Báo cáo khoa học: UDP-galactose 4-epimerase from Kluyveromyces fragilis – catalytic sites of the homodimeric enzyme are functional and regulated docx

UDP-galactose 4-epimerase from Kluyveromyces
fragilis – catalytic sites of the homodimeric enzyme
are functional and regulated
Amrita Brahma*, Nupur Banerjee* and Debasish Bhattacharyya
Structural Biology and Bioinformatics Division, Indian Institute of Chemical Biology (CSIR), Jadavpur, Kolkata, India
Introduction
UDP-galactose 4-epimerase, hereafter called epimerase,
is an essential and ubiquitous enzyme that reversibly
converts UDP-Gal to UDP-Glc. The epimerase from
the yeast Kluyveromyces fragilis is a homodimer of
nearly 75 kDa per subunit, and contains bound
NAD
+
acting as cofactor [1–3]. Epimerases from Esc-
herichia coli [4–6], Saccharomyces cerevisiae [7] and
human sources [8] have been cloned and sequenced,
and their X-ray crystallographic structures are known.
The bacterial enzyme has two NAD
+
-binding sites
Keywords
catalytic sites; inhibitor; multimeric enzyme;
regulation; UDP-galactose 4-epimerase
Correspondence
D. Bhattacharyya, Structural Biology and
Bioinformatics Division, Indian Institute of
Chemical Biology (CSIR), 4, Raja S. C.
Mallick Road, Jadavpur, Kolkata 700 032,
India
Fax: +91 33 2473 5197 ⁄ 0284
Tel: +91 33 2499 5764

E-mail:
*These authors contributed equally to this
work
(Received 20 July 2009, revised 20 August
2009, accepted 16 September 2009)
doi:10.1111/j.1742-4658.2009.07386.x
UDP-galactose 4-epimerase from Kluyveromyces fragilis is a homodimer
containing one catalytic site and one NAD
+
as cofactor per subunit. One
5¢-UMP, a competitive inhibitor, binds per dimer of epimerase as isolated
and causes inactivation. Addition of 0.2 mm inhibitor to the enzyme in vitro
leads to three sequential steps: first, the inhibitor binds to the unoccupied
site; second, the inhibitor bound ex vivo is displaced allosterically; and
finally, both sites are occupied by the inhibitor. These reactions have been
monitored by kinetic lag in substrate conversion, coenzyme fluorescence,
protection against trypsin digestion, and reductive inhibition. The transi-
tion profiles indicate the existence of a stable intermediate with one inhibi-
tor-binding site remaining unoccupied. Reductive inhibition of this
intermediate reduced the activity to 58% ± 2%, with modification of one
catalytic site. A change of conformation of the epimerase upon binding
with substrate or inhibitor was evident from fluorescence emission spectra.
The epimerase demonstrated a biphasic Michaelis–Menten dependency.
The epimerase devoid of 5¢-UMP showed a Michaelis–Menten dependency
that can be explained by assuming simultaneous operation of two catalytic
sites. A monomeric form of the epimerase was devoid of such regulation.
The inhibitory profile of 5¢-UMP also suggested negative cooperativity.
Incubation of the epimerase with combinations of substrate analogs ren-
dered one of the sites inactive, supporting the presence of two functional
and regulated catalytic sites. Dissimilar kinetic patterns of the reconstituted

enzyme after treatment with p-chloromercuribenzoate indicated stability of
the dimeric enzyme against fast association–dissociation, which could
otherwise generate multiple forms of the enzyme with functional
heterogeneity.
Abbreviations
CHD, 1,2-cyclohexanedione; GG, glycylglycine; pCMB, p-chloromercuribenzoate; STI, soybean trypsin inhibitor.
FEBS Journal 276 (2009) 6725–6740 ª 2009 Council of Scientific and Industrial Research, New Delhi. Journal compilation ª 2009 FEBS 6725
away from the subunit contact region, and its mono-
mers are functional [9].
The molecular mass of the yeast enzyme is almost
double that of the bacterial and human epimerases. A
blast search of the yeast epimerase revealed two fea-
tures: its N-terminal half showed strong homology
with the E. coli epimerase, and the C-terminal half
showed homology with mutarotase [10]. The predicted
mutarotase activity in K. fragilis epimerase was later
demonstrated [11,12]. Furthermore, the enzyme can be
cleaved by trypsin into two parts in the presence of
epimerase and mutarotase inhibitors. They can func-
tion independently as an epimerase and a mutarotase
[11]. Interestingly, when trypsin digestion is performed
in the presence of only the epimerase inhibitor, the
mutarotase domain is fragmented, yielding a 45 kDa
monomeric epimerase [11,13].
The yeast enzyme exhibits a stoichiometry of two
NAD
+
molecules per dimer, similarly to E. coli and
human epimerase, raising the possibility of the exis-
tence of two catalytic sites. Binding of one molecule of

nondialyzable 5¢-UMP, a competitive inhibitor, renders
the enzyme inactive. This led to the assumption that
there is regulation between the catalytic sites. Func-
tionality could be restored after replacement of the
inhibitor by a substrate [14]. Therefore, either the
unbound site is nonfunctional or the bound inhibitor
regulates its functionality. The second possibility is
favored, as there is evidence that the activities of many
metabolic enzymes are controlled in vivo. In the pres-
ence of excess 5¢-UMP in vitro, the epimerase is com-
pletely inactivated, signifying that the inhibitor blocks
its two catalytic sites [15].
The background literature on epimerase shows that
the nature of binding of 5¢-UMP in vivo and in vitro is
different. First, the inhibitor bound as isolated, but
not the one bound in vitro, shows a kinetic lag in
catalysis [14]. Second, for removal of a nondialyzable
inhibitor by its own counterpart added extraneously,
there must be another binding site of the inhibitor in
the enzyme. This site is presumably the second cata-
lytic site. This observation supports the idea that the
stoichiometry of bound 5¢-UMP ex vivo is indeed £ 1
per dimer. As the externally added inhibitor interacts
specifically at the unoccupied binding site, a long-range
interaction between the occupied and unoccupied sites
is predicted (Scheme 1). Third, there is an arginine at
the catalytic site of epimerase that can be modified by
Scheme 1. Proposed model of conversion of E
1
to E

4
.E
1
, epimerase containing one 5¢ -UMP per dimer bound as isolated (native epimer-
ase); [E
2
], an intermediate of the conversion where the unoccupied 5¢-UMP-binding site of E
1
is occupied by the added 5¢-UMP (the bracket
indicates its transient character); E
3
, stable intermediate where the 5¢-UMP bound ex vivo to E
1
is replaced allosterically by the added
5¢-UMP; E
4
, epimerase where both the 5¢-UMP binding sites are occupied by added 5¢-UMP; [E
2A
], product of reductive inhibition of [E
2
] with
L(+)-arabinose (the bracket indicates uncertainty about its existence); E
3A
, product of reductive inhibition of [E
3
] with L(+)-arabinose. The two
lobes in all the structures indicate homodimeric epimerase; the flange at the middle of each lobe separates the epimerase (upper) and
mutarotase (lower) domains of a monomer; the rectangular denting of the upper domains of each lobe indicates the binding site of 5¢-UMP
ex vivo; the shaded rectangle indicates 5¢-UMP bound as isolated; the open rectangle indicates added 5¢-UMP;
•

, NAD
+
, , NADH, s,
arginine at the active site; s over the 5¢-UMP binding site indicates protection against trypsinization; circumference of the lobe next to the
site indicates susceptibility to the protease; +UMP and –UMP indicate its association and dissociation; the arrow on the top of the scheme
indicates the direction of 5¢-UMP-dependent conversion of epimerase. The symmetrical pattern of the dimeric enzyme as shown is a
working model only.
Regulation of catalytic sites of yeast epimerase A. Brahma et al.
6726 FEBS Journal 276 (2009) 6725–6740 ª 2009 Council of Scientific and Industrial Research, New Delhi. Journal compilation ª 2009 FEBS
1,2-cyclohexanedione (CHD), leading to inactivation
[15]. Whereas 5¢-UMP added in vitro could completely
protect the arginine from modification, the inhibitor
bound ex vivo is incapable of doing so [16]. Fourth,
degradation of epimerase by trypsin is initiated from
the arginine of the catalytic site, leading to further
unraveling of the molecule [11]. Trypsin digestion of
epimerase could be completely prevented if the said
arginine were protected by 5¢-UMP in vitro. However,
the inhibitor bound ex vivo is incapable of preventing
trypsin digestion [15]. Fifth, 5¢-UMP bound ex vivo
does not participate in reductive inhibition (‘reductive
inhibition’ is a specific reaction whereby epimerase
bound NAD
+
, acting as cofactor, is reduced to
NADH in the presence of 5¢-UMP, a competitive
inhibitor, and l(+)-arabinose, a reducing sugar, lead-
ing to complete inactivation of the enzyme [24] –
NAD
+

as free nucleotide or bound to other enzymes
is not sensitive to this reaction), whereas the inhibitor
bound in vitro does [14].
Here, we provide a hypothesis for the pathway fol-
lowed by epimerase as isolated (E
1
) during its satura-
tion with extraneously added 0.5 mm 5¢-UMP (E
4
)
(Scheme 1). An essential feature in this proposal is that
the nature of the binding of 5¢-UMP in E
1
and that in
E
4
are different, evidence for which has been men-
tioned above. On the basis of the situation in
Scheme 1, the minimum requirement for the conver-
sion is the existence of two intermediates, E
2
and E
3
.
In E
1
, one inhibitor-binding site is occupied and the
other is vacant. In E
2
, the added 5¢-UMP binds to the

unoccupied site of E
1
.InE
3
, the added inhibitor has
removed the 5¢-UMP bound ex vivo by allostericity. A
higher concentration of the inhibitor leads to the final
product E
4
Thus, the model predicts that, in E
3
, the
inhibitor occupies the high-affinity site, leaving the
low-affinity site vacant. In summary, the conversion
involves three steps: association of added inhibitor at
low concentration; dissociation of the inhibitor bound
ex vivo; and, finally, association of added inhibitor at
the unoccupied site (Scheme 1).
Theoretical considerations predict that E
2
should
exist only as a transient intermediate. If E
2
is stable, it
will prevent the spontaneous formation of E
3
.AsE
2
has no additional 5¢-UMP-binding site, its conversion
to E

