow) and green in the NAD+-free subunit A
of crystal form I.

The two residues are thus more mobile than in the
presence of NAD+, but less than in the alternate conformations observed in subunits IB and IIA.
NAD+ interacts, via a hydrogen bond of one of its
ribose hydroxyl oxygens, with the peptide carbonyl
group of Asn87 (Fig. 4). The loss of this interaction
may be an important factor for the change of main
chain conformation of this residue in the absence of
NAD+. The present data thus indicate that residues
87–90 have a higher mobility in the absence of bound
cosubstrate (as indicated by higher B factors) and they
can adopt alternative conformations.
Translation, libration and screw (TLS) refinement
At the end of the refinement procedure, anisotropic
displacement parameters were determined by a TLS
refinement [12,13]. Translation (T) and libration (L)
tensors describe the anisotropic motion of groups in
the crystal. Besides improving the fit of the model to
the observed data, the TLS tensors may allow a
description of correlated motions in the crystal. It
must be stressed, however, that a fit of the TLS model
to the observed structure factor amplitudes implies
nothing about the relative phases of the atomic displacements within the group [13].

FEBS Journal 274 (2007) 5767–5779 ª 2007 The Authors Journal compilation ª 2007 FEBS


K. S. Paithankar et al.

All three crystal forms were subject to TLS refinement with the substrate-binding loop region (186–212)
and the catalytic domain (2–185, 213–256) in each subunit as two distinct TLS groups. By TLS refinement,

the R factors (Rfree factors) improved by 2.2% (3.3%),
0.7% (1.0%) and 1.7% (2.9%) for crystal forms I, II
and III, respectively.
The total B factor (Btotal) of each atom is the sum of
the TLS contribution (BTLS, from the rigid body
motion) and the residual B factor (Bresidual, the individual mobility independent of the rigid bodies). The
residual B factors of the substrate-binding loop were
relatively constant and had values similar to those
of other regions (data shown in the supplementary
Fig. S1). The significantly higher B factors of the substrate-binding loop are predominantly caused by a
rigid body motion of the loop. For all crystal forms, a
significant librational movement is present for the substrate-binding loop (data shown in the supplementary
Table S1). Furthermore, the libration is quite anisotropic in nature (Fig. 5).
How does the anisotropic librational motion of the
substrate-binding loop compare with the closing
motion of this loop? A superposition of the main axis

Fig. 5. Principal axis (green) of the libration tensor of the substratebinding loop of subunit IIIA. The anisotropic movement of the Ca
atoms, as derived from the TLS tensors, is depicted by thermal
ellipsoids for the loop (red) and for the catalytic subdomain (blue).
For orientation, a superimposed NAD+ molecule is shown in yellow,
although it is not bound in this crystal form.

D-3-Hydroxybutyrate dehydrogenase

of the L tensors for all subunits showed that all L tensors roughly describe a similar libration movement in
which the main rotation component is an axis approximately parallel to the two helices of the loop (Fig. 5,
superposition not shown). The second largest rotational component describes the closure motion of the
loop.
Movement of the substrate-binding loop

In order to characterize the nature of the movement of
the substrate-binding loop, the structures of open and
closed forms were compared using program dyndom
to determine dynamic regions that move as pseudorigid structures, termed ‘dynamic domains’ (Fig. 6).
For stretches of five amino acid residues, the rotational
movement between two enzyme conformations was
analyzed. Residues that belong to one rigid body show
a similar rotation in the superposition and thus form a
cluster, as shown in Fig. 6B. The analysis revealed two
dynamic domains: residues 4–183 and 212–254 form
one domain; and residues 187–210 of the substratebinding loop form the other domain. However, within
the substrate-binding loop, three subclusters were
defined: residues 187–199 of helix aFG1; residues 204–
210 of helix aFG2; and residues 200–203 of the loop
connecting the two helices. Thus, the substrate-binding
loop moves largely as a rigid body; however, the internal structure of the loop changes slightly by a re-orientation of the two helices. Residues 184–186 and 211–
213 are the bending residues that connect the two
dynamic domains (Fig. 6).
The substrate-binding loops in the open and closed
˚
conformations superimposed with an rmsd of 1.4 A.
The movement corresponded to a rotation by 46° with
˚
a small translational component of 0.06 A. It is typical
for hinge-bending movements that the rotation axis
passes near the bending residues, which thus act as a
mechanical hinge (Fig. 6) [14]. An analysis of the main
chain conformations in the different loop structures
showed that several small changes of main-chain torsion angles of the bending residues allowed the movement of the substrate-binding loop, but no large
changes of the main chain conformation were observed

