Diol dehydratase-reactivating factor is a
reactivase – evidence for multiple turnovers and subunit
swapping with diol dehydratase
Koichi Mori, Yasuhiro Hosokawa, Toshiyuki Yoshinaga and Tetsuo Toraya
Department of Bioscience and Biotechnology, Graduate School of Natural Science and Technology, Okayama University, Japan
Keywords
adenosylcobalamin; coenzyme B
12
; diol
dehydratase; diol dehydratase-reactivating
factor; reactivase
Correspondence
T. Toraya, Department of Bioscience and
Biotechnology, Graduate School of Natural
Science and Technology, Okayama
University, Tsushima-naka, Kita-ku,
Okayama, 700-8530, Japan
Fax: +81 86 251 8264
Tel: +81 86 251 8194
E-mail:
(Received 26 July 2010, revised
17 September 2010, accepted 1 October
2010)
doi:10.1111/j.1742-4658.2010.07898.x
Adenosylcobalamin-dependent diol dehydratase (DD) undergoes suicide
inactivation by glycerol, one of its physiological substrates, resulting in the
irreversible cleavage of the coenzyme Co–C bond. The damaged cofactor
remains tightly bound to the active site. The DD-reactivating factor reacti-
vates the inactivated holoenzyme in the presence of ATP and Mg
2+
by

Q59470) physically interact (MI:0915)bycomigration in
non denaturing gel electrophoresis (
MI:0404)
l
MINT-7997157: Diol Dehydratase alpha (uniprotkb:Q59470), Diol Dehydratase beta (uni-
protkb:
Q59471), Diol Dehydratase gamma (uniprotkb:Q59472), Reactivase beta (uniprotkb:
O68196) and Reactivase alpha (uniprotkb:O68195) physically interact (MI:0915)bymolecular
sieving (
MI:0071)
Abbreviations
AdePeCbl, adeninylpentylcobalamin; AdoCbl, adenosylcobalamin or coenzyme B
12
; CN-Cbl, cyanocobalamin; DD, diol dehydratase.
FEBS Journal 277 (2010) 4931–4943 ª 2010 The Authors Journal compilation ª 2010 FEBS 4931
Introduction
Adenosylcobalamin (AdoCbl)-dependent enzymes cata-
lyze chemically difficult reactions by the use of highly
reactive radicals. The homolytic cleavage of the Co–C
bond of the coenzyme forms a Co(II) species and an
adenosyl radical, which triggers the reactions [1].
Although enzymes generally deal with highly reactive
intermediates by ‘negative catalysis’ [2], cobalamin
enzymes tend to undergo mechanism-based inactiva-
tion because of the involvement of highly reactive radi-
cal intermediates during catalysis [3]. Diol dehydratase
(DD) (EC 4.2.1.28) catalyzes the AdoCbl-dependent
conversion of 1,2-propanediol, glycerol and 1,2-ethane-
diol to the corresponding aldehydes [4,5]. Its physio-
logical substrates are 1,2-diols, such as 1,2-propanediol

ADP-bound form of the reactivating factor has a high
affinity for the enzyme, and interacts with the inacti-
vated holoenzyme to form a tight apoenzymeÆreactivat-
ing factor complex, with the concomitant release of the
damaged cofactor. The reactivating factor reverts to a
low-affinity form through the replacement of bound
ADP by free ATP, resulting in the dissociation of the
apoenzymeÆreactivating factor complex into apoenzyme
and the reactivating factor. Active holoenzyme is then
reconstituted from apoenzyme and free AdoCbl. DD
does not form a complex with the reactivating factor
while it exists as an active holoenzyme. The glycerol
dehydratase-reactivating factor reactivates the inacti-
vated hologlycerol dehydratase in a similar manner.
Both dehydratase-reactivating factors exist as a
2
b
2
heterotetramers [a, DdrA or GdrA (DhaF); b, DdrB or
GdrB (DhaG)] [16,19,20]. Liao et al. reported the
crystal structure of the nucleotide-free form of glycerol
dehydratase-reactivating factor [24]. Independently, we
solved the crystal structures of the DD-reactivating
factor in both the ADP-bound and nucleotide-free
forms [25]. The structures of both reactivating factors
are similar. Their a subunits have a structural feature
common to the ATPase domains of actin superfamily
proteins, including Hsp70 molecular chaperones.
Interestingly, their b subunits have similar folds to the
b subunits of diol and glycerol dehydratases. Such

