Assignment of molecular properties of a superactive coagulation
factor VIIa variant to individual amino acid changes
Egon Persson
1
and Ole H. Olsen
2
1
Haemostasis Biology and
2
Medicinal Chemistry Research IV, Novo Nordisk A/S, Ma
˚
løv, Denmark
The most active factor VIIa (FVIIa) variants identified to
date carry concurrent substitutions at positions 158, 296 and
298 with the intention of generating a thrombin-mimicking
motif, optionally combined with additional replacements
within the protease domain [Persson, E., Kjalke, M. &
Olsen, O. H. (2001) Proc. Natl Acad. Sci. USA 98, 13583–
13588]. Here we have characterized variants of FVIIa
mutated at one or two of these positions to assess the relative
importance of the individual replacements. The E296V and
M298Q mutations gave an increased intrinsic amidolytic
activity (about two- and 3.5-fold, respectively) compared
with wild-type FVIIa. An additive effect was observed upon
their combination, resulting in the amidolytic activity of
E296V/M298Q-FVIIa being close to that of the triple
mutant. The level of amidolytic activity of a variant was
correlated with the rate of inhibition by antithrombin (AT).
Compared with wild-type FVIIa, the Ca
2+
dependence of

triggering and appropriate location of FVIIa haemostatic
activity upon vascular lesion and concomitant TF exposure.
The three-dimensional structure of the protease domain
of free FVIIa is, apart from certain loop regions, virtually
identical to that of thrombin and other constitutively active
and homologous serine proteases [2,3]. In addition, the
structural differences between free [3–5] and TF-bound
FVIIa [6,7] are subtle; thus the details in molecular
architecture that restrict the activity of free FVIIa remain
elusive. However, the high degree of similarity may be due
to the presence of an active site inhibitor in the structure of
the free FVIIa. The crystal (or solution) structure of
noninhibited FVIIa is presumably needed to reveal the
structural differences between ÔlatentÕ (zymogen-like) and
ÔactiveÕ FVIIa. However, information possibly pertaining to
the latent conformation of free FVIIa has been obtained
from the crystal structure of zymogen FVII [8]. This
structure suggests that relative b strand movements and a
hydrogen bond involving Glu296{154} (chymotrypsinogen
numbering is given in curly brackets to facilitate compar-
isons with homologous enzymes) regulate the activity state
of FVIIa.
Recent advances in our understanding of the mechanisms
regulating the activity of FVIIa have pinpointed side chains
that function as zymogenicity determinants in the free
enzyme. Replacements of these amino acid residues have
resulted in FVIIa molecules with improved intrinsic
(TF-independent) catalytic efficiency [9–11]. The relatively
high intrinsic activity of some of these FVIIa variants
suggests that the zymogen-like conformation of free factor

Wild-type FVIIa [12] and sTF [13] were prepared according
to published procedures. The concentrations of FVIIa and
sTF were determined by ELISA and spectrophotometry,
respectively, as described [9]. SDS/PAGE was run on
8–25% gradient gels using the PhastSystem (Amersham
Pharmacia) and followed by silver staining to check the
purity of the FVIIa mutants and to verify their conversion
to the activated, two-chain form. Factor X (FX) and factor
Xa were from Enzyme Research Laboratories (South Bend,
IN, USA) and antithrombin (AT) from Hematologic
Technologies (Essex Junction, VT, USA). Unfractionated
heparin was from Leo Pharmaceutical Products (Ballerup,
Denmark), potassium cyanate (KOCN) from Merck and
the chromogenic substrates S-2288 (
D
-Ile-Pro-Arg-p-nitro-
anilide) and S-2765 (benzyloxycarbonyl-
D
-Arg-Gly-Arg-
p-nitroanilide) from Chromogenix (Mo
¨
lndal, Sweden).
Mutagenesis and preparation of FVIIa mutants
The FVII expression plasmid pLN174 [14] was used as the
template for site-directed mutagenesis using the Quik-
Change kit (Stratagene, La Jolla, CA, USA). The primer
(only sense primer is given) used to introduce the E296V
mutation, with base substitution in italic and the affected
codon underlined, was GCC ACG GCC CTG
GTGCTC

