Báo cáo khoa học: The natural mutation by deletion of Lys9 in the thrombin A-chain affects the pKa value of catalytic residues, the overall enzyme’s stability and conformational transitions linked to Na+ binding pdf

The natural mutation by deletion of Lys9 in the
thrombin A-chain affects the pKa value of catalytic
residues, the overall enzyme’s stability and conformational
transitions linked to Na+ binding
Raimondo De Cristofaro1, Andrea Carotti2, Sepideh Akhavan3,*, Roberta Palla3, Flora Peyvandi3,
Cosimo Altomare2 and Pier Mannuccio Mannucci3
1 Haemostasis Research Centre, Institute of Internal Medicine and Geriatrics, Catholic University School of Medicine, Rome, Italy
2 Department of Pharmaceutical Chemistry, University of Bari, Italy
3 Angelo Bianchi Bonomi Hemophilia and Thrombosis Center and Fondazione Luigi Villa, IRCCS Maggiore Hospital University of Milan, Italy

Keywords
allostery; molecular dynamics; pKa values;
stability; thrombin
Correspondence
R. De Cristofaro, Haemostasis Research
Centre, Institute of Internal Medicine and
Geriatrics, Catholic University School of
Medicine, Largo F. Vito 1, 00168 Rome,
Italy
Fax: +39 6 30 155 915
Tel: +39 6 30 154 438
E-mail:
C. Altomare, Department of Pharmaceutical
Chemistry, University of Bari, Via E.
Orabona 4, 70125 Bari, Italy
Fax: +39 80 544 2230
Tel: +39 80 544 2781
E-mail:
*Present address
´
INSERM E0348, Faculte Xavier Bichat,

University Paris 7, France
(Received 2 September 2005, revised 12
October 2005, accepted 7 November 2005)
doi:10.1111/j.1742-4658.2005.05052.x

The catalytic competence of the natural thrombin mutant with deletion of
the Lys9 residue in the A-chain (DK9) was found to be severely impaired,
most likely due to modification of the 60-loop conformation and catalytic
triad geometry, as supported by long molecular dynamics (MD) simulations in explicit water solvent. In this study, the pH dependence of the
catalytic activity and binding of the low-molecular mass inhibitor N-a-(2naphthylsulfonyl-glycyl)-4-amidinophenylalanine-piperidine (a-NAPAP) to
the wild-type (WT) and DK9 thrombin forms were investigated, along with
their overall structural stabilities and conformational properties. Two ionizable groups were found to similarly affect the activity of both thrombins.
The pKa value of the first ionizable group, assigned to the catalytic His57
residue, was found to be 7.5 and 6.9 in ligand-free DK9 and WT thrombin,
respectively. Urea-induced denaturation studies showed higher instability
of the DK9 mutant compared with WT thrombin, and disulfide scrambling
experiments proved weakening of the interchain interactions, causing faster
release of the reduced A-chain in the mutant enzyme. The sodium ion binding affinity was not significantly perturbed by Lys9 deletion, although the
linked increase in intrinsic fluorescence was lower in the mutant. Essential
dynamics (ED) analysis highlighted different conformational properties of
the two thrombins in agreement with the experimental conformational
stability data. Globally, these findings enhanced our understanding of the
perturbations triggered by Lys9 deletion, which reduces the overall stability
of the molecule, weakens the A–B interchain interactions, and allosterically
perturbs the geometry and protonation state of catalytic residues of the
enzyme.

Recently, a homozygous deletion mutation of one of the
two contiguous Lys9 ⁄ Lys10 residues in the A-chain of
a-thrombin (DK9) was identified in patients with severe

prothrombin deficiency and hemorrhagic diathesis [1,2].

Compared with the wild-type (WT) form, the specificity
constants of hydrolysis by DK9 of the synthetic
substrate d-Phe-Pip-Arg-pNA and fibrinopeptide A
were found to be 18- and 60-fold lower, respectively.

Abbreviations
a-NAPAP, N-a-(2-naphthylsulfonyl-glycyl)-4-amidinophenylalanine-piperidine; Bis-Tris, (2-hydroxyethyl)iminotris(hydroxymethyl)methane;
CHES, 2-(N-cyclohexylamino)ethanesulfonic acid; ED, essential dynamics; DK9, Lys9 deleted mutant; MD, molecular dynamics; Pip,
pipecolyl; pNA, para-nitroanilide; SAS, solvent-accessible surface; WT, wild-type.

FEBS Journal 273 (2006) 159–169 ª 2005 FEBS No claim to original US government works

159


Functions of the thrombin A-chain

R. De Cristofaro et al.

Interaction with antithrombin was also reduced in the
mutant, the association rate being % 20-fold lower than
in WT thrombin. Moreover, DK9 showed very weak
platelet-activating capacity, whereas binding to the
platelet glycoprotein Iba and thrombomodulin was
unaffected. At variance with these findings, inhibitors
showed better binding to DK9 than to the WT form. A
long-term molecular dynamics (MD) simulation of DK9
thrombin in explicit water solvent supports the role of

the A-chain in affecting the conformation and catalytic
properties of the B-chain, particularly in some insertion
loops of the enzyme, such as the 60-loop, as well as in
the geometry of the catalytic triad residues. Our MD
analysis highlighted relevant modifications within the
so-called ‘aryl-binding site’, in particular, expulsion ⁄
rearrangement of the W60d side chain (S2 site) and
shifting of W215 (S3). Functional and computational
data show that the catalytic cycle and efficient interaction with substrates and natural inhibitors by DK9
undergo a severe impairment, likely due to propagation
to the active site residues of structural and conformational perturbations caused by Lys9 deletion in the
A-chain [2].
These findings prompted us to further investigate
the pH dependence of the catalytic activity and stability of the DK9 mutant in comparison with WT thrombin, using experimental techniques in conjunction with
computational approaches to prove the effects of the
K9 deletion on the ionization of catalytic residues and
the overall stability of the enzyme. This investigation
contributes to the unraveling of the mechanisms
responsible for both the impaired catalytic activity of

the K9-deleted natural mutant of thrombin in vitro
and its hemorrhagic phenotype in vivo.

