ANALYTICAL

BIOCHEMISTRY
Analytical Biochemistry 347 (2005) 244–253
www.elsevier.com/locate/yabio

b-D-Glucosidase reaction kinetics from isothermal titration

microcalorimetry
Tina Jeoh *, John O. Baker, Mursheda K. Ali, Michael E. Himmel, William S. Adney
National Renewable Energy Laboratory, National Bioenergy Center, 1617 Cole Blvd., Golden, CO 80401, USA

Received 9 July 2005

Available online 13 October 2005

Abstract

The cellobiase activities of nine thermal stable mutants of Thermobifida fusca BglC were assayed by isothermal titration microcalori-
metry (ITC). The mutations were previously generated using random mutagenesis and identified by high-temperature screening as
imparting improved thermal stability to the b-D-glucosidase enzyme. Analysis of the substrate–saturation curves obtained by ITC for
the wild-type enzyme and the nine thermally stabilized mutants revealed that the wild type and all the mutants were subject to binding
of a second substrate molecule. Furthermore, the ‘‘inhibited’’ enzyme–substrate complexes were shown to retain catalytic activity. In the
case of three of the BglC mutants (N178I, N317Y/L444F, and N317Y/L444F/A433V), binding of a second substrate molecule resulted in
improved cellobiose turnover rates at lower substrate concentrations. No correlation between denaturation temperatures of the mutants
and activity on cellobiose at 25
C was evident. However, one particular mutant, BglC S319C, was significantly improved in both thermal
tolerance and cellobiase activity with respect to those of the wild-type BglC. The triple mutant, N317Y/L444F/A433V, had a 5
C
increase in denaturation temperature while maintaining activity levels similar to that of the wild type at higher substrate concentrations.
ITC provided a highly sensitive and nondestructive means to continuously monitor the reaction of BglC with cellobiose, resulting in
abundant data sets that could be rigorously analyzed by fitting to known enzyme kinetics models. One distinct advantage of using data
from the ITC was the empirical validation of the pseudo steady state assumption, a necessary condition for obtaining solutions to the
proposed mechanisms.
 2005 Elsevier Inc. All rights reserved.

Keywords: b-Glucosidase; Cellobiase; Isothermal titration microcalorimetry; Enzyme kinetics; Substrate inhibition

b-D-Glucosidase (EC 3.2.1.21) is a critical enzyme in cel-
lulase mixtures designed to efficiently break down cellulose
in biomass. The hydrolysis of cellobiose to glucose by b-D-
glucosidases in the context of enzymatic cellulose depoly-
merization serves two key purposes: first to minimize
end-product inhibition of the cellulases and second to pro-
vide a glucose stream for conversion to ethanol and other
industrially important chemicals by fermentative microor-
ganisms. Commercially available cellulase preparations
are often supplemented with b-D-glucosidase to boost over-
all cellulolytic activity on pretreated biomass [1–3].
*
Corresponding author. Fax: +1 303 384 7752.
E-mail address: tina_jeoh@nrel.gov (T. Jeoh).
Highly efficient b-D-glucosidases are essential for opti-
mizing biomass utilization processes. One strategy for
improving b-D-glucosidases is to increase the thermal sta-
bility of the enzyme by protein engineering [4–6]. Higher
process temperatures can improve overall economics by
limiting microbial contamination and improving the reac-
tion kinetics [7]. A recent study identified several mutations
of an actinomycete b-D-glucosidase, Thermobifida fusca
BglC, that imparted improved thermal stability to the
enzyme [6]. T. fusca BglC, first characterized by Spiridonov
and Wilson [8], is a family 1 b-D-glucosidase that is active
on cellobiose, cellotriose, and cellotetraose in addition to
sophorose, a b-(1 fi 2)-linked conformer of cellobiose. A
critical concern when attempting to improve the thermal
tolerance of b-D-glucosidases is that the alteration does
not negatively impact the intended activity of the enzyme.

0003-2697/$ - see front matter  2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.ab.2005.09.031
 b-D-Glucosidase kinetics from titration calorimetry / T. Jeoh et al. / Anal. Biochem. 347 (2005) 244–253
A thorough analysis of the activities of the thermal stable
BglC mutants on cellobiose is reported here.
A sensitive and nondisruptive assay to determine the
activities of these T. fusca BglC mutants uses isothermal
titration microcalorimetry (ITC)1 to follow the hydrolysis
of glycosidic bonds by ITC. ITC was originally developed
to measure binding energies [9] and was recently intro-
duced as a powerful tool for assaying enzyme kinetics in
soluble systems [10–13] characterized by the mechanism
k1 k cat
ðMechanism 1Þ E þ S () ES ! E þ P;
k2
where enzyme (E) binds reversibly to the substrate (S) to
form an enzyme–substrate complex (ES), which then
undergoes an essentially irreversible reaction characterized
by rate constant (kcat) to form product (P) and release the
enzyme. This reaction can be followed in the ITC, which
records the heat evolved or absorbed by the binding and
the catalysis reactions. The ITC consists of a reference cell
and a sample cell in a feedback system which continuously
adjusts heating rates to either cell to maintain isothermal
conditions [9]. A baseline is established when the power re-
quired to maintain the two cells at the preset reaction tem-
perature is constant. When a disturbance in the sample cell
(such as a binding event and/or a catalytic event) produces
or absorbs heat, the heating rates are adjusted to maintain
the two cells at the required reaction temperature. The
instrument records the rate of heating as the thermal power
(dQ/dt). In a strategy described by Todd and Gomez [11],
enzyme kinetics can be successfully determined under con-
ditions conducive to Michaelis–Menten kinetics [14], where
there is a large excess of substrate over enzyme and the
reaction rate is limiting. Addition of successive small ali-
quots of substrate to the reaction, therefore, results in a ser-
ies of steady state conditions, during each of which the ES
concentration is essentially constant. In such cases, the
thermal power is maintained at a new level reflecting the
constant thermal power required to counter the heat pro-
duced or absorbed by a constant rate of reaction (Eq. (1)),
d½ P