3
should be independent of added 5¢-UMP. A cor-
ollary of the prediction is that half of the catalytic sites
of E
1
and E
3
remain bound to 5¢-UMP, whereas both
of the catalytic sites in E
2
and E
4
are occupied by the
inhibitor. Thus, after incubation with trypsin, the
residual activities of E
1
,E
2
(if it exists), E
3
and E
4
are
expected to be 0%, 50%, 50% and 100%, respectively,
on the basis of selective protection by the added inhib-
itor. Similarly, the formation of NADH and residual
activity after reductive inhibition in the presence of
only l(+)-arabinose for these enzyme–inhibitor com-
plexes are expected to be 0, 1, 1 and 2 mol per dimer
and 100%, 50%, 50% and 0%, respectively. To be

more precise, as the catalytic sites are supposed to be
regulated, inactivation of E
2
and E
3
after trypsiniza-
tion or reductive inhibition leading to E
2A
and E
3A
(Scheme 1) may not be exactly 50%.
While supporting Scheme 1, two physical properties
of epimerase have been established. These are the stoi-
chiometry of bound cofactor NAD
+
being two per
dimer, and the stability of its dimeric structure with
regard to fast dissociation–association. In the absence
of this information, Scheme 1 is not valid. Subse-
quently, we verified whether the Michaelis–Menten
relationship is or is not maintained by the enzyme.
When product inhibition and secondary reactions
are not applicable for an enzyme, its deviation from
the Michaelis–Menten relationship is a strong indication
of allosteric regulation. Specific inactivation of one
catalytic site of epimerase with substrate analogs has
also been investigated to provide supportive evidence.
Results
Stoichiometry of bound NAD
+

Conversion of NAD
+
to NADH by reductive inhibi-
tion of epimerase offered a sensitive method for deter-
mination of stoichiometry of the bound cofactor.
Reductive inhibition was applied to 33.3 nmol (5 mg)
of epimerase. A control set of the enzyme under identi-
cal conditions but in the absence of the reducing
agents [5¢-UMP and l(+)-arabinose] retained 98% of
the activity. Complete inactivation of the enzyme
ensured quantitative reduction of NAD
+
to NADH.
Dissociation of NADH from the enzyme was achieved
with 8 m urea [17], and was quantified spectrofluori-
metrically. Recovery of NADH was 63.0 ± 4.0 nmol,
yielding a stoichiometry of 1.89 ± 0.12 per dimer, or
close to 1.0 NADH per subunit (n = 4). Further
improvement in quantification was restricted by the
uncertainty of protein concentration determination and
the incompleteness of conversion of NAD
+
to NADH.
Therefore, with respect to the composition of cofactor,
the catalytic sites of epimerase remain indistinguish-
able.
Stability of subunits
According to Scheme 1, the subunits of epimerase (E
1
)

are asymmetric, of the type a:b, on the basis of bind-
A. Brahma et al. Regulation of catalytic sites of yeast epimerase
FEBS Journal 276 (2009) 6725–6740 ª 2009 Council of Scientific and Industrial Research, New Delhi. Journal compilation ª 2009 FEBS 6727
ing of 5¢-UMP. Epimerase exists as a stable dimer [18],
but this does not exclude fast association–dissociation
of the dimer, leading to three types of functional entity
(Eqn 1). Depending on the magnitude of the rate con-
stants involved, there might be undetectable amounts
of the monomers at equilibrium. Thus, noncompliance
with the Michaelis–Menten relationship of epimerase,
as described later, might originate from heterogeneity
of the enzyme without regulatory behavior.
2a : b Ð 2a þ 2b Ð a : a þ b : b ð1Þ
To check this, epimerase was treated with p-chloro-
mercuribenzoate (pCMB), leading to inactivation and
dissociation of the subunits without denaturation.
Functionality was restored with reconstitution of the
dimeric structure after reduction of the modified
enzyme with dithiothreitol and NAD
+
[19]. The dura-
tions of kinetic lag of catalysis for 0.004 units of native
and reconstituted epimerase were found to be 83 s and
17 s, respectively, whereas the rates of substrate con-
version in the steady state were very similar, being
2.52 lmol and 2.63 lmol of UDP-Gal per min, respec-
tively (Fig. 1A). Furthermore, the Michaelis–Menten
patterns of the native and reconstituted epimerase were
constructed and were found to be entirely different
(Fig. 1B). The kinetic features of the native enzyme

remained unchanged when the enzyme was incubated
in the assay mixture without the substrate for 30 min
at 25 °C prior to activity measurements. These obser-
vations collectively indicate that epimerase does not
undergo rapid exchange of subunits during catalysis.
Characterization of epimerase–inhibitor
complexes
The equilibrium intermediates formed during conver-
sion of the inhibitor complex E
1
to E
4
in the presence
of 0–0.6 mm 5¢-UMP were characterized by a kinetic
lag in catalysis, ‘coenzyme fluorescence’ (described
later), and inactivation by trypsin (Fig. 2A), and two
other parameters of reductive inhibition, namely inacti-
vation and formation of NADH (Fig. 2B). The last
two parameters are sensitive enough to be measured
with an accuracy of ± 0.25%.
The dependency of the lag in catalysis followed a
sharp decrease of 100% to 10% ± 2% in the presence
of 0–0.2 mm 5¢-UMP. This indicated removal of the
inhibitor bound as isolated in E
1
or modification of
the inhibitor-binding site. A corresponding conforma-
tional change at the cofactor-binding site was moni-
tored from coenzyme fluorescence. There was an initial
25% rise in emission intensity in the presence of

0–0.2 mm 5¢-UMP. The intensity gradually returned to
its original value in the presence of 0.5 mm inhibitor.
The maximum emission was seen in the presence of
0.15 mm 5¢-UMP, which corresponded closely to the
transition midpoint of abolition of the kinetic lag. In
the case of inactivation by trypsin, the profile distinctly
indicated the existence of a stable intermediate between
0.1 mm and 0.2 mm of the inhibitor. This intermediate
retained 58% ± 2% of the residual activity, indicating
that the added inhibitor could protect half of the
catalytic sites, as expected for E
3
. All transitions were
A
B
Fig. 1. Differences in catalytic properties of native and reconsti-
tuted epimerase after dissociation with pCMB. (A) Pre-steady-state
kinetics of 0.004 units of native (1) and reconstituted (2) epimerase.
The horizontal dotted lines indicate initial absorbance of native (1)
and reconstituted (2) epimerase. Durations of initial lag of catalysis
are indicated by vertical dotted lines. The parallel nature of curves 1
and 2 indicate that, at steady state, the catalytic efficiencies of the
two forms of epimerase are equal. (B) Michaelis–Menten plots of
native (s—s) and reconstituted (
•
—
•
) epimerase. The bar indi-
cates variation of results (n = 3). Results for the native epimerase
are elaborated in Fig. 5. Inset: Lineweaver–Burk plot of the recon-

stituted epimerase. Units of the ordinate and abscissa are lmolÆ-
min
)1
and mM
)1
, respectively; values correspond to the original
plot.
Regulation of catalytic sites of yeast epimerase A. Brahma et al.
6728 FEBS Journal 276 (2009) 6725–6740 ª 2009 Council of Scientific and Industrial Research, New Delhi. Journal compilation ª 2009 FEBS
completed with 0.5 mm 5¢-UMP, as higher concentra-
tions of the inhibitor failed to cause additional change
(Fig. 2A).
Proteolysis of native epimerase by trypsin is initiated
from an arginine residue located at the catalytic site
[11]. The degradation can be prevented either by modi-
fication of this residue by CHD or by incubation with
5mm 5¢-UMP [16]. 5¢-UMP can protect E
4
against
trypsin, but not E
1
. Furthermore, as the arginine at
the catalytic site of E
1
does not appear to be protected
by 5¢-UMP, the amino acid could be modified by
CHD, which in turn is expected to resist trypsin diges-
tion. SDS ⁄ PAGE of CHD-modified E
1
showed genera-