(data not shown).
A movement of the substrate-binding loop upon cosubstrate binding has not been observed before for
other SDR enzymes. Only for human estrogenic
17b-hydroxysteroid dehydrogenase [5,15], Drosophila
lebanonensis alcohol dehydrogenase [16], Datura stramonium tropinone reductase [17] and Escherichia coli
b-keto acyl carrier protein reductase [18,19] have structures of the apoenzyme and of the binary complex

FEBS Journal 274 (2007) 5767–5779 ª 2007 The Authors Journal compilation ª 2007 FEBS

5773


D-3-Hydroxybutyrate dehydrogenase

K. S. Paithankar et al.

200

200

A

203

203

213

211
184

B

211

213

186

186
184

201
200

202
203

205
206
207 208
209
210
204
211

189
194 195
193
190 196

191
188
192
197
187
199

198
186

212
185
213

184

with cosubstrate been determined, without significant
differences in the position of the substrate-binding
loops. In the crystal structure of 3a-hydroxysteroid
dehydrogenase ⁄ carbonyl reductase from Comamonas
testosteroni the substrate-binding loop is largely disordered in the absence and presence of bound
NAD+[20]. There are, however, crystal structures of
SDR enzymes in the closed form available for binary
complexes with cosubstrate, such as the structure of
5774

Fig. 6. Dynamic domains of P. putida D-3hydroxybutyrate dehydrogenase (PpHBDH)
on the basis of a comparison of conformers
ID and IIA, which contain a fully defined
substrate-binding loop. (A) The two dynamic

domains are colored blue and red for the
fold of conformer ID (closed) and the bending residues are shown in green. (B) Clustering of the rotational movements of
stretches of five residues, on which the
assignment of the domains moving as rigid
bodies is based in program DYNDOM [31].
Each sphere represents the rotation vector
of a five-residue stretch. For the substratebinding loop and the bending residues, the
rotation vectors are labeled according to the
residue in the center of the stretch.

human DHRS6 in complex with NAD [11]. A structure of the DHRS6 apoenzyme is not yet available.
Residues inducing closure movement
In an analysis of ligand-induced domain movements in
other proteins, it was shown that a small number of
residues from the closing domain interact with the
ligand bound to the binding domain in the open

FEBS Journal 274 (2007) 5767–5779 ª 2007 The Authors Journal compilation ª 2007 FEBS


K. S. Paithankar et al.

D-3-Hydroxybutyrate dehydrogenase

conformation to initiate and drive domain closure [21].
In this model, specific interactions of the ligand with
residues on the coenzyme-binding domain produce a
torque about the hinge axis driving the domain closure. Residues are assumed to induce closure movement if they satisfy specific conditions. These residues
are usually located in the bending regions or in the
closing domain. Additionally, thermal motions may

also contribute to the closure movement.
A large number of interactions are involved in the
binding of NAD+ to the catalytic subdomain of
PpHBDH. However, there are only a few interactions
of the substrate-binding loop with NAD+ that might
cause the conformational change. A crucial residue for
the coenzyme-induced conformational change might be
Thr187, which forms hydrogen bonds to the nicotinamide NH2 and to a phosphate oxygen of NAD+ via
its hydroxyl group (Fig. 7). In the absence of NAD+,
˚
the alcoholic oxygen is displaced by about 1.5 A, such
˚ distant from the position of its putative
that it is  4 A
hydrogen-bonding partners. This interaction might
trigger the closure motion of the substrate-binding
loop in the presence of NAD+ because Thr187 is
located just behind the bending residues 184–186 and
the torque produced by the interaction of Thr187 with
NAD+ drives or supports the loop movement about
the hinge residues. A further, nonpolar interaction of
the substrate-binding loop with the coenzyme is mediated by the side chain of Leu189, which makes hydrophobic contacts with the ribose group and
nicotinamide ring of NAD+ in the closed conformation but is faced towards the solvent in the open conformation (Fig. 7). Both residues are conserved in the
bacterial HBDHs as part of a ‘TPLV’ motif. Also, the
interaction of the main chain NH and CO groups of