¼ k cat ½ES
¼ constant at steady state;
dt
where d[P]/dt is rate of product formation (reaction rate)
(mM/s).
As long as the enzyme is not saturated with substrate,
subsequent titrations with substrate aliquots will result in
increased bound enzyme concentrations in the cell and thus
the differential thermal power level at each steady state will
change accordingly. This phenomenon is demonstrated by
Todd and Gomez [11] and Lonhienne and co-workers [12].
Once saturation is achieved, the thermal power does not
change with increased substrate concentrations in the cell
and instead remains constant as long as the enzymes main-
1
Abbreviations used: ITC, isothermal titration microcalorimetry; Mops,
4-morpholinepropanesulfonic acid; DSC, differential scanning
microcalorimetry.
245
tain activity and the substrate to enzyme ratio remains
high. The kinetics of an enzyme reacting with a particular
substrate can thus be assayed by measuring the differential
thermal power levels in reaction and reference cells at each
steady state between titrations and by the relationship con-
verted to the rate of reaction;


d½ P
1 dQ
¼ ; ð2Þ
dt Vol DH app dt
where Vol is the volume of the sample cell = 1.431 mL,
DHapp is the molar enthalpy of the reaction determined
experimentally (cal/mole), and dQ/dt is the thermal power
recorded by the instrument (lcal/s).
The reaction rate obtained from the thermal power at
each steady state (Eq. (2)) can be characterized by applying
the pseudo steady state assumption [15],
d½ P
V max ½S

¼ ; ð3Þ
dt K m þ ½S

where Vmax = kcat[Et] is the maximum rate of reaction
(mM/s), [Et] = [ES] + [E] is the total enzyme
 concentration

in the closed system (mM), and K m ¼ k2 þk cat
is the half-
k1
saturation constant (mM).
If the enzyme system obeys the assumptions of Michae-
lis–Menten kinetics, then Eq. (3) will fit the rate data
obtained from an ITC assay. Deviations from the fit indi-
cate complications in the reaction mechanism such as sub-
strate inhibition.
The b-D-glucosidase enzyme is an ideal candidate for
calorimetric kinetic assays because the primary function
of this enzyme is to hydrolyze the b-(1 fi 4) glycosidic
bond of a soluble cellobiose molecule. In this paper, we
describe the use of ITC to determine the reaction mecha-
nism for the catalysis of cellobiose hydrolysis by T. fusca
BglC and the nine thermal stable mutants. We also discuss
the effect of the mutations on the activity of BglC.
Materials and methods
Enzyme production and purification
ð1Þ
The production and purification of the Escherichia coli-
expressed wild-type BglC and related mutants were per-
formed as previously described [6,8]. Protein purity was
confirmed using SDS–PAGE analysis and the concentra-
tions of purified proteins were determined by absorbance
at 280 nm using a calculated extinction coefficient and
molecular weight obtained by using the Web-based ProtPa-
ram tool [16].
Determination of molar enthalpy of cellobiose hydrolysis
The enthalpy associated with hydrolyzing 1 mol of cello-
biose was determined calorimetrically in a MicroCal VP-
ITC isothermal titration calorimeter (MicroCal, LLC,
Northampton, MA). All buffers and solutions were
degassed immediately prior to conducting the assay. The
246 b-D-Glucosidase kinetics from titration calorimetry / T. Jeoh et al. / Anal. Biochem. 347 (2005) 244–253

buffers were degassed at 25
C immediately prior to use.

In each experiment, the entire system consisting of the ref-
erence cell, sample cell, and titration syringe was equilibrat-
ed at reaction temperature while stirring at 290 rpm, so
that the thermal power baseline was allowed to equilibrate
before the initial titration of substrate into the sample cell.
Enzyme and substrate concentrations were adjusted for
high signal to noise ratio in the thermal power baseline.
Based on the time required for the first titration to reach
steady state, the time between titrations was subsequently
adjusted. Experimental conditions for each enzyme are list-
ed in Table 1. The thermal power was averaged over the 15-
s period prior to the subsequent injection (inset, Fig. 2) and
converted to reaction rate using Eq. (2). The substrate con-
centration corresponding to the reaction rate is calculated
Fig. 1. Thermogram of cellobiose hydrolysis by T. fusca BglC, one of
three replicates conducted to determine molar enthalpy of cellobiose
as the sum on the input substrate minus the consumed sub-
hydrolysis (DHapp). Conditions used were 1 mg/mL BglC with 0.5 lmol of strate and adjusted for dilution. The reaction rates
cellobiose in 20 mM Mops buffer, pH 7.2, at 25
C under constant stirring achieved at each steady state are plotted (Fig. 2) against
at 290 rpm. the corresponding total substrate concentration in the cell
to produce a curve describing saturation kinetics of the
reference cell was filled with 20 mM Mops buffer, pH 7.2,
the sample cell was filled with 1 mg/mL T. fusca BglC,
and the titration syringe was filled with 10 mM cellobiose
solution in 20 mM Mops, pH 7.2. The system was equili-
brated at 25
C with stirring at 290 rpm prior to the addi-
tion of 50 lL of the cellobiose solution. Before
termination of the experiment, the thermal power was
allowed to return to the level matching the original base-
line, indicating complete hydrolysis. The area under the
baseline (Fig. 1) was integrated and divided by the total
amount of substrate added to the cell to determine the
molar enthalpy of reaction. The final molar enthalpy value
used in the kinetic assays was obtained from averaging
three independent experiments.