tion of a stable 45 kDa fragment after trypsin
digestion, which was presumably the N-terminal
domain of the epimerase (Fig. 2A, inset). Thus, suscep-
tibilities of different epimerase–UMP complexes to
trypsin may be used to follow interactions of the inhib-
itor at the catalytic site.
The protective effect of 5¢-UMP against modification
of the arginine located at the catalytic site by CHD
was similar to that observed for trypsin digestion.
Native epimerase was modified by CHD in the pres-
ence of 0, 0.1–0.2 and 0.5–1.0 mm 5¢-UMP, and the
excess reagent was removed with a spin column. The
results showed that, in these ranges, the enzymes were
inactivated by 95% ± 5%, 48% ± 2%, and
8% ± 3%, respectively. This accords with the idea
that externally added 5¢-UMP can protect the said
arginine against modification but the inhibitor bound
ex vivo is unable to do so, which is consistent with
Scheme 1.
The existence of an isolable intermediate in the con-
version of E
1
to E
4
was also indicated by the profiles
generated from reductive inhibition of E
1
in the pres-
ence of 0–0.5 mm 5¢-UMP and 10 mml(+)-arabinose
(Fig. 2B). These clearly indicate a three-state transition

with a stable intermediate in the presence of 0.1–
0.2 mm 5¢-UMP. Formation of E
3A
from E
3
represents
the reductive inhibition of the stable intermediate. This
intermediate possesses residual activity of 58% ± 2%
as compared with E
1
, and NADH fluorescence of
63% ± 2% as compared with E
4
. The NADH fluores-
cence was measured under denaturing conditions to
remove interference from coenzyme fluorescence. This
indicated that under the defined conditions of reduc-
tive inhibition, one of the two NAD
+
molecules of the
dimeric enzyme was converted to NADH. In control
experiments, it was verified that the reagents carried
over to the assay mixture did not cause inactivation of
the coupled enzyme. Therefore, the results of
Fig. 2A,B are in agreement with Scheme 1.
Irreversible conversion
If the conversion of Scheme 1 were reversible, the
enzyme–inhibitor complexes would become unstable
while the excess inhibitor was removed. The complexes
E

3
and E
4
were dialyzed extensively against 50 mm
A
B
Fig. 2. Dependence of physicochemical properties of native epim-
erase (0.25–0.5 mgÆmL
)1
) in the presence of 0–6 mM added
5¢-UMP at pH 8.0. (A) The kinetic lag in converting UDP-Gal to
UDP-Glc was followed. Coenzyme fluorescence was measured
without dilution. Inactivation by trypsin was followed from residual
epimerase activity after protease digestion for 4 h under the stipu-
lated conditions. Results are expressed by taking activity of the
native epimerase as 100%. Inset: SDS ⁄ PAGE of epimerase (20 lg)
digested with trypsin (50 : 1, w ⁄ w) at pH 8.0 and at 4 °C. Lane 1:
after 30 s of protease pulse. Lane 2: after 4 h of protease pulse.
Lane 3: in the presence of 5 m
M 5¢-UMP for 4 h. Lane 4: as lane 2,
except that the epimerase used was modified by CHD. The upper
and lower arrows indicate the positions of BSA (66 kDa) and
ovalbumin (45 kDa) in the electrogram. (B) Reductive inhibition was
performed with 5 m
ML(+)-arabinose for 1 h at 25 °C, in the
presence of varying concentrations of 5¢-UMP. Fluorescence was
measured after treatment with 8
M urea (n = 2–3). In all measure-
ments, baseline correction was performed.
A. Brahma et al. Regulation of catalytic sites of yeast epimerase

FEBS Journal 276 (2009) 6725–6740 ª 2009 Council of Scientific and Industrial Research, New Delhi. Journal compilation ª 2009 FEBS 6729
sodium phosphate (pH 7.5) at 4 °C to remove
unbound inhibitor. It was verified that the native epim-
erase (E
1
) could withstand inactivation due to dialysis
under such conditions. The presence of 5¢-UMP in all
of the dialyzed samples was confirmed by MS analysis
(Fig. 3). The dialyzed enzyme complexes were sub-
jected to reductive inhibition in the presence of 10 mm
l(+)-arabinose. The residual activities of E
1
,E
3
and
E
4
were 80% ± 5%, 55% ± 5% and 8% ± 5%,
respectively (n = 2). The kinetic lag in catalysis as
observed with E
1
could not be reproduced with the
dialyzed samples of E
3
and E
4.
This indicated that the
inhibitor-binding steps are irreversible.
Change in tertiary structure
A change in conformation of an enzyme is an obligatory

requirement for regulation of its activity. In epimerase,
one of four tryptophans per subunit is located at the
catalytic site [20]. Thus, perturbation of the catalytic
sites is likely to be reflected in the fluorescence spectra
of its tryptophans (excitation at 295 nm). The change of
conformation of epimerase during its interaction with
0–0.6 mm 5¢-UMP and 0–0.4 mm UDP-Gal for 5 min
at 25 °C in two separate experimental sets was followed.
The substrate served as a catalytic site-directed ligand.
As compared with the native enzyme, the maximum
emission of the treated samples remained unaltered, at
343.8 ± 0.5 nm, indicating retention of the micro-
environment of the tryptophans. However, in the
presence of maximum concentrations of 5¢-UMP and
UDP-Gal as applied in this study, the emission intensity
was reduced by 12% ± 2% and 17% ± 2%, respec-
tively, indicating conformational change (Fig. 4A,B).
Application of higher concentrations of the inhibitor or
the substrate could not alter the extent of quenching.
The conformers also attained equilibrium, as no
further change in emission intensity was observed with
increasing incubation period.
Kinetic patterns
To investigate whether the catalytic sites of epimerase
are distinguishable on the basis of turnover, three
forms of the enzyme, namely dimeric native epimerase
(E
1
), dimeric inhibitor free epimerase (E
0

), and mono-
meric epimerase (E
M
), were used for kinetic analysis in
the presence of 0–0.35 mm substrate (Figs 5–7).
Fig. 3. MS analysis of the dissociated ligand of native epimerase. The observed peaks have been assigned as follows: 5¢-UMP, 2Na
+
,
H
+
= 369.1 (obs. 368.99); 5¢-UMP, 2Na
+
,2H
+
= 370.1 (obs. 369.87); 5¢-UMP, 3Na
+
= 391.1 (obs. 390.96); and 5¢-UMP, 3Na
+
,H
+
= 392.1
(obs. 391.87). Commercially available 5¢-UMP-disodium salt, under similar experimental conditions, showed an identical mass pattern. The
spectral zone of NAD
+
has not been included.
Regulation of catalytic sites of yeast epimerase A. Brahma et al.
6730 FEBS Journal 276 (2009) 6725–6740 ª 2009 Council of Scientific and Industrial Research, New Delhi. Journal compilation ª 2009 FEBS
Regulation of catalytic activity has been clearly dem-
onstrated in the case of the native epimerase. The
Michaelis–Menten relationship showed hyperbolic

dependencies in the substrate concentration ranges
0–0.075 mm and 0.2–0.35 mm, whereas there was no
significant variation of reaction rate between 0.075 mm
and 0.2 mm (Fig. 5). Thus, the low-affinity site (the
high-affinity and low-affinity sites of epimerase referred
to in the text are related to the substrate UDP-Gal –
A
B
Fig. 5. Michaelis–Menten plot of native epimerase with UDP-Gal
as substrate. The enzyme concentration was 3.3 n
M. (A, B) Linewe-
aver–Burk plots with 0–0.05 m
M and 0.2–0.35 mM substrate. Units
of the ordinate and abscissa are lmolÆmin
)1
and mM
)1
, respec-
tively; values correspond to the original plot. Derived values of K
m
and V
max
are presented in Table 1. Solid and hatched bars
represent the presence of bound 5¢-UMP ex vivo and the gradual
disappearance of the initial lag of catalysis, respectively.
A
B
Fig. 4. Change of conformation of native epimerase in the pres-
ence of (A) 0–0.6 m
M 5¢-UMP and (B) 0–0.4 mM UDP-Gal. The

emission intensity of native epimerase is considered to be 100% in
either set. The ligands had no emission in this spectral zone.
AB
Fig. 6. Michaelis–Menten plot of inhibitor-free epimerase with
UDP-Gal as substrate. The enzyme concentration was 1.65 n
M. The
open circles (s—s) indicate experimentally observed points. (A, B)
Lineweaver–Burk plots with 0–0.05 m
M and 0.05–0.35 mM sub-
strate. Units of the ordinate and abscissa are lmolÆmin
)1
and
m
M
)1
, respectively; values correspond to the original plot. Derived
values of K
m
and V
max
are presented in Table 1. The line (
•
—
•
)
was constructed according to Eqn (2), using the parameters of
Table 1. The other line is the best fit joining the experimentally
observed points (s—s).
Fig. 7. Michaelis–Menten plot of monomeric epimerase with
UDP-Gal as substrate. Inset: Lineweaver–Burk plot with 0–0.35 m

M
substrate. Units of the ordinate and abscissa are lmolÆmin
)1
and
m
M
)1
, respectively; values correspond to the original plot. The
enzyme concentration was 0.5 n
M. Derived values of K
m
and V
max
are presented in Table 1.
A. Brahma et al. Regulation of catalytic sites of yeast epimerase
FEBS Journal 276 (2009) 6725–6740 ª 2009 Council of Scientific and Industrial Research, New Delhi. Journal compilation ª 2009 FEBS 6731
high-affinity and low-affinity sites of 5¢-UMP have no
relationship with the corresponding UDP-Gal-binding
sites) became functional at a substrate concentration
much higher than that required for saturation of the
high-affinity site. Lineweaver–Burk plots were con-
structed for the two dependencies (Fig. 5A,B). Derived
K
m
and V
max
values were 0.01 mm and 2.88 lmolÆ
min
)1
Æmg