Val185 might contribute to the conformational change
by binding to the NAD+ nicotinamide group. There
are no further interactions that might explain the conformational change of the substrate-binding loop upon
NAD+ binding. The rest of the substrate-binding loop
is also quite diverged, with the exception of Gln193. In

the human enzyme DHRS6, Thr187 is conserved and
makes a polar contact to NAD+, as in the bacterial
enzyme. Leu189 is replaced by a serine, which is in
hydrogen-bonding distance to a phosphate oxygen
atom of NAD+ and may thus also be involved in the
induction of closure movement via cosubstrate binding. The interaction of Val186 (Val185 in PpHBDH)
and Thr188 (Thr187) of 3a-hydroxysteroid dehydrogenase with the cofactor has also been discussed to stabilize the substrate-binding loop in the loop–helix
transition observed in this enzyme [7].
Substrate binding
A model for the binding mode of the substrate D-3-hydroxybutyrate to HBDH has been suggested based on
a homology model of PpHBDH and molecular modelling techniques [9]. In this model, the side chain of
Gln193 belonging to the substrate-binding loop forms
hydrogen bonds to the carboxylate group of the substrate. Figure 8 shows the modelled substrate in the

Lys149

Gln193

Gln91

His141

Tyr152

Leu189

Leu189

Thr187

Val185

Fig. 7. Interactions of NAD+ with P. putida D-3-hydroxybutyrate
dehydrogenase (PpHBDH), which might drive a cosubstrate-induced
conformational change of the substrate-binding loop. The closed
conformation is depicted in green and the open conformation in
grey. Also shown is the rotation axis in blue.

Fig. 8. A model for the binding mode of substrate D-3-hydroxybutyrate (cyan) to P. putida D-3-hydroxybutyrate dehydrogenase
(PpHBDH). Shown are the enzyme conformations in the open
(green) and closed (red) forms, and selected residues, as discussed
in the main text. The cosubstrate NAD+ is shown with yellow carbon atoms. Hydrogen bonds are shown as dashed lines, and the
red dashed line between the nicotinamide C4 atom and the substrate carbon atom bound to the substrate alcohol group marks the
distance between the reactive centers.

FEBS Journal 274 (2007) 5767–5779 ª 2007 The Authors Journal compilation ª 2007 FEBS

5775


D-3-Hydroxybutyrate dehydrogenase

K. S. Paithankar et al.

crystal structure of PpHBDH in the closed form. The
side chain of Gln193 was shown to contribute significantly to substrate binding because the Km value
increased from 0.6 mm for the wild-type enzyme to
about 70 mm in a Gln193Ala mutant, whereas the kcat
value decreased from 432Ỉs)1 in the wild-type enzyme
to 215Ỉs)1 in the same mutant. Gln193 was in hydrogen-bonding distance to the carboxylate group of the

substrate in the closed conformation of the substratebinding loop. In all subunits where this residue was
defined in subunits with bound NAD+ (D, E and G
of crystal form II as well as B and C of crystal form II), this residue was positioned to interact with the
substrate. This finding indicates that the loop is indeed
in a position competent for catalysis upon NAD+
binding, even in the absence of the bound substrate.
Gln193 is the only residue of the substrate-binding
loop that forms polar interactions with the substrate.
In addition, the side chain of Leu189 has hydrophobic
contacts to the methyl group of the substrate model.
Further polar interactions to the substrate are formed
by Lys149, Gln91, His141 and Tyr152. Of these residues, the important function for substrate binding has
been demonstrated for a Gln91Ala mutant with Km
51 mm and kcat 411Ỉs)1 and for a Lys149Ala mutant
that was essentially inactive [9]. His141 appears to be
also important for efficient catalysis because in a
His141Ala mutant the Km increased to only 4 mm,
whereas the kcat decreased significantly to 13Ỉs)1.
Tyr152 is known to be a core catalytic residue. It is
assumed to be present as a tyrosinate and to accept a
proton from the alcohol group in order to facilitate H–
transfer to NAD+. Our crystallographic study confirmed the repositioning of the substrate-binding loop
in the closed conformation, as obtained from a molecular dynamics simulation of a PpHBDH model [9].
A superposition of the active-site structures of
PpHBDH and DHRS6 showed that the human
enzyme developed a different environment for substrate
binding: His141 was replaced by an alanine, Lys149 by
an arginine, Gln91 by a valine and Gln193 (from the
substrate-binding loop) by an arginine (data not
shown). These replacements might account for the significant differences in the kinetic data of both enzymes.

