Potent α4β1 peptide antagonists as potential anti-inflammatory agents

Article abstract of DOI:10.1021/jm970175s

The migration, adhesion, and subsequent extravasation of leukocytes into inflamed tissues contribute to the pathogenesis of a variety of inflammatory diseases including asthma, rheumatoid arthritis, inflammatory bowel disease, and multiple sclerosis. The integrin adhesion receptor α4β1 expressed on leukocytes binds to the extracellular matrix protein fibronectin and to the cytokine inducible vascular cell adhesion molecule-1 (VCAM-1) at inflamed sites. Binding of α42b1 to VCAM-1 initiates firm adhesion of the leukocyte to the vascular endothelium followed by extravasation into the tissue. Monoclonal antibodies generated against either α4β1 or VCAM-1 can moderate this inflammatory response in a variety of animal models. Recently peptides containing a consensus LDV sequence based on the connecting segment-1 (CS-l) of fibronectin and cyclic peptides containing an RCD motif have shown promise in modulating leukocyte migration and inflammation presumably by blocking the interaction of α42b1 with VCAM-1. Here we describe novel, highly potent, cyclic peptides that competitively inhibit α42b1 binding to VCAM-1 and fibronectin at sub nanomolar concentrations. The structure of a representative analog was determined via NMR spectroscopy and used to facilitate optimization of peptide leads. The peptides discussed here utilize similar functional groups as the binding epitope of VCAM-1, inhibit lymphocyte migration in viva, and are highly selective for α4β1. Furthermore the structure-activity relationships described here have provided a template for the structure-based design of small molecule antagonists of α4β1-mediated cell adhesion processes.

Full text of DOI:10.1021/jm970175s

J . Med. Chem. 1997, 40, 3359-3368
3359
P oten t r4â1 P ep tid e An ta gon ists a s P oten tia l An ti-In fla m m a tor y Agen ts
David Y. J ackson,*,† Clifford Quan,† Dean R. Artis,† Thomas Rawson,† Brent Blackburn,† Martin Struble,†
Geraldine Fitzgerald,† Kathryn Chan,† Sheldon Mullins,† J . P. Burnier,† Wayne J . Fairbrother,‡ Kevin Clark,§
Maureen Berisini,§ Henry Chui,| Mark Renz,| Susan J ones,| and Sherman Fong|
Departments of Bioorganic Chemistry, Immunology, Protein Engineering, and Analytical Methods Development,
Genentech Inc., 460 Point San Bruno Boulevard, South San Francisco, California 94080
Received March 17, 1997X
The migration, adhesion, and subsequent extravasation of leukocytes into inflamed tissues
contribute to the pathogenesis of a variety of inflammatory diseases including asthma,
rheumatoid arthritis, inflammatory bowel disease, and multiple sclerosis. The integrin adhesion
receptor R4â1 expressed on leukocytes binds to the extracellular matrix protein fibronectin
and to the cytokine inducible vascular cell adhesion molecule-1 (VCAM-1) at inflamed sites.
Binding of R4â1 to VCAM-1 initiates firm adhesion of the leukocyte to the vascular endothelium
followed by extravasation into the tissue. Monoclonal antibodies generated against either R4â1
or VCAM-1 can moderate this inflammatory response in a variety of animal models. Recently
peptides containing a consensus LDV sequence based on the connecting segment-1 (CS-1) of
fibronectin and cyclic peptides containing an RCD motif have shown promise in modulating
leukocyte migration and inflammation presumably by blocking the interaction of R4â1 with
VCAM-1. Here we describe novel, highly potent, cyclic peptides that competitively inhibit R4â1
binding to VCAM-1 and fibronectin at sub nanomolar concentrations. The structure of a
representative analog was determined via NMR spectroscopy and used to facilitate optimization
of peptide leads. The peptides discussed here utilize similar functional groups as the binding
epitope of VCAM-1, inhibit lymphocyte migration in vivo, and are highly selective for R4â1.
Furthermore the structure-activity relationships described here have provided a template for
the structure-based design of small molecule antagonists of R4â1-mediated cell adhesion
processes.
In tr od u ction
tory bowel disease (IBD).13,14 These antibodies have also
been shown to peripheralize hemopoietic progenitors in
primates.15,16 Small molecule antagonists of R4â1/
VCAM-1 might therefore be useful for treatment of
chronic inflammatory diseases, be useful for mobiliza-
tion of bone marrow stem cells, or even find utility as
antitumor agents.