)1
for the high-affinity site, and 1.0 mm and
5.56 lmolÆmin
)1
Æmg
)1
for the low-affinity site, respec-
tively. The presence of 5¢-UMP bound to the enzyme
as isolated was detected from the kinetic lag in cataly-
sis and the characteristic MS pattern. When these anal-
yses were performed, the enzyme was incubated with
variable concentrations of the substrate for 1 min
under the assay conditions, and was passed through a
spin column to separate unbound ligands from the
eluted enzyme. It was observed that, in the presence of
up to 0.06 mm UDP-Gal, the inhibitor remained
bound to the enzyme (Fig. 5, solid and hatched bars).
Thus, the catalytic site of the native epimerase, which
was free of the inhibitor, was nonfunctional at low
substrate concentrations.
In the case of inhibitor-free epimerase, the depen-
dency of reaction velocity on substrate concentration
cannot be represented by a single Michaelis–Menten
relationship over the range of substrate concentrations
used, because the corresponding Lineweaver–Burk plot
had a poor correlation (R
2
= 0.8473, where R
2
is the

regression coefficient). However, when two Linewe-
aver–Burk plots were constructed for 0–0.02 mm and
0.05–0.2 mm substrate, a significant improvement in
linear dependency was observed, R
2
being 0.960 and
0.999, respectively. These yielded K
m
and V
max
values
of 0.011 mm and 2.08 lmolÆmin
)1
mg
)1
for the high-
affinity site, and 0.178 mm and 1.76 lmolÆmin
)1
Æmg
)1
for the low-affinity site, respectively (Fig. 6A,B). It is
noteworthy that the Lineweaver–Burk plot for the
high-affinity site showed a downward curvature at a
higher substrate concentration (Fig. 6A). This is an
indication of the presence of a second operational site
of low efficiency; otherwise, the plot would follow the
linear trend [21]. It has been calculated that the low-
affinity site contributed at most 10% towards the turn-
over efficiency in the presence of 0.02 mm UDP-Gal.
The Michaelis–Menten relationship of the mono-

meric epimerase showed hyperbolic dependency and a
linear Lineweaver–Burk plot between 0 mm and
0.35 mm UDP-Gal. Derived K
m
and V
max
values were
0.01 mm and 2.52 lmolÆmin
)1
Æmg
)1
, respectively, sug-
gesting that the catalytic site was similar to the high-
affinity site of the native and inhibitor-free epimerase
(Fig. 7 and inset). All kinetic parameters are summa-
rized in Table 1.
Assessment of kinetic data
When the epimerase reaction did not show a Michaelis–
Menten relationship, it was assumed that the cata-
lytic sites were operating simultaneously at unequal
efficiencies. Under such conditions, the rate of an
enzyme reaction (V) can be expressed as the sum of two
Michaelis–Menten dependencies, as in Eqn (2) [21].
V ¼
V
max ðHÞ
½S
K
m ðHÞ
þ½S

þ
V
max ðLÞ
½S
K
m ðLÞ
þ½S
ð2Þ
where the subscripts H and L refer to high-affinity and
low-affinity sites for the substrate. V
max
of the enzyme
was obtained from the Lineweaver–Burk plot at infi-
nite substrate concentration, because, under this condi-
tion, both of the catalytic sites were operating at
maximum efficiency; that is, V
max
= V
max (H)
+ V
max (L)
. V
max (H)
was calculated after extrapolation of
the linear portion of the Lineweaver–Burk plot using
data points from the low substrate concentration.
V
max (L)
was obtained by subtracting V
max (H)

from
V
max
.
From the values of K
m
and V
max
(Table 1), the
dependency of V on [S] was calculated between 0 mm
and 0.35 mm UDP-Gal and compared with the experi-
mental data. For inhibitor-free epimerase, the correla-
tion was quite satisfactory (R
2
= 0.982) (Fig. 6). In
the case of native epimerase, Eqn (2) was not expected
to be valid, as the catalytic sites were not operating
simultaneously (Fig. 5). Analysis of Fig. 5 showed that
contributions by the high-affinity and low-affinity sites
to overall turnover were 34.2% and 65.8%, respec-
tively, when maximum turnover by the enzyme was
achieved. This is in agreement with the profiles of
Fig. 2A,B, where inactivation of one catalytic site by
trypsinization or reductive inhibition led to residual
activities of 61.3% and 59.9%, respectively. Deviation
from equal catalytic efficiency of the two functional
Table 1. Kinetic properties of different forms of epimerase. The
high-affinity and low-affinity sites refer to the substrate UDP-Gal.
Results shown are within ± 5% error.
Epimerase

K
m
(mM UDP-Gal)
V
max
(lmolÆmin
)1
Æmg
)1
)
Native epimerase
High-affinity site 0.01 2.88
Low-affinity site 1.0 5.56
Inhibitor-free epimerase
High-affinity site 0.011 2.08
Low-affinity site 0.178 1.76
Monomeric epimerase 0.01 2.52
Regulation of catalytic sites of yeast epimerase A. Brahma et al.
6732 FEBS Journal 276 (2009) 6725–6740 ª 2009 Council of Scientific and Industrial Research, New Delhi. Journal compilation ª 2009 FEBS
sites is in agreement with Scheme 1 and Fig. 2, where
inactivation of E
3
by trypsinization and reductive inhi-
bition was 55% ± 5% and 58% ± 2%, respectively.
Consistent deviation from inactivation by 50% in these
reactions indicated that the catalytic sites of epimerase
are nonidentical.
Equation (2) was used further to assess the perfor-
mance of the catalytic sites of the inhibitor-free epim-
erase. It has been assumed that the efficiency of the

catalytic sites at infinite substrate concentration
reached 100%, although these values are different in
absolute terms because of regulation. The analysis
shows that raising the substrate concentration from
0.001 mm to 0.025 mm increased the activities of the
high-affinity and low-affinity sites from 0.90% to
69.4% and from 0.06% to 12.26%, respectively. When
the substrate concentration was further increased from
0.025 mm to 0.35 mm, the corresponding increases
were 69.4–96.96% and 12.26–66.27%, respectively.
Effects of inhibitor
The range of inhibitor concentrations and the pattern
of dependency of inhibition of regulatory enzymes dif-
fer from those of Michaelis–Menten-type enzymes [21].
Competitive inhibition of epimerase by 5¢-UMP is
known [14–16,20,22]. Typical plots of residual activities
of native and monomeric epimerase versus inhibitor
concentration show that the profiles are widely differ-
ent (Fig. 8). In the case of native epimerase, no inhibi-
tion was observed up to 0.8 mm 5¢-UMP, as compared
with 62.5% inhibition for the monomeric epimerase.
At 20 mm 5¢-UMP, the monomeric epimerase showed
76.3% inhibition, the native epimerase showed 85%
inhibition. Dixon plots (inverse of rate versus inhibitor
concentration) of the monomeric and native epimerase
were hyperbolic and parabolic (results not shown). The
hyperbolic dependency indicated partial inhibition
from a single binding site of the inhibitor in mono-
meric epimerase. The parabolic dependency indicated
two binding sites of the inhibitor in native epimerase

that are regulatory in nature. These patterns are simi-
lar to those of the nonregulatory and regulatory types
of enzyme [21]. UDP and UTP are also competitive
inhibitors of epimerase, but have weaker affinity than
5¢-UMP [15,22]. Thus, they are expected to remove the
5¢-UMP of native epimerase. Abolition of the lag in
catalysis of the native epimerase after interaction with
these inhibitors was correlated with their inhibitor con-
stants [23]. The values of the residual lag, correspond-
ing inhibitor concentration and K
i
for 5¢-UMP, UDP
and UTP were 10%, 0.4 mm, and 0.15 mm, 13%,
6mm, and 0.37 mm, and 25%, 6 mm, and 0.60 mm,
respectively. Hence, the ability of the inhibitors to
remove the kinetic lag was inversely related to their K
i
,
and they also showed specificity of such substitution
according to their K
i
values.
Selective inactivation of one catalytic site
5¢-UMP and l(+)-arabinose are the most effective
reagents for reductive inhibition. A combination of
uridine nucleotides such as UDP or UTP and reducing
sugars such as galactose or glucose can also induce
reductive inhibition, but with lower efficiency. Native
or inhibitor-free epimerase was incubated with 0.2 mm
UDP or UTP along with 2 mmd(+)-Gal or d(+)-Glc

at 4 °C for 40 h at pH 7.5. Whereas native epimerase
without any reagent retained 96% ± 2% of its activ-
ity, incubation with any combination of reagents
reduced the activity to 65% ± 5%, with a distinctly
different pattern in the Michaelis–Menten plot. There
was a hyperbolic dependency between 0 mm and
0.075 mm UDP-Gal, after which there was no signifi-
cant change in the catalytic rates up to 0.35 mm sub-
strate. This evidently indicates inactivation of the
second site (Fig. 9A). Inhibitor-free epimerase incu-
bated with various combinations of reagents demon-
strated the same biphasic Michaelis–Menten
dependency as that shown in Fig. 6, with 80% ± 3%
recovery of residual activity (Fig. 9B).
Discussion
Allosteric regulation of the epimerase from K. fragilis
has not been investigated with confidence before. That
there is deviation from the Michaelis–Menten relation-
ship during the reversible conversion of UDP-Gal to
Fig. 8. Inhibitory profiles of native and monomeric epimerase by
5¢-UMP. The enzyme and substrate concentrations were 1.65 n
M
and 0.1 mM, respectively. 5¢-UMP has no effect on the coupling
enzyme. The enzyme activity in the absence of the inhibitor is
considered to be 100%.
A. Brahma et al. Regulation of catalytic sites of yeast epimerase
FEBS Journal 276 (2009) 6725–6740 ª 2009 Council of Scientific and Industrial Research, New Delhi. Journal compilation ª 2009 FEBS 6733
UDP-Glc before product equilibration is attained has
been known for a long time [24]. The allosteric kinetics
of the epimerase were reported more recently [25,26].