Figure 9 shows the molecular surface of HBDH in
the open conformation, together with the model for
the substrate binding mode. The NAD+ coenzyme
taken from the structures of PpHBDH in the closed
form was included in the calculation of the surface.
Thus, the surface represented the protein in a state
where NAD+ has just bound, but the loop is still in
the open conformation. The substrate was bound in a
deep groove formed between the substrate-binding
5776

Fig. 9. Molecular surface of P. putida D-3-hydroxybutyrate dehydrogenase (PpHBDH) in the open conformation (conformer IIA) colored
by the electrostatic potential. Positive potential is depicted in blue
and negative potential in red. Also shown are the model for substrate binding in green and the substrate-binding loop conformation
in the closed form in yellow.

loop and the catalytic subdomain. The carboxylate
group of the substrate was located at a region with
positive potential, which is mainly caused by Lys149.
Also shown is the substrate-binding loop in the closed
conformation. In particular, residues Leu189 and
Gln193 would block the entrance to the substratebinding pocket in the closed conformation. A molecular surface drawn for the enzyme in the closed form
revealed no access to the buried substrate-binding
pocket (data not shown). Kinetic studies demonstrated
that HBDH, similarly to other NAD+-dependent dehydrogenases, has an ordered sequential binding mechanism of cosubstrate binding followed by substrate
binding (M. Grunow, unpublished results). Therefore,
although the binding of the coenzyme obviously
induces a change of the enzyme to the closed form, the
loop must exist in an equilibrium with the open form
to enable binding of the substrate and release of the

products. This flexibility is demonstrated by a partial
disorder of the substrate-binding loop in some subunits
of NAD+-bound subunits of PpHBDH and also by
the strong disorder of the loop in the NAD+–
PfHBDH complex [10].
In conclusion, the crystallographic analysis of
PpHBDH in different crystal forms and in the presence and absence of bound NAD+ showed that the
presence of the cosubstrate alone is able to induce a
conformation of the substrate-binding loop that is
competent for catalysis. Such a conformational
change of the substrate-binding loop has not been
observed in previous crystallographic studies on other
SDR enzymes. Our results are in agreement with the

FEBS Journal 274 (2007) 5767–5779 ª 2007 The Authors Journal compilation ª 2007 FEBS


K. S. Paithankar et al.

role of the cosubstrate in promoting a loop–helix
transition in 3a-hydroxysteroid dehydrogenase [7]. A
cosubstrate-induced conformational change is also in
agreement with spectroscopic studies [8]. These studies
demonstrate an important contribution of cosubstrate
binding in shifting the equilibrium between disordered
and ordered, as well as open and closed, conformations of the substrate-binding loop. Thus, the equilibrium is progressively shifted towards the closed
conformation – initially upon cosubstrate binding and
then by substrate binding. Cosubstrate and substrate
have interactions with the substrate-binding loop in
the closed conformation. It is possible that the shift

towards the closed conformation induced by cosubstrate binding is necessary for substrate binding and
ensures the sequential binding mechanism (i.e. that
substrate binding does not occur before cosubstrate
binding). As outlined above, the current data strongly
indicate that the conformational change is a result of
cosubstrate binding and that it is not a result of crystal packing interactions. The nature of the conformational change could be characterized in detail by a
comparison of the different loop conformers and by
an analysis of the loop mobility in the crystal. In first
approximation, the loop moves as a rigid body, with
minor re-arrangements of the relative orientation of
the two helices (aFG1 and aFG2). Thus, NAD+
binding increases the flexibility of the substrate-binding loop and shifts the equilibrium between the open
and closed forms towards the closed form. In all
subunits the loop is in either the open or the closed
conformation; no intermediate conformations have
been observed. Information on substrate binding is
currently based on modeling of the substrate binding
mode; however, it may be possible that the interactions between the enzyme and the substrate or analogues inhibitors can be studied by cocrystal
structures in further studies.