reactivating factor-mediated reactivations of DD
during the dehydration of glycerol [16]. The number of
reactivations per molecule of DD was calculated to be
approximately six under conditions where the reacti-
vating factor was added to a 10-fold molar excess rela-
tive to the enzyme. This indicates that the enzyme
undergoes multiple inactivation–reactivation cycles. On
the other hand, the maximum number of reactivations
per molecule of the reactivating factor was observed to
be approximately two, at a molar ratio of the reacti-
vating factor to the enzyme of 0.5. As the reactivating
factor exists as a dimer of ab heterodimers, i.e. (ab)
2
,
it remained unclear whether the reactivating factor-
mediated reactivation of inactivated holoenzymes is
catalytic or stoichiometric. It is experimentally not
possible to demonstrate multiple turnovers for the
reactivating factor in this reactivation assay, probably
because of the inhibition of the holoenzyme by accu-
mulated 3-hydroxypropionaldehyde.
To avoid this difficulty, we examined whether the
reactivating factor can mediate multiple turnovers of
the replacement of tightly bound cyanocobalamin
(CN-Cbl) (an inactive coenzyme analog lacking the ade-
nine ring in the upper axial ligand; a model of damaged
cofactors) for free adeninylpentylcobalamin (AdePeCbl)
(an inactive coenzyme analog containing the adenine
ring in the upper axial ligand; a model of intact coen-
zyme, AdoCbl) in the presence of ATP and Mg

Fig. 2. Evidence for the catalytic turnover of the DD-reactivating
factor. (A) The reactivating factor-mediated replacement of enzyme-
bound CN-Cbl with free AdePeCbl in the presence of ATP and
Mg
2+
was analyzed by the spectral change of enzyme-bound cobal-
amin. A 10-fold excess of the enzymeÆCN-Cbl complex over the
reactivating factor was used. Experimental details are described in
the text. After removal of unbound cobalamin at 0 min (thick solid
line), 30 min (thin solid line), 60 min (thin long-dashed line),
120 min (thin short-dashed line), 240 min (thin dotted line) and
360 min (thick dotted line) of incubation, the spectrum of enzyme-
bound cobalamin was measured. Inset: spectra of enzyme-bound
CN-Cbl (solid line) and AdePeCbl (dotted line). (B) Experimental con-
ditions were the same as in (A), except that spectra were taken
after 360 min of incubation in the absence of ATP and Mg
2+
(thick
solid line) or without the reactivating factor in the presence of ATP
and Mg
2+
(thick dotted line). (C) Time course of the reactivating fac-
tor-mediated exchange of enzyme-bound CN-Cbl for AdePeCbl. The
extent of exchange was determined from the change in absor-
bance at 364 nm. The total amount of enzyme-bound cobalamin
was determined spectrophotometrically after conversion to the
dicyano form. Inset: a semilogarithmic plot.
K. Mori et al. Reactivase for coenzyme B
12
-dependent enzyme

2
dimer, it can be
assumed that the reactivase mediates the exchange of
enzyme-bound damaged cofactor for intact AdoCbl
with a rate constant (k
cat,cbl-release
) of 0.14 min
)1
.
Kinetic parameters of the reactivase for ATP in
DD (re)activation and ATP hydrolysis
Kinetic constants for ATP in the reactivation of glyc-
erol-inactivated holoenzyme and the activation of the
enzymeÆCN-Cbl complex by the reactivase were mea-
sured (Table 1). K
m
values for ATP in the reactivation
and the activation were essentially the same:
6.9 ± 0.4 mm and 6.8 ± 1.6 mm, respectively. This is
reasonable, because these two events are different
aspects of the same phenomenon [14,16]. K
m
values for
the ATPase activity were also measured in the presence
and absence of equimolar apoenzyme (Table 1). The K
m
for ATP in the ATPase activity in the absence of enzyme
was 61 ± 14 lm, i.e. two orders of magnitude smaller
than that in the DD (re)activation. Moreover, the K
m