However, when the subunit-sharing model of a single
catalytic site of the dimeric enzyme was proposed, the
relevance of allostericity could not be explained [3].
Now it is known that the epimerase contains two
NAD
+
molecules per dimer, and the subunit-sharing
model of the catalytic site is invalid [7,11,13]. Also,
binding of 1 mol of 5¢-UMP per dimer as isolated,
leading to inactivation of the enzyme, indicates regula-
tory behavior, provided that the catalytic sites are
functional [14]. These findings have revived interest in
exploring the regulatory behavior of this enzyme. To
avoid misinterpretation of the results, the composition
and stoichiometry of the bound cofactor(s) and the
stability of the dimeric structure of epimerase with
regard to fast association–dissociation were ascer-
tained.
The epimerase from E. coli can accommodate
NADH in place of NAD
+
when overexpressed from a
plasmid [27]. As E. coli and yeast epimerase are similar
in many respects, there remains a possibility that the
yeast enzyme can recruit NADH instead of NAD
+
,
leading to partial inactivation as well as functional het-
erogeneity. To verify this, the native epimerase was
treated with 8 m urea to dissociate the cofactor(s) [17].

The resulting solution had no characteristic NADH
fluorescence to the limit of detection (< 0.01 mol per
dimer). Incomplete recruitment of NAD
+
and reacti-
vation of the enzyme during the assay after it has
absorbed NAD
+
from the assay mixture can also
cause functional heterogeneity. This was ruled out, as
the enzyme preincubated with 0.05 mm NAD
+
for
15 min prior to the assay did not show enhancement
of activity. Reductive inhibition of epimerase followed
by quantification of dissociated NADH (as illustrated
in Experimental procedures) showed that the stoichi-
ometry was nearly 2.0 per dimer or 1 per catalytic site.
Earlier, maximum recovery of 1.70 ± 0.10 mol of
NAD per dimer was reported, based on dissociation of
the nucleotide by trichloroacetic acid or heat, where
partial coprecipitation of the holoenzyme with the
apoenzyme is suspected [14].
The stability of dimeric structure of epimerase with
regard to rapid association–dissociation was estab-
lished from complete and reversible dissociation of the
subunits after modification with pCMB, followed by
reduction under nondenaturing conditions [19]. The
kinetic parameters of the reconstituted enzyme were
distinctly different from those of the native enzyme

(Fig. 1A,B). The native enzyme could never attain this
property of the reconstituted enzyme. This indicates
that the equilibrium described in Eqn (1) is not valid
for epimerase.
As kinetic data cannot predict the number of cata-
lytic sites of an enzyme, selective inactivation of one
site of epimerase appeared to be the only answer to
this question. Such a proposition remained elusive, as
the catalytic sites were found to be identical with
regard to several modification reagents. Reductive inhi-
bition offered a unique opportunity to address this
issue. It was proposed that reductive inhibition could
reduce one NAD
+
of the enzyme–inhibitor complex
E
3
, leading to E
3A
. As a consequence, E
3
would be
inactivated by 50%. In reality, such experiments
A
B
Fig. 9. Michaelis–Menten plots of epimerase preincubated with
substrate analogs. (A) Native epimerase.
•
—
•

, epimerase incu-
bated without any substrate analog.
, enzyme preincubated
with 0.2 m
M UDP + 2 mMD(+)-Gal. , enzyme preincubated
with 0.2 m
M UDP + 2 mMD(+)-Glc. (B) Inhibitor-free epimerase.
•
—
•
, dependencies of inhibitor-free epimerase preincubated
without substrate analogs.
, dependencies of inhibitor-free
epimerase preincubated with 0.2 m
M UDP + 2 mMD(+)-Gal.
Regulation of catalytic sites of yeast epimerase A. Brahma et al.
6734 FEBS Journal 276 (2009) 6725–6740 ª 2009 Council of Scientific and Industrial Research, New Delhi. Journal compilation ª 2009 FEBS
yielded a residual activity of 58% ± 2% (n = 6). This
indicates that the catalytic sites of the enzyme are
functional and are distinguishable on the basis of bind-
ing with the inhibitor. Careful analysis of Fig. 2 shows
that inactivation of E
3
by trypsin and reductive inhibi-
tion consistently deviated from 50%, in spite of modi-
fication of one catalytic site of two. Such inequality
between two identical catalytic sites is usually achieved
by allosteric regulation. A perceivable change in the
conformation of the enzyme in the presence of
5¢-UMP and UDP-Gal was demonstrated by trypto-

phan fluorescence emission (Fig. 4). Furthermore, it is
now comprehensible why repeated attempts to purify
the native epimerase using a UMP-agarose affinity
matrix had failed (A. Brahma, unpublished observa-
tions). In the native epimerase, one inhibitor-binding
site is blocked by 5¢-UMP, whereas the other is
insensitive to this inhibitor until its concentration in
solution reaches 0.1 mm. Certainly, this condition is
not met with affinity chromatography.
Allosteric enzymes display sigmoidal rather than
hyperbolic Michaelis–Menten dependencies, arising
from cooperative interaction between ligand-binding
sites. In the case of epimerase, if the catalytic sites
operate independently with different efficiencies, the
resultant Michaelis–Menten profile will be a combina-
tion of two hyperbolic dependencies (Eqn 2). Depend-
ing on the difference between their efficiencies, a
biphasic pattern might result, as observed for native
epimerase (Fig. 5). This is convincing evidence that the
catalytic sites are operating at distinctly nonoverlap-
ping zones of substrate concentration. As 5¢-UMP
bound ex vivo inactivates the epimerase, it is of interest
to determine which site offers catalysis at lower sub-
strate concentrations. It has been confirmed from MS
analysis and the kinetic lag in catalysis that the inhibi-
tor remained associated with the enzyme in the pres-
ence of low concentrations of UDP-Gal (Fig. 5).
Analysis of the Michaelis–Menten dependency of
inhibitor-free epimerase also indicated that its catalytic
sites are operating independently with unequal efficien-

cies. The nonidentical nature of the Michaelis–Menten
relationships of inhibitor-free and native epimerase is
an indication of the regulatory role of the bound
5¢-UMP in the latter. The Michaelis–Menten relation-
ship of the monomeric epimerase displays the usual
monophasic hyperbolic dependency, which further vali-
dates cooperativity between the two catalytic sites
(Fig. 7). In this context, we reviewed why previously
reported K
m
values varied between 0.1 mm and
0.13 mm [15,16,24]. In those cases, a double reciprocal
plot of the native epimerase was constructed between
0.025 mm and 0.5 mm UDP-Gal, a 20-fold variation in
concentration, but without attention being paid to
even lower concentrations of the substrate. The exis-
tence of inhibitor-free epimerase was unknown at that
time.
The deviation from the Michaelis–Menten relation-
ship, including sigmoidal kinetics, is a necessary but
not sufficient condition for allostericity. Several fac-
tors, including impurities from reagents or the presence
of interfering enzymes, could alter the enzyme kinetics
[28]. The monophasic Michaelis–Menten relationship
of the monomeric epimerase demonstrated the absence
of such artefacts and the requirement for a dimeric
structure to explain the regulation. The difference in
V
max
values among different forms of the epimerase is

a reflection of change of conformation (Table 1)
[29,30]. The functional distinction between native and
monomeric epimerase was also apparent from the
inhibitory pattern with 5¢-UMP (Fig. 8). At low con-
centrations of the inhibitors, a regulatory enzyme
shows greater resistance to inhibition than a Michael-
is–Menten type of enzyme [21]. Binding of 5¢-UMP to
native epimerase as isolated should not be considered
as inhibitory, because it is quickly replaced by the sub-
strate. Interdomain regulation, however, does not
operate, as the mutarotase activity of the C-terminal
part of epimerase remains unaltered after addition of
the highest concentration of 5¢-UMP as applied here
[11]. The significance of dimerization may rest on this
regulation, which is only possible through allostericity
like phenomena [31].
Finally, a correlation between the presented results
and known X-ray crystallographic structures of yeast
epimerase was shown. In the absence of crystallo-
graphic data for K. fragilis epimerase, a comparison
was made with a similar epimerase from S. cerevisiae
[7]. In the S. cerevisiae enzyme, the two subunits are
connected in a symmetric fashion, and the epimerase
catalytic site remains close to the subunit contact
region. ‘Electron density in this area is broken and
impossible to interpret with any certainty but is indica-
tive of multiple conformations’ [7]. Features such as
the close proximity of epimerase catalytic sites and
flexibility of the UDP-Gal-binding regions are in
agreement with the allosteric relationship between