Results and Discussion
Effect of pH on thrombin catalytic activity and
inhibition
The pH-dependent steady-state amidase activities of
WT and K9-deleted mutant thrombins were studied in
the pH range 5.5–10, using previously reported experimental and theoretical approaches [3,4]. Kinetic
schemes and equations (Eqns 2–4), allowing the effects

of pH on the Michaelis–Menten parameters kcat and Km
to be assessed, are reported in Experimental Procedures.
Although protons globally affected the amidase
activity of both WT and DK9 thrombin forms in a
similar way (Fig. 1A–C), the pKa values of the ionizable groups involved in the catalytic cycle were found
to be significantly different. As reported in Table 1, we
found an appreciable increase in the pKa value of the
first ionizable group, which in the free enzyme showed
a pKa value of 7.53 in the DK9 mutant and 6.86 in
WT thrombin. Because previous studies have assigned
this group to the active site His57 side chain [3–5], this
finding suggests that Lys9 deletion allosterically affects
the protonation equilibrium of the active site His57,
enhancing its affinity for protons both in free and substrate-bound forms of the enzymes. The second pKa
value was assigned to the N-terminal group of Ile16
(NTIle), which holds Asp194 in a salt bridge as a
result of zymogen activation [3,5,6]. The NTIle pKa

Fig. 1. Analysis of pH dependence of
Michaelis–Menten constants of D-Phe-PipArg-pNA hydrolysis (A–C) by WT (d) and
DK9 thrombin (s), along with the Ki values
of NAPAP binding (D) at 25 °C and 0.15 M
NaCl. Continuous lines were drawn according to the best-fit parameters values of
Eqns (2–4) and listed in Table 1. The vertical
bars are the standard errors of the determinations.

160

FEBS Journal 273 (2006) 159–169 ª 2005 FEBS No claim to original US government works

R. De Cristofaro et al.

Functions of the thrombin A-chain

Table 1. Best-fit pKa values of the ionizable groups of both WT and DK9 mutant thrombins (A), along with the best-fit kinetic parameters
contained in Eqns (2–4) involved in the hydrolysis of the synthetic substrate D-Phe-Pip-Arg-pNA (B) at 25 °C in the presence of 0.15 M NaCl.
The best-fit pKa values of the ionizable groups of both WT and DK9 mutant thrombins, calculated using Eqn (2), are reported in C.
(A)
Enzymes

Group 1

Group 2

Free

kcat s)1

Free

Bound

6.86 ± 0.06
7.53 ± 0.12

WT
DK9

Bound

6.30 ± 0.06
7.13 ± 0.09

8.45 ± 0.06
8.87 ± 0.09

9.04 ± 0.05
9.36 ± 0.12

kcat s)1

(B) Enzymes

0

1

WT
DK9

31.9 ± 4
1.02 ± 0.4

76.3 ± 3
3.82 ± 0.3

(C)
Enzymes

2

kcat s)1

32 ± 4
2.01 ± 0.2

°Km lM

r0 · 106 (M)1Ỉs)1)

r1 · 106 (M)1Ỉs)1)

r2 · 106 (M)1Ỉs)1)

6.6 ± 0.3
3.6 ± 0.3

5 ± 0.6
0.31 ± 0.04

42 ± 3
3.1 ± 0.3

5.3 ± 0.6
0.69 ± 0.07

Group 1

Group 2

Free

a

Ki° ¼ 3.9 ± 0.10 nM.

b

Free

Bound

6.97 ± 0.05
7.53 ± 0.05

WTa
DK9b

Bound
6.52 ± 0.05
6.59 ± 0.05

8.63 ± 0.05
8.99 ± 0.06

9.00 ± 0.05
9.85 ± 0.07

Ki° ¼ 2.7 ± 0.13 nM.

value in the mutant thrombin undergoes a moderate
increase from 8.45 to 8.87 in the free enzyme and from
9.04 to 9.36 in the substrate-bound form, for WT
thrombin and DK9 mutant, respectively.
Table 1B reports changes in the kcat and kcat ⁄ Km
values at the three protonation levels of WT and DK9
thrombins. The kcat and kcat ⁄ Km values show a drastic
decrease, mostly due to a net fall in the kcat value,
which expresses the acylation rate. By contrast, the
decrease in the DK9 Km value is consistent with better
accommodation of the substrate into the catalytic
pocket of the unprotonated form of the mutant, as
shown recently [2].
Interestingly, the log kcat ⁄ Km values pertaining to
WT and DK9 forms with protonated His57 (r1 of
Eqn 4) in the presence of 0.15 m NaCl were inversely
related to the respective His57 pKa values. Based on a
Brønsted mechanism [7], this observed relation is consistent with a transition state for breakdown of the tetrahedral intermediate involving partial carbon–nitrogen
(C–N) bond cleavage, which is stabilized by H-bonding
to the His57 imidazolium Ne nitrogen (see Experimental
procedures Scheme 2). The imidazolium form would act
as a general acid to facilitate amine expulsion (kA) from
the tetrahedral intermediate (TI). Such a conclusion is
in agreement with recent findings obtained with proton
inventory studies of a-thrombin-catalyzed hydrolysis of
amide substrates [6]. Thus, in human thrombin the ratelimiting step in the acylation reaction is the breakdown
of tetrahedral intermediate, being the different affinity
for protons of His57 imidazole Ne nitrogen inversely
related to the specificity constant of the amidase activity
of the two thrombin forms.

