A novel gain-of-function mutation of the integrin a2 VWFA domain
Alexis Aquilina
1
, Michelle Korda
1
, Jeffrey M. Bergelson
2
, Martin J. Humphries
1
, Richard W. Farndale
3
and Danny Tuckwell
1
1
School of Biological Sciences, University of Manchester, Manchester, UK;
2
The Children’s Hospital of Philadelphia, Philadelphia,
PA, USA;
3
Department of Biochemistry, University of Cambridge, Cambridge, UK
Integrin a2b1 is the major r eceptor for collagens in human
tissues, being involved in cell adhesion and the control of
collagen and collagenase gene expression. The collagen
binding site of a2b1 h as been localized to the a2 von Wille-
brand Factor type A (VWFA) domain (A-domain or
I-domain) and the residues responsible for the interaction
with collagen have been mapped. We report a study of a2
VWFA domain in which residue E318, which lies outside the
collagen binding site, is mutated to tryptophan, showing that
this is a gain-of-function mutation. Recombinant a2-E3 18W
VWFA domain showed elevated and specific b inding to

generation by cells in collage n gels [13] and may be involved
in matrix assembly [14,15]. In vivo, a2b1 plays an important
role in platelet adhesion to collagens during thrombus
formation [ 16], and a lthough a2b1 i s probably not the
major collagen receptor on platelets [17], a genetic predis-
position to increased levels of platelet a2b1maybearisk
factor for stroke [18], myocardial i nfarction [19], and
diabetic retinopathy [20] (although see [21,22]). a2b1has
also been found to be involved in the r egulation of
inflamatory responses in experimental models of hypersen-
sitivity and arthritis [23].
The ligand binding site of a2b1 is located in the 200
amino acid v on Willebrand Factor type A domain (VWFA
domain, also known a s the A- or I-domain) of t he a2
subunit. The a2 VWFA domain can be produced as a
recombinant protein, which reproduces the ligand specifi-
city, affinity and cation preferences of the complete molecule
[8,24–29]. The recognition sequence for a2 VWFA domain
on collagen I has been identified and contains an essential
GER sequence [ 30]. The determination of the structure of
a2 VWFA domain complexed with the c ollagen peptide
demonstrates that the E of the peptide coordinates with a
cation bound to the VWFA domain, forming a metal-ion-
dependent adhesion site (MIDAS) [31,32]. Recognition
sequences for the other noncollagenous ligands of a2b1
have yet to be determined.
Comparison of the structure of the a2 VWFA domain
complexed with collagen (ÔopenÕ conformation) and alone
(ÔclosedÕ conformation) [31,32] indicates that collagen bind-
ing is accompanied by important conformational changes in

and exploited this property to study a2 VWFA domain
interactions with noncollagenous ligands of a2b1. These
data indicate that E318W has a similar effect on a2 as F302
does on aM and that residue E318 has an important role in
modulating a2 VWFA domain function.
MATERIALS AND METHODS
General reagents
Acid soluble rat tail type I c ollagen and EHS-laminin w ere
obtained from S igma-Aldrich. Collagen peptides were pre-
pared as described in [30]. Collagen IV fragment CB3 was the
kind gift of K. Ku
¨
hn, Max-Planck-Institute for Biochemist-
ry, Martinsreid, Germany [38]; collagen C-propeptide was
prepared as described in [ 7,8]; t he E- cadherin-COMP
construct comprising the five extracellular domains of mouse
E-cadherin fused to the assembly domain of rat cartilage
oligomeric matrix protein (COMP) was the kind gift of
J. Engel, Bı
´
ozentrum, University of Basel, Switze rland.
Mutagenesis and production of recombinant VWFA
domains
The generation of the wild-type a2 VWFA domain
construct has been described previously [26]. Mutagenesis
of this construct was carried out using the mutagenesis
protocols described by Kunkel [39] a nd the BIO-RAD
Muta-Gene mutagenesis kit. Essentially, single-stranded
uracil-containing DNA was generated by helper phage
infection o f Escherichia coli strainCJ236andthenusedas

