(Received for publication, September 4, 1996, and in revised form, November 19, 1996)
From the Roon Research Center for Arteriosclerosis and Thrombosis, Division of Experimental Hemostasis and Thrombosis, Department of Molecular and Experimental Medicine, and the Department of Vascular Biology, The Scripps Research Institute, La Jolla, California 92037
The Arg-Tyr-Asp (RYD) and Arg-Gly-Asp (RGD)
sequences within the third complementarity-determining region of the
heavy chain (H3) of murine recombinant Fab molecules OPG2 and AP7,
respectively, are responsible for their specific binding to the
platelet integrin IIb
3. In this study, we
evaluated the influence of divalent cation composition and single amino
acid substitutions at key positions within H3 on the selectivity of
these Fab molecules for integrin
IIb
3
versus the vitronectin receptor
V
3. The parent Fab molecule OPG2 (H3
sequence, HPFYRYDGGN) binds selectively to
IIb
3 and not at all to any other
RGD-cognitive integrin, particularly
V
3,
under any divalent cation conditions. The binding of the AP7 Fab
molecule (HPFYRGDGGN) to
IIb
3 is not
affected by the relative composition of calcium, magnesium or
manganese. However, AP7 binding to
V
3,
either expressed by M21 cells or as the purified integrin, is supported
by manganese and inhibited by calcium. If the flanking asparagine 108 residue within the AP7 H3 loop is replaced by alanine (HPFYRGDGGA), the
resulting Fab molecule AP7.4 binds selectively to
V
3 in a cation-dependent manner, but does not bind at all to
IIb
3
under any conditions. AP7.4 binding to
V
3
is supported by manganese, completely inhibited by calcium, and largely
unaffected by magnesium. This behavior mimics that of the adhesive
protein, osteopontin, another ligand that binds preferentially to
V
3. Despite these differences in specificity for
IIb
3 and
V
3, AP7 and AP7.4 remain selective for
the
3 integrins and do not bind to cell lines that
express the RGD-cognitive integrins
V
5 or
5
1. These results confirm that subtle
changes in the amino acid composition immediately flanking the RGD or
RYD motifs can have a profound effect on
3 integrin
specificity, most likely because they influence the juxtaposition of
the arginine and aspartate side chains within the extended RGD loop
sequence.
The two members of the 3 integrin subgroup are
IIb
3, which is required for platelet
cohesion mediated by the binding of fibrinogen or von Willebrand factor
(1-3); and
V
3, the ubiquitous vitronectin receptor that mediates a variety of cellular processes, including migration, tumor cell metastasis, and angiogenesis (4-7). The distinctive
subunits,
IIb and
V,
are relatively unique within the integrin family and exhibit only 36%
sequence identity (8).
Although both 3 integrins recognize the RGD motif, each
exhibits a preference for certain RGD-containing ligands. An example is
the snake venom disintegrin barbourin (9), containing the KGD sequence,
which binds with greater affinity to the platelet integrin
IIb
3. Smith et al. (10) have
exploited this fact to alter the specificity of an engineered
RGD-containing Fab molecule and increase its selectivity for
IIb
3 over
V
3. In addition, this fundamental
observation has led to the synthesis of a cyclic homoarginine-Gly-Asp
(cHarGD)1 peptide that has one of the
highest differential affinities for
IIb
3
versus
V
3 (11).
Nonetheless, each of these ligands, including this peptide, retain some
affinity for
V
3, so that they still
inhibit cell adhesion mediated by
V
3
(11). The relative specificity of ligands for
IIb
3 versus
V
3 is also markedly influenced by
divalent cations. For example, fibrinogen binds to
V
3 in the presence of Mn2+
but not in the presence of Ca2+ (12).
To address these factors, Suehiro et al. (13) compared
carefully the binding of various RGD ligands and peptides to the 3 integrins as a function of divalent cation
composition, and grouped ligands into four classes. Class I,
represented by RGD peptides and vitronectin, bind equivalently to
IIb
3 and
V
3. Class II, represented by cHarGD,
fibrinogen, or fibrinogen
-chain peptides, bind to both integrins in
the presence of Mn2+, but only to
IIb
3 in the presence of Ca2+.
Class III, such as barbourin, bind exclusively to
IIb
3 under any condition. Class IV,
represented by osteopontin, bind primarily to
V
3.