The integrins are R/â-heterodimeric cell surface re-
ceptors involved in numerous cellular processes from
cell adhesion to gene regulation.1,2 Several different
integrins have been implicated in disease processes and
have generated widespread interest as potential targets
for drug discovery.3 In the immune system, integrins
are involved in leukocyte trafficking, adhesion, and
infiltration during inflammatory processes.4 Differen-
tial expression of integrins regulates the adhesive
properties of leukocytes, and different leukocytes are
involved in different inflammatory responses.5 The
integrin R4â1 is expressed primarily on monocytes,
lymphocytes, eosinophils, basophils, and macrophages
but not on neutrophils.6 The primary ligands for R4â1
are the endothelial surface protein vascular cell adhe-
sion molecule-1 (VCAM-1) and the extracellular matrix
protein fibronectin.7 The binding of the R4â1 receptor
to cytokine-induced VCAM-1 expressed on high-endo-
thelial venules (HEVs) at sites of inflammation results
in firm adhesion of the leukocyte to the vascular
endothelium followed by extravasation into the inflamed
tissue.8 Binding of R4â1 to VCAM-1 also plays a key
role in stem cell adhesion to bone marrow stromal cells9
and may also be involved in tumor cell metastasis.10
Monoclonal antibodies directed against R4â1 or VCAM-1
have been shown to be effective modulators in animal
models of chronic inflammatory diseases such as
asthma,11 rheumatoid arthritis (RA),12 and inflamma-
Recently cyclic peptide antagonists of GPIIbIIIa/
fibrinogen based on an RGD motif common in many
integrin ligands have been reported.17 These cyclic
peptides led to the structure-based design of novel
peptidomimetic antagonists that inhibit platelet ag-
gregation and may be useful for treatment of heart
disease.18 Subsequently, cyclic peptide inhibitors con-
taining RCD or LDV sequences were shown to block
R4â1 and R5â1 binding to fibronectin at low micromolar
concentrations.19 Indeed, many integrin ligands display
similar sequences in their binding sites including fi-
brinogen (binds RIIbâIIIa),20 vitronectin (binds Rvâ3),21
and collagen (binds R2â1),22 and compounds containing
such motifs will likely become useful therapeutic agents.
The integrin-binding sites of VCAM are thought to
reside in the first and fourth domains of the seven-
domain protein.23 The crystal structure of domains 1
and 2 of VCAM, solved by two independent groups,24,25
reveals an immunoglobulin-like fold which had been
previously predicted based on sequence homology with
the IgG CH1 domain and molecular modeling.26 The
proposed VCAM-1 integrin-binding epitope, which con-
tains the sequence R36TQID40SPLN44 in a surface-
exposed loop connecting two â-strands (the CD-loop),
suggested Arg36 and Asp40 as potentially important
binding residues. Mutation of either residue to alanine
in a VCAM-1/IgG chimera resulted in decreased binding
affinity for cells expressing R4â1.26 We have recently
* Author to whom correspondence should be addressed.
† Department of Bioorganic Chemistry.
‡ Department of Protein Engineering.
§ Department of Analytical Methods Development.
| Department of Immunology.
X Abstract published in Advance ACS Abstracts, September 1, 1997.
S0022-2623(97)00175-1 CCC: $14.00 © 1997 American Chemical Society

3360 J ournal of Medicinal Chemistry, 1997, Vol. 40, No. 21
J ackson et al.
Ta ble 1. Inhibition of R4â1/VCAM-1 by (X)CDPC Peptides
reported the design of cyclic amide R4â1 peptide inhibi-
tors (IC50s in the low micromolar range) based on the
VCAM CD-loop sequence TQIDSP.27 The structures of
these peptides were found to mimic that of the VCAM-1
CD-loop via NMR spectroscopy. In addition, amino acid
substitutions in the peptides produced changes in af-
finity which paralleled the structure-activity relation-
ship (SAR) of existing protein mutagenesis studies and
were predictive of the activities of new mutant pro-
teins.27
IC50 (µM)
no.