10 n
M
FVIIa and 50 n
M
sTF) with 20 lL10m
M
S-2288 at
25 °C. The measurement with free FVIIa was also per-
formed in an assay buffer where CaCl
2
was replaced by
EDTA. The ability of the FVIIa variants to activate FX (the
proteolytic activity) was studied by incubating 10 n
M
(M298Q-, V158D/M298Q-, E296V/M298Q- and V158D/
E296V/M298Q-FVIIa) or 50 n
M
(wild-type, V158D-,
E296V-, and V158D/E296V-FVIIa) FVIIa variant alone
or 5 n
M
FVIIa variant plus 100 n
M
sTF with various
concentrations of FX (0.1–4.8 l
M
)for20minatambient
temperature (22 ± 1 °C). Buffer containing S-2765 was
then added to give a chromogenic substrate concentration
of 0.5 m

were introduced individually into FVIIa, V158D had no
significant effect on the amidolytic activity, whereas E296V
and M298Q yielded approximately two- and 3.5-fold
enhancement, respectively (Table 1). The result with
M298Q-FVIIa agrees with earlier reports [9,10]. The double
mutant E296V/M298Q-FVIIa had an amidolytic activity
six times higher than that of wild-type FVIIa and close to
that of V158D/E296V/M298Q-FVIIa. In addition, V158D/
E296V-FVIIa had significantly lower amidolytic than
E296V-FVIIa and V158D/M298Q-FVIIa had similar or
slightly lower activity than M298Q-FVIIa. This shows that
the simultaneous presence of the E296V and M298Q
mutations suffices to achieve an amidolytic activity similar
Table 1. Enzymatic activity of free FVIIa variants. All values are
means ± SD (n ¼ 3). The amidolytic activity is given as the ratio
between the activity of mutant and wild-type FVIIa.
FVIIa variant
Amidolytic activity
(mutant/wt)
FX activation
(k
cat
, · 10
)3
s
)1
)
Wild-type 0.088 ± 0.006
V158D 1.0 ± 0.1 0.069 ± 0.008
E296V 2.1 ± 0.3 0.083 ± 0.010

Ca
2+
was omitted from the assay buffer; E296V-FVIIa
(retained 24% of the activity), V158D/E296V-FVIIa (29%),
E296V/M298Q-FVIIa (39%) and, in particular, V158D/
E296V/M298Q-FVIIa (64%). This shows that the substitu-
tion of Val for Glu296{154}, which contacts the acidic Ca
2+
binding loop in the protease domain, attenuates the Ca
2+
dependence of FVIIa and confirms that this replacement is
responsible for the diminished Ca
2+
requirement observed
for V158D/E296V/M298Q-FVIIa [9]. The additional sub-
stitution of Gln for Met298{156}, especially in combination
with the replacement of Val158{21} by Asp, further
attenuates the Ca
2+
dependence.
We have previously shown that, in the absence of TF, the
k
cat
values for FX activation by M298Q-FVIIa and V158D/
E296V/M298Q-FVIIa were increased 5.5- and 28-fold
compared with that of wild-type FVIIa, respectively [9]. In
agreement with these results, the new batches of the two
variants displayed seven- and 25-fold higher values, respect-
ively (Table 1). E296V/M298Q-FVIIa and V158D/M298Q-
FVIIa activated FX five to seven times more rapidly than