Fig. 3. Time course of the reactivation of the glycerol-inactivated
holoenzyme by DD reactivase. The glycerol-inactivated holoenzyme
formed as described in the text was subjected to ultrafiltration on a
Microcon YM-10 microconcentrator (Millipore). To a concentrated
protein fraction containing 1.2 nmol of glycerol-inactivated holoen-
zyme, we added 2.3
M 1,2-propanediol, 38 lM AdoCbl, 19 mM ATP
and 19 m
M MgCl
2
in 0.02 M potassium phosphate buffer (pH 8.0)
without or with 0.12 nmol of reactivase to a total volume of
160 lL. After incubation at 37 °C for the indicated time periods,
20 lL aliquots were withdrawn, and the amount of DD reactivated
was measured by the 3-methyl-2-benzothiazolinone hydrazone
method [33] after appropriate dilution.
Table 1. Kinetic parameters of the reactivase for ATP.
K
m
for
ATP (m
M)
V
max
(lmol
propionaldehyde
formed in 10 min) k
cat
(min
)1

zyme (5.0 units, 0.20 nmol) in 50 lL of 0.01
M potassium phosphate
buffer (pH 8.0) containing 0.3–10 m
M each of [
32
P]ATP[cP] and
MgCl
2
. ATPase activity was measured as described in the text.
Reactivase for coenzyme B
12
-dependent enzyme K. Mori et al.
4934 FEBS Journal 277 (2010) 4931–4943 ª 2010 The Authors Journal compilation ª 2010 FEBS
divalent cations tested did not inhibit the DD activity at
3mm in the standard assay conditions (data not shown).
Mn
2+
was most effective for the activation (81% as
compared with the apoenzyme control; 185% relative to
Mg
2+
). Co
2+
and Ni
2+
were also effective (97% and
48%, respectively, relative to Mg
2+
), whereas Ca
2+

2+
(Cr
3+
) and Cu
2+
had little or no effect on the ATPase
activity as compared with the control without divalent
metal ions. Irrespective of the presence of divalent
cations, ATP was hydrolyzed to ADP + P
i
by the reac-
tivase (data not shown).
Analysis of complex formation between DD and
its reactivase by gel filtration
Apoenzyme was incubated with the reactivase in the
presence of ADP or ATP and Mg
2+
, and then subjected
to gel filtration on a Superose 6 column that had
been preliminarily equilibrated with nucleotide⁄ Mg
2+
-
containing buffer (Fig. 4). In the presence of ATP, the
enzyme and the reactivase eluted separately at their
respective retention times. In contrast, a peak of the free
reactivase decreased and a new peak appeared in the
presence of ADP. The latter peak was eluted with a
Table 2. Nucleotide and divalent cation specificities of the reactivase for the activation of the enzymeÆCN-Cbl complex. The enzymeÆCN-Cbl
complex (DDÆCN-Cbl) (61 pmol) was incubated at 37 °C for 10 min with and without 0.30 nmol of reactivase in 50 lL of 0.02
M potassium

2+
2.9 ± 0.6 22 55
DDÆCN-Cbl + 3¢-dATP Mg
2+
3.4 ± 0.1 25 64
2 apoDD )))13.7 ± 1.8 100 )
DDÆCN-Cbl + ATP ) 0.0 ± 0.0 0 )
DDÆCN-Cbl + ATP Mg
2+
6.0 ± 0.2 44 100
DDÆCN-Cbl + ATP Ca
2+
0.1 ± 0.1 1 2
DDÆCN-Cbl + ATP Cr
2+a
0.1 ± 0.1 1 2
DDÆCN-Cbl + ATP Mn
2+
11.1 ± 0.4 81 185
DDÆCN-Cbl + ATP Co
2+
5.8 ± 0.1 42 97
DDÆCN-Cbl + ATP Ni
2+
2.9 ± 0.7 21 48
DDÆCN-Cbl + ATP Cu
2+
0.1 ± 0.0 1 2
a
Cr

Mn
2+
2.23 ± 0.01 112
Co
2+
2.14 ± 0.04 107
Ni
2+
2.42 ± 0.09 122
Cu
2+
0.80 ± 0.05 40
– 0.54 ± 0.02 27
a
Cr
2+
added might be oxidized to Cr
3+
by air in the reaction
mixture.
K. Mori et al. Reactivase for coenzyme B
12
-dependent enzyme
FEBS Journal 277 (2010) 4931–4943 ª 2010 The Authors Journal compilation ª 2010 FEBS 4935
retention time that was slightly shorter than that of the
free enzyme and much shorter than that of the reactivas-
e. When this peak was subjected to SDS⁄ PAGE, the
peak comprised all of the subunits from the enzyme
(a, b, and c) and the reactivase (a and b) (data not
shown). It was thus evident that this peak contained the