According to our recent report, DK9 compared with
WT thrombin achieves a better interaction with the
low-molecular mass inhibitor N-a-(2-naphthylsulfonylglycyl)-4-amidinophenylalanine-piperidine (a-NAPAP)
[2,8]. As shown in Fig. 1D, the pH dependence of
a-NAPAP binding was characterized by a bell-shaped
curve. The best-fit pKa values calculated from this data
set (Table 1C) were very close to those calculated from
the pH-dependent enzyme activity profiles (Table 1A).
The pKa values of the DK9 residues obtained in the
two data sets showed almost the same experimental
error. This may be because the synthetic substrate used
in the experiments is not a ‘sticky’ substrate for DK9
thrombin, as shown recently [2]. In fact, according to
the known following relation [3,5]
pKðobsÞ ẳ pK log1 ỵ k2 =k1 ị

1ị

in case of the interaction between thrombin and a
<
nonsticky substrate, where k2 < k1, pK(obs) should be
equal to the true pK value, as found experimentally.
Slightly higher pKa values were obtained in a-NAPAP
than in the substrate data set for WT thrombin (6.97
vs. 6.86 and 8.63 vs. 8.45, for the first and second
ionizable group, respectively). This is likely because
d-Phe-Pip-Arg-pNA is a ‘sticky’ substrate for WT
thrombin [2] and thus, according to Eqn (1), the
observed pKa value from steady-state kinetic experiments should be lower than the true pKa value by a

factor equal to log(1 ) k2 ⁄ k1), as seen experimentally.
In analogy with the values calculated in enzymatic
experiments, an increase of % 0.5 pK units of the
His57 in DK9 mutant was also calculated analyzing the
NAPAP data set (Table 1C). In fact, His57 undergoes

FEBS Journal 273 (2006) 159–169 ª 2005 FEBS No claim to original US government works

161


Functions of the thrombin A-chain

R. De Cristofaro et al.

a decrease in pKa value upon NAPAP binding in the
reversible complex formation, both in WT and DK9
thrombins.
The change in pKa value of the catalytic His residue
in the mutant thrombin implies that the deletion of
Lys9 allosterically affects the conformational state of
relevant domains of the catalytic B-chain. Thus, in an
attempt to unravel the mechanisms responsible for the
observed effects, we investigated the sodium-binding
properties and conformational stability of the mutant
enzyme in comparison with the WT thrombin form.
Binding of Na+ to thrombin
Sodium ion binding to WT thrombin was characterized by a saturable increase in fluorescence at 342 nm
(Fig. 2). The apparent equilibrium dissociation constant of Na+ binding was calculated using a single
site binding equation and was 22.0 ± 1.5 and

24 ± 2.6 mm for WT and DK9 thrombins, respectively, in reasonable agreement with previous results [9].
However, the intrinsic fluorescence of the mutant was
% 20% lower than that pertaining to WT thrombin
and the magnitudes of the fluorescence change differed
significantly, being equal to approximately +18 and
+9% for WT and DK9 mutant thrombin, respectively.
These results suggest that the Na+-binding loop in
the K9-deleted mutant retains the intrinsic affinity for
the cation, although the conformational transitions
linked to its binding are of more limited extension

Fig. 2. Titration by steady-state fluorescence of Na+ binding to
75 nM WT (d) and DK9 thrombin (s). Na+ binding was investigated
at 25 °C at ionic strength of 0.2, pH 8.00. Excitation wavelength
was 280 nm. Continuous lines were drawn according to single-site
binding isotherms with best-fit Kd values of 22 ± 1.5 and
24 ± 2.6 mM for WT and DK9 thrombin, respectively. The vertical
bars are the standard errors of the measurements.

162

compared with those of the WT form. In other words,
in the mutant thrombin the binding of sodium is not
intrinsically perturbed, but should be uncoupled from
the specific conformational transitions occurring in the
WT form [9]. This implies that the conformational
transitions induced in the B-chain by Lys9 deletion are
unique, being different from that of either the ‘fast’ or
‘slow’ form of WT thrombin [10].
Essential dynamics (ED) analysis [11] was applied to

key regions of WT and DK9 thrombin forms, with the
aim of identifying motions relevant for their folding,
separating them from those describing irrelevant local
fluctuations. ED analysis has proven to be a valid
method allowing the correlation between motions of
different parts of the protein to be assessed, overcoming possible artifacts which could derive from a simple
rmsd analysis. As shown in Fig. 3, the A-chain, as well
as the Na+-binding site, proved more flexible in WT
than in the K9-deleted mutant. The Na+-binding site
undergoes a significant conformational transition after
% 10 ns in DK9 thrombin, whereas this phenomenon
occurs after % 5 ns in WT thrombin. The motions of
the Na+-binding site were found to be correlated with
those of regions of particular importance, such as the
S3 specificity site (Trp215–Ile174), the Trp60 insertion
loop, the Cys168–Cys182 disulfide bond, and the
fibrinogen secondary binding exosite, all occurring
after % 10 ns (results not shown). A previous X-ray
diffraction study demonstrated that the Cys168–
Cys182 disulfide bond undergoes a re-registration upon
sodium binding [10]. In particular, the distance
between the sulfur atom of Cys182 and the Cb of
˚
Tyr225 was reduced by % 1 A in WT thrombin by
+
Na binding and mediated the conformational transitions in the catalytic pocket of the enzyme responsible
for the enhanced catalytic activity [10]. The results of
our calculations were in qualitative agreement with this
behavior of the WT form, whereas the reduction of
that distance in DK9 thrombin upon Na+ binding,

along the whole productive MD, was found about half
of that of WT (data not shown), highlighting reduced
conformational mobility of the Cys168–Cys182 disulfide bond linked to Na+ binding.
It had been shown by others that there is an inverse
relation between fluorescence intensity and exposure to
solvent in a number of Trp residues (60d, 96, 148, 207,
and 215) [12]. In particular, Trp207 and Trp29 residues,
located at the boundary between the A- and B-chains
and interacting with Arg137 via three water structural
molecules having low mobility (w321, w325, and w454)
[12], have been shown to contribute about 35 and 9%,
respectively, to the total intrinsic fluorescence of WT
thrombin [12]. In an attempt to understand the