them.
Biological significance
It is pertinent to ask why the native epimerase is the
only isolable form of the enzyme from yeast cells har-
vested near termination of growth. The content of
inhibitor-free epimerase is gradually reduced with the
concomitant rise of the native form in a time-depen-
A. Brahma et al. Regulation of catalytic sites of yeast epimerase
FEBS Journal 276 (2009) 6725–6740 ª 2009 Council of Scientific and Industrial Research, New Delhi. Journal compilation ª 2009 FEBS 6735
dent manner from initiation to termination of cell
growth [14]. In vitro experiments show that inhibitor-
free epimerase cannot be reversibly converted to the
native form. It therefore appears that, at a late phase
of cell growth, the enzyme remains inactive after bind-
ing to substoichiometric amounts of 5¢-UMP. On the
other hand, the concentration of free nucleotides in
microbial cells is 25–50 nmolÆmg
)1
of protein [29]. This
concentration is 100-fold lower than that applied to
convert the inhibitor-free epimerase to the native form.
Consequently, spontaneous conversion of native epim-
erase to E
3
or E
4
in vivo is a remote possibility. This
consideration is important in maintaining the
UDP-Gal or UDP-Glc pool in the cell, as well as in
ascertaining the homogeneity of native epimerase in its

isolable form. Inhibitor-free epimerase is thermally
unstable as compared with the native form. It will be
interesting to see whether the epimerase refolded
reversibly in vitro can capture 5¢-UMP and show
similar regulation of activity as the native epimerase.
Experimental procedures
Enzymes
Epimerase, UDP-galactose 4-epimerase (EC 5.1.3.2); ‘native’
epimerase or E
1
, dimeric epimerase of 75 kDa per subunit
that contains < 1.0 mol of 5¢-UMP per dimer and shows a
kinetic lag in catalysis, and the N-terminal and C-terminal
domains of which are responsible for epimerase and mutaro-
tase activities, respectively; E
4
, dimeric epimerase that
contains 2 mol of 5¢-UMP per dimer; E
2
and E
3
, inter-
mediates that are presumably formed during the conversion
of E
1
to E
4
; inhibitor-free epimerase or E
0
, dimeric epi-

merase that is devoid of 5¢-UMP; monomeric epimerase
or E
M
, a 45 kDa truncated and monomeric functional
form of native epimerase that is devoid of the C-terminal
mutarotase domain. The notations E
1
–E
4
have been used in
the context of binding of the inhibitor 5¢-UMP to epimerase.
The terms native, inhibitor-free and monomeric epimerase
have been used in the context of kinetic properties of the
enzyme.
Reagents
d(+)-Gal, d(+)-Glc, l(+)-arabinose, glycylglycine (GG),
CHD, b-NAD
+
, UDP-Gal, UDP-Glc, 5¢-UMP, UDP,
UTP, DTT, pCMB, soybean trypsin inhibitor (STI), trypsin
(bovine pancreas) and hydroxyapatite were from Sigma
(USA). Urea (GR) was recrystallized from hot ethanol. The
yeast K. fragilis (ATCC 10022, renamed Kluyveromy-
ces marxianus var. marxianus) was purchased from the
Microbial Type Collection Center and Gene Bank, IM-
TECH, Chandigarh, India. YNB (yeast nitrogen base) was
from Hi-media, Mumbai, India. UDP-Glc dehydrogenase
(EC 1.1.1.22) was partially purified from beef liver up to
the heat denaturation step [32]. This enzyme was left for
15 days at ) 20 °Cin50mm sodium acetate (pH 5.5),

whereby contaminating epimerase activity was lost.
Cell growth
Epimerase being an inducible enzyme, K. fragilis cells were
grown in 0.67% YNB and 1.5% galactose medium for 16 h
at 30 °C under aerobic conditions with shaking. Cells were
harvested at the early stationary phase, when the culture
exhibited a turbidity of 0.65 at 650 nm after 10-fold
dilution with water. Cells grown to this extent produced
epimerase associated with inhibitor.
Purification of epimerase (E
1
)
Purification of epimerase involved crude cell extraction,
55% ammonium sulfate fractionation, hydroxyapatite treat-
ment, and DEAE–cellulose chromatography [11]. Homoge-
neity of the enzyme was checked by SDS ⁄ PAGE and
PAGE. The preparation after hydroxyapatite treatment was
free from protease (equivalent to < 0.001 lg of trypsin
activity, with azoalbumin as substrate). SDS ⁄ PAGE of the
enzyme after incubation under various conditions did not
show degradation of the enzyme. This enzyme showed a
kinetic lag in the conversion of UDP-Gal because of its asso-
ciation with 5¢-UMP [14]. The specific activity of the enzyme
was 65–75 unitsÆmg
)1
, where 1 unit is defined as the amount
of enzyme that catalyzes the conversion of 1 lmol of
UDP-Gal per min at 25 °C under standard assay conditions.
Preparation of inhibitor-free epimerase (E
0

)
Epimerase (0.5 mgÆmL
)1
in 50 mm GG, pH 8.8) was trea-
ted with 1 mm UDP-Gal for 15 min at 25 °C, whereby the
bound inhibitor was replaced by the substrate [14]. Excess
substrate and inhibitor were removed by passage through a
Sephadex G-50 spin column. MS analysis showed that this
enzyme was devoid of 5¢-UMP.
Preparation of monomeric epimerase (E
M
)
Epimerase (1 mgÆmL
)1
) was treated with trypsin (50 : 1,
w ⁄ w) in 20 mm potassium phosphate (pH 8.0) at 4 °C for
4 h in the presence of 0.5 mm 5¢-UMP. Residual trypsin
was inactivated by the addition of a two-fold molar excess
of STI. The digest was passed through a Waters Protein
Pak 125 size exclusion HPLC column equilibrated with
20 mm sodium phosphate (pH 7.5) at a flow rate of
0.5 mLÆmin
)1
, and the major peak, corresponding to
45 kDa, was collected. Alternatively, 100 lL of the digest
was passed through a Sephadex G-50 spin column to
Regulation of catalytic sites of yeast epimerase A. Brahma et al.
6736 FEBS Journal 276 (2009) 6725–6740 ª 2009 Council of Scientific and Industrial Research, New Delhi. Journal compilation ª 2009 FEBS
remove reagents and small peptides. The recovery of mono-
meric epimerase from the spin column was 95% in terms of

activity and 60% in terms of mass calculated on the basis
of the dimeric enzyme [11].
Models of inhibitor-free and monomeric epimerase are
shown in Scheme 2.
Preparation of epimerase-2–UMP complex (E
4
)
Epimerase (1 mgÆmL
)1
in 20 mm potassum phosphate,
pH 8.0) was treated with 0.5 mm 5¢-UMP for 15 min at
25 °C. Excess 5¢-UMP was removed by passage through a
Sephadex G-50 spin column equilibrated with the same buf-
fer [33]. It was confirmed by MS analysis that 5¢-UMP
bound to E
1
or E
4
could not be removed after extensive
dialysis or passage through a spin column.
Enzyme assay
Epimerase activity was measured by coupled assay, in
which conversion of UDP-Gal to UDP-Glc was continu-
ously monitored at 340 nm and 25 °C by the coupling
enzyme UDP-Glc dehydrogenase in the presence of NAD
+
[14]. The assay mixture contained 0.1 m GG (pH 8.8),
0.5 mm NAD
+
, 0.1 mm UDP-Gal and eight units of UDP-

Glc dehydrogenase in 1 mL. It was incubated for 10 min at
25 °C to remove contaminating UDP-Glc present in the
UDP-Gal. Unless preincubation was performed, an initial
burst phase was observed as an artefact. The assay was ini-
tiated by adding 3–15 nmol of the epimerase. The rate of
formation of product showed linear dependency for at least
8 min, when the absorbance change per minute was within
0.03. The rate of product formation also showed linear
dependency when the epimerase concentration was
3–15 nm. In that profile, the extrapolated line passed
through the origin, indicating that the coupling enzyme was
free from epimerase activity. By varying the volume of
enzyme added between 10 lL and 400 lL instead of water
in the assay mixture and observing the linear progress curve
of D0.005–0.03 absorption unitsÆmin
)1
, the assay permits
detection of as little as 0.25% of enzyme activity. To study
the effects of inhibitor, the assay mixture was incubated
with 0–20 mm 5¢-UMP for 10 min at 25 °C; the reaction
was initiated by adding the epimerase. During the epimer-
ase assay, the UDP-Gal concentration could not be raised
above 0.5 mm, where the coupling enzyme became limiting.
The coupling enzyme UDP-Glc dehydrogenase was assayed
with UDP-Glc as substrate in the presence of NAD
+
at
340 nm, under the same conditions as those used for the
epimerase assay [15]. The points presented in all kinetic
experiments (Figs 4–9) are the average of three sets, where

the variation of results was within ± 10%.
Under the stated assay conditions, only the native epim-
erase (E
1
) showed an initial lag in catalysis. The duration
of lag (in seconds) was calculated from the time axis by
extrapolating the linear portion of the progress curve of the
coupled assay. The initial rate could be defined when pro-
gress to 30 s was obtained from the spectrophotometer.
The span of the lag is dependent on the enzyme and sub-
strate concentrations; therefore, these parameters were
maintained at 0.33 nm and 0.2 mm in the assay in all sets
of experiments [14]. Under this condition, the observed lag
was distinct and accurately measurable.
Dissociation by pCMB and reconstitution
Epimerase (0.5 mgÆmL
)1
)in20mm Tris ⁄ HCl (pH 8.0) was
incubated with 0.5 mm pCMB at 25 °C for 30 min,
Scheme 2. Proposed models of preparation
of inhibitor-free and monomeric epimerase.
E
1
, dimeric epimerase containing one 5¢-
UMP per dimer bound as isolated (native
epimerase); E
4
, epimerase where both the
5¢-UMP-binding sites are occupied by added
5¢-UMP; E