(+ATP). Positions of the free apoenzyme (DD), reactivase (DD-R)
and the enzymeÆreactivase complex(es) (DDÆDD-R) are indicated on
the tops of chromatograms.
A
B
C
66 K–
66 K–
45 K–
36 K–
29 K–
24 K–
20 K–
14 K–
Fig. 5. Analysis of the enzymeÆreactivase complex by PAGE. Exper-
imental details are described in the text. (A) Nondenaturing PAGE
in the presence of ADP and Mg
2
+
. Lanes: (a) apoenzyme; (b) reacti-
vase; (c) enzymeÆCN-Cbl complex + reactivase (+ADP); (d) enzy-
meÆCN-Cbl complex (+ADP); (e) apoenzyme + reactivase (+ADP);
(f) apoenzyme (+ADP); (g) reactivase (+ADP). +ADP indicates that
samples were incubated with ADP ⁄ MgCl
2
. (B, C) Bands i–vi in (A)
were excised and subjected to SDS ⁄ PAGE on 6% (B) and 14% (C)
gels. Lanes i–vi correspond to bands i–vi in (A). Lane D: purified
enzyme. Lane R: purified reactivase. Positions of DD and its subun-
its (a

spective of the presence of enzyme (Fig. 5A, lanes b, c,
e, and g). To determine their subunit compositions,
bands i–vi were excised and subjected to SDS⁄ PAGE
on 6% and 14% gels (Fig. 5B,C), followed by densito-
metric analyses. Both band i and band vi of the
enzymeÆreactivase complexes comprised all of the
subunits from the enzyme (a, b, and c) and the reacti-
vase (a and b) (Fig. 5B,C, lanes i and vi). Subunit
compositions of the bands were the same when either
the enzymeÆCN-Cbl complex or apoenzyme was used
(data not shown). If the a, b and c subunits of the
enzyme are abbreviated as a
D
, b
D
, and c
D
, respectively,
and the a and b subunits of the reactivase are abbrevi-
ated as a
R
and b
R
, respectively, molar ratios of a
D
, b
D
,
c
D

Æ(a
R
Æa
R
b
R
)
2
complexes, respectively. We
named the former the enzymeÆreactivase (1 : 1) complex
and the latter the enzymeÆreactivase (1 : 2) complex.
When bands ii and iii of Fig. 5A, i.e. two thick bands of
the reactivase, were subjected to SDS ⁄ PAGE, they con-
tained both a
R
and b
R
, although the ratios of subunits
were different (Fig. 5C, lanes ii and iii). Densitometric
analysis indicated that molar ratios of a
R
to b
R
for
bands ii and iii were approximately 2 : 1 and 1 : 1,
respectively. These results indicated that the upper and
lower bands of the reactivase correspond to the a
R
Æa
R

2+
, the
enzymeÆreactivase complex was formed in small
amounts from apoenzyme and the reactivase and not at
all from the enzymeÆCN-Cbl complex and the reactivase
(data not shown).
Affinity of the reactivase for DD
Figure 6 shows the dependence of enzymeÆreactivase
complex formation on reactivase concentration. The
A
B
Fig. 6. Dependence of complex formation
on reactivase concentration at a fixed
enzyme concentration. (A) Apoenzyme +
reactivase (left, none; right, +ADP).
(B) EnzymeÆCN-Cbl complex + reactivase
(left, none; right, +ADP). Experimental condi-
tions were similar to those for nondenaturing
PAGE in Fig. 5, except that the reactivase
and enzyme concentrations were varied and
fixed (1 l
M), respectively. The number on the
top of each lane indicates the reactivase
concentration (l
M). Lane R: reactivase 2 lM.
Positions of the enzyme, reactivase and the
enzymeÆreactivase complexes are indicated
on the right of the gels: (i) enzymeÆreactivase
(1 : 2) complex; (ii) enzymeÆreactivase (1 : 1)
complex; (iii) enzyme; (iv) reactivase