FEBS Journal 273 (2006) 159–169 ª 2005 FEBS No claim to original US government works


R. De Cristofaro et al.

Functions of the thrombin A-chain

A

B

Fig. 3. Essential dynamics analysis of (A)
A-chain and (B) Na+-binding site of WT ( )
and DK9 thrombin (—). Motions along the
first eigenvector of the selected protein
regions in the time frame of the MD simulation are reported.

observed difference in the fluorescence properties of the
WT and DK9 thrombins, we calculated the average solvent (water)-accessible surface areas (SAS) of the above
residues along the whole MD simulations and found
that Trp207, whose predominant contribution to the
total fluorescence is well established [12], exposes its
surface to solvent about three times more in the DK9
˚
mutant (average SAS ¼ 24.9 A2) than in the WT form
˚ 2). A similar trend was observed
(average SAS ¼ 9.3 A
˚
for Trp29 (average SAS ¼ 9.6 and 14.1 A2 in WT and
DK9, respectively), whereas the other Trp residues
(60d, 96, 148 and 215), contributing < 11% to the total
fluorescence [12], vary in their average SASs by < 25%
in DK9 compared with WT thrombin. Furthermore, it
is likely that as a consequence of Lys14 deletion, the
hydrogen bond between Glu8 and Trp207 is lost, as
our MD calculations proved, with a consequent
decrease in tryptophan fluorescence, in agreement with
the known inverse relationship between the intrinsic
fluorescence of a molecule and its conformational
mobility [13]. These computational results suggest that
perturbation in the polarity and ⁄ or flexibility of the
environment of Trp207 and 29 may significantly affect
the intrinsic fluorescence of the DK9 variant, the higher
the surface exposure and flexibility of their side-chains
to the solvent the smaller the intrinsic fluorescence of
the enzyme.

The smaller gain in the intrinsic fluorescence of DK9
upon Na+ concentration (9 vs. 18% in WT) may reflect
not only the lower flexibility of the Na+-binding loop,
but also a different conformational rearrangement of

Trp215 (as indicated by the ED results), which contributes solely to the gain in fluorescence observed in the
Na+-bound conformer of thrombin [10,14].
Thrombin stability studies
Urea at 6 m concentration induced complete denaturation of both WT and DK9 thrombin (Fig. 4). The
denaturating process at pH 6.80 was monitored by the

Fig. 4. Urea-induced denaturation of WT (d) and DK9 thrombin (s).
Measurements were performed at 25 °C in 10 mM Bis ⁄ Tris, 0.15 M
NaCl, pH 6.80. Continuous lines were drawn according to the bestfit EC50 values of urea-induced denaturation: 2.94 ± 0.05 and
2.49 ± 0.04 M for WT and DK9 thrombin, respectively.

FEBS Journal 273 (2006) 159–169 ª 2005 FEBS No claim to original US government works

163


Functions of the thrombin A-chain

R. De Cristofaro et al.

decrease in fluorescence of the intrinsic enzyme and
was almost perfectly reversible, because a 10-fold dilution of the thrombin samples in the same buffer without urea resulted in recovery (after correction for the
dilution factor) of the fluorescence in the absence of
urea. The process was highly cooperative for both
thrombin species (slope factor % 8 in both cases), and

the concentration of urea inducing the 50% effect on
the fluorescence signal, [urea]1 ⁄ 2, was determined as
2.94 ± 0.05 m for WT and 2.49 ± 0.04 m for DK9
thrombin. These values suggested that, in the presence
of a saturating Na+ concentration, the mutant species
was less stable than WT thrombin, which in turn
showed a behavior similar to that reported previously
[15]. The higher sensitivity to urea denaturation shown
by the mutant may explain why DK9 thrombin is clinically characterized by a much lower phenotypic
expression in vivo, likely as a consequence of intracellular precipitation or enhanced degradation [2].
The stability of the two thrombin variants was further investigated using a more complex denaturation
procedure, referred to as disulfide scrambling [16]. One
disulfide bond connects covalently the A- and B-chain
(Cys1–Cys122) in the thrombin molecule, whereas the
B-chain is stabilized by three intrachain disulfide bonds
(Cys42–Cys58, Cys168–Cys182 and Cys191–Cys220)
[17]. In disulfide scrambling, urea breaks noncovalent
interactions between the two chains, subsequently
enhancing the susceptibility of the connecting disulfide
bonds to the reductive action of b-mercaptoethanol.
However, the low reducing agent concentration allows
the disulfide bonds to scramble and rearrange according to conformational changes induced by the denaturant. This allowed us to assess, better than by the
simple urea-induced denaturation, whether the deletion
of the K9 residue in the A-chain could affect the conformation stability of the whole thrombin molecule,
as a consequence of perturbed intra- and interchain
bonds.
After 180 min denaturation with 6 m urea and
0.2 mm b-mercaptoethanol, the A-chain was released
and the the native enzyme form disappeared for both
WT and mutant thrombin (Fig. 5). The kinetic rate

constant of free A-chain release was 1.69 ± 0.03 ·
10)2Ỉmin)1 in the WT form and 2.96 ± 0.05 ·
10)2Ỉmin)1 in DK9 thrombin (Fig. 6). The disappearance of the intact enzyme (A- + B-chains) was characterized by a first-order rate constant equal to 1.69 ±
0.06 and 3.01 ± 0.02 · 10)2Ỉmin)1 for WT and DK9
thrombins, respectively. Furthermore, the lag time for
the early appearance of the stable isomer ‘3’ shown in
Fig. 6 is shorter in DK9 (20 min) than in the WT form
(33 min).
164

Fig. 5. HPLC chromatograms of disulfide scrambling of WT and
DK9 thrombin. Disulfide scrambling was obtained under 6 M urea
and 0.2 mM b-mercaptoethanol at 0.5 min (upper) and 180 min
(remaining chromatograms). The HPLC chromatogram pertaining to
WT thrombin after 180 min of treatment is reported in the middle
of the figure, whereas the chromatogram of DK9 form is given at
the bottom. The primed numbers refer to the stable B-chain isomers of DK9 thrombin.