X
ELISA
Ramos
The dissociation constants of ligand/integrin interac-
tions in solution are typically weak, ranging from high
micromolar to high nanomolar.1,2 The additivity of the
multiple adhesive interactions at a cell surface (avidity),
however, provides the necessary binding energy to
anchor leukocytes to the vascular endothelium. We
hypothesized that a therapeutically useful R4â1 antago-
nist would need to have substantially higher affinity for
the integrin receptor than its native protein ligands in
order to block a high percentage of the cell surface
interactions involved in cell adhesion. Clearly, different
integrins can selectively bind a variety of ligands in
solution, as part of a matrix or attached to a cell
surface,1,2 and peptide inhibitors with different struc-
tures can selectively bind to different integrins.17,19
These observations led us to speculate that epitope
conformation may be more important for binding selec-
tivity than for high affinity.28-30 With this in mind, we
set out to find novel R4â1 selective antagonists which
also possess the necessary affinity for in vivo inhibition
of leukocyte adhesion and infiltration.
1
2
3
4
5
6
7
8
Arg
Ac-Arg
Lys
3.7
1.3
∼250
0.13
21.0
2.1
80
40
∼1000
Fmoc-Arg
Leu-Arg
Phe-Arg
Trp-Arg
Tyr-Arg
Tyr-Lys
Tyr-Ala
Tyr
15
250
100
100
6
5
5
1.2
0.11
0.15
0.21
0.004
9
10
11
0.4
Ta ble 2. Inhibition of R4â1/VCAM-1 by XC(Z)PC Peptides
IC50 (µM)
no.
X
Z
ELISA
Ramos
8
Tyr-Arg
Tyr-Arg
Tyr-Arg
Tyr-Arg
Tyr-Arg
Ac-Tyr
Ac-Tyr
Ac-Tyr
Ac-Tyr
Ac-Tyr
Asp
Glu
Asn
Gln
Ala
Asp
Ser
Thr
Tyr
Ala
0.11
0.09
0.04
0.05
6.0
30
7.0
9.0
Resu lts a n d Discu ssion
12
13
14
15
16
17
18
19
20
(X)CDP C P ep tid es. In order to more clearly define
the functional binding moieties of RCD peptide inhibi-
tors for R4â1, we made a series of peptide analogs based
on the previously described â1 selective peptide antago-
nist cyclo-RCD(thioP)C which, when immobilized to an
affinity column, bound R4â1 and R5â1.19,30 As a pri-
mary assay we measured the IC50 for inhibition of
purified R4â1 binding to immobilized VCAM-1 in a 96-
well ELISA format (see Experimental Methods). As a
secondary assay, we measured the IC50 for inhibition
of fluorescently labeled R4â1 positive Ramos cells bind-
ing to immobilized VCAM-1 (see Experimental Meth-
ods). The higher IC50s generally observed in this cell-
based assay relative to the primary assay probably
result from the large number of receptors on a cell
surface (>10 000/cell)26 that must be blocked in order
to inhibit cell adhesion.
Our first set of analogs were designed to assess the
effect of N-terminal substitution in the cyclic pentapep-
tide RCDPC (1) which has an IC50 of 3.7 µM by protein-
based ELISA (Table 1). Previous mutagenesis studies
of VCAM-1 showed that mutation of Arg36 to Ala
resulted in decreased binding affinity, implicating Arg36
as an important contact residue.26 If the cyclo-RCDPC
peptide (1) mimics the VCAM-binding epitope for R4â1,
then Arg substitution in the peptide should result in
an analogous loss of affinity. Accordingly, substitution
of Lys for Arg (3) resulted in decreased binding affinity
(Table 1).
0.06
8.0
0.003
0.002
0.005
0.003
0.004
0.30
0.12
0.32
0.23
0.31
extension of cyclo-RCDPC (1) with the hydrophobic
residues Leu, Phe, and Trp (5-7) had little effect on
binding affinity (Table 1). However, aminoacylation of
Arg with Tyr (8) resulted in a substantial improvement
in affinity (IC50 ) 0.11 µM). The Tyr hydroxyl appears
to contribute the majority of this affinity enhancement
because the corresponding Phe analog (6) had reduced
affinity. Surprisingly, substitution of Arg in YRCDPC
(8) with Lys (9) or Ala (10) resulted in little change in
affinity. In fact deletion of Arg to afford 11 led to
further improvement of affinity (IC50 ) 4 nM) indicating
that Arg in these peptides is neither an essential nor
optimal residue in the binding of these molecules to
R4â1. Because 11 also inhibits binding of a competitive
R4 specific monoclonal antibody but does not inhibit a
competitive VCAM-1 specific antibody from binding to
VCAM (data not shown), it is likely that inhibition is
due to selective binding of 11 to R4â1 but not VCAM-1.