Calorimetric enzyme kinetics assays

All reactions were carried out at 25
C in 20 mM Mops
buffer, pH 7.2. The reference cell was filled with buffer, the
sample cell was filled with the enzyme solution, and the syr-
inge was filled with cellobiose solution. All solutions and
Fig. 2. ITC assay of wild-type T. fusca BglC with cellobiose. Reaction
conditions are given in Table 1. Reaction rates were calculated from the
thermal power averaged over 15 s (of the typically 2-min intervals) prior to
the subsequent titration. The solid curve shows the curve fit to Eq. (3).
Inset shows the raw thermogram recorded during the experiment.

Table 1
Reaction parameters for assaying the kinetics of T. fusca BglC kinetics in the ITC
Mutant Concentration of enzyme Amount of cellobiose Amount of time between
in sample cell (lg/mL) per titration (nmol) titrations (s)
Wild-type 1.0 200 120
N439S 0.5 200 120
L444F 2.0 100 120
N317Y 1.0 500 720
A433V 1.0 200 120
S319C 0.25 250 120
N178I 0.5 100 120–180
N317Y/N439S (Com1) 2.0 100 120
N317Y/L444F (Com2) 1.0 100 120
N317Y/L444F/A433V (Com5) 1.0 250 120
 b-D-Glucosidase kinetics from titration calorimetry / T. Jeoh et al. / Anal. Biochem. 347 (2005) 244–253
b-D-glucosidase. The data were then fit to Eq. (3) to obtain
estimates of the half saturation constant, Km, and the reac-
tion rate constant, kcat. Curve fitting was conducted in
KaleidaGraph ver. 3.6 (Synergy Software), which utilizes
the Levenberg–Marquardt algorithm.
Colorimetric (conventional) enzyme kinetic assays
Initial velocities on cellobiose were determined by the
direct measurement of glucose formation using the Trinder
reaction [17] in which the end product is a quinoneimine
dye measured spectrophotometrically at 500 nm. Reaction
mixtures were prepared in 20 mM Mops buffer, pH 7.0,
with 0.068 to 8.5 mM cellobiose and 0.2 lg enzyme in
1.0-mL final reaction volumes. The reactions were incubat-
ed at 40
C for 10 min and stopped by boiling for 7 min,
followed by 8 to 10 min on ice. To assay for glucose,
100 lL of each reaction was transferred to a microtiter
plate, incubated at 38
C for 7 min, and followed by the
addition of 100 lL Stanbio Glucose Liquicolor reagent
(Stanbio Laboratory). The plate was then incubated at
38
C for 10 min and read at 490 nm in a Spectramax190
plate reader (Molecular Devices). Glucose concentrations
were quantitated against a glucose standard curve in the
range of 0.005 to 0.05 mM assayed under the same condi-
tions as those of the reaction samples.
Differential scanning calorimetry (DSC)
DSC thermograms were obtained by means of a VP-
DSC scanning microcalorimeter (Microcal) operated at a
scan rate of 60
C/h. Protein loading was typically 50 lg/
mL in 20 mM Bis–Tris, pH 6.2 (adjusted at 23
C), con-
taining 100 mM NaCl.
Results
Molar enthalpy of cellobiose hydrolysis
For the experiments to determine the molar enthalpy of
cellobiose hydrolysis, a high concentration of BglC (1 mg/
mL) was used to ensure complete hydrolysis of the sub-
strate in the reaction cell. The hydrolysis reaction is exo-
thermic, thus producing a large negative deflection in
thermal power (Fig. 1). The return of thermal power to
the level matching the initial baseline indicates complete
hydrolysis of cellobiose in the sample cell. The area under
the baseline caused by the deflection of thermal power is
equivalent to the total heat produced in hydrolyzing the
total amount of cellobiose. The molar enthalpy of cellobi-
ose hydrolysis, therefore, is the total heat produced divided
by the total moles of cellobiose hydrolyzed. There is no net
change in proton concentration in the reaction at the cur-
rent experimental conditions since the product, glucose, is
a weak acid with a pKa = 12.3 at 25
C [18]. Thus the
experimental molar enthalpy measured here is independent
of the heat of ionization of the buffer. The average of three
247
replicates gave a molar enthalpy (DHapp,CB) of
627.8 ± 5.25 cal/mol or 2.63 ± 0.022 kJ/mol. This val-
ue is in good agreement with a previously reported value
of 2.43 ± 0.31 kJ/mol [19]. For additional comparison,
the molar enthalpy of the hydrolysis of gentiobiose, a b-
(1 fi 6) isomer of cellobiose, has been reported to be
2.26 ± 0.48 kJ/mol [19] and the value for chitobiose, a
cellobiose analogue composed of N-acetyl glucosamine
monomers, has been reported to be 2.44 ± 0.05 kJ/mol
[12].
Assaying the kinetics of wild-type and mutant T. fusca BglC
on cellobiose
The critical criteria for obtaining data that can be pro-
cessed using the pseudo steady state assumption are (1)
high substrate to enzyme ratio and (2) substrate titrations
ranging from sufficiently below saturation to near satura-
tion. The thermogram of the kinetics of wild-type BglC at
25
C is shown in the inset of Fig. 2. Each titration of
200 nmol of cellobiose resulted in an initial deflection
due to the heat of mixing, with the tracing then rising
to a plateau at pseudo steady state. Even for the initial
titration ([S] = 0.14 mM), which is below saturation, the
difference between thermal power maintained after addi-
tion of the substrate and that at baseline was very small
(Fig. 2). This is probably due to the low BglC concentra-
tion in the reaction cell to limit the extent of reaction cou-
pled with low molar enthalpy of cellobiose hydrolysis. By
applying Eq. (2), the average thermal power over the 15 s
prior to the subsequent substrate titration is converted to
the rate of cellobiose hydrolysis. Fig. 2 shows the satura-
tion curve obtained from the raw thermogram. The data
shown in the saturation curve were fit to Eq. (3) to obtain
estimates of Km = 0.22 ± 0.02 mM and kcat = 16.3 ±
0.16 s1. The Km value agrees well with previous estimates
of 0.35 and 0.19 mM reported by Spiridonov and Wilson
[8] and Jarvis and co-workers [6], respectively. The kcat
value is also consistent with that obtained by Jarvis
et al. [6] of 10.4 s1.
The nine mutants of T. fusca BglC, all previously select-
ed as having higher thermal tolerance than the wild-type
enzyme, were next tested for activity on cellobiose. The
reaction concentrations used in the experiments (Table 1)
were initially optimized for the wild-type enzyme and
adjusted as needed for each mutant to maximize signal to
noise ratio. For each of the mutants assayed, the concen-
trations established for the wild-type assay were used in
the first run. For the mutants with lower turnover rates,
however, higher enzyme concentrations had to be used to
obtain measurable differences in thermal power between
steady states. Conversely, lower enzyme concentrations
had to be used for the highly active mutants to minimize
hydrolysis extents and maintain high substrate concentra-
tions in the cell. Parameter estimates obtained from fitting
Eq. (3) to the saturation curves of each enzyme are given in
Table 2.
248 b-D-Glucosidase kinetics from titration calorimetry / T. Jeoh et al. / Anal. Biochem. 347 (2005) 244–253