Characterization of stabilizing interactions
between A- and B-chains
In WT thrombin, the A-chain assumes an overall
boomerang-like shape interacting with the B-chain surface opposite to the active site [17]. Stabilization within
the A-chain and between the A- and B-chains occurs
mainly through salt bridges and H bonds involving
charged side chains. The A-chain is intramolecularly
cross-linked by five side-chain electrostatic interactions
grouped into three separate clusters (D1a–K9, K14a–
D14–R4–E8 and E13–R14d). Besides the covalent
disulfide connection between Cys1 and Cys122, seven
salt bridges, grouped into five clusters (D1a–R206,

E8–K202–E14c, D14–R137, K135–E14e–K186d and
K14a–E23), interconnect A- with B-chain; almost 90%

FEBS Journal 273 (2006) 159–169 ª 2005 FEBS No claim to original US government works


R. De Cristofaro et al.

Fig. 6. Kinetics of disulfide scrambling of WT and DK9 thrombin.
Kinetics of reduced A-chain release from WT (s) and DK9 thrombin
(h). Continuous lines were drawn according to a single exponential
equation with the best-fit first order rate constant equal to
1.69 ± 0.03 · 10)2Ỉmin)1 for the WT form, and 2.96 ± 0.05 ·
10)2Ỉmin)1 for DK9 thrombin. The kinetics of disappearance of the
intact adduct of A with B chain for WT (d) and DK9 thrombin (n) is
also shown. The single exponential decay rate constant was equal
to 1.70 ± 0.06 and 3.01 ± 0.02 · 10)2Ỉmin)1 for WT and DK9
thrombin, respectively. The vertical bars are the standard errors of
the experimental measurements.

of the total electrostatic energy of the A–B-chain interaction has been calculated to be due to these salt clusters [17]. We calculated the total number of stabilizing
A–B interchain electrostatic ⁄ H-bonding interactions
along the whole MD simulations, by using gromacs
routines (g_saltbr and g_hbond), and found that in the
WT form they are % 10% more than in the DK9
mutant (25969 and 23471 in WT and DK9, respectively). These computational results suggest that in this
case the higher the flexibility of the A-chain the higher
the number of electrostatic ⁄ H-bonding contacts
between the A- and B-chain in the WT thrombin, in
agreement with slower release of the light chain under

mild reducing conditions, as shown by the disulfide
scrambling experiments.

Conclusions
The results obtained in this study provide knowledge
about the perturbations triggered by deleting the Lys9
residue in the A-chain of thrombin [1,2]. Measurements of the pH dependence of both steady-state amidase activity and binding of the high-affinity inhibitor
a-NAPAP showed pKa values of the catalytic His57
higher in DK9 mutant than in WT thrombin. Application of the Brønsted theory on acid ⁄ base-catalyzed

Functions of the thrombin A-chain

reactions indicated that in the thrombin amidase cycle
the His57 imidazolium form acts as a general acid to
facilitate amine expulsion from the tetrahedral intermediate, and this process is the rate-limiting step for the
overall acylation reaction. The increased basicity of the
His57 N (nitrogen in the DK9 mutant would oppose
this function, resulting in a decrease in its catalytic
competence, as shown experimentally by both in vitro
and in vivo data).
Based on disulfide scrambling denaturation experiments, we inferred that in the DK9 mutant a refolded
A-chain should reduce the structural stability of the
whole a-thrombin molecule, weakening A–B interchain
contacts. Previously reported MD simulations showed
a transition of the A-chain from a boomerang-like
shape (WT) to a handle-like shape (DK9) [2]. Computational studies highlighted lower conformational flexibility in the A-chain resulting in fewer A–B interchain
electrostatic ⁄ H-bonding contacts in the DK9 mutant.
These A-chain folding effects should be allosterically
transmitted to the active site cleft: (a) altering the
geometry and protonation state of the residues involved

in catalysis and inhibitor binding, and (b) limiting the
allosteric effects triggered by sodium binding.
X-ray structures of human thrombin show that the
A-chain closely follows the contour of the catalytic
B-chain, hinging the two interacting six-stranded barrel-like domains of the B-chain [17]. A well-structured
network of ionic and H-bond interactions stabilize the
correct orientation of the two barrels in the catalytic
B-chain. Deletion of Lys9 may cause a re-registration
of this ionic network. In particular, in the WT forms
Asp14 makes a very strong salt bridge with Arg137. In
the DK9 mutant this salt bridge should be severely perturbed, because Asp14, as a consequence of Lys9 deletion, preferentially interacts with Lys202, which is
electrostatically linked to Glu14C in WT thrombin
[17]. Destruction of the salt bridge Asp14–Arg137
could alter the environment in which two Trp residues,
namely Trp207 (at vdW distance from Arg137) and
Trp29, are located, modifying its polarity and inducing
conformational changes in the two Trp side chains
which would be in DK9 more exposed to the solvent,
as our calculations on MD conformations proved. It is
known that the higher the conformational flexibility of
a fluorescent molecule the lower its fluorescence quantum yield [13], and Trp207 predominantly contributes
to the global fluorescence of thrombin [12]. Taking
these reports and our data into account, it is reasonable to hypothesize that even a subtle perturbation in
the polarity and ⁄ or flexibility of the environment of
the Trp residues could significantly affect the fluorescence of the DK9 thrombin. Moreover, the hydropho-

FEBS Journal 273 (2006) 159–169 ª 2005 FEBS No claim to original US government works