M
, monomeric epimerase where
there is only one 5¢-UMP-binding site occu-
pied by added 5¢-UMP and the mutarotase
part is degraded. E
0
, dimeric epimerase
where both the catalytic sites are occupied
by UDP-Gal (also referred to as inhibitor-free
epimerase). The solid rectangle indicates
added UDP-Gal bound to the epimerase; s,
arginine at the active site; s over UDP-Gal
binding site indicates protection against
trypsinization in the presence of UDP-Gal.
All other notations are as in Scheme 1.
A. Brahma et al. Regulation of catalytic sites of yeast epimerase
FEBS Journal 276 (2009) 6725–6740 ª 2009 Council of Scientific and Industrial Research, New Delhi. Journal compilation ª 2009 FEBS 6737
whereby the dimeric enzyme was dissociated with release of
the cofactors [19]. It was passed through a Sephadex G-50
spin column to remove excess reagents and free NAD
+
.
Reconstitution of the dimeric holoenzyme was initiated by
adding 50 mm dithiothreitol and 1 mm NAD
+
, and
90–100% reactivation was achieved by 30 min at 25 °C.
Coenzyme fluorescence
Yeast epimerase possesses an inherent NADH-like charac-
teristic fluorescence (excitation, 353 nm; emission, 400–

600 nm) arising from noncovalent interaction between
NAD
+
and the SH group of a cysteine, termed coenzyme
fluorescence. It has been used extensively as a sensitive
probe to monitor the integrity of the active site
[3,16,20,22,34]. When epimerase is treated with 8 m urea at
pH 7.5 for 10 min at 25 °C, unfolding of the enzyme occurs
with dissociation of the cofactor. This leads to abolition of
98% of coenzyme fluorescence, indicating that the enzyme-
bound NAD
+
is not a fluorophore by itself [17,34,35].
Reductive inhibition
The epimerase (0.50–0.25 mgÆmL
)1
in 50 mm potassium
phosphate, pH 7.5) was inactivated after incubation with
0.5 mm 5¢-UMP and 5 mml(+)-arabinose for 1 h at 25 °C
[36]. A control set under the same experimental conditions
but in the absence of these reagents retained 98% of
enzyme activity. For the expression of results, the residual
activity of native epimerase has been considered to be
100%. No baseline corrections were made, as 5¢-UMP had
no emission in this spectral zone. Whereas coenzyme
fluorescence originates from NAD, reductive inhibition
actually produces NADH; the difference is explained in
Scheme 3.
Estimation of NAD
+

Quantitative conversion of epimerase bound NAD
+
to
NADH was performed by reductive inhibition. Completion
of the reaction was indicated by complete inactivation of
the enzyme. The reduced cofactor was dissociated from the
enzyme after incubation with 8 m urea at pH 7.5 for
10 min. The concentration of the reduced cofactor was
determined from the fluorescence intensity with respect to a
calibration curve. The curve correlated the fluorescence
intensity of NADH in the presence of 8 m urea with its
concentration (0–10 lm), where a linear dependency was
observed (R
2
= 0.9983). The presence of globular proteins
without visible chromophore did not interfere with NADH
emission. In all fluorescence experiments, the inner filter
effect was allowed for [17,34,35].
Trypsin digestion
Digestion of epimerase by trypsin was performed essentially
as described for preparation of monomeric epimerase,
except for the variation in 5 ¢-UMP concentration. Unbound
5¢-UMP and small peptides so formed were separated
from the undigested epimerase by passage through a
Sephadex G-50 spin column. The functional enzyme was
recovered from the eluent for assay.
Characterization of epimerase–inhibitor
complexes
Epimerase (0.25–0.5 mgÆmL
)1

)in50mm sodium phosphate
(pH 7.5) was incubated with 0–6 mm 5¢-UMP for 15 min at
25 °C. Previous observations relating to inhibition studies
indicated that the duration of incubation was sufficient to
obtain equilibrium conformers (A. Brahma, unpublished
data). Aliquots were transferred to assay mixtures for esti-
mation of the kinetic lag in catalysis. In control experi-
ments, the enzyme–inhibitor complexes so formed were
passed through Sephadex G-50 spin columns to remove
excess inhibitor. Assay of the eluted enzyme showed that
the 5¢-UMP that was carried over to the assay mixture in
the previous experimental sets did not affect the rate of
catalysis, including the extent of kinetic lag. Coenzyme fluo-
rescence of the enzyme–inhibitor complexes was measured
without removal of unbound inhibitor, as the inhibitor did
not interfere with the emission. Trypsin digestion of the
complexes was performed as stated earlier, and proteolysis
was arrested by adding a two-fold molar excess of STI over
trypsin. Aliquots from the protease digest were tested for
residual epimerase activity. Reductive inhibition of enzyme–
inhibitor complexes was performed after incubation with
5mml(+)-arabinose under conditions as described above.
Time-dependent inactivation of epimerase during reductive
inhibition was monitored. After complete inactivation had
Scheme 3. Distinction between coenzyme fluorescence and NADH fluorescence of epimerase. The underlining indicates the enzyme sur-
face. Binding of NAD
+
⁄ NADH to the enzyme is noncovalent. The hatched sign indicates weak spatial interaction between NAD
+
and a cyste-

ine of the enzyme, leading to NADH-like coenzyme fluorescence. The subscript ‘free’ denotes dissociated nucleotides. Enzyme-bound and
free NADH fluorescence are similar but not identical.
Regulation of catalytic sites of yeast epimerase A. Brahma et al.
6738 FEBS Journal 276 (2009) 6725–6740 ª 2009 Council of Scientific and Industrial Research, New Delhi. Journal compilation ª 2009 FEBS
been ensured, the NADH fluorescence of incubates was
measured. It was verified that none of the reagents used in
these experiments interfered with NADH fluorescence.
MS
A Q-TOF micro (Micromass) instrument with microchannel
plate detectors was used. Positive ionization electrospray
mode at a desolvation temperature of 200 °C was applied.
Argon, at a pressure of 2 kgÆcm
)2
, with a collision energy
of 10 eV, was used as collision gas. Epimerase
(0.05 mgÆmL
)1
) was dialyzed extensively against water at
4 °C, lyophilized, and reconstituted in 10 mm potassium
phosphate (pH 7.0) at a concentration of 0.5 mgÆmL
)1
. The
sample was heated at 100 °C for 5 min to dissociate 5¢-
UMP and NAD
+
. The solution was centrifuged at 5585 g
for 15 min to remove precipitated protein. The components
in the supernatant were separated by RP-HPLC before
mass analysis. 5¢-UMP and NAD
+

were also heat treated
under identical conditions before MS. The following MS
data (fast atom bombardment) were obtained: m ⁄ z 368.99
(5¢-UMP, 2Na
+
,H
+
), 369.87 (5¢-UMP, 2Na
+
,2H
+
),
390.96 (5¢-UMP, 3Na
+
), and 391.87 (5¢-UMP, 3Na
+
,H
+
).
Analysis for NAD
+
mass has not been included here. Nei-
ther NAD
+
nor other UMP derivatives yielded fragments
that corresponded to those of 5¢-UMP [14].
Inactivation
Native or inhibitor-free epimerase (0.25–0.5 mgÆmL
)1
) was

incubated with its substrate analogs, which were combina-
tions of 0.25 mm uridine nucleotides (UMP, UDP, or UTP)
and 2 mm reducing sugars [d(+)-Gal, d(+)-Glc, or l(+)-
arabinose] in 0.05 m potassium phosphate (pH 7.5) at 4 °C
for 40 h. As a positive control, an enzyme without any sub-
strate analog was incubated under identical conditions,
where 2–3% of inactivation was observed. After incubation,
the rates of enzymatic conversion were measured in the
presence of 0–0.35 mm UDP-Gal.
Other methods
A UV–visible recording spectrophotometer (Specord 200;
Analytical Jena, Germany) was used for enzyme assays.
Other optical measurements were performed with a Bio-
chrom S2000 diode array spectrophotometer (UK). All flu-
orescence measurements were performed with a
Hitachi F4500 recording spectrofluorimeter, using 700 lL
quartz cuvettes with excitation and emission slit widths of
5 nm each. Arginines of epimerase (0.5 mgÆmL
)1
) were
modified with 2 mm CHD in 0.2 m sodium borate (pH 9.0)
for 3 h at 37 °C [3]. Protein estimation was performed
according to the method of Lowry et al. [37] or with
Bio-Rad Protein Assay Reagent, as per the manufacturer’s
protocol (catalog no. 10044; Bio-Rad Laboratories), using
BSA as reference. The following extinction coefficient val-
ues were used: NADH, e
340 nm
= 6.3 · 10
3

m
1
Æcm
)1
;
NAD
+
, e
260 nm
= 17.8 · 10
3
m
)1
Æcm
)1
; uridine nucleotides
and their derivatives, e
260 nm
=10· 10
3
m
)1
Æcm
)1
.
Acknowledgements
N. C. Price (University of Glasgow, UK) extensively
revised the manuscript, and B. Achariya (Emeritus
Scientist, IICB) edited the text. K. Sarkar (IICB)
provided MS data. This work was supported by the