vase in the presence of ADP and Mg
2+
, whereas it was
observed at ‡ 2 lm reactivase in the absence of ADP
and Mg
2+
. Moreover, although the enzymeÆreactivase
(1 : 2) species was the only enzymeÆreactivase complex
observed at ‡ 4 lm reactivase in the presence of ADP
and Mg
2+
, some enzymeÆreactivase (1 : 1) complex
remained even at the highest concentration of reactivase
tested (20 lm) in the absence of ADP and Mg
2+
. The
apparent K
D
values of the reactivase for formation of
the enzymeÆreactivase complex were 0.4 lm and 3 lm in
the presence and absence of ADP and Mg
2+
, respec-
tively. When similar experiments were carried out with
the enzymeÆCN-Cbl complex in place of apoenzyme,
essentially no complex formation was observed, even at
20 lm reactivase, in the absence of ADP and Mg
2+
.In
contrast, in the presence of ADP and Mg

concentration (20–180 mm), as the reactivation was
monitored by product formation from reactivated
holoenzyme at high concentrations of the enzyme and
the reactivase. It would be easier to demonstrate the
multiple turnovers of DD reactivase and glycerol
dehydratase reactivase in the in situ reactivation,
because the reactivation takes place in toluene-treated
cells, where local concentrations of the enzyme and the
reactivase are high enough for reactivation, and an
inhibitory product, b-hydropropionaldehyde, diffuses
away.
From the initial rate of exchange of enzyme-bound
CN-Cbl for AdePeCbl, the rate constant of the reacti-
vase in cobalamin release (k
cat,cbl-release
) for CN-Cbl
was calculated to be 0.27 min
)1
at 37 °C. From the
initial rate of reactivation of the glycerol-inactivated
holoenzyme, the rate constant of the reactivase in
the reactivation (k
cat,react
) was calculated to be
0.071 ± 0.008 min
)1
at 37 °C. Considering that the
enzyme contains two cobalamin-binding sites in the
(abc)
2

) [17],
but about five-fold and 10-fold larger than the rate
constants for the release of CN-Cbl (0.27 min
)1
) and
the damaged cofactor (0.14 min
)1
), respectively.
Therefore, ATP hydrolysis and cobalamin release or
reactivation might be not very tightly coupled. The
reactivation of the inactivated holoenzymes by the
reactivase seems to be physiologically relevant, because
k
cat,cbl-release
is much larger than the rate constant for
bacterial growth on glycerol.
The reactivase exhibited broad specificities for nucle-
otides and divalent metal cations, both of which are
absolutely required for the in vitro activation of the
enzymeÆCN-Cbl complex. We have previously reported
similar specificities in the in situ reactivation of the
glycerol-inactivated hologlycerol dehydratase with
K. pneumoniae cells [13]. It was established that the
reactivase-mediated reactivation of the inactivated
holoenzymes with ATP and Mg
2+
takes place in two
steps: (a) ADP-dependent cobalamin release with
Reactivase for coenzyme B
12

Alaa461 in glycerol dehydratase reactivase. Further-
more, 2¢-dATP retained half of the efficacy of ATP in
the activation of the enzymeÆCN-Cbl complex. It was
therefore concluded that these hydrogen bonds are not
essential for (re)activation. Similarly, 3¢-dATP retained
half of the efficacy of ATP in the activation. In the crys-
tal structure of the reactivase, no amino acids were
found to be hydrogen bonded to O3¢ of ADP. Thus, no
requirement for the 3¢-OH group seems to be reasonable
from its crystal structure.
The reactivase has two distinct divalent metal ion-
binding sites in the ab heterodimeric unit [25]. One of
them is present in the interface between the a and
b subunits. This metal ion is coordinated by four
amino acids (Aspa166, Aspa183, Thra105, and
Glub31), all of which are completely conserved in both
reactivases for diol and glycerol dehydratases. These
coordinations are maintained in the reactivase, irre-
spective of the ADP binding. The crystal structure of
the DD reactivase suggested that this metal ion is
Mg
2+
in the ADP-bound form, whereas it is Ca
2+
in
the nucleotide-free form. It might be possible that
Mg
2+
occupies this site in vivo and is replaced by
Ca