165

Functions of the thrombin A-chain

R. De Cristofaro et al.

bic cluster below Arg137, comprising the side chains of
Phe181, Phe199, Phe227, and Tyr228, close to the active site, may be destabilized, with repercussions for the
catalytic pocket, as shown in crystal structure studies
[10]. In particular, a perturbed conformational change
of Trp215, linked to Na+ binding and likely responsible for the reduced fractional change in fluorescence
in the Na+-bound DK9 conformer, may be another
evidence that the conformational transition caused by
Lys9 deletion in the A-chain are sensed by catalytic
subsites of the mutant enzyme. These findings lead us
to propose that a significant uncoupling between Na+
binding and conformational changes sensed by the
fluorescence change of Trp215 takes place in DK9
thrombin.
Globally taken, the experimental and computational
studies reported herein provide mechanistic support to
the phenomenological evidence that the A-chain would
affect both the conformation and the catalytic activity
of the thrombin B-chain, thus corroborating the belief
of an extraordinary conformational plasticity of this
enzyme.

Experimental procedures
Site-directed mutagenesis and construction
of expression vectors
Site-directed mutagenesis, expression, activation and purification of WT and DK9 thrombin form were obtained as

recently detailed [1,2]. The active-site titration of thrombin
forms, obtained by using p-nitro-phenyl guanidinobenzoate
gave a concentration of 95 ± 5% with respect to that
measured spectrophotometrically at 280 nm using E ẳ
1.83 mgặmL)1. SDS PAGE showed a single band of
~ 36 kDa for all the thrombin forms.

Effect of pH on thrombin amidase activity and
binding of the inhibitor a-NAPAP to the thrombin
active site
The effects of pH (5.5–10) on the hydrolysis of d-Phe-PipArg-pNA substrate by both WT and DK9 mutant thrombin
and binding of the inhibitor a-NAPAP to the enzyme active
site were analyzed. The experiments were carried out in an
appropriate triple buffer (25 mm Bis ⁄ Tris, 25 mm Tris,
50 mm CHES, 0.15 m NaCl, 0.1% PEG 6000). This buffer
system allowed us to keep the ionic strength of the solution
nearly constant over the entire pH range [18]. Calculations
performed with a program written in basic showed that
using the triple-buffer system in presence of 0.15 m monovalent salts, the ionic strength falls to a value % 2.5% lower
than that at extreme pH values. The Michaelis–Menten

166

Scheme 1.

constants of d-Phe-Pip-Arg-pNA (Instrumentation Laboratory, Milan, Italy) hydrolysis, as well as the equilibrium
dissociation constant of a-NAPAP (Sigma-Aldrich, St.
Louis, MO) binding to thrombin, were calculated as
detailed previously [8]. The kinetic scheme for the catalytic
cycle of thrombin in the steady-state analysis is given in

Scheme 1.
E, ES and EP are the free, Michaelis–Menten and acylated enzyme forms, whereas k1, k-1, k2 and k3 are the kinetic
constants for substrate binding, dissociation, acylation and
deacylation, respectively. Within the overall acylation step
of the canonical Scheme 1 (k2), tetrahedral intermediate
(TI) formation (kB) or breakdown (kA) may be rate limiting
[7,19–21] (Scheme 2).
Recently, a kinetic study showed that DK9 thrombin
catalyses the hydrolysis of a synthetic amide substrate with
<
a k2 < k3 [2]. Under these conditions, k2 % kcat. Furthermore, for both WT and DK9 thrombin Km % Kd, that is
the equilibrium dissociation constant of substrate’s binding
to thrombin. The effect of protons on this kinetic scheme
was analyzed by an appropriate extension of Scheme 1,
whereby binding and dissociation of protons were considered much more rapid than all binding and catalytic steps
of the substrate [3,4]. Thus the kinetic Scheme 1 was expanded, assuming the existence of two ionizable thrombin
groups involved in catalysis, as emerged by a best-fit minimization procedure of the experimental data taken over a
5.5–10 pH range. Accordingly, Scheme 1 was expanded as
shown in Scheme 3.
In Scheme 3 1K and 2K are the equilibrium association
constant for proton binding to the unprotonated and
mono-protonated thrombin form, respectively, whereas the
‘s’ and ‘p’ subscript refer to ES and EP thrombin species,
and ‘1’ and ‘2’ superscript refer to the mono- and diprotonated thrombin species, respectively. Using the linkage
scheme analysis detailed previously [3,4], the pH effects on
the Michaelis–Menten parameters kcat, Km, and kcat ⁄ Km of
WT and DK9 mutant thrombin hydrolysis of d-Phe-PipArg-pNA were analyzed as follows:
obs

Km ẳ 0Km

1 ỵ 1 KHị1 ỵ 2 KHị
1 ỵ 1 K s Hị1 ỵ 2 K s HÞ

ð2Þ

where Km values were approximated to the equilibrium dissociation constant of the substrate binding to thrombin [2],
0
Km is the asymptotic Km value in absence of protons, and
1
K and 2K are the equilibrium association constant of proton (H) binding to the first and second ionizable group for
free, and ES thrombin species (the latter denoted by the

FEBS Journal 273 (2006) 159–169 ª 2005 FEBS No claim to original US government works


R. De Cristofaro et al.

Functions of the thrombin A-chain

Scheme 2.

was studied in the presence of seven fixed NAPAP concentrations (1–64 nm) and were simultaneously analyzed using
a simple scheme, whereby both binding of NAPAP and
d-Phe-Pip-Arg-pNA to the thrombin active site were mutually exclusive [8]. In the analysis of pH effects on NAPAP
binding, the Km values were replaced in Eqn (2) by the Ki
values, calculated by the above competitive scheme [8].
Fitting of catalytic parameters was constrained to an
internally consistent picture, such that, at the three protonation levels, values of kcat, Km, and kcat ⁄ Km must be
closely related according to Eqns (2–4). Global and simultaneous analysis of the experimental data by grafit software allowed computation of the pKa values of the two

groups both in the free and substrate-bound species of
thrombin forms, along with the kinetic parameters’ values
pertaining to the three protonated thrombin species. The
values of standard errors (± SD) were also obtained in the
fitting procedure.