To gain further insight into the molecular basis for differences in
ligand specificity, we exploited our well characterized, recombinant
RGD-containing Fab molecule AP7, and its RYD-containing progenitor OPG2
(14). By the criteria of Suehiro et al. (13), OPG2 is a
Class III ligand, binding only to IIb
3
under any condition, while AP7 is a Class II ligand, binding to
V
3 in Mn2+ but not
Ca2+ and to
IIb
3 in either
cation. The single amino acid replacement Asn
Ala within the
RGD-containing loop of AP7 further changes the specificity of the
mutagenized Fab molecule. The new Fab molecule, which we designate
AP7.4, binds exclusively to
V
3 and
belongs to the Class IV RGD ligand group. These findings confirm that selectivity for the
3 integrins is determined by precise
juxtapositions of the Arg and Asp side chains that are significantly
affected by flanking amino acid sequences.
The
term Fd denotes the segment of the Ig heavy chain that includes the
VH domain, the C1 domain, and a portion of the hinge region up to and including the cysteine residue, which participates in
a disulfide bond with the carboxyl-terminal cysteine residue of the
light chain (14). Fab molecules represent disulfide-linked heterodimers
composed of Fd plus
chains. Hexahistidine-tagged Fd cDNAs and
chain cDNA were prepared as described previously (14-16). In
each Fd construct, the oligonucleotide sequence (CATCAC-)3 was inserted upstream from a TGA stop codon so as to encode a carboxyl-terminal (His)6 sequence used to purify the Fab or
Fd molecules by nickel affinity matrix chromatography, as described (15, 16).
The Fd construct AP7.4 was generated by splice overlap extension PCR
(14, 16), using AP7 Fd cDNA (14) as a template. The 5 cDNA
fragment of AP7.4 Fd was obtained as a
BglII/SacII fragment from AP7 Fd cDNA. The
primer 5
-CTTCTACCGCGGCGACGGGGGAGCTTACTATGCTATGG-3
(AP74FOR) changes
Asn108 within AP7 H3 to Ala, while retaining a
SacII site. AP74FOR was used in combination with the
oligonucleotide 5
-CTA
TCAATCCCTGGGCAC-3
(HREV) to
produce the 3
fragment of AP7.4 Fd cDNA. Both fragments were then
digested with SacII and ligated. Ligated cDNA served as
template for subsequent PCR reactions using the oligonucleotide 5
-CTC
ACAATGGACTTCGGGCTC-3
(HFOR) and HREV to
amplify AP7.4 Fd cDNA. A BglII-NcoI digest of
the cDNA product was then ligated into pVL1392 containing a portion
of the murine constant region, the carboxyl-terminal cysteine, and the
(CATCAC-)3 sequence followed by the TGA stop codon. The
cloned, recombinant virus containing this new Fd construct is
designated pVL7.4Fd.
The Fd construct of AP7.7 was generated by PCR-based overlap extension
using FdOPG2 as template. The 5 fragment was produced with the primer
HFOR and the oligonucleotide 5
- GGTCCATAGCATAGTAAGCTCCCCCGTCGTACC-3
(AP77REV). The 3
fragment was generated using the oligonucleotide 5
-GGTACGACGGGGGAGCTTACTATGCTATGGACC-3
(AP77FOR) and the primer HREV.
The cDNA products were gel-purified and added together in a PCR
reaction with no additional oligonucleotide primers. The DNA fragments
were allowed to anneal and complimentary sequences were produced by
extension from the overlapping sequences for 10 thermocycles.
Subsequently, the primers HFOR and HREV were added, and an additional
25 cycles were carried out. The product was then digested with
BglII and NcoI, purified, and ligated into the
transfer vector employed for AP7.4 cDNA, as described above. The
cloned, recombinant virus containing this insert is designated pVL7.7Fd.
Recombinant viruses were cloned by infection of Sf9
cells (Invitrogen) (2 × 106 in 2 ml of complete
Grace's medium) seeded in T25 culture flasks, as described (14-16).
The sequence of each recombinant clone was confirmed prior to its use,
using Sequenase version 2.0 (U.S. Biochemicals, Inc.). Recombinant
viruses were used to coinfect High Five insect cells (Invitrogen,
Inc.), and Fab molecules were harvested from the media, normally after
72-h cultures, as described (14-16). Recombinant Fd and chains
were detected by a quantitative Western blot assay using rabbit
polyclonal anti-murine Fd+
antibody, developed in our laboratory
(14, 16). Fab molecules were purified by adsorption to Ni-NTA resin
(QIAGEN, Chatsworth, CA) and elution with imidazole buffer, as
described (16). The purity of Fab molecules was assessed by silver
staining of eluted proteins separated by electrophoresis on 10%
SDS-polyacrylamide gel electrophoresis slab gels (16). Purified Fab
molecule concentration was determined by optical density at 280 nm
using an extinction coefficient of 1.4.