Overall, the simple substitution of Arg to Tyr (11) in
the parent RCDPC cyclic peptide 1 enhanced the bind-
ing affinity for R4â1 by nearly 1000-fold.
XC(Z)P C P ep tid es. In addition to Arg, Asp has also
been implicated as an important residue for R4â1-
mediated cell adhesion to fibronectin and VCAM-1.
Mutation of Asp40 in VCAM-1 to Ala virtually elimi-
Acylation of the N-terminal Arg of cyclo-RCD(thioP)C
with hydrophobic groups had previously been shown to
enhance binding affinity.19,30 We found that N-terminal

R4â1 Peptide Antagonists as Anti-Inflammatory Agents
J ournal of Medicinal Chemistry, 1997, Vol. 40, No. 21 3361
F igu r e 1. Representative low-energy structures (stereo) from the NMR ensemble of 5.
Ta ble 3. Inhibition of R4â1/VCAM-1 by Cyclic XCA(Z)C
nated binding to R4â1.26 In addition, substitution of Asp
in RCD peptides19 or in fibronectin CS-1-based peptide
inhibitors containing a concensus LDV motif31 abolished
activity of the peptides. Using the cyclic peptides
YRCDPC (8) or Ac-YCDPC (16) as references, analogs
were made to assess the importance of aspartic acid in
these peptide inhibitors. From the data shown in Table
2, it is apparent that a wide variety of substitutions are
permitted at this position with little effect on binding
affinity. This strongly suggests that the cyclic peptides
do not bind in the same orientation as the native protein
ligands. On the basis of the data presented thus far,
we conclude that RCD is sufficient for inhibitory activity
but neither Arg nor Asp is essential or optimal for potent
inhibition of R4â1 binding to VCAM-1. Because the
effects of Arg or Asp substitution in 8 on inhibition of
R4â1/fibronectin are similar (data not shown), the
peptides probably bind the R4â1 receptor but do not bind
its ligands VCAM-1 or fibronectin.
NMR Str u ctu r a l An a lysis. In order to investigate
possible structural similarities between the cyclic pep-
tides and the VCAM-1 CD-loop, we determined the
structure of a representative peptide (LRCDPC, 5) via
NMR spectroscopy (Figure 1). The final set of refined
structures contains no distance restraint violations >
0.07 Å and no dihedral restraint violations > 3.4°. The
cyclic CDPC core of 5 was found to adopt a well-ordered
structure in solution dominated by a distinct conforma-
tion. In contrast the N-terminal Leu-Arg residues are
highly disordered. The constraint imposed by the
disulfide bond was found to fold the peptide into a very
tight turn marked by a cis-amide bond at the Asp-Pro
linkage. Interestingly, analysis of coupling constant and
NOE data indicated that the ø1 of the Asp side chain
was constrained to values of ∼ -60°. While the appar-
ent disorder of the N-terminal residues would appear
to limit the role Arg might be playing in this conforma-
tion, an Arg-Asp salt bridge appears possible and may
be one of the factors contributing to the limited rotation
at the Asp ø1, as some of the structures consistent with
the NMR restraints exhibit this feature. The lack of
structural similarity between peptide 5 and the VCAM-1
CD-loop (discussed later), and the fact that neither Arg
nor Asp is essential for peptide binding to R4â1, led us
to conclude that the peptides are not straightforward
mimics of VCAM-1. These observations coupled with
Peptides
IC50 (µM)
no.
X
Z
ELISA Ramos
20 Ac-Tyr Pro
21 Ac-Tyr Gly
22 Ac-Tyr N-methylglycine
23 Ac-Tyr Pip
0.004
22.0
0.71
0.001
0.076
0.005
0.023
0.034
2.80
0.31
ND
ND
0.02
33
0.12
0.94
10.5
182
4.7
24 Ac-Tyr phenyl-Gly
25 Tyr
26 Tyr
27 Tyr
28 Tyr
29 Tyr
30 Tyr
31 Tyr
32 Tyr
33 Tyr
34 Tyr
Pro
Ile
Leu
Phe
Oic
0.02
γ-hydroxy-Pro
O-benzyl-γ-hydroxy-Pro
cyclohexyl-Ala
cyclohexyl-Gly
2-aminoindane-2-carboxylate
0.005
0.04
0.14
ND
ND
25
0.23
36.0
ND
ND
the importance of the N-terminal Tyr of 11 (which is
not present in the CD-loop) led us to hypothesize that
the cyclic peptides might bind R4â1 in a different
orientation or at a different site relative to the VCAM-1
CD-loop.