Table 2
Kinetic parameters of T. fusca BglC thermal tolerant mutants on cellobiose at 25
C estimated by fitting Eqs. (3) and (6) to saturation curves obtained in
the ITC
Mutant No substrate inhibition With substrate inhibitiona
Km (mM) kcat (s1) Km (mM) kcat (s1) K 0I ðmMÞ kcat2 (s1)
Wild-type 0.22 (0.02) 16.3 (0.16) 0.66 (0.02) 24.2 (0.39) 2.30 (0.27) 11.8 (0.28)
N439S 0.23 (0.03) 29.5 (0.60) 0.91 (0.06) 56.7 (2.2) 1.67 (0.31) 13.6 (1.6)
L444F 0.31 (0.03) 1.6 (0.03) 1.91 (0.55) 4.5 (0.96) 0.55 (0.29) 0.9 (0.12)
N317Y 0.21 (0.02) 14.0 (0.16) 0.39 (0.03) 17.0 (0.38) 1.95 (1.0) 11.3 (0.86)
A433V 0.19 (0.02) 12.3 (0.15) 0.64 (0.02) 19.0 (0.28) 2.83 (0.28) 8.0 (0.22)
S319C 0.40 (0.04) 111.3 (1.7) 1.37 (0.03) 198.6 (2.3) 3.75 (0.29) 42.2 (3.2)
N178I 0.63 (0.03) 22.3 (0.31) 0.99 (0.22) 17.0 (5.8) 0.40 (0.08) 18.5 (0.61)
N317Y/N439S (Com1) 0.66 (0.04) 4.3 (0.09) 4.14 (1.5) 11.8 (3.9) 0.16 (0.07) 3.2 (0.15)
N317Y/L444F (Com2) 0.41 (0.02) 16.8 (0.18) 0.80 (0.04) 16.3 (1.2) 0.31 (0.02) 14.1 (0.14)
N317Y/L444F/ A433V (Com5) 1.01 (0.04) 19.2 (0.17) 1.76 (0.06) 24.7 (0.6) 6.49 (2.4) 12.3 (1.8)
Standard errors of parameter estimates are in parentheses.
a
Fitted to Eq. (6).

Determining the true reaction mechanism of T. fusca BglC
on cellobiose
Upon close inspection of the curve fits, it is apparent
that Eq. (3) does not adequately describe the data obtained
by the ITC kinetics assay. Fig. 3 showcases BglC A433V,
one example from this set of mutants where Eq. (3) was
not able to fit the saturation data (solid line in Fig. 3A).
In fact, the curve-fitting attempt emphasizes the decrease
in reaction rate at higher substrate loadings, a pattern typ-

Fig. 3. ITC assay of T. fusca BglC mutant A433V on cellobiose with evidence for substrate inhibition. (A) Calculated reaction rates with curve fits to Eq.
(3) (solid line), Eq. (4) (dotted line), and Eq. (6) (dashed line) shown. Inset shows the raw thermogram obtained from kinetic assay in the ITC (reaction
conditions are as described under Materials and methods and Table 1). (B) Lineweaver–Burk plot showing the inverse of d[P]/dt (1/V) with respect to the
inverse of the initial substrate concentration (1/S). Solid line indicates linear regression to the data set. (C) Residuals from v2 minimized fits of Eq. (3)
(filled circles), Eq. (4) (open triangles), and Eq. (6) (solid squares) to the saturation curve in A (lines are drawn only to guide the eye).
 b-D-Glucosidase kinetics from titration calorimetry / T. Jeoh et al. / Anal. Biochem. 347 (2005) 244–253 249

ical of substrate inhibition. The contribution of substrate

inhibition to the kinetics is further evident from the nonlin-
ear Lineweaver–Burk plot (Fig. 3B).
According to the model that we have adopted to explain
the data, substrate inhibition occurs by an uncompetitive
mechanism, in which a second substrate molecule binds
reversibly to the enzyme–substrate complex. The resulting
SES complex may be incapable of breaking down to form
enzyme and product, as shown in Mechanism 2.
k3
ðMechanism 2Þ ES þ S () SES.
k4