Integrin
IIb
3 was purified as a functional
heterodimer from human platelets as described by Fitzgerald et
al. (17), except that protease inhibitors were included in the
final buffer, namely 0.4 mM phenylmethylsulfonyl fluoride,
100 µg/ml leupeptin, 0.02 µg/ml pepstatin A, and 10 mM
benzamidine. The vitronectin receptor
V
3
was purified as a functional heterodimer from human placentas by
immunoaffinity chromatography using the murine monoclonal antibody LM609, as described (18). Purified integrin heterodimers were adsorbed
onto the wells of Immulon II microtiter plates (Dynatech, Inc.,
Chantilly, VA), and the ability of murine monoclonal Fab molecules or
recombinant proteins to bind to each integrin was assessed by ELISA
(19).
Platelet-rich plasma was obtained by differential
centrifugation of ACD(A)-anticoagulated whole blood. Platelet-rich
plasma was harvested, and platelets were gently pelleted by
centrifugation at 950 × g for 11 min at ambient
temperature. The pellet was immediately resuspended in HEPES-modified,
Tyrode's buffer, pH 6.5, containing 0.1% bovine serum albumin and
0.1% dextrose. The platelet suspension was applied to a Sepharose 2B
column, and fractions containing platelets were collected. The
recombinant Fab molecules in Tyrode's buffer were added to 5 × 105 platelets in the presence of either 20 ng/ml
prostaglandin E1 or 0.2 µM phorbol myristate.
After a 15 min incubation at ambient temperature, fluorescein
isothiocyanate-labeled goat anti-mouse IgG F(ab)2 (Jackson
Immunoresearch Laboratories, Inc., West Grove, PA) was added. After an
additional 15-min incubation in the dark, samples were diluted 10-fold
with Tyrode's buffer and analyzed on a Becton Dickinson FACScan
apparatus, as described (14, 15).
Cell lines that differentially express the integrins of interest to
this study were maintained as described previously (20). These include:
M21, which expresses V
3 and
5
1 (21, 22); M21-L, which expresses
5
1 but not
V
3 (21, 22); and UCLA-P3, which expresses
V
5 and not
V
3 (23). M21 and UCLA-P3 cells were used
with permission of Dr. D. L. Morton (UCLA, Los Angeles, CA), and M21-L
cells were used with permission of Dr. D. A. Cheresh (The Scripps
Research Institute, La Jolla, CA). Cells were washed twice in 0.5 mM EDTA and resuspended in 0.02 M Tris, 0.15 M NaCl, pH 7.4, containing the desired combination of 1 mM CaCl2, 1 mM MgCl2,
or 0.1 mM MnCl2. The cell suspensions were then
incubated for 30 min at ambient temperature with selected murine
monoclonal antibodies: 50 µg/ml recombinant OPG2, AP7, AP7.2, or
AP7.4 Fab molecules; 10 µg/ml anti-
5
1
(P1D6) (24), anti-
V
3 (LM609) (25), or
anti-
V (AV-8) (26). To these mixtures, 50 µl of secondary antibody were added (fluorescein isothiocyanate-labeled goat
anti-mouse Fab; 1:50 dilution in Tris-buffered saline; Jackson Immunoresearch Laboratories, Inc., West Grove, PA). Following another
30 min of incubation, bound Fab molecules were detected by flow
cytometry, as described (16, 20).
The RGD-containing Fab ligand AP7 binds the integrin
IIb
3 equivalently on both resting or
activated platelets (14, 16). On the premise that flanking sequences
can influence integrin specificity, our goal was to determine the
smallest possible mutation within the AP7 H3 loop that would change its
specificity from
IIb
3 to
V
3.
Since each Fd construct employed in this study contains the
(His)6 coding sequence, Fab molecules were purified to
99% homogeneity by Ni-NTA resin chromatography. Purified protein
concentrations were determined by adsorption at 280 nm. The average
yield of Fab molecules per 72 h of culture was essentially equal
for each construct: OPG2, 20.1 ± 2.1 µg/ml (mean ± S.D.,
n = 5) AP7, 19.4 ± 1.1 µg/ml (n = 5); AP7.4, 16.4 ± 3.4 µg/ml (n = 2); and
AP7.7, 20.9 ± 0.9 µg/ml (n = 2).
The binding
of AP7, AP7.4, and AP7.7 Fab molecules to gel-filtered platelets was
compared by flow cytometry as a function of extracellular divalent
cation composition and platelet activation with phorbol myristate.