site-directed) and potassium cyanate (N-terminal carbamy-
lation), was investigated. The rate of inhibition by anti-
thrombin reflects the reactivity of the active site and has
previously been found to nicely correlate to the level of
amidolytic activity of FVIIa variants [9]. The results herein
show that the new variants also obey this rule, with a strong
relationship between amidolytic activity enhancement and
increased inhibition rate (Table 2). The inhibition resulting
from potassium cyanate-mediated, N-terminal carbamyla-
tion reflects the degree of exposure of the protease domain
N-terminus. When compared with that of wild-type FVIIa,
the susceptibility to carbamylation was found to be
strikingly reduced for V158D/E296V/M298Q-FVIIa (and
reduced to some extent also for M298Q-FVIIa) indicative of
a more buried N-terminal amino group [9]. A majority
of the present FVIIa variants exhibits an intermediate level
of protection from carbamylation (Table 2). E296V-FVIIa
retains about half of its activity after incubation with
potassium cyanate, which is slightly more than wild-type
FVIIa. V158D/E296V-FVIIa, V158D/M298Q-FVIIa and
E296V/M298Q-FVIIa all retain about 60% of their activity,
which is similar to the residual activity of M298Q-FVIIa but
considerably less than that of V158D/E296V/M298Q-
FVIIa. This indicates that no single mutation is particularly
efficient in terms of promoting the insertion of the
N-terminus and, importantly, demonstrates that all three
mutations are needed for stable burial of the N-terminus,
most likely through salt bridge formation with
Asp343{194}).
DISCUSSION

5952 E. Persson and O. H. Olsen (Eur. J. Biochem. 269) Ó FEBS 2002
property in the presence of valine at position 296. It is
possible that the increased activity of FVIIa after Ca
2+
binding to the acidic 210–220 {70–80} loop [18] is induced,
at least to some extent, by a conformational change
resulting from charge neutralization. The altered local
charge distribution upon replacement of Glu296 by Val
might in part mimic this effect and contribute to an
increased activity and reduce the positive effect of Ca
2+
binding (Fig. 1A). The relative intrinsic proteolytic activity,
i.e. the degree of enhanced catalysis of FX activation,
follows a pattern slightly different from that of the
amidolytic activity. The M298Q mutation, in contrast to
V158D and E296V, enhances the proteolytic activity and is
indeed present in all members of this family of FVIIa
variants with increased proteolytic activity. Moreover,
V158D/E296V/M298Q-FVIIa is far superior to the other
variants, indicating that the three mutations work together
in a concerted fashion to dramatically boost the proteolytic
activity. The fact that FX itself binds Ca
2+
precludes direct
studies of the influence of Ca
2+
on the proteolytic activity.
An intriguing property of V158D/E296V/M298Q-FVIIa is
the nonparallel increase in amidolytic and proteolytic
activity compared with wild-type FVIIa. Such a behaviour,

P4-P1¢ to P4-P7¢ of FX) by V158D/E296V/M298Q-FVIIa
as compared with that of wild-type FVIIa was constant,
indicative of that the three mutations do not simply increase
the accessibility of the substrate binding cleft to longer
substrates (E. Persson, A. M. Hansen, K. Madsen and O.
H. Olsen, unpublished observation). However, the peptides
may not correctly mimic the corresponding sequences when
part of FX. Finally, the high proteolytic activity of V158D/
E296V/M298Q-FVIIa appears to be accompanied by a
stabilized salt bridge between Ile153{16} and Asp343{194}.
Crystallographic data and molecular dynamics simula-
tions of FVII suggest that the purpose of the activation to
FVIIa is to maturate and open the substrate binding site, in
particular the S1 pocket, whereas an appropriate catalytic
triad geometry appears to be preformed in the zymogen
[8,20]. However, even after conversion to FVIIa the
conformational equilibrium appears to be shifted toward
an enzymatically latent form. Thus, the role of TF, apart
from localizing FVIIa to the site of vascular injury,
optimally positioning the active site [21] and contributing
to an extended, specificity-determining, factor IX/X binding
surface [22–24], is to stabilize the active FVIIa conforma-
tion. Strong evidence supports that Met306 in FVIIa is the
starting point for the TF-mediated effect on the FVIIa
conformation leading to allosteric stimulation of the
enzymatic activity [6,15,25,26]. Recently, site-directed muta-
genesis on FVIIa has been able to mimic the effect of TF
binding, at least in part, resulting in FVIIa molecules with
enhanced intrinsic activity [9–11,19]. Two published hypo-
theses accommodate an instrumental role of the activation