D-3-Hydroxybutyrate dehydrogenase

10 mm acetoacetate. Acetoacetate was added in order to
study substrate binding. However, under the specified
conditions it did not bind to the enzyme. The same crystals were also obtained in the absence of acetoacetate
under identical conditions. The presence of the substrate
in the crystallization buffer had no influence on the
refined structure, in particular concerning the position of
the substrate-binding loop. We present the structure
refined from crystals obtained in the presence of acetoacetate because the best data were obtained from these

crystals. Many crystals of PpHBDH suffered from a significant orientational disorder of the PpHBDH molecules
in the crystals, which resulted in high Wilson B factors
and poor density. This phenomenon was independent of
the presence of acetoacetate. For crystal forms I and II,
2 mm NAD+ was added. Crystals appeared within a couple of days. The data sets of crystal forms I and II, used
for refinement of the structures presented here, were in
fact obtained from two fragments of the same crystal
that broke apart upon transfer into the cryobuffer. The
smaller fragment belonged to crystal form I and the larger fragment to crystal form II. Crystals for which data
were collected at room temperature (22 °C) belonged to
crystal form II. Thus, crystal form I is probably the
result of a phase change upon crystal cooling.

Data collection
Crystals of size 400 · 150 · 150 lm were cryoprotected in
the crystallization buffer, which included 15% (v ⁄ v) glycerol, and flash-frozen in a N2 stream at 100 K in a nylon
loop. The intensity data were collected using MAR345
image plate detectors (Mar Research Inc., Norderstedt,
Germany) mounted to an Rigaku RU-H3R rotating anode
generator (Rigaku Corp., Tokyo, Japan) or a Bruker
FR591 Microstar generator (Bruker AXS, Delft, the Netherlands). Data processing and reduction was carried out
using the hkl software, version 1.97.9 (HKL Research Inc.,
Charlottesville, VA, USA) [22]. All crystals belonged to
monoclinic space group C2. Details of data collection and
refinement are listed in Table 1.

Experimental procedures
Enzyme purification and crystallization
Recombinant PpHBDH was expressed in E. coli XL1Blue strain and purified by charge-controlled hydrophobic
chromatography, as described previously [9]. Before crystallization, PpHBDH was further purified by Sephadex

G-100 size-exclusion chromatography (GE Healthcare
Bio-Sciences, Uppsala, Sweden). For crystallization using
the hanging drop vapor diffusion method, 3 lL of
PpHBDH (10 mgỈmL)1) was mixed with an equal volume
of crystallization buffer containing 17–20% polyethylene
glycol 1500, 0.1 m Tris-HCl, pH 7.1, 0.2 mm CaCl2 and

Structure determination and refinement
The phase problem was solved by molecular replacement
with molrep [23] using the PfHBDH structure (Protein
Data Bank accession code: 1WMB) as a search model.
Model building was performed using o [24] and crystallographic refinement with refmac [25]. Before refinement of
˚
the TLS parameters, the B factors were set to 20 A2. The
program tlsanl was used to calculate the principal axes of
the translation and libration tensors. The quality of the
models was assessed with Ramachandran plots using the
program procheck [26].

FEBS Journal 274 (2007) 5767–5779 ª 2007 The Authors Journal compilation ª 2007 FEBS

5777


D-3-Hydroxybutyrate dehydrogenase

K. S. Paithankar et al.

Structure analysis and generation of figures
Electrostatic potential maps were calculated by solution of

the Poisson–Boltzmann equation in a continuum electrostatic model, as implemented in the program delphi [27]. A
˚
probe radius of 1.4 A was used, and the dielectric constant
was set to 4 for the protein region and to 80 for the solvent. Full charges of Asp, Glu, Lys and Arg were used, in
addition to the charges of the terminal amino acids. The
molecular surface was generated with the program msms
[28]. The molecular figures were generated with programs
molscript [29] and raster3d [30].

Protein Data Bank accession codes
The crystallographic models are available from the RCSB
Protein Data Bank under the accession codes 2Q2Q (crystal
form I), 2Q2V (crystal form II) and 2Q2W (crystal
form III).