KCl, 1.5 m
M
KH
2
PO
4
,0.9m
M
CaCl
2
,
0.5 m
M
MgCl
2
, pH 7.4 (NaCl/P
i
+
wherethe+sign
indicates the presence of Mg and Ca ions) overnight at
4 °C. The following day, protein solutions were removed
and wells blocked with 200 lL50mgÆmL
)1
BSA,in40 m
M
Tris/HCl, 150 m
M
NaCl, pH 7.4 (NaCl/Tris), for 2 h at
room temperature. Wells were then washed three times
with 200 lL N aCl/Tris, 1 mgÆmL

reagent). Absorbance was measured at 405 nm on a plate
reader. For experiments measuring collagen b inding in the
presence of different cation concentrations, the protocol was
carried out as above up to the blocking stage, then wells
were washed three times with 200 lL NaCl/Tris,
1mgÆmL
)1
BSA, then 50 lLcation(inNaCl/Tris,
1mgÆmL
)1
BSA at double the final concentration) was
added followed by 50 lL b iotinylated c ollagen I (also i n
NaCl/Tris, 1 mgÆmL
)1
BSA at double the final concentra-
tion). All subsequent steps were as above.
For the measurement o f VWFA domain binding to
immobilized collagen I, collagen IV CB3, C-propeptide,
laminin, E-cadherin or polylysine, Immulon 4 microtitre
plates (Dynex) were coated with 100 lL c ollagen I or other
ligands diluted in NaCl/P
i
+
overnight at 4 °C. The follow-
ing day, protein solutions were removed and wells blocked
as above. We lls were then washed three times with buffer A,
andtoeachwellwasadded50lL inhibitor (in buffer A at
double t he final concentration) followed by 50 lL VWFA
domain (also in buffer A at double the final concentration),
for 3 h at room temperature. Wells were then washed three

that buffer A contained 1 m
M
MgCl
2
and 1 m
M
MnCl
2
.
Assays to measure antibody binding to recombinant
VWFA domains w ere carried out after the method of
Ó FEBS 2002 Integrin a2 gain-of-function mutation (Eur. J. Biochem. 269) 1137
Brookman et al. [40]. Immulon 4 microtitre plates were
coated with 100 lL5lgÆmL
)1
fusion protein in NaCl/P
i
+
,
overnight at 4 °C. Wells were the n washed twice with
200 lLNaCl/P
i
–
(NaCl/P
i
+
without Ca
2+
or Mg
2+

colour developed by the addition of ABTS reagent as above.
RESULTS
The E318W mutation in a2 increases collagen binding
The introduction of the mutation F302W into aM has been
showntoresultinagainoffunctioninboththeisolatedaM
VWFA domain and in aMb2 [36]. Comparison of the a2
and aM sequences and structures indicated that the
homologous residue in a2isE318(Fig.1A),aresiduein
the a7 helix which, like aM F302, undergoes a large
displacement on collagen binding and moves fro m a buried
to an exposed position (Fig. 1B and C). a2E318was
therefore mutated to tryptophan, the recombinant mutant
VWFA domain gen erated, and t he a2 E318W VWFA
domain tested in solid phase binding assays. a2-E318W
VWFA domain showed enhanced binding to collagen I
compared with w ild-type VWFA domain, both w hen
binding of biotinylate d collagen to i mmobilized VWFA
domain (Fig. 2A) and binding of VWFA domain to
immobilized collagen was measured (Fig. 2B). The inter-
action of a2-E318W VWFA domain with collagen was
specific as it could be inhibited by EDTA (Fig. 2 A and B) as
well as by a collagen peptide containing the a2 recognition
sequence GFOGER (Fig. 2C). The data from the binding
of biotinylated collagen t o VWFA domains were a nalysed
by curve fitting a nd double reciprocal plots (analyses were
carried out on the results of four independent experiments):
the apparent affinities for the wild-type and mutant V WFA
domains were 3 .3 lgÆmL
)1
and 0.5 lgÆmL

The E318W mutation affects the MIDAS and helix a3
The interac tion of c ollagens with the a2 VWFA domain
requires the MIDAS c ation and we have previously shown
that either Mg
2+
or Mn
2+
will support VWFA domain–
collagen binding [9,26]. The effects of the E318W mutation
on the requirements for cations in collagen binding were
therefore investigated. The wild-type and mutant VWFA
domains showed similar curves f or binding of biotinylated
collagen i n t he presence of Mg
2+
but differed in their
binding in the p resence of Mn
2+
,withthea2-E318W
VWFA binding curve shifted to the left compared w ith the
wild-type, which displayed a complex profile ( Fig. 4A
and B). Similar profiles for a2-E318W and wild-type w ere
seen over a range of cation and fusion protein concentra-
tions (data not shown). A ssays in which t he immobilized
and soluble components were swapped (measuring the
binding of VWFA domain to collagen-coated plates) also
gave very similar profiles (Fig. 4C,D). The a2-E318W
mutation therefore affects t he formation of t he collagen–
cation–VWFA domain complex at the MIDAS, l eading to
altered cation preferences compared with wild-type. This is
despite the mutation being topologically distinct from the