Denaturation by urea

Scheme 3.

‘s’ subscript), respectively. Likewise, the observed kcat values were analyzed as a function of pH as follows [3,4]:
obs

0

kcat ẳ

kcat ỵ 1 kcat 1 K s þ 2Ks ÞH þ 2 kcat ð1 K s 2 K s ịH2
1 ỵ 1 K s Hị1 ỵ 2 K s HÞ

ð3Þ

where 0kcat, 1kcat, and 2kcat refer to the kcat value pertaining
to unprotonated, mono- and diprotonated thrombin form,
respectively. Hence, the pH dependence of the observed
kcat ⁄ Km value (referred to as r) is:
obs

r ẳ ẵr0 ỵ r1 1 K Hị ỵ r2 1 K 2 K H2 ފ=Z

ð4Þ

where the superscript 0, 1, and 2 refer to the unprotonated,
mono- and diprotonated thrombin form, respectively, and
Z ¼ (1 +1K H) (1 +2K H).
In the data sets of NAPAP inhibition of d-Phe-Pip-ArgpNA hydrolysis at each pH value, the steady-state velocity
of cleavage of seven substrate concentrations (0.5–32 lm)

Urea-induced denaturation curves of both WT and DK9
thrombin were obtained in 10 mm Bis ⁄ Tris, 0.15 m NaCl,
pH 6.80, by monitoring the fluorescence emission at
342 nm (excitation at 280 nm) in a Varian Eclipse spectrofluorometer (Leini, Italy). Spectra were taken with an excitation ⁄ emission slit of 5 nm. The results were expressed as
the percentage of the measure fluorescence at any urea concentration compared with that obtained in the absence of
the denaturating agent.

Denaturation by disulfide scrambling
Denaturation of both WT and DK9 thrombin was performed by disulfide scrambling, that is the unfolding of
thrombin by urea in the presence of low concentrations of
the reducing agent b-mercaptoethanol, as reported previously [16]. Briefly, both WT and DK9 thrombin
(50 lgỈmL)1 in 0.1 m Bis ⁄ Tris buffer, pH 6.8) were treated
with 6 m urea in the presence of 0.2 mm b-mercaptoetha-

FEBS Journal 273 (2006) 159–169 ª 2005 FEBS No claim to original US government works

167


Functions of the thrombin A-chain

R. De Cristofaro et al.

nol. Denaturation was performed at 25 °C for 3 h to allow
the reaction to reach equilibrium. To monitor the kinetics
of denaturation, 50 lL of the samples were removed at different time intervals, mixed with an equal volume of 4%
TFA, and analyzed by RP-HPLC, using a Bio-Rad C18
Hi-Pore RP-318, 250 · 4.6 mm column. The eluant was
composed of solvent A (0.1%) trifluoroacetic acid, and
solvent B was acetonitrile ⁄ water (9 : 1 v ⁄ v) containing
0.08% trifluoroacetic acid. The gradient was 20–40% solvent B for 10 min, linearly increased from 40 to 55% for
40 min, kept at 55% for 10 min, and reduced to 20% solvent B in 20 min, while the flow rate was 0.5 mLỈmin)1.
The HPLC instrument was a PU-2080 instrument connected to a UV-2075 spectrophotometer (Jasco Europe s.r.l.,
Cremella, Italy). The chromatographic peaks were analyzed
and quantified using borwin-1 software (Jasco Europe).
The results were expressed as a percentage with respect to
the peak area measured at the start. The kinetic data referring to the disappearance of the native enzyme, [A + B%],
were fitted to the single exponential decay equation:
ẵA ỵ B%t ẳ 100  expðÀktÞ

ð5Þ

where [A + B%]t is the percentage of the native enzyme
present at time t, and k is the first-order rate constant of
this process. The appearance of the free A-chain at time t,
referred to as [A%]t, was fitted to single exponential relation:
ẵA%t ẳ 100 1 expktịị

Computational methods
gromacs 3.2.1 software [22], running on a Linux PC cluster, was used for the MD simulations and analysis of the
trajectories.
Using the gromacs utilities, the ED analysis was carried

out in to separate the motions of the examined protein
models into an essential subspace, describing most of the
functional motions, and into a physically constrained subspace, describing irrelevant local fluctuations. Eigenvectors
defining the direction of higher displacement are extracted
in decreasing order of the corresponding eigenvalue, and
the first few eigenvectors capture most of the essential
motions of the proteins. The cross-correlation function
between the projection of protein segments pairs onto the
first eigenvector obtained from the ED analysis of both segments was also calculated to estimate correlation of the
essential motions.

ð6Þ

where k is the first-order rate constant of the A chain
release.
At the end of the denaturation process three stable B
chain isomers were produced, in agreement with previous
studies, although obtained at different pH [16].
The kinetic Eqns (5–6) were fitted to the experimental
data using the grafit program (Erithacus Software Ltd,
Staines, UK).

Acknowledgements
Financial support from Italian Ministry of Education,
Universities and Research (MIUR) is gratefully
acknowledged by CA and RDC (PRIN 2003, Grant
no. 2003064812). We thank Dr Vincenzo De Filippis
(University of Padova, Italy) for helpful comments
and critical reading of the manuscript.