XCA(Z)C An a logs. Pro42 was previously found to
be an important binding determinant of the VCAM-1
CD loop.26 The peptides cyclo-Ac-YCAPC (20) and cyclo-
YCAPC (25) were chosen as references for evaluating
the importance of proline within the cyclic peptide core
(Table 3). The large increase in IC50 (∼5000-fold) upon
substitution of proline (20) with glycine (21) demon-
strates that proline also plays a significant role in the
interaction of these cyclic peptides with R4â1 (Table 3).
Because proline often plays a unique conformational role
in protein and peptide backbone structure, it is possible
that the loss of affinity upon proline substitution is due
to a conformational change in the CAPC core as well as
loss of a potential hydrophobic interaction. Consistent
with this notion, substitution of Pro with N-methylgly-
cine (22) only partially restored affinity relative to 21

3362 J ournal of Medicinal Chemistry, 1997, Vol. 40, No. 21
Ta ble 4. Effect of C-Terminal Carboxylic Acid Substitutions
J ackson et al.
more dramatic (100-1500-fold) effect was seen for
peptides lacking aspartic acid (20, 38-40). These
results demonstrate that the C-terminal carboxylate
group of RCXPC-like peptides is an important binding
determinant, and the loss of affinity upon carboxamide
substitution can be partially recovered if Asp is present
at the X position. Accordingly, Asp substitution in the
native protein ligands26 and in cyclic peptides based on
the VCAM-1 CD-loop sequence cyclo-TQIDSPXX-amide27
resulted in relatively large decreases in affinity.
no.
sequence
RCDPC-acid
IC50 (µM)a
OH/NH2
1
4.0
35
36
37
38
39
20
40
RCDPC-amide
50.0
0.04
0.30
0.008
12.0
0.003
0.320
12.5
7.5
1-FCA-RCDPC-acid
1-FCA-RCDPC-amide
YC(m-aminobenzoyl)C-acid
YC(m-aminobenzoyl)C-amide
Ac-YCAPC-acid
1500
107
Ac-YCAPC-amide
a
IC50s are from the protein ELISA.
The results from Table 4 have led us to hypothesize
that RCDPC-like peptides containing Asp might bind
in two possible orientations: one in which the confor-
mationally rigid C-terminal carboxylate group prefer-
entially fills a putative electrostatic Asp-binding site or
a second less important orientation in which the Asp
side chain fills the carboxylate-binding site similar to
the native protein ligands. Peptides not containing Asp
would be restricted to a single orientation utilizing the
C-terminal carboxylate group resulting in a more sig-
nificant decrease in binding affinity upon conversion to
a carboxamide. Further studies are being conducted to
test this hypothesis and more clearly define the interac-
tion of the cyclic peptides with R4â1.
Com p a r ison of P ep t id e a n d P r ot ein Bin d in g
P r op er ties. Previous mutagenesis studies on VCAM-1
have implicated Arg36, Asp40, and Pro42 as important
residues for binding to R4â1.26,28 The X-ray structure
of the first domain of VCAM-1 highlighting the CD-loop
(depicted in white) and Arg36, Asp40, and Pro42
(depicted in green) is shown in Figure 2.24 For com-
parison, representative structures from the NMR en-
semble of 5 (from Figure 1) are also shown. Because
the CD-loop of VCAM-1 is involved in crystal-packing
interactions with three other molecules of the protein,
the actual receptor-binding conformation of the CD-loop
in VCAM-1 is likely to be different from that depicted
in the X-ray structure.24 We have recently designed
cyclic amide peptides of the general sequence TQIDSP
based on the VCAM CD-loop sequence and X-ray
structure. Using NMR and binding displacement as-
says, we showed that these peptides mimic the VCAM
CD-loop structure and bind with low micromolar affini-
ties.27 Inspection of Figure 2 led us to conclude that
although the cyclic disulfide peptides described here
share little sequence homology with the VCAM CD-loop
region, they demonstrate some structural similarities
primarily near the proline- and carboxylate-binding
moieties implicating a likely overlap in binding sites.