colla-
gen I and binding of a2-E318W or wild-type a2(2lgÆmL
)1
)measured
in the p resence of 1 m
M
MgCl
2
(dark bars) or 1 m
M
MgCl
2
/10 m
M
EDTA (light bars). Data are means ± SD; n ¼ 7 from three experi-
ments (wild-type) and n ¼ 5 from t wo experiments (a2-E318W).
(C) Microtitre plates were c oated with 1 lgÆmL
)1
collagen I and the
binding of 0.5 lgÆmL
)1
a2-E318W measured in the presence of 1 m
M
MgCl
2
with the ad dition of 100 lgÆmL
)1
GFOGER collagen peptide
(which carries the a2 binding site), or co ntrol collagen peptide (which
does not carry the a2 binding site).

transmitted to the MIDAS and the JA203 epitope.
The consequences of the E318W mutation for a2 function
a2b1 is a recep tor not only for collage n I but also for
collagen I V, collagen I C-propeptide, laminin a nd E-cad-
herin [5,7–9,38], and so the effect of the E318W mutation on
the interaction of the a2 VWFA domain with other ligands
was investigated. a2-E318W VWFA domain showed
elevated specific binding to the integrin-binding CB3
fragment of collagen IV and to all three noncollagen
proteins ( Fig. 7A). No similar i ncrease i n binding to the
control p roteins, fibrinogen or the 50 k Da fragmen t of
fibronectin was seen. This indicated t hat the enhancing
effect of the E318W mutation was not confined to collagens
alone. Little i s known about the molecular basis of a2
VWFA domain binding to its noncollagenous ligands and
the binding of wild-type a2 VWFA domain to laminin and
E-cadherin is typically much lower than to collagen I.
However, the elevated binding seen with a2-E318W allowed
us to study these otherwise weak interactions. Binding of
Fig. 4. a2-E318W and wild-type a2 differ in their cation preferences for binding to collagen I. (A, B) Microtitre plates were coated with 0.2 lgÆmL
)1
a2-E318W (A) or 0.5 lgÆmL
)1
wild-type a2 (B) and the binding of 1 lgÆmL
)1
biotinylated collagen I measured in the presence of Mn
2+
(d), Mg
2+
(m), or EDTA (e). (C, D) Micro titre plates were coated with collagen I (1 lgÆmL

antibody JA203. These data suggested t hat the mutation
exerted its effect by inducing conformational changes in the
VWFA domain. We a lso describe f urther mutations of
E318 which help to define the molecular basis of the gain-of-
function effect as well as fu nctional studies showing that
noncollagenous a2b1 ligands bind to the mutant VWFA
domain at an elevated level relative to wild-type a nd by the
same mech anism a s c ollagen I. The E318W mutation in a2
therefore has a similar effect as the equivalent mutation
F302W in the aM VWFA domain, which also showed
increased binding [36].
The molecular basis of the interaction between a2 VWFA
domain and collagen I is now well understood following the
solution of the X-ray crystal structure of the VWFA
domain–collagen c ocrystal [32]. This interaction involves a
discrete set of residues clustering round the cation binding
site, as well as the cation itself. It is o f interest t hat the
E318W mutation has such a large effect on the binding of
collagenous and non collagenous ligands but does not form
part of the ligand binding site. In addition to the effects on
ligand binding, the mutation also affected the use of cations
by the ligand binding site. Although the precise nature of
the atomic events responsible for the shifts between
preferences for Mg
2+
and M n
2+
are n ot clear, change s in
Fig. 6. The epitope for JA203 is located in the a3 helix. The JA203 epitope was mapped using human -mouse a2 VWFA domain chimeras [41]. The
figure shows the mutations introduced to convert stretches of human to mouse sequence; the percentage binding of JA203, compared with wild-type