Equivalent results were obtained in the presence of either 10 µM Mn2+ (Fig. 1) or 1 mM Ca2+ plus 1 mM Mg2+
(data not shown). As we have shown in the past (14, 16), the binding of
AP7 Fab molecules to platelets was saturable and could be completely
inhibited by RGDW (10 µM) or EDTA (5 mM) (data not shown). AP7.4 and AP7.7, on the other hand, do not bind to
platelets in the presence of any combination of divalent cations and
regardless of the extent of platelet activation (Fig. 1).
Binding of Recombinant Fab Molecules to Human Cell Lines
The
parent Fab molecule OPG2 fails to bind to
V
3 expressed by the melanoma cell M21
under any conditions (Fig. 2A), a finding that is consistent with our previous reports that this antibody is
selective for the integrin
IIb
3 (27, 28).
However, the comparative binding of AP7 and its derivatives to M21
cells is more complex and obviously influenced by divalent cation
composition. While AP7.7 Fab molecules also fail to bind to M21 cells
under any conditions (data not shown), differential binding of both AP7
and AP7.4 Fab molecules is observed. In general, we found that 1 mM Ca2+ exerts an inhibitory effect on the
binding of either AP7 or AP7.4 Fab molecules, that the binding of each
is markedly augmented by the presence of
100 µM
Mn2+, and that 1 mM Mg2+ has
neither a supportive nor inhibitory influence (Fig. 2A). In
the case of AP7, a synergistic effect is observed in the presence of
both 1 mM Mg2+ and 100 µM
Mn2+, so that maximal binding equivalent to that seen with
AP7.4 is observed. This is not true of AP7.4 itself, which shows
maximal binding in the presence of either 100 µM
Mn2+ alone or both 1 mM Mg2+ and
100 µM Mn2+. In any case, binding that would
otherwise be supported by Mn2+ is significantly attenuated
in the presence of 1 mM Ca2+. Under optimum
divalent cation conditions, the binding of either AP7 or AP7.4 Fab
molecules to M21 melanoma cells is completely inhibited in a
dose-dependent manner by GRGDSPK but not GRGESPK (Fig. 2B).
Despite the observed changes in selectivity for
IIb
3 versus
V
3, the AP7 and AP7.4 Fab molecules
remained selective for only these
3 integrins and did
not bind to either M21-L cells, which express
5
1 but not
V
3, or UCLA-P3 cells, which express
V
5 but not
V
3 (data not shown).
The selectivity of each Fab
molecule for IIb
3 or
V
3 and the dependence of binding on the
RGD sequence were further investigated using the purified integrins in
an ELISA (Fig. 3).
In the presence of 1 mM each of Ca2+ and
Mg2+, AP7 or OPG2 Fab molecules exhibit a strong affinity
for IIb
3 (Fig. 3A) but fail to
bind to
V
3 (Fig. 3B). Binding
of either Fab molecule is completely inhibited by 1 mM EDTA
or
10 µM RGDW but not by
2 mM RGEW (data not shown). Under the same divalent cation conditions, both AP7.4 and
AP7.7 Fab molecules fail to bind to either
IIb
3 (Fig. 3A) or
V
3 (Fig. 3B).
In the presence of 10 µM Mn2+, the binding
of each Fab molecule to purified
IIb
3 is
unchanged (Fig. 3C). However, AP7 and AP7.4 Fab molecules
now bind strongly to
V
3 in the presence of Mn2+, while OPG2 Fab molecules fail to bind (Fig.
3D). The binding of AP7 or AP7.4 Fab molecules to
V
3 in this cell-free system is completely
inhibited by 1 mM EDTA or
10 µM
RGDW, but not at all by up to 2 mM RGEW (data not shown).
As a negative control, AP7.7 Fab molecules (50 µg/ml) fail to bind to
either integrin under any conditions (Fig. 3, A-D).
Using the recombinant murine Fab molecule OPG2 as a versatile
framework, our results validate the hypothesis that the amino acid
composition immediately flanking an RGD tripeptide can profoundly influence the specificity and divalent cation modulation of ligand binding to 3 integrins. The relevant sequences of the
recombinant Fab molecules, OPG2, AP7, AP7.4, and AP7.7, and their
comparative specificities, as determined by this study, are summarized
in Table I.
|
The model that we have developed using OPG2 and the AP7 series of
recombinant Fab molecules provides a unique opportunity to compare both
RGD and RYD analogs of the same ligand and to predict the impact of
single amino acid substitutions on specificity based upon known side
chain interactions defined by x-ray crystallography of the parent Fab
molecule (28). In the case of each Fab molecule in our series, a single
amino acid substitution within the third complementarity-determining
region of the heavy chain results in a profound change in specificity.