In addition, we have recently used a fluorescently
labeled BCK(N-fluorescene)PC peptide (77) (ELISA IC50
) 8 nM) in displacement assays to confirm that the
cyclic disulfides described here can be competitively
displaced from the R4â1 receptor by both VCAM and
the aforementioned cyclic amide CD-loop-based pep-
tides. These results (discussed in ref 27) further support
the hypothesis that all three bind a similar overlapping
site on the receptor.
and substitution of Pro in 25 with Ile (26) and Leu (27)
resulted in slight decreases in affinity.
Substitution with L-pipecolic acid (Pip) to afford 23
actually improved affinity in both protein- and cell-
based assays and was the only Pro substitution which
did not adversely affect activity (Table 3). Notably,
substitution with Phe (28) resulted in a dramatic
decrease in affinity. Constraint of the Phe side chain
via substitution by 2-aminoindane-2-carboxylate (34)
increased the IC50 an additional 10-fold. Extension of
Pro to the bicyclic octahydroindolecarboxylate (Oic) (29)
produced a modest increase in the IC50. This was
increased a further 7-fold by substitution of cyclohexyl-
Ala (32), indicating some measure of the value of the
side chain constraint for this system. Substitution of
γ-hydroxyproline for proline (30) resulted in virtually
no change in IC50, and the O-benzyl analog (31) of this
compound showed a modest 8-fold decrease in affinity,
indicating the putative site for the proline had neither
a strict hydrophobic requirement nor a completely
restrictive steric boundary.
These results led us to speculate that the proline
might be interacting with a region of relatively open
hydrophobic surface, possibly by stacking against a
delocalized π-system on the receptor. The surprising
decrease in affinity (20-fold) between the largely iso-
steric cyclohexyl-Ala (32) and Phe (28) residues sug-
gested the latter might be positioned such that the
interaction between π-systems was repulsive. Modeling
studies based on the core residues from the NMR
structure determined above suggested an acceptable
repositioning of the aromatic ring could be achieved
using phenylglycine (24). This substitution provided a
compound which exhibited a 40-fold improvement over
the phenylalanine analog 28 and was 3-fold better than
the corresponding cyclohexylglycine-containing peptide
33.
C-Ter m in a l Ca r boxyla te Su bstitu tion s. Asp40 of
VCAM-1 was previously shown to be an important
integrin-binding residue as determined by alanine scan-
ning mutagenesis,26 yet Asp is not an essential residue
in the cyclic peptides discussed here (see Table 2).
Because the importance of proline in both peptide and
protein binding was well established and the link
between protein and peptide SAR suggested they bind
an overlapping site on the receptor, we hypothesized
that the C-terminal carboxylate in the peptides might
occupy a site on the receptor normally filled by Asp40
of VCAM-1. To test this hypothesis we prepared several
C-terminal carboxamide analogs and compared them
with the corresponding C-terminal acids (Table 4).
Notably, C-terminal carboxamide substitution in pep-
tides containing Asp (1, 35-37) resulted in a moderate
(∼10-fold) reduction in binding affinity, while a much
(X)CSP C P ep tid es. The data presented thus far
have demonstrated that potent inhibition is achieved
by the consensus sequence YCXPC with the N-terminal
Tyr, the endocyclic Pro, and the C-terminal carboxylate
contributing most of the binding energy. The data
presented below (Table 5) show the results of a similar
effort to optimize the N-terminal Tyr for stability and

R4â1 Peptide Antagonists as Anti-Inflammatory Agents
J ournal of Medicinal Chemistry, 1997, Vol. 40, No. 21 3363
F igu r e 2. Comparison of the VCAM-1 X-ray structure24 (left) and the NMR structure of 5 (right). The VCAM-1 CD-loop (white)
with Arg36, Asp40, and Pro42 (green) is shown.
Ta ble 5. N-Terminal Optimization of Cyclic (X)CSPC Peptides
position (49), a substantial decrease in affinity was
observed. In addition, a dramatic loss of potency is seen
when Tyr is replaced by Phe (42), demonstrating the
importance of the Tyr hydroxyl for high-affinity binding.