Acknowledgements
The Deutsche Forschungsgemeinschaft is acknowledged for funding to MG and NS. The BESSY synchrotron in Berlin, Germany, is acknowledged for
beamtime at the PSF beamlines.

References
1 Tanaka N, Nonaka T, Nakamura KT & Hara A (2001)
SDR: structure, mechanism of action, and substrate
recognition. Curr Org Chem 5, 89–111.
2 Jornvall H, Persson B, Krook M, Atrian S, GonzalezDuarte R, Jeffery J & Ghosh D (1995) Short-chain
dehydrogenases ⁄ reductases (SDR). Biochemistry 34,
6003–6013.
3 Oppermann U, Filling C, Hult M, Shafqat N, Wu X,
Lindh M, Shafqat J, Nordling E, Kallberg Y, Persson B
et al. (2003) Short-chain dehydrogenases ⁄ reductases
(SDR): the 2002 update. Chem Biol Interact 143–144,

247–253.
4 Ghosh D, Weeks CM, Grochulski P, Duax WL, Erman
M, Rimsay RL & Orr JC (1991) Three-dimensional
structure of holo 3 alpha,20 beta-hydroxysteroid dehydrogenase: a member of a short-chain dehydrogenase
family. Proc Natl Acad Sci USA 88, 10064–10068.
5 Ghosh D, Pletnev VZ, Zhu DW, Wawrzak Z, Duax
WL, Pangborn W, Labrie F & Lin SX (1995) Structure
of human estrogenic 17 beta-hydroxysteroid dehydrogenase at 2.20 A resolution. Structure 3, 503–513.
6 Tanaka N, Nonaka T, Tanabe T, Yoshimoto T, Tsuru D
& Mitsui Y (1996) Crystal structures of the binary and
ternary complexes of 7 alpha-hydroxysteroid dehydrogenase from Escherichia coli. Biochemistry 35, 7715–
7730.

5778

7 Nakamura S, Oda M, Kataoka S, Ueda S, Uchiyama S,
Yoshida T, Kobayashi Y & Ohkubo T (2006) Apo- and
holo-structures of 3alpha-hydroxysteroid dehydrogenase
from Pseudomonas sp. B-0831. Loop-helix transition
induced by coenzyme binding. J Biol Chem 281, 31876–
31884.
8 Li B & Lin SX (1996) Fluorescence-energy transfer in
human estradiol 17 beta-dehydrogenase-NADPH complex and studies on the coenzyme binding. Eur J Biochem 235, 180–186.
9 Feller C, Gunther R, Hofmann HJ & Grunow M (2006)
ă
Molecular basis of substrate recognition in D-3-hydroxybutyrate dehydrogenase from Pseudomonas putida.
Chembiochem 7, 1410–1418.
10 Ito K, Nakajima Y, Ichihara E, Ogawa K, Katayama
N, Nakashima K & Yoshimoto T (2006) D-3-hydroxybutyrate dehydrogenase from Pseudomonas fragi: molecular cloning of the enzyme gene and crystal structure of
the enzyme. J Mol Biol 355, 722–733.

11 Guo K, Lukacik P, Papagrigoriou E, Meier M, Lee
WH, Adamski J & Oppermann U (2006) Characterization of human DHRS6, an orphan short chain dehydrogenase ⁄ reductase enzyme: a novel, cytosolic type 2 Rbeta-hydroxybutyrate dehydrogenase. J Biol Chem 281,
10291–10297.
12 Winn MD, Murshudov GN & Papiz MZ (2003) Macromolecular TLS refinement in REFMAC at moderate
resolutions. Methods Enzymol 374, 300–321.
13 Winn MD, Isupov MN & Murshudov GN (2001) Use
of TLS parameters to model anisotropic displacements
in macromolecular refinement. Acta Crystallogr D Biol
Crystallogr 57, 122–133.
14 Hayward S (1999) Structural principles governing
domain motions in proteins. Proteins 36, 425–435.
15 Mazza C, Breton R, Housset D & Fontecilla-Camps JC
(1998) Unusual charge stabilization of NADP+ in
17beta-hydroxysteroid dehydrogenase. J Biol Chem 273,
8145–8152.
16 Benach J, Atrian S, Gonzalez-Duarte R & Ladenstein R
(1998) The refined crystal structure of Drosophila
lebanonensis alcohol dehydrogenase at 1.9 A resolution.
J Mol Biol 282, 383–399.
17 Nakajima K, Yamashita A, Akama H, Nakatsu T,
Kato H, Hashimoto T, Oda J & Yamada Y (1998)
Crystal structures of two tropinone reductases: different
reaction stereospecificities in the same protein fold.
Proc Natl Acad Sci USA 95, 4876–4881.
18 Price AC, Zhang YM, Rock CO & White SW (2004)
Cofactor-induced conformational rearrangements establish a catalytically competent active site and a proton
relay conduit in FabG. Structure 12, 417–428.
19 Price AC, Zhang YM, Rock CO & White SW (2001)
Structure of beta-ketoacyl-[acyl carrier protein] reductase from Escherichia coli: negative cooperativity and its
structural basis. Biochemistry 40, 12772–12781.