Researchers have observed, however, that in a number of

cases the SES complex is, in fact, capable of breaking down
to product (Mechanism 3), albeit with a rate altered with
respect to that of the rate of breakdown of ES [20,21]:
k3 k cat2
ðMechanism 3Þ ES þ S () SES ! ES þ P.
k4

The noncatalytic inhibition mechanism, Mechanism 2, and

the catalytic inhibition mechanism, Mechanism 3, were Fig. 4. Goodness-of-fit parameters (minimized v2 values) for fits of Eq. (3)
tested for ability to fit the saturation data obtained by (white bars), Eq. (4) (striped bars), and Eq. (6) (cross-hatched bars) to ITC
kinetic data of wild-type T. fusca BglC and the nine thermal tolerant
ITC for the BglC mutants. Incorporating Mechanism 2
mutants on cellobiose. The maximum of the graph on the y axis was
with Mechanism 1 results in the rate equation truncated at 2.0e9 to focus on comparisons between the fits using Eqs. (4)
d½ P
V max ½S
and (6).
¼  2 ; ð4Þ
dt K m þ ½S
þ ½KS
I
 
where K I ¼ kk43 is the inhibition constant (mM). The fit appears to further improve with Eq. (6), suggesting
To obtain a simplified approximation to the reaction that the SES complex does indeed remain catalytically
rate for the catalytic inhibition mechanism (Mechanism active. The residuals from the three curve fits (Fig. 3C) con-
3), the pseudo steady state assumption must hold for both firm the poor fit of Eq. (3), which shows large residuals, with
the SES complex and the ES complex. The raw data a definite dependence on substrate concentration. The resid-
(Fig. 3A, inset) shows plateaus in the power curve (dQ/ uals from fitting Eqs. (4) and (6) both scatter about zero.
dt) prior to each substrate titration, thus confirming that A comparison of the goodness-of-fit (minimized v2 val-
the reaction does achieve steady state as stated in Eq. (5): ues) obtained from fitting all three equations (Eqs. (3),
(4), and (6)) to the saturation curves for the wild-type BglC
d½P
and the nine mutants (Fig. 4) revealed that the best fits for
¼ k cat ½ES
þ k cat2 ½SES
¼ constant at steady state.
dt the enzymes were achieved by Eq. (6). In some cases (such
ð5Þ as that of the mutants BglC L444F and BglC N317Y), the
improvements in v2 values from fitting Eq. (6) over Eq.
The following rate equation incorporates Mechanism 3 (4) were small. To determine whether the additional param-
with Mechanism 1, describing the overall reaction rate eter in Eq. (6) is justified, an F test comparing fits to the two
when the inhibited enzyme (SES) continues to be catalyti- equations (Eqs. (4) and (6)) was conducted [22]. The mini-
cally active: mum calculated Fv statistic was 5.2, thus confirming that,
 2 for all the BglC enzymes, the fit to Eq. (6) was significantly
0 ½S

d½ P
V max ½S
þ V max K0
¼  2 I ; ð6Þ (p > 0.99) improved over the fit to Eq. (4) [22]. This analysis
dt K m þ ½S
þ ½S
0 therefore revealed that the BglC mutants and the wild-type
KI
enzyme are subject to substrate inhibition and that the SES
0
where V max¼ k cat2 ½Et
is the maximum rate of Mechanism 3 complexes are catalytic. The parameters estimated from fit-
(mM/s) and K 0I ¼ k4 þk k3
cat2
is the inhibition constant (mM) ting Eq. (6) to the ITC kinetic assay data of the set of ther-
describing binding of a second molecule of substrate to ES mal stable BglC mutants are presented in Table 2.
to form a productive (i.e., product-forming) SES.
Fig. 3A shows the fit of Eqs. (3), (4), and (6) to the satu- Extents of substrate inhibition
ration curve of BglC A433V. Visual inspection of Fig. 3A
shows that Eq. (4), representing the noncatalytic substrate The extent to which substrate inhibition affects the activ-
inhibition mechanism (Mechanism 2), fits the data better ity of the enzyme can be quantitated as the reduction in the
than Eq. (3), where no substrate inhibition is assumed. overall product formation rate due to the inhibition reac-
250 b-D-Glucosidase kinetics from titration calorimetry / T. Jeoh et al. / Anal. Biochem. 347 (2005) 244–253

tion as characterized by Eq. (6). In the absence of substrate
inhibition, Eq. (3) describes the product formation rate.
Thus, the ratios of the rates for a given substrate concen-
tration can provide insight into the effect that the inhibition
mechanism has on the overall kinetics of the enzyme. The
ratio of Eq. (6) to Eq. (3) (using in both cases the values
for Km and kcat derived from fitting the data to Eq. (6))
gives the equation
n o
dP
j r  fk cat2 K m þ ½S
ðk cat2  k cat Þg kcat½S
K 0
dt Equation 6 i I
¼ ¼1þ
dP
j r K m þ ½S
þ 0 ½S
2
dt Equation 3 KI
ð7Þ
where ri is the product formation rate in the presence of
substrate inhibition (mM/s) and r is the product formation
rate predicted by the fit to Eq. (6) in the absence of sub-
strate inhibition (mM/s).
Note that if the SES complex is not catalytically active
(i.e., kcat2 = 0; Mechanism 2), Eq. (7) simplifies to the ratio
of Eq. (4) to Eq. (3), with KI substituted for K 0I . In Eq. (7),
the value of ri/r deviates further from 1.0 as the effect of
substrate inhibition increases.
As can be observed in Eq. (7), the effect of substrate
inhibition on the overall rate of product formation is
dependent on the substrate concentration present in the
reaction. At high substrate concentrations, however, the
ratio approaches a constant value. The ratios, calculated
at two substrate concentrations, 2.9 and 29 mM, are shown
in Fig. 5. The two concentrations correspond to 1 and 10 g/L
cellobiose, which represent ballpark concentrations that a
b-glucosidase enzyme typically encounters in biomass
depolymerization reactions [23,24]. In almost all cases,
incorporation of the substrate inhibition mechanism had
a negative effect on the rate of product formation. Three
exceptions are immediately obvious from Fig. 5, BglC
N178I, BglC N317Y/L444F, and BglC N317Y/L444F/
A433V, where the product formation rates were enhanced
by the substrate inhibition mechanism. In all three of
these cases, the inhibition effect is evident at lower sub-
strate concentrations. For BglC N317Y/L444F and BglC
;
N317Y/L444F/A433V, the ri/r is in fact less than one at
the higher substrate concentration. In general, the effect
of substrate inhibition on the reaction rates was more
severe at the higher substrate concentrations. The effect
of substrate inhibition leveled off beyond 30 mM for the
BglC enzymes.
Comparing the activities of T. fusca BglC mutants
The ratio of the catalytic constant, kcat, to the half sat-
uration constant, Km, is often used to compare the relative
effectiveness (or specificity for a given substrate) of
enzymes in soluble systems, with higher kcat/Km ratio indi-
cating a more effective/specific enzyme. The kcat/Km ratios
from estimates obtained from fitting Eqs. (4) and (6) are
shown in Fig. 6. This data set is compared with a data
set obtained from end-point colorimetric assays at 40
C.
Initial rate data from the colorimetric assays were also fit-
ted to Eq. (6) to estimate kinetic constants of the reaction
at the elevated temperature. This data set was more scat-
tered with fewer data and thus was more difficult to fit to