References

Binding of sodium ion
Steady-state fluorescence titration measurements were carried out using 500 nm WT and mutant thrombin, as described previously [9]. Fluorescence emission spectra (kex ¼
280 nm) were recorded at 25 °C in a 1 cm quartz cell, using
a Varian Eclipse spectrofluorometer (Leini, Italy), in
5 mmolỈL)1 Tris, pH 8.0 and increasing amount of NaCl.
Emission spectra between 300 and 400 nm did not show
any significant peak shift. Titrations were performed by
acquiring the changes in fluorescence intensity at the peak
of the emission spectrum at 342 nm. The concentration
of Na+ was increased by aspirating a defined volume of
thrombin solution dissolved in the cuvette in the above buffer containing 0.2 m tetramethylammonium chloride and by
adding the same volume of a thrombin solution (at the
same concentration) dissolved in the same buffer but con-

168

taining 0.2 m NaCl. This procedure allowed us to keep constant both the concentration of thrombin and the ionic
strength of the solution. The decrease in fluorescence signal,
usually defined as ‘bleaching effect’, due to the iterative
exposure of the sample to high intensity light beam, was
restricted to < 3% of the initial intensity and was always
taken into consideration in the analysis of the titration
data.

1 Akhavan S, Mannucci PM, Lak M, Mancuso G,
Mazzucconi MG, Rocino A, Jenkins PV & Perkins SJ
(2000) Identification and three-dimensional structural
analysis of nine novel mutations in patients with

prothrombin deficiency. Thromb Haemost 84,
989–997.
2 De Cristofaro R, Akhavan S, Altomare C, Carotti A,
Peyvandi F & Mannucci PM (2004) A natural prothrombin mutant reveals an unexpected influence of
A-chain structure on the activity of human alphathrombin. J Biol Chem 279, 13035–13043.
3 Di Cera E, De Cristofaro R, Albright DJ & Fenton JW
II (1991) Linkage between proton binding and amidase
activity in human a-thrombin: effect of ions and temperature. Biochemistry 30, 7913–7924.

FEBS Journal 273 (2006) 159–169 ª 2005 FEBS No claim to original US government works


R. De Cristofaro et al.

4 De Cristofaro R & Di Cera E (1990) Effect of protons
on the amidase activity of human a-thrombin. Analysis
in terms of a general linkage scheme. J Mol Biol 216,
1077–1086.
5 Stone SR, Betz A & Hofsteenge J (1991) Mechanistic
studies on thrombin catalysis. Biochemistry 30, 9841–
9848.
6 Enyedy EJ & Kovach IM (2004) Proton inventory studies
of alpha-thrombin-catalyzed reactions of substrates with
selected P and P’ sites. J Am Chem Soc 126, 6017–6024.
7 Fersht A (1985) Enzyme Structure and Mechanism, 2nd
edn. W.H. Freeman, New York.
8 De Cristofaro R & Landolfi R (1994) Thermodynamics
of substrates and reversible inhibitors binding to the
active site cleft of human alpha-thrombin. J Mol Biol
239, 569–577.

9 Wells CM & Di Cera E (1992) Thrombin is a Na+-activated enzyme. Biochemistry 31, 11721–11730.
10 Huntington JA & Esmon CT (2003) The molecular
˚
basis of thrombin allostery revealed by a 1.8 A structure
of the ‘slow’ form. Structure 11, 469–479.
11 Amadei A, Linssen AB & Berendsen HJ (1993) Essential dynamics of proteins. Proteins Struct Funct Genet
17, 412–425.
12 Bell R, Willem K, Stevens WK, Jia Z, Samis J, Cote
ˆ ´
HCF, MacGillivray RTA & Nesheim ME (2000) Fluorescence properties and functional roles of tryptophan
residues 60d, 96, 148, 207, and 215 of thrombin. J Biol
Chem 275, 29513–29520.
13 Edison AS, Abildgaard F, Westler WM, Mooberry ES,
Woody RW & Dunker AK (1996) Aromatic and cystine
side chain circular dichroism in proteins. In Circular
Dichroism and the Conformational Analysis of
Biomolecules (Fasman GD, ed.), pp. 136–144. Plenum
Press, New York.

Functions of the thrombin A-chain

14 Arosio D, Ayala YM & Di Cera E (2000) Mutation of
W215 compromises thrombin cleavage of fibrinogen,
but not of PAR-1 or protein C. Biochemistry 39, 8095–
8101.
15 De Filippis V, De Dea E, Lucatello F & Frasson R
(2005) Effect of Na+ binding on the conformation, stability and molecular recognition properties of thrombin.
Biochem J 390, 485–492.
16 Singh RR & Chang J-Y (2003) Structural stability of
a-thrombin studied by disulfide reduction and scrambling. Biochim Biophys Acta 1651, 85–92.

17 Bode W, Turk D & Karshikov A (1992) The refined
˚
1.9-A X-ray crystal structure of d-Phe-Pro-Arg chloromethylketone-inhibited human alpha-thrombin: structure analysis, overall structure, electrostatic properties,
detailed active-site geometry, and structure–function
relationships. Protein Sci 1, 426–471.
18 Ellis KJ & Morrison JF (1982) Buffers of constant ionic
strength for studying pH-dependent processes. Methods
Enzymol 87, 405–426.
19 Satterthwait AC & Jencks WP (1974) The mechanism
of the aminolysis of acetate esters. J Am Chem Soc 96,
7018–7031.
20 Hiroara H, Bender ML & Stark RS (1974) Acylation of
alpha-chymotrypsin by oxygen and sulfur esters of specific substrates: kinetic evidence for a tetrahedral intermediate. Proc Natl Acad Sci USA 71, 1643–1647.
21 Hegazi MF, Quinn DM & Schowen RL (1976) Transition-state properties in acyl and methyl transfer. In
Transition States of Biochemical Processes (Gandour
RD & Schowen RL, eds), pp. 355–423. Plenum Press,
New York.
22 Lindahl E, Hess B & van der Spoel D (2001) Gromacs
3.0: a package for molecular simulation and trajectory
analysis. J Mol Mod 7, 306–317.

FEBS Journal 273 (2006) 159–169 ª 2005 FEBS No claim to original US government works

169