,Co
2+
and Ni
2+
enhanced the ATPase
activity of the reactivase, although the reactivase can
hydrolyze ATP to ADP even without divalent metal
ions. These metal ions were effective in the activation
of the enzymeÆCN-Cbl complex by the reactivase in the
presence of ATP, although relative efficiencies were
not always correlated. On the other hand, the reacti-
vase was unable to activate the enzymeÆCN-Cbl com-
plex even in the presence of ATP with Ca
2+
,Cr
2+
or
Cu
2+
or without divalent metal ions. These metal ions
had little or no enhancing effect on the ATPase activ-
ity of the reactivase. Thus, the reactivase-mediated
activation of the enzymeÆCN-Cbl complex absolutely
requires the hydrolysis of ATP in the presence of diva-
lent metal ions. The reactivase does not form the
enzymeÆreactivase complexes from the enzymeÆCN-Cbl
complex in the presence of ADP without divalent
cations. These results suggest that the binding of ADP
alone to the ATPase domain of the reactivase a sub-
unit is not sufficient to cause a conformational change

b
R
)
2
into
K. Mori et al. Reactivase for coenzyme B
12
-dependent enzyme
FEBS Journal 277 (2010) 4931–4943 ª 2010 The Authors Journal compilation ª 2010 FEBS 4939
a
R
Æa
R
b
R
and b
R
in the presence of ADP and Mg
2+
was observed even without the enzyme. The crystal
structure of the reactivase also suggested that the
interactions between the reactivase a and b subunits
are weakened at least partially by the ADP binding
[25]. The space that is opened by the dissociation of
the reactivase b subunit would most likely be occupied
by the enzyme b subunit, as these subunits have simi-
lar folds [25,28]. The docking model of the a
D
b
D

lacking cobalamins, and thus allows the damaged
cofactor to pass through it. Intact cofactor, an ade-
nine-containing cobalamin, is not released from the
enzyme by the reactivase. One reason might be its
larger size, and the other possible reason might be that
the additional interaction between its adenine moiety
and the enzymes’s adenine-binding pocket stabilizes
the interaction between the enzyme a and b subunits.
In contrast, even in the absence of ADP, the reactivase
forms the enzymeÆreactivase complex with apoenzyme.
However, it does not release the damaged cofactor
from the inactivated holoenzymes under this condi-
tion. This may be because the steric repulsion is less
or is canceled by the conformational flexibility in the
absence of ADP ⁄ Mg
2+
. In order to prove or disprove
these predictions, we have to await the structural
analysis of a real enzymeÆreactivase complex.
Experimental procedures
Materials
Crystalline AdoCbl was a gift from Eisai (Tokyo, Japan).
CN-Cbl was obtained from Glaxo Research Laboratories
(Greenford, UK). AdePeCbl was synthesized according to
published procedures [30]. [
32
P]ATP[cP] was obtained from
PerkinElmer (Waltham, MA, USA). 2¢-DeoxyATP and
3¢-deoxyATP were obtained from Sigma-Aldrich (St Louis,
MO, USA). All other chemicals were commercial products

(A) and the existence of a cavity between DD a (pink) and b (green)
subunits (B). a, a
D
(pink) or a
R
(light blue) subunit; b, b
D
(green) or
b
R
(orange) subunit; c, c
D
subunit (dark blue).
Reactivase for coenzyme B
12
-dependent enzyme K. Mori et al.
4940 FEBS Journal 277 (2010) 4931–4943 ª 2010 The Authors Journal compilation ª 2010 FEBS
Glycerol-inactivated holoenzyme and
enzymeÆCN-Cbl complex
The glycerol-inactivated holoenzyme was formed by incuba-
tion of apoenzyme (70 units, 2.8 nmol) with 38 lm AdoCbl
at 37 °C for 30 min in 1.7 mL of 0.05 m potassium phos-
phate buffer (pH 8.0) containing 19% glycerol. Glycerol
and excess AdoCbl were removed by dialysis at 4 °C for
49 h against 600 volumes of 0.05 m potassium phosphate
buffer (pH 8.0) containing 2% 1,2-propanediol with a
buffer change. The enzyme Æ CN-Cbl complex was prepared
by incubation of apoenzyme (30–45 units, 1.2–1.8 nmol)
with 13 lm CN-Cbl at 37 °C for 30 min in 540 lLof
0.05 m potassium phosphate buffer (pH 8.0) containing 2%