Fig. 6. Comparing activities of thermal stable BglC mutants based on the

Fig. 5. Extents of substrate inhibition affecting each T. fusca BglC mutant,
as quantitated as the ratio of ri, the product formation rate in the presence
of substrate inhibition (Eq. (6)), to r, the product formation rate in the
absence of substrate inhibition (Eq. 3), (ri/r) at 2.9 mM (black bars) and
29 mM (white bars) substrate concentrations. A dashed line is shown at
ri/r = 1.
ratio of kcat to Km estimated from fitting Eq. (6) to ITC data at 25
C
(white bars) and on data obtained from end-point colorimetric assays at
40
C (striped bars). The solid and dotted lines indicate the magnitudes of
the ratio for wild-type BglC as determined for the ITC data and for the
end-point assay data, respectively. The error bars for N439S are beyond
the range of the axis shown.
 b-D-Glucosidase kinetics from titration calorimetry / T. Jeoh et al. / Anal. Biochem. 347 (2005) 244–253
all three equations. Systematic curve-fitting analysis as
described for the ITC data determined that the catalytic
substrate inhibition model best fits the data obtained by
the end-point assays, as evidenced by the large error bars
in Fig. 6; however, the fitting procedure was not able to
obtain good estimates of the parameters. Despite the large
errors in the estimates obtained from the colorimetric
assay, we can draw three general conclusions from Fig. 6.
First, similar activity trends are observed for the BglC
mutants by the two methods; second, the activity changes
conferred by the mutations trend the same way at the
two temperatures (25
C for ITC and 40
C for the end-
point kinetics); third, the ITC method provides a stronger
data set for parameter estimation.
Since this set of enzymes is affected by substrate inhibi-
tion to varying degrees, it is insufficient to compare overall
activities on the basis of kcat and Km only. The kinetic
parameters obtained from the curve fits (Table 2) were used
to calculate overall product formation rates for each BglC
enzyme at three substrate concentrations (0.29, 2.9, and
29 mM) normalized to 1 mM enzyme loading (Fig. 7).
Two of the mutants, BglC N439S and BglC S319C, stand
out in Fig. 7 with the most significant activity improve-
ments over that of the wild-type enzyme at the three sub-
strate loadings. The activity improvement of BglC S319C
was an approximately four- to five fold improvement over
that of the wild-type BglC, and N439S showed up to two-
fold improvement. Conversely, two of the mutants, BglC
L444F and BglC N317Y/N439S, had significantly lower
activity than the wild-type enzyme. Generally, the activity
Fig. 7. Product formation rates (d[P]/dt) were calculated for the BglC
enzymes using Eq. (6). Parameter estimates given in Table 2 with a total
enzyme concentration of 1 mM were used for the calculations at three
substrate concentrations: 0.29 mM (white bars), 2.9 mM (striped bars),
and 29 mM (cross-hatched bars). The three substrate concentrations
correspond to 0.1, 1, and 10 g/L cellobiose, respectively.
251
of the enzymes improved with higher substrate loadings,
with some showing decreases at the highest loading of
29 mM as the negative effect of substrate inhibition
becomes apparent.
Discussion
Effect of substrate inhibition on T. fusca BglC mutants
Strong substrate inhibition of the b-D-glucosidases iso-
lated from other microorganisms, such as Aspergillus niger
[21,25,26] and Trichoderma viride [20], has previously been
reported. The effect of substrate inhibition on a plot of
reaction rate versus substrate concentration is typically
observed as a reduction in the reaction rate [15], especially
at higher substrate concentrations. In cases of severe sub-
strate inhibition, the slope of the curve becomes negative
at higher substrate concentrations. When the substrate
inhibition effect is subtle, however, there may not be an
obvious decrease in reaction rate at the higher substrate
concentrations. This was the case with some of the T. fusca
BglC mutants tested in this study. The effect of substrate
inhibition on the reaction mechanisms of BglC was deter-
mined by regression analysis of ITC kinetics data. Based
on kinetics data obtained at 40
C by an end-point color-
imetric assay, BglC is similarly affected by substrate inhi-
bition at the higher temperature. In one case, A. niger
b-D-glucosidase was reported to be substrate inhibited
up to 60
C, with a similar assay at 70
C showing no
signs of inhibition [21]. Since wild-type BglC appears to
have a temperature optimum at 50
C [8], further analy-
sis of the influence of temperature on substrate inhibition
of this enzyme might be of interest.
Is there a correlation between thermal tolerance and activity
of BglC?
Improvements in the thermal stability of enzyme cata-
lysts using directed evolution is a common goal of protein
engineering, with single mutations generally conferring
Tmax increases in the range of 1 to 2
C [6,27]. In general,