In general, other Tyr substitutions resulted in loss of
affinity; however, halogen substitution ortho to the Tyr
hydroxyl group (50-52) had a minimal effect on po-
tency. Other phenol analogs also demonstrated de-
creased affinity relative to 17 highlighting the essential
role of Tyr in this position. Substitution of the Tyr
N-acetyl with a phenyl group (56) resulted in only a
slight decrease in binding affinity consistent with the
earlier observation that removal of the Tyr N-acetyl
group (25, Table 3) also had little effect on binding
affinity.
IC50 (µM)
no.
X
ELISA Ramos
17 Ac-Tyr
41 D-Tyr
42 Phe
0.002
0.89
1.50
0.134
4.60
0.06
126
15.0
1.9
50
43 benzoyl
44 isonipecotyl
45 4-methoxyphenylacetyl
46 1-fluorenylcarboxyl
47 1-naphthoyl
48 2-naphthoyl
49 3-hydroxy-Phe
50 3-iodo-Tyr
51 3-fluoro-Tyr
With these results in mind, we synthesized the amino
acid analog B (Scheme 1) in which the Tyr side chain
conformation is constrained by a two-carbon bridge
between the â-carbon and the amino group. We hoped
that the further constraint of the phenol side chain
imposed in B would result in a favorable entropic effect
on binding. The synthetic route to B (Scheme 1) was
based on previously described 3-substituted proline
derivatives.32 The amino acid analog B was synthesized
as a mixture of cis/ trans-isomers which could be
resolved by preparative C18 HPLC and incorporated into
peptides to yield the corresponding peptide analogs 57
and 58 (Table 5). Analog 57, which contains (()-trans-
B, showed 8-fold enhanced affinity relative to 17 in the
cell-based assay, while the corresponding (()-cis-isomer
58 had slightly reduced affinity relative to 17. Because
0.25
0.14
3.0
1.5
6.3
3.6
25
0.2
0.5
0.6
120
60
0.097
0.054
0.25
0.004
0.008
0.007
1.65
52 3-chloro-Tyr
53 4-(4-hydroxyphenyl)benzoyl
54 2,3,5,6-tetrafluoro-Tyr
55 6-hydroxynaphthoyl
56 2-phenyl-3-(4-hydroxyphenyl)propionyl 0.013
57 (()-trans-B
58 (()-cis-B
0.53
1.10
40
2.0
0.008
ND
0.0005
0.004
potency. The cyclic disulfide Ac-YCSPC (17) was chosen
as a reference compound for comparative purposes.
When the hydroxyl group of Tyr was moved to the meta

3364 J ournal of Medicinal Chemistry, 1997, Vol. 40, No. 21
J ackson et al.
Sch em e 1. Synthesis of
N-Acetyl-3-(4-hydroxyphenyl)proline (B)
Ta ble 6. Inhibition of R4â1/VCAM-1 by Cyclic YC(X)C
F igu r e 3. Additivity of functional group optimizations at the
Arg, Asp, and Pro positions of the lead peptide RCDPC (1).
length of five atoms between the two cystines was
determined based on data for the simple carbon analogs
60, 62, 64, and 65. The 3-aminobenzoyl analog 38,
which is also a five-atom linker, shows that aromatic
substitutions are well tolerated without significant loss
of activity. Overall the peptide backbone bridging the
two cystines appears to contribute substantially to the
activity of the cyclic peptides: with length, conforma-
tion, and hydrophobicity all being important factors.
Ad d itivity of Op tim iza tion s. In an effort to further
enhance the potency of this class of cyclic peptides,
optimizations at the N-terminus and the endocyclic
proline were combined (Figure 3). The effects of N-
terminal substitution with (()-trans-B (57) and Pro
substitution with the six-membered pipecolic acid (Pip)
analog 23 were essentially additive yielding 74 which
demonstrated subnanomolar potency in both the pro-
tein- and cell-based assays (Figure 3). Overall, 74 was
approximately 106 times more potent that the original
RCDPC lead peptide (1)19 and over 500 times more
potent than the native R4â1 ligand VCAM (ELISA IC50
∼ 50 nM). These results further demonstrate that
although Arg and Asp motifs may be functionally
important epitopes in the native R4â1 ligands VCAM-1
and fibronectin, they are not optimal for high-affinity
binding of RCDPC-like cyclic peptides to the R4â1
integrin receptor.