Department of Science and Technology (Grant
No. SR ⁄ SO ⁄ BB-66 ⁄ 2005) awarded to D. Bhattachar-
yya. A. Brahma and N. Banerjee were supported by
CSIR-SRF and UGC NET-SRF, respectively.
References
1 Frey PA (1987) Complex pyridine nucleotide-dependent
transformations. In Pyridine nucleotide coenzymes:
chemical, biochemical and medical aspects (Dolphin D,
Poulson R & Avarmovie O, eds), Vol. 2B, pp. 461–511.
Wiley, New York, NY.
2 Bhattacharjee H & Bhaduri A (1992) Distinct functional
roles of two active site thiols in UDP-galactose 4-epim-
erase from Kluyveromyces fragilis. J Biol Chem 267,
11714–11720.
3 Frey PA (1996) The Leloir pathway: a mechanistic
imperative for three enzymes to change the stereochemi-
cal configuration of a single carbon in galactose.
FASEB J 10, 461–470.
4 Thoden JB, Frey PA & Holden HM (1996) High-resolu-
tion X-ray structure of UDP-galactose 4-epimerase
complexed with UDP-phenol. Protein Sci 5, 2149–2161.
5 Thoden JB, Frey PA & Holden HM (1996) Crystal
structure of oxidized and reduced forms of UDP-galac-
tose 4-epimerase isolated from E. coli. Biochemistry 35,
2557–2566.
6 Thoden JB & Holden HM (1998) Dramatic differences
in the binding of UDP-glucose and UDP-galactose to
UDP-galactose 4-epimerase from Escherichia coli.
Biochemistry 37, 11469–11477.
7 Thoden JB & Holden HM (2005) The molecular

architecture of galactose mutarotase ⁄ UDP-galactose
4-epimerase from Saccharomyces cerevisiae. J Biol Chem
280, 21900–21907.
8 Thoden JB, Wohlers TM, Fridovich-Keil JL & Holden
HM (2000) Crystallographic evidence for Tyr 157
functioning as the active site base in human UDP-
galactose 4-epimerase. Biochemistry 39, 5691–5701.
9 Nayar S & Bhattacharyya D (1997) UDP-galactose
4-epimerase from E. coli: existence of a catalytic
monomer. FEBS Lett 409, 449–451.
A. Brahma et al. Regulation of catalytic sites of yeast epimerase
FEBS Journal 276 (2009) 6725–6740 ª 2009 Council of Scientific and Industrial Research, New Delhi. Journal compilation ª 2009 FEBS 6739
10 Thoden JB, Frey PA & Holden HM (1996) Molecular
structure of the NADH ⁄ UDP-glucose abortive complex
of UDP-galactose 4-epimerase from E. coli: implication
for the catalytic mechanism. Biochemistry 35, 5137–
5144.
11 Brahma A & Bhattacharyya D (2004) UDP-galactose
4-epimerase from Kluyveromyces fragilis: evidence for
independent mutarotation site. Eur J Biochem 271,
58–68.
12 Majumdar S, Ghatak J, Mukherji S, Bhattacharjee H &
Bhaduri A (2004) UDPgalactose 4-epimerase from
Saccharomyces cerevisiae: a bifunctional enzyme with
aldose 1-epimerase activity. Eur J Biochem 271, 753–
759.
13 Brahma A & Bhattacharyya D (2004) UDP-galactose
4-epimerase from Kluyveromyces fragilis: existence of
subunit independent functional site. FEBS Lett 257,
34–41.

14 Nayar S, Brahma A, Barat B & Bhattacharyya D
(2004) UDP-galactose 4-epimerase from Kluyveromy-
ces fragilis: analysis of its hysteretic behavior during
catalysis. Biochemistry 43, 10212–10223.
15 Darrow RA & Rodstrom R (1968) Purification and
properties of uridine diphosphate galactose 4-epimerase
from yeast. Biochemistry 7, 1645–1654.
16 Mukherjee S & Bhaduri A (1986) UDP-glucose 4-epim-
erase from Saccharomyces fragilis: presence of an essen-
tial arginine residue at the substrate-binding site of the
enzyme. J Biol Chem 261, 4519–4524.
17 Bhattacharyya D (1993) Reversible folding of UDP-
galactose 4-epimerase from yeast Kluyveromyces fragilis.
Biochemistry 32, 9726–9734.
18 Darrow RA & Rodstrom R (1970) Uridine diphosphate
galactose 4-epimerase from yeast: studies on the rela-
tionship between quaternary structure and catalytic
activity. J Biol Chem 245, 2036–2042.
19 Majumdar S, Bhattacharjee H, Bhattacharyya D &
Bhaduri A (1998) UDP-galactose 4-epimerase from
Kluyveromyces fragilis: reconstitution of holoenzyme
structure after dissociation with para-
chloromercuribenzoate. Eur J Biochem 257, 427–433.
20 Roy S, Mukherjee S & Bhaduri A (1995) Two trypto-
phans at the active site of UDP-glucose 4-epimerase
from Kluyveromyces fragilis . J Biol Chem 270, 11383–
11390.
21 Roberts DV (1977) Regulatory Enzymes and their
Kinetic Behavior.InEnzyme Kinetics. pp 168–227.
Cambridge University Press, Cambridge.

22 Mukherjee S & Bhaduri A (1992) An essential histidine
residue for the activity of UDP-glucose 4-epimerase
from Kluyveromyces fragilis . J Biol Chem 267
, 11709–
11713.
23 Pal DK & Bhaduri A (1971) Nucleotide inhibition of
UDP-glucose 4-epimerase from Saccharomyces fragilis
and goat liver. Biochim Biophys Acta 250, 588–591.
24 Maxwell ES (1957) The enzymatic interconversion of
uridine diphospho-galactose and uridine diphospho-glu-
cose. J Biol Chem 229, 139–151.
25 Ray M & Bhaduri A (1975) UDP-glucose 4-epimerase
from Saccharomyces fragilis: allosteric kinetics with
UDP-glucose as substrate. J Biol Chem 250 , 4373–4375.
26 Ray M & Bhaduri A (1978) UDP-glucose 4-epimerase
from Saccharomyces fragilis: asymmetry in allosteric
properties led to unidirectional catalysis. Biochem
Biophys Res Commun 85, 242–248.
27 Vanhooke JL & Frey PA (1994) Characterization and
activation of naturally occurring abortive complexes of
UDP-galactose 4-epimerase from E. coli. J Biol Chem
269, 31496–31504.
28 Westley J (1969) Regulatory enzymes and sigmoid
kinetics. In Enzyme Catalysis. pp 169–178. Harper and
Row Publishers, New York, NY.
29 Kim M-K, Lee I-Y, Ko J-H, Rhee Y-H & Park Y-H
(1999) Higher intracellular levels of UMP under
nitrogen-limited conditions enhance metabolic flux of
curdlan synthesis in Agrobacterium sp. Biotechnol
Bioeng 62, 317–323.

30 Citri N (1973) Conformational adaptability in enzymes.
Adv Enzymol Relat Areas Mol Biol 37, 397–648.
31 Traut TW (1994) Dissociation of enzyme oligomers: a
mechanism for allosteric regulation. Crit Rev Biochem
Mol Biol 29 , 125–163.
32 Zalitis J, Uram M, Bowser AM & Feingold DS (1972)
UDP-glucose dehydrogenase from beef liver. Methods
Enzymol 28, 430–435.
33 Nath S, Brahma A & Bhattacharyya D (2003) Extended
application of gel-permeation chromatography by spin
column. Anal Biochem 320, 199–206.
34 Dutta S, Maity NR & Bhattacharyya D (1997) Multiple
unfolded states of UDP-galactose 4-epimerase from
yeast K. fragilis: involvement of proline cis–trans isom-
erization in reactivation. Biochim Biophys Acta 1343,
251–262.
35 Maiti NR, Barat B & Bhattacharyya D (1999) UDP-
glucose 4-epimerase from Kluyveromyces fragilis: equi-
librium unfolding studies. Indian J Biochem Biophys 36,
433–441.
36 Kalckar HM, Bertland AU & Bagge B (1970) The
reductive inactivation of UDP-galactose 4-epimerase
from yeast and E. coli. Proc Natl Acad Sci USA 65,
1113–1119.
37 Lowry OH, Rosenbrough NJ, Farr AL & Randall RJ
(1951) Protein measurement with the folin-phenol
reagent. J Biol Chem 193, 265–276.
Regulation of catalytic sites of yeast epimerase A. Brahma et al.
6740 FEBS Journal 276 (2009) 6725–6740 ª 2009 Council of Scientific and Industrial Research, New Delhi. Journal compilation ª 2009 FEBS