the combination of individual stabilizing mutations is
reported to be additive [6,27]. The current study gives
emphasis to the notion that, while individual mutations
may confer higher thermal stability, there may or may
not be an overall performance advantage over the wild-
type enzyme when a careful kinetics analysis is carried
out. Based on activity comparisons (Figs. 6 and 7), the
mutant BglC S319C appeared to show marked improve-
ment relative to the wild-type form on cellobiose. However,
in the initial screens, the thermal stability improvement of
this enzyme was found to be very limited, with only 1 to
2
C improvements [6]. In contrast, BglC Com5 (N317Y/
L444F/A433V), the mutant that showed the most promise
as an improved, thermal tolerant form of BglC by virtue of
its 7 to 8
C improvement [6], showed limited activity
improvements (Fig. 7).
252 b-D-Glucosidase kinetics from titration calorimetry / T. Jeoh et al. / Anal. Biochem. 347 (2005) 244–253
Fig. 8. Comparing activity changes (quantitated as the product formation
rate of the mutant normalized to that of the wild-type BglC enzyme)
conferred by the BglC mutations to the denaturation temperatures
determined by DSC. Calculations for three substrate loadings, 0.29 mM
(filled squares), 2.9 mM (open diamonds), and 29 mM (open triangles), are
shown. Com1, N317Y/N439S; Com2, N317Y/L444F; Com5, N317Y/
L444F/A433V. A dashed line is shown at r/rwild-type = 1.
Jarvis and co-workers [6] reported denaturation temper-
atures of wild-type BglC and three of the BglC mutants,
N317Y, L444F, and Com2, which incorporate the two sin-
gle mutations. Differential scanning calorimetry analyses of
the remaining mutants were conducted for this study to
compare denaturation temperatures with the activities of
the entire set of BglC mutants. The activities of the mutants
normalized to that of the wild-type enzyme (ratios of prod-
uct formation rates calculated at 1 mM enzyme loading)
were plotted against the denaturation temperatures
(Fig. 8), where a value of one for r/rwild-type indicates no
change in the activity of the mutant as compared to the
wild-type BglC. No correlation was observed between
denaturation temperature and activity of the mutants.
According to the plot in Fig. 8, there are two BglC mutants
(N439S and S319C) that show promise as forms of BglC
with improved thermal tolerance and cellobiase activity.
BglC S319C stands out as the mutant with the greatest
improvement for both thermal tolerance and activity. This
mutant had 4.5- to 6.5-fold improvement in activity as
compared to the wild-type enzyme at 25
C. Coupled with
a 4.2
C improvement in denaturation temperature, this
mutant appears to be a promising alternative to the wild-
type BglC enzyme. The data in Fig. 8 reemphasize the
importance of the substrate concentration at which the
enzymes are being compared. For example, the activity of
the triple mutant Com5 (N317Y/L444F/A433V) with the
highest denaturation temperature of 63.2
C (a 5
C
improvement over that of wild type) was minimally affected
at two of the substrate loadings. However, at 0.29 mM sub-
strate concentration, the activity of Com5 was less than
half that of the wild-type BglC. The Com5 mutant, there-
fore, remains an attractive choice for applications requiring
elevated operating temperatures as long as the substrate
concentration within the reaction is maintained at higher
levels.
The advantages of using ITC to compare activities of the
BglC mutants
As demonstrated in this study, the ITC method offered
significant advantages for obtaining data for a large num-
ber of purified b-D-glucosidases. The kinetic assays used
were continuous with real-time data acquisition, which
allowed monitoring the progress of the reaction during
the assay. [The solutions to the kinetic mechanisms
applied in this paper could not be obtained in the simpli-
fied forms presented in Eqs. (3), (4), and (6) without
applying the assumption that the reaction is at steady
state between substrate additions]. In our view, the raw
data obtained from the ITC experiments verify the valid-
ity of making this particular assumption. In conventional
kinetic enzyme assays, several control experiments have to
be conducted to determine that the reaction is indeed at
steady state at the assay reaction times. To be rigorous,
the control experiments must be repeated for each mutant
tested. For the ITC experiments described here, the steady
state can be observed and verified during the experiment
itself. ITC, therefore, is an invaluable tool for studying
the kinetics of any soluble enzyme system. In addition,
large degrees of freedom due to the large number of mea-
surements possible improve parameter estimates in subse-
quent data regression. Furthermore, subtle substrate
inhibition exhibited for some of the thermal tolerant BglC
mutants would most likely have been undetected with
fewer data points, where a poorer fit could be attributed
to scatter in the data.