natant was determined by liquid scintillation counting, and
ATPase activity was obtained by subtracting the radioactiv-
ity of a minus reactivase control. In some experiments,
MgCl
2
was replaced with other divalent metal chlorides.
Reactivase-mediated exchange of enzyme-bound
CN-Cbl for AdePeCbl
The enzymeÆCN-Cbl complex was formed by incubation of
30 units of apoenzyme (1.2 nmol) with 80 lm CN-Cbl at
37 °C for 60 min in 75 lL of 0.04 m potassium phosphate
buffer (pH 8.0) containing 2% 1,2-propanediol and 0.5%
Brij35. To the resulting mixture were added 19 lg of reacti-
vase (0.12 nmol), together with 40 lm AdePeCbl, 20 mm
ATP and 20 mm MgCl
2
in 0.03 m potassium phosphate
buffer (pH 8.0) containing 1% 1,2-propanediol and 0.1%
Brij35, in a total volume of 150 lL. After incubation at
37 °C for appropriate periods, the exchange reaction was
terminated by addition of 150 lL of 0.01 m potassium
phosphate buffer (pH 8.0) containing 2% 1,2-propanediol,
0.2% Brij35, and 60 mm EDTA. The resulting mixture was
then subjected to ultrafiltration on a Microcon YM-10
microconcentrator (Millipore, Billerica, MA, USA) to
remove unbound cobalamins. The protein fraction retained
on the filter was washed twice by the addition of 150 lLof
0.01 m potassium phosphate buffer (pH 8.0) containing 2%
1,2-propanediol, 0.2% Brij35, and 10 mm EDTA, and this
was followed by ultrafiltration. The spectrum was measured

column, using an FPLC system (GE Healthcare, Little
Chalfont, UK). In the presence and absence of 21 mm
adenine nucleotide (ATP or ADP) and 21 mm MgCl
2
, apo-
enzyme (12 units, 0.48 nmol) were incubated at 37 °C for
60 min with 0.27 mg (1.7 nmol) of reactivase in 170 lLof
0.04 m potassium phosphate buffer (pH 8.0) containing
1.4% 1,2-propanediol and 0.7% Brij35. The resulting mix-
ture was applied to a column that had been equilibrated
with 0.05 m potassium phosphate buffer (pH 8.0) contain-
ing 2% 1,2-propanediol and 0.5% Brij35 with or without
the corresponding adenine nucleotide (ADP or ATP) and
MgCl
2
(1 mm each). The column was developed with the
same buffer at a flow rate of 0.4 mLÆmin
)1
. The enzyme
and the reactivase alone were also applied under the same
conditions as controls. The elution of proteins was moni-
tored by the absorbance at 280 nm.
Analysis of the enzymeÆreactivase complexes by
PAGE
The reactivase (12 lg, 76 pmol) was incubated with 10 mm
dithiothreitol at 30 °C for 30 min in 5 lL of 0.01 m potas-
sium phosphate buffer (pH 8.0). ADP and MgCl
2
(10 mm
each) were added to the resulting mixture in a total volume

with the SDS-containing sample buffer. Then excised gels
were subjected to SDS ⁄ PAGE on 14% and 6% gels under
the conditions described by Laemmli [37]. Protein was
stained again with Coomassie Brilliant Blue R-250. Densi-
tometric analysis of gels was performed with a Print-
graph AE-6911CX system (ATTO, Tokyo, Japan) and
nih-image, version 1.6.3 (National Institutes of Health).
Model figure
Figure 7B was generated with chimera [38], using the atomic
coordinates for the enzymeÆCN-Cbl complex (Protein Data
Bank accession code: 1EGM).
Acknowledgements
This work was supported in part by Grants-in-Aid for
Scientific Research [(B) 13480195 and 17370038 and
Priority Areas 753 and 513 to T. Toraya], a Grant-in-
Aid for Young Scientists [(B) 18770111 to K. Mori]
from the Japan Society for Promotion of Science and
the Ministry of Education, Culture, Sports, Science and
Technology, Japan, and a Grant in Aid for Natural
Sciences Research (to T. Toraya) from the Asahi Glass
Foundation, Tokyo, Japan. We thank Y. Kurimoto for
her assistance with manuscript preparation.
References
1 Banerjee R, ed (1999) Chemistry and Biochemistry of
B
12
. Wiley, New York, NY.
2Re
´
tey J (1990) Enzymic reaction selectivity by negative