)1
; collagen I C-propeptide,
10 lgÆmL
)1
; laminin, 2 0 lgÆmL
)1
;E-cadherin-COMP,10lgÆmL
)1
;
50 kDa fragment of fibronectin, 10 lgÆmL
)1
; fibrinogen, 10 lgÆmL
)1
)
and the binding of 0.5 lgÆmL
)1
fusion protein measured.
Ó FEBS 2002 Integrin a2 gain-of-function mutation (Eur. J. Biochem. 269) 1141
the conformation of the cation coordinating residues seem
likely. These data suggest that there is a general alteration in
the structure or environment of the ligand binding site as a
result of the E318W mutation, affecting the way i n which
the t ertiary complex of VWFA domain, c ation and ligand
are formed. This was accomplished without any a lteration
in the specificity of t he interaction. The differential binding
of antibody JA203 to a2 E318W compared with wild-type
clearly identified this region as un dergoing conformational
changes as a result of the mutation. It is likely that this is due
to alterations at the MIDAS, as t he C-terminal region of the
JA203 epitope falls within the ligand binding site. These

mational change b etween the open a nd closed forms can
also result in a gain of function in a2. This is t he first report
of a deliberate engineering strategy being u sed for a2and
since t he E318 is conserved in a1, a10 and a11, the gain-of-
function property seen here should be reproducible in these
other integrins.
To account for the molecular events resulting from the
F302W mutation [36], it was suggested that the increased
bulk of the tryptophan side chain drove residue 302 from its
buried location in the closed form, to the solvent-exposed
location seen in the open form, thus promoting the open
form of the whole domain [36]. Residue E318 in a2VWFA
domain is, like aM F302, buried in the closed f orm and
exposed in the open form, but our data indicate that the side
chain bulk is not important in the gain-of-function effect.
However, the side chain of E318 forms a hydrogen bond
with R288 in the closed form, and this bond is brok en on
moving to the open form. The loss of the hydrogen bond in
the E318I/Y/W mutations, coupled with the hydrophobic
character of the mutation, may facilitate conformational
changes in the domain, for example by lowering t he energy
barrier separating the two forms and thus promoting the
open state.
a2-E318 VWFA domain showed increased binding to
collagen I C-propeptide, laminin an d E-cadherin, compared
with wild-type a2, indicating that the mutation also affected
binding to these noncollagenous ligands. The interactions
with laminin and E-cadherin are normally very weak and in
consequence hard to study. However, the elevated binding
seen with the mutant made possible i nhibition studies and

receptor. Am. J. Pathol. 137, 113–120.
2. Tuckwell, D.S. & Humphries, M.J. (1996) Integrin-collagen
binding. Seminars Cell Dev Biol. 7 , 649–657.
3. Dickeson, S.K. & Santoro, S.A. (1998) Ligand recognition by I
domain-containing integrins. Cell. M ol. Life Sci. 54, 556–566.
4. Heino, J. (2000) The collagen receptor integrins have distinct
ligand recognition and s ignaling functions. Ma trix Bi ol. 19,
319–323.
5. Elices, M .J. & Hemler, M.E. (1989) The human integrin VLA-2 is
a collagen receptor on some cells and a collagen/laminin receptor
on others. Proc. Natl Acad. Sci. USA 86, 9906–9910.
6. Bergelson, J.M., Shepley, M.P., Chan, B .M.C., Hemler, M.E. &
Finberg, R.W. (1 992) Identification of the integrin VLA-2 as a
receptor for echovirus 1. Science 255, 1718–1720.
7. Weston, S.A., Hulmes, D .J.S., Mould, A.P., Watson, R .B. &
Humphries, M.J. (1994) Identification of the integrin a2b1asacell
surface receptor for the C-prop eptide of type I procollagen. J. Biol.
Chem. 269, 20982–20986.
8. Davies, D., Tuckwell, D.S., Calderwood, D. A., Weston, S .A.,
Takigawa, M . & Humphries, M.J. (1997) M olecular characteri-
sation of integrin-procollagen C-propeptide interactions. Eur. J.
Biochem. 246, 274–282.
1142 A. Aquilina et al. ( Eur. J. Biochem. 269) Ó FEBS 2002
9. Tuckwell, D.S., Smith, L.C., Korda, M., Askari, J.A., Santoso, S.,
Barnes, M., Farndale, R.W. & Humphries, M.J. (2000) Mono-
clonal antibodies i dentify residues 1 99–216 of the integrin a2
vWFA domain as a functionally important region w ithin a2b1.
Biochem. J. 350 , 485–493.
10. Eble, J.A., Beermann, B., Hinz, H J. & Schmidt-Hederich, A.
(2001) a2b1 integrin is not recognized by rhodocytin, but is the