Acknowledgments
The authors thank Dr. David Wilson for providing the
gene-coding T. fusca BglC, Dr. A. Brad Anton for critical
analysis of the data reduction methodology, and the U.S.
DOE Office of the Biomass Program for funding this work.
References
[1] F. Teymouri, L. Laureano-Perez, H. Alizadeh, B.E. Dale, Ammonia
fiber explosion treatment of corn stover, Appl. Biochem. Biotechnol.
113 (16) (2004) 951–963.
[2] T.H. Kim, J.S. Kim, C. Sunwoo, Y.Y. Lee, Pretreatment of corn
stover by aqueous ammonia, Bioresource Technol. 90 (2003) 39–47.
[3] S. Takashima, A. Nakamura, M. Hidaka, H. Masaki, T. Uozumi,
Molecular cloning and expression of the novel fungal beta-glucosi-
dase genes from Humicola grisea and Trichoderma reesei, J. Biochem.
125 (1999) 728–736.
[4] S. Wolosowska, J. Synowiecki, Thermostable beta-glucosidase with a
broad substrate specifity suitable for processing of lactose-containing
products, Food Chem. 85 (2004) 181–187.
[5] P. Murray, N. Aro, C. Collins, A. Grassick, M. Penttila, M.
Saloheimo, M. Tuohy, Expression in Trichoderma reesei and charac-
terisation of a thermostable family 3 beta-glucosidase from the
moderately thermophilic fungus Talaromyces emersonii, Protein Expr.
Purif. 38 (2004) 248–257.
 b-D-Glucosidase kinetics from titration calorimetry / T. Jeoh et al. / Anal. Biochem. 347 (2005) 244–253
[6] E.E. Jarvis, W.S. Adney, S.R. Decker, J.O. Baker, M.E. Himmel, in:
W.P.M. Van Swaaij, T. Fjallstrom, P. Helm, A. Grassi (Eds.), Second
World Biomass Conference: Biomass for Energy, Industry and
Climate Protection, Rome, Italy, 2004, pp. 1512–1516.
[7] S.G. Burton, D.A. Cowan, J.M. Woodley, The search for the ideal
biocatalyst, Nat. Biotechnol. 20 (2002) 37–45.
[8] N. Spiridonov, D.B. Wilson, Cloning and biochemical characteriza-
tion of BglC, a beta-glucosidase from the cellulolytic actinomycete
Thermobifida fusca, Curr. Microbiol. 42 (2001) 295–301.
[9] M.J. Blandamer, P.M. Cullis, J. Engberts, Titration microcalorime-
try, J. Chem. Soc. Faraday Trans. 94 (1998) 2261–2267.
[10] P.E. Morin, E. Freire, Direct calorimetric analysis of the enzymatic
activity of yeast cytochrome c oxidase, Biochemistry 30 (1991) 8494–
8500.
[11] M.J. Todd, J. Gomez, Enzyme kinetics determined using calorimetry:
a general assay for enzyme activity? Anal. Biochem. 296 (2001) 179–
187.
[12] T. Lonhienne, E. Baise, G. Feller, V. Bouriotis, C. Gerday, Enzyme
activity determination on macromolecular substrates by isothermal
titration calorimetry: application to mesophilic and psychrophilic
chitinases, Biochim. Biophys. Acta 1545 (2001) 349–356.
[13] N. Karim, S.-I. Kidokoro, Precise and continuous observation of
cellulase-catalyzed hydrolysis of cello-oligosaccharides using iso-
thermal titration microcalorimetry, Thermochim. Acta 412 (2004)
91–96.
[14] L. Michaelis, M.L. Menten, Biochem. Z. 49 (1913) 333–369.
[15] M.L. Shuler, F. Kargi, Bioprocess Engineering: Basic Concepts,
Prentice-Hall P T R, Englewood Cliffs, NJ, 1992.
[16] E. Gasteiger, C. Hoogland, A. Gattiker, S. Duvaud, M.R. Wilkins,
R.D. Appel, A. Bairoch, in: J.M. Walker (Ed.), The Proteomics
Protocols Handbook, Humana Press, Totowa, NJ, 2005.
253
[17] P. Trinder, Determination of glucose in blood using glucose oxidase
with an alternative oxygen acceptor, Ann. Clin. Biochem. 6 (1969) 24.
[18] J.A. Dean (Ed.), LangeÕs Handbook of Chemistry, McGraw-Hill,
New York, 1999.
[19] Y.B. Tewari, R.N. Goldberg, Thermodynamics of hydrolysis of
disaccharides—cellobiose, gentiobiose, isomaltose, and maltose, J.
Biol. Chem. 264 (1989) 3966–3971.
[20] J. Hong, M.R. Ladisch, C.S. Gong, P.C. Wankat, G.T. Tsao,
Combined product and substrate-inhibition equation for cellobiase,
Biotechnol. Bioeng. 23 (1981) 2779–2788.
[21] V. Bravo, M.P. Paez, A.A. Reyes, The influence of temperature upon
the hydrolysis of cellobiose by beta-1,4-glucosidases from Aspergillus
niger, Enzyme Microb. Technol. 26 (2000) 614–620.
[22] P.R. Bevington, D.K. Robinson, Data Reduction and Error Analysis
for the Physical Sciences, WCB/McGraw-Hill, New York, 1992.
[23] P. Ghosh, N.B. Pamment, W.R.B. Martin, Simultaneous saccharifi-
cation and fermentation of cellulose-effect of beta-D-glucosidase
activity and ethanol inhibition of cellulases, Enzyme Microb. Tech-
nol. 4 (1982) 425–430.
[24] G.P. Philippidis, T.K. Smith, C.E. Wyman, Study of the enzymatic-
hydrolysis of cellulose for production of fuel ethanol by the
simultaneous saccharification and fermentation process, Biotechnol.
Bioeng. 41 (1993) 846–853.
[25] W. Grous, A. Converse, H. Grethlein, L. Lynd, Kinetics of cellobiose
hydrolysis using cellobiase composites from Trichoderma reesei and
Aspergillus niger, Biotechnol. Bioeng. 27 (1985) 463–470.
[26] L.P.V. Calsavara, F.F. De Moraes, G.M. Zanin, Modeling cellobiose
hydrolysis with integrated kinetic models, Appl. Biochem. Biotechnol.
77 (9) (1999) 789–806.
[27] O. Kuchner, F.H. Arnold, Directed evolution of enzyme catalysts,
Trends Biotechnol. 15 (1997) 523–530.