Together, the three mutations result in a highly active
FVIIa molecule that is more comfortable in the ÔactiveÕ b
strand registration and with a buried N-terminus.
Moreover, the mentioned b strand B2 and the preceding
loop contain Glu296{154} and Arg290{147} which have
been shown to be important for FX activation [25,27]. This
might explain why an ordering of this region selectively
increases the proteolytic activity more than the amidolytic
activity. In accordance with this, displacement of this region
by a peptide exosite inhibitor causes a larger effect on the
proteolytic activity of FVIIa than on its amidolytic activity
[5].
ACKNOWLEDGEMENT
We thank Anette Østergaard and Helle Bak for excellent technical
assistance.
REFERENCES
1. Pedersen, A.H., Nordfang, O., Norris, F., Wiberg, F.C., Chris-
tensen, P.M., Moeller, K.B., Meidahl-Pedersen, J., Beck, T.C.,
Norris, K., Hedner, U. & Kisiel, W. (1990) Recombinant human
extrinsic pathway inhibitor. Production, isolation, and character-
ization of its inhibitory activity on tissue factor-initiated coagu-
lation reactions. J. Biol. Chem. 265, 16876–16793.
2. Bode, W., Brandstetter, H., Mather, T. & Stubbs, M. (1997)
Comparative analysis of haemostatic proteinases: structural
aspects of thrombin, factor Xa, factor IXa and protein C. Thromb.
Haemostasis 78, 501–511.
3. Pike, A.C.W., Brzozowski, A.M., Roberts, S.M., Olsen, O.H. &
Persson, E. (1999) Structure of human factor VIIa and its
implications for the triggering of blood coagulation. Proc. Natl
Acad. Sci. USA 96, 8925–8930.

for leucine 305 in factor VIIa increases the intrinsic enzymatic
activity. J. Biol. Chem. 276, 29195–29199.
12. Thim, L., Bjoern, S., Christensen, M., Nicolaisen, E.M., Lund-
Hansen,T.,Pedersen,A.H.&Hedner,U.(1988)Aminoacid
sequence and posttranslational modifications of human factor
VIIa from plasma and transfected baby hamster kidney cells.
Biochemistry 27, 7785–7793.
13. Freskga
˚
rd, P O., Olsen, O.H. & Persson, E. (1996) Struc-
tural changes in factor VIIa induced by Ca
2+
and tissue factor
5954 E. Persson and O. H. Olsen (Eur. J. Biochem. 269) Ó FEBS 2002
studied using circular dichroism spectroscopy. Protein Sci. 5,
1531–1540.
14. Persson, E. & Nielsen, L.S. (1996) Site-directed mutagenesis but
not c-carboxylation of Glu-35 in factor VIIa affects the association
with tissue factor. FEBS Lett. 385, 241–243.
15. Persson, E., Nielsen, L.S. & Olsen, O.H. (2001) Substitution of
aspartic acid for methionine-306 in factor VIIa abolishes the
allosteric linkage between the active site and the binding interface
with tissue factor. Biochemistry 40, 3251–3256.
16. Butenas, S., Lawson, J.H., Kalafatis, M. & Mann, K.G. (1994)
Cooperative interaction of divalent metal ions, substrate, and
tissue factor with factor VIIa. Biochemistry 33, 3449–3456.
17. Persson, E. & Petersen, L.C. (1995) Structurally and functionally
distinct Ca
2+
binding sites in the c-carboxyglutamic acid-con-

24. Zhong, D., Bajaj, M.S., Schmidt, A.E. & Bajaj, S.P. (2002) The
N-terminal epidermal growth factor-like domain in factor IX and
factor X represents an important recognition motif for binding to
tissue factor. J. Biol. Chem. 277, 3622–3631.
25. Dickinson, C.D., Kelly, C.R. & Ruf, W. (1996) Identification of
surface residues mediating tissue factor binding and catalytic
function of the serine protease factor VIIa. Proc. Natl Acad. Sci.
USA 93, 14379–14384.
26. Dickinson, C.D. & Ruf, W. (1997) Active site modification of
factor VIIa affects interactions of the protease domain with tissue
factor. J. Biol. Chem. 272, 19875–19879.
27. Ruf, W. (1994) Factor VIIa residue Arg
290
is required for efficient
activation of the macromolecular substrate factor X. Biochemistry
33, 11631–11636.
Ó FEBS 2002 Dissection of a superactive FVIIa variant (Eur. J. Biochem. 269) 5955