ssler, R. (2001) G lycoprotein VI but
not a2b1 integrin is essential for platelet int eraction with collagen.
EMBO J. 9, 2120–2130.
18. Carlsson, L.E., Santoso, S., Spitzer, C., Kessler, C. & Greinacher,
A. (1999) The a2 gene coding sequence T807/A873 of the platelet
collagen receptor integrin a2b1 might be a genetic risk factor
for t he development of stroke in younger patients. Blood 93,
3583–3586.
19. Moshfegh, K., Wuillemin, W.A., Redondo, M., La
¨
mmle, B., Beer,
J.H., L iechti-Gallati, S . & Meyer, B.J. (1999) Association o f two
silent polymorphisms of platelet glycoprotein Ia/Iia receptor with
risk of myocardial infarction : a case con trol study. Lancet 353,
351–353.
20. Matsubara, Y., Mur ata, M., Maruyama, T., Handa, M.,
Yamagata, N ., Watanabe, G ., Saruta, T. & Ikeda, Y. (2000)
Association between diabetic retin opathy and genetic variation s
in a2b1 i ntegrin, a p latelet receptor for collagen. Blood 95 ,
1560–1564.
21. Carlsson, L.E., Greinacher, A., Spitzer, C., Walther, R. & Kessler,
C. (1997) Polymoprhisms of the human platelet a ntigens HPA-1,
HPA-2, HPA-3 and HPA-5 on the platelet receptors for fibrinogen
(GPIIb/IIIa), von Willebrand factor (GPIb/IX), and collagen
(GPIa/Iia) are not correlated with an increased risk for stroke.
Stroke 28, 1392–1395.
22. Croft, S.A., Hampton, K.K., Sorrell, J.A., Steeds, R.P., Channer,
K.S., Samani, N.J . & Daly, M.E. (199 9) The GPIa C807T
dimorphism asso ciated with p latelet collagen receptor density is
not a risk factor for myocardial infarction. Br. J. Haematol. 106,

Messent, A.J., Smethurst, P.A., Tuckwell, D.S., Farndale, R.W. &
Barnes, M.J. (1998) Identification in collagen type I of an integrin
a2b1-binding site containing an essential GER sequence.
J. Biol. Chem. 273, 33287–33294.
31. Emsley, J., King, S.L., Bergelson, J.M. & Liddington, R.C. (1997)
Crystal structure of the I-domain from from integrin a2b1. J. Biol.
Chem. 272, 28512–28517.
32. Emsley, J., K night, C.G., Farndale, R.W., Barnes, M.J. &
Liddington, R.C. (2000) Structural basis of collagen recognition
by integrin a2b1. Cell 10 0 , 47–56.
33. Lee, J O., Bankston, L.A., Arnaout, M.A. & L iddington, R.C.
(1995) Two c onformations of the integrin A-domain (I-domain):
a pathway for a ctivation? Structure 3, 1333–1340.
34. Lee, J O., Rieu, P., Arnaout, M.A. & L iddington, R.C. (1995)
Crystal structure of the A-domain from the a subunit of integrin
CR3 (CD11b/CD18). Cell 80, 631–638.
35. Huth, J.R., Olejniczak, E.T., Mendoza, R., Liang, H., Harris,
E.A.S., Lupher, M.L., Wilson, A.E., Fesik, S.W. & Staunton,
D.E. (2000) NMR and mutagenesis evidence for an I d omain
allosteric site that regulates lymphocyte f unction-associat ed
antigen 1 ligand binding. Proc. Natl Acad. Sci. USA 97,
5231–5236.
36. Li, R., Rieu, P., Griffith, D., Scott, D. & Arnaout, A. (1998) Two
functional states of the CD11b A-domain: C orrelations with key
features of two M n
2+
-complexed crystal structures. J. Cell Biol.
143, 1523–1534.
37. Shimaoka, M., Shifman, J.M., Jing, H., Takagi, J., Mayo, S.L. &
Springer, T.A. (2000) Computational design of an integrin I

among i ntegrin domains. Proc. Natl A cad. Sci. USA 98, 2393–
2398.
45. Karecla, P.I., Green, S.J., Bowden, S.J., C oadwell, J. & Kilshaw,
P.J. (1996) Identification o f a binding site for integrin aEb7in
the N-terminal domain of E-cadherin. J. Biol. Chem. 271,
30909–30915.
1144 A. Aquilina et al. ( Eur. J. Biochem. 269) Ó FEBS 2002