of apoenzyme synthesis in Klebsiella pneumoniae
(Aerobacter aerogenes ) ATCC 8724. J Bacteriol 135,
726–729.
9 Toraya T, Kuno S & Fukui S (1980) Distribution of
coenzyme B
12
-dependent diol dehydratase and glycerol
dehydratase in selected genera of Enterobacteriaceae
and Propionibacteriaceae. J Bacteriol 141, 1439–1442.
10 Bachovchin WW, Eagar RG Jr, Moore KW & Richards
JH (1977) Mechanism of action of adenosylcobalamin:
glycerol and other substrate analogues as substrates and
inactivators for propanediol dehydratase – kinetics,
stereospecificity, and mechanism. Biochemistry 16,
1082–1092.
11 Schneider Z & Pawelkiewicz J (1966) The properties of
glycerol dehydratase isolated from Aerobacter aeroge-
nes, and the properties of the apoenzyme subunits. Acta
Biochim Pol 13, 311–328.
12 Poznanskaya AA, Yakusheva MI & Yakovlev VA
(1977) Study of the mechanism of action of adenosylco-
balamin-dependent glycerol dehydratase from Aerobac-
ter aerogenes. II. The inactivation kinetics of glycerol
dehydratase complexes with adenosylcobalamin and its
analogs. Biochim Biophys Acta 484, 236–243.
13 Honda S, Toraya T & Fukui S (1980) In situ reactiva-
tion of glycerol-inactivated coenzyme B
12
-dependent
enzymes, glycerol dehydratase and diol dehydratase.

adenosylcobalamin-dependent glycerol dehydratase.
J Bacteriol 181, 4110–4113.
19 Kajiura H, Mori K, Tobimatsu T & Toraya T (2001)
Characterization and mechanism of action of a
reactivating factor for adenosylcobalamin-dependent
glycerol dehydratase. J Biol Chem 276, 36514–36519.
20 Seifert C, Bowien S, Gottschalk G & Daniel R (2001)
Identification and expression of the genes and
purification and characterization of the gene products
involved in reactivation of coenzyme B
12
-dependent
glycerol dehydratase of Citrobacter freundii. Eur J
Biochem 268, 2369–2378.
21 Pierik AJ, Graf T, Pemberton L, Golding BT & Re
´
tey
J (2008) But-3-ene-1,2-diol: a mechanism-based active
site inhibitor for coenzyme B
12
-dependent glycerol
dehydratase. Chembiochem 9, 2268–2275.
22 Toraya T, Tamura N, Watanabe T, Yamanishi M,
Hieda N & Mori K (2008) Mechanism-based
inactivation of coenzyme B
12
-dependent diol
dehydratase by 3-unsaturated 1,2-diols and thioglycerol.
J Biochem 144, 437–446.
23 Kajiura H, Mori K, Shibata N & Toraya T (2007)

ethanolamine ammonia-lyase complexed with
coenzyme B
12
analogs and substrates. J Biol Chem 285,
26484–26493.
30 Sando GN, Grant ME & Hogenkamp HPC (1976) The
interaction of adeninylalkylcobalamins with ribonucleo-
tide reductase. Biochim Biophys Acta 428, 228–232.
31 Tobimatsu T, Sakai T, Hashida Y, Mizoguchi N,
Miyoshi S & Toraya T (1997) Heterologous expression,
purification, and properties of diol dehydratase, an
adenosylcobalamin-dependent enzyme of Klebsiella
oxytoca. Arch Biochem Biophys 347, 132–140.
32 Mori K, Hieda N, Yamanishi M, Shibata N &
Toraya T (2005) Crystallization and preliminary X-ray
analysis of molecular chaperone-like diol dehydratase-
reactivating factor in ADP-bound and nucleotide-free
forms. Acta Crystallogr F 61, 603–605.
33 Toraya T, Ushio K, Fukui S & Hogenkamp HPC
(1977) Studies on the mechanism of the adenosylcobala-
min-dependent diol dehydrase reaction by the use of
analogs of the coenzyme. J Biol Chem 252, 963–970.
34 Gill SC & von Hippel PH (1989) Calculation of protein
extinction coefficients from amino acid sequence data.
Anal Biochem 182 , 319–326.
35 Barker HA, Smyth RD, Weissbach H, Munch-Petersen
A, Toohey JI, Ladd JN, Volcani BE & Wilson RM
(1960) Assay, purification, and properties of the
adenylcobamide coenzyme. J Biol Chem 235, 181–190.
36 Davis BJ (1964) Disc electrophoresis. II. Method and