FEBS Journal 274 (2007) 5767–5779 ª 2007 The Authors Journal compilation ª 2007 FEBS


K. S. Paithankar et al.

20 Grimm C, Maser E, Mobus E, Klebe G, Reuter K &
Ficner R (2000) The crystal structure of 3alphahydroxysteroid dehydrogenase ⁄ carbonyl reductase
from Comamonas testosteroni shows a novel
oligomerization pattern within the short chain
dehydrogenase ⁄ reductase family. J Biol Chem 275,
41333–41339.
21 Hayward S (2004) Identification of specific interactions
that drive ligand-induced closure in five enzymes with
classic domain movements. J Mol Biol 339, 1001–
1021.
22 Otwinowski Z & Minor W (1997) Processing of X-ray
diffraction data collected in oscillation mode. Methods
Enzymol 276, 307–326.
23 Vagin A & Teplyakov A (2000) An approach to multicopy search in molecular replacement. Acta Crystallogr
D Biol Crystallogr 56, 1622–1624.
24 Jones TA, Zou JY, Cowan SW & Kjeldgaard M (1991)
Improved methods for building protein models in electron density maps and the location of errors in these
models. Acta Crystallogr A 47, 110–119.
25 Murshudov GN, Vagin AA & Dodson EJ (1997)
Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallogr D Biol Crystallogr 53, 240–255.
26 Laskowski RA, MacArthur MW & Thornton JM
(1998) Validation of protein models derived from experiment. Curr Opin Struct Biol 8, 631–639.
27 Gilson MK, Sharp KA & Honig BH (1987) Calculating
the electrostatic potential of molecules in solution:

method and error assessment. J Comput Chem 9, 327–
335.
28 Sanner MF, Olson AJ & Spehner JC (1996) Reduced
surface: an efficient way to compute molecular surfaces.
Biopolymers 38, 305–320.

D-3-Hydroxybutyrate dehydrogenase

29 Kraulis PJ (1991) MOLSCRIPT: a program to produce
both detailed and schematic plots of protein structures.
J Appl Crystallogr 24, 946–950.
30 Merritt EA & Murphy ME (1994) Raster3d Version
2.0. A program for photorealistic molecular graphics.
Acta Crystallogr D Biol Crystallogr 50, 869–873.
31 Hayward S & Lee RA (2002) Improvements in the analysis of domain motions in proteins from conformational
change: DynDom version 1.50. J Mol Graph Model 21,
181–183.

Supplementary material
The following supplementary material is available
online:
Fig. S1. B factor plot for the residues of subunit IIIA.
Shown is the total B factor and its contribution from
the rigid body movements (BTLS, obtained from a
refinement of translation, libration and screw tensors)
and from the individual atom or residue movement
(Bresidual).
Table S1. Magnitudes of the translation and libration
tensors for the catalytic domain and the substratebinding loop of the different subunits in crystal forms
I to III. For the libration tensor of the substrate-binding loop the eigenvalues of the tensor are listed and

the mean value shown in brackets.
This material is available as part of the online article
from
Please note: Blackwell Publishing is not responsible
for the content or functionality of any supplementary
materials supplied by the authors. Any queries (other
than missing material) should be directed to the corresponding author for the article.

FEBS Journal 274 (2007) 5767–5779 ª 2007 The Authors Journal compilation ª 2007 FEBS

5779