For example, there is the complete loss of binding to either
3 integrin that is characteristic of AP7.7 created by
the replacement of Asn108 by an Alanine within the OPG2 H3
sequence. The major reason for our selection of Asn108 as a
target for substitution is the fact that, in the crystal structure of
the OPG2 Fab molecule (28), the side chain of Asn108 is in
a position to form a hydrogen bond with that of Asp105,
i.e. the distance between Asp105-OD1 and
Asn108-ND1 is 2.8 or 3.2 Å in each of two alternate
conformers of the H3 loop. We reasoned that disruption of such a side
chain interaction would likely influence the juxtaposition of the
remaining side chains, particularly those of Arg103 and
Asp105. This hypothesis is borne out by our experimental
evidence, and our results provide strong support for the presence of a
hydrogen bond between these side chains. It follows that this side
chain interaction probably holds the Asp105 carboxyl group
in a unique orientation with respect to the amino group of
Arg103 such that OPG2 is recognized exclusively by
IIb
3.
The most dramatic and novel finding of our study is the change in
specificity created by the engineering of the AP7.4 Fab molecule. The
sole difference between the Fab molecule AP7, which binds
preferentially to IIb
3 in the presence of
calcium ions, and AP7.4, which binds solely to
V
3 in the presence of manganese ions, is
a single amino acid substitution adjacent to the RGD motif within the
H3 loop (HPFYRGDGGN in AP7 versus
HPFYRGDGGA in AP7.4). To our knowledge, this is the first
published report of a complete change in specificity of an RGD ligand
from
IIb
3 to
V
3 as a result of a single amino acid
substitution. While others have shown that single amino acid
differences in RGD peptides or an Fab molecule can increase their
relative affinities for
IIb
3 (10, 11, 29,
30), the engineered AP7.4 molecule represents the first instance in
which a dramatic decrease in affinity for
IIb
3 and reciprocal increase in affinity
for
V
3 has been produced. This change in
relative affinities is so extreme that the binding of the Fab molecule
to
IIb
3 has fallen below the level of
detection. Apparently, because Tyr104 of OPG2 has been
replaced by Gly in AP7, the subsequent replacement of
Asn108 with Ala no longer results in the dramatic loss of
binding to either integrin that was observed with the creation of AP7.7
(see above). In the case of AP7.4, the elimination of the putative hydrogen bond and the accommodation of this change by the mutated H3
loop must increase the flexibility of the Arg103 and
Asp105 side chains and facilitate the increased selectivity
of AP7.4 for the integrin
V
3. Our results
would argue that both
IIb
3 and
V
3 are highly restrictive with respect to
the Arg and Asp side chain orientations that each will recognize.
There are at least two mechanisms that may be involved in the divalent
cation regulation of ligand binding to integrins. On one hand, integrin
conformation is likely influenced by cations, particularly
Mn2+. As an example, the monoclonal antibody 9EG7 binds to
a Mn2+-induced epitope and stimulates 1
integrin functions (31). On the other hand, there is substantial
evidence that divalent cations support an initial ternary complex with
integrin and ligand. As the ligand-integrin binding becomes stabilized,
the divalent cation is displaced (32). Replacement of
Asn108 with Ala may eliminate the ability of AP7.4 to
disrupt cation coordination upon contact with an integrin. This would
not explain, however, why AP7.4 then binds selectively to
V
3, an integrin whose recognition of RGD
ligands is equally regulated by divalent cations.
Osteopontin was the first RGD ligand identified that has a substantial
preference for V
3 relative to
IIb
3 (33). It is particularly relevant
that Ca2+ is a strong inhibitor of osteopontin binding to
V
3, while Mn2+ enhances this
interaction (33). In the microenvironment of bone tissue, osteoclasts
liberate Ca2+ from demineralized bone during resorption
such that levels of free Ca2+ increase locally. This would
then favor detachment of osteoclasts from bone by inhibition of the
osteopontin binding to
V
3. Conversely, increases in the relative level of Mg2+ compared to
Ca2+ in areas of bone growth would favor
osteopontin-mediated cell attachment. Micromolar levels of
Mn2+ found in many tissues, including bone and liver, would
support cell attachment via
V
3 (12).
Clearly, the engineered, recombinant Fab molecule AP7.4 behaves
precisely as does osteopontin with respect to its selectivity for the
integrin
V
3 and the influence of divalent
cations on its binding properties. Thus, AP7.4 is a powerful new tool
to investigate the role of the integrin
V
3 in various biological processes,
including bone resorption.
We thank Dr. E. Wayner (University of Minnesota, St. Paul, MN) for monoclonal antibody P1D6 and Dr. D. Cheresh (The Scripps Research Institute) for monoclonal antibody LM609.