IC50 (µM)
no.
X
ELISA
Ramos
25
59
60
61
62
63
64
65
66
67
38
68
69
70
71
72
73
Ala-Pro
Pro-Ala
Gly
Ala
â-Ala
0.005
0.230
5.80
9.40
4.20
0.047
0.038
0.410
0.031
1.30
0.12
14.0
250
350
210
2.7
1.0
17
3-aminoisobutyroyl
4-aminobutyroyl
5-aminopentanoyl
isonipecotyl
1.5
80
2-aminobenzoyl
3-aminobenzoyl
4-aminobenzoyl
0.008
0.16
0.42
0.12
4.2
15
4-(aminomethyl)benzoyl
3-aminomethyl)benzoyl
4-(aminomethyl)cyclohexylcarbonyl
cyclohexyl-Gly
6.00
210
19
ND
ND
0.100
115.0
18.00
cyclohexyl-Ala
substitution of Tyr with D-Tyr (41) also resulted in a
substantial loss of affinity, we hypothesized that the
more active of the two trans-diastereomers present in
57 probably has an L-configuration at the R-carbon.
YC(X)C P ep tid es. With the relative individual
contributions of the Asp-Pro in RCDPC being well
established (Tables 2 and 3), analogs were designed to
probe the possibility of replacing both residues simul-
taneously with natural or synthetic amino acids. Our
goal was to reduce the cyclic peptides to a minimal high-
affinity inhibitor. As a reference compound for com-
parative purposes we chose YCAPC (25) (Table 6).
Significantly, compound 59 in which Ala and Pro were
swapped demonstrates the necessity of Pro being in the
correct position. Other substitutions in which backbone
length and side chains were varied generally yielded less
active compounds (Table 6). An optimal backbone
Selectivity for r4â1. The previously reported pep-
tide RCD(thioP)C was shown to bind both R4â1 and
R5â1 suggesting selectivity for â1 integrins.19 We tested
the cyclic peptide 17 (Ac-YCSPC) in two other previously
described in vitro ELISA assays including fibronectin/
R5â134 and fibrinogen/IIbIIa.17 Compound 17 did not
inhibit fibrinogen binding to GPIIbIIIa at concentrations
up to 500 µM. Inhibition of fibronectin binding to R5â1
was observed with 17 (IC50 ) 2.5 µM) which probably
reflects the binding component of 17 for the â1-subunit.

R4â1 Peptide Antagonists as Anti-Inflammatory Agents
J ournal of Medicinal Chemistry, 1997, Vol. 40, No. 21 3365
VCAM-1 and fibronectin. These peptides share impor-
tant functional groups with the first domain of VCAM-1
but show little direct structural similarity to the protein.
Significantly, the peptides can inhibit lymphocyte hom-
ing in vivo, making them potential therapeutic agents
for treatment of a variety of inflammatory diseases.
Furthermore, we have used these peptides as templates
for the structure-based design of potent small molecule
antagonists (which are also active in vivo) of the VCAM/
R4â1 interaction; reports on the design, structure, and
activities of these novel small molecules are forthcom-
ing.
From an evolutionary point of view, it is possible that
selective pressure on cell adhesion molecules leads to
optimization of specificity rather than affinity. The
charged epitopes found in many integrin ligands may
possess the necessary properties. One can imagine
slight variations in conformation leading to a diverse
range of integrin-binding specificities, allowing very
similar adhesion molecules to recruit a variety of
different inflammatory cells at different stages of in-
flammation. The evolution of high-affinity interactions
between cell adhesion molecules and their cell surface
integrin receptors may have proven undesirable by
preventing mobilization or extravasation, with the leu-
kocytes being virtually stuck to the endothelium unable
to carry out their functions.
F igu r e 4. Inhibition of lymphocyte homing by 17 versus the
known R4 specific antibody (PS/2)26 and the negative control
peptide YASPA (75).
Although therapeutic strategies involving integrin
antagonists are relatively new, antibodies, peptides, and
small molecules that block integrin interactions are
currently under clinical investigation for a variety of
indications, and new strategies are constantly being
proposed. Recently, Pasqualini et al. showed that a
soluble high-affinity form of fibronectin (superfibronec-
tin) could inhibit tumor cell metastasis presumably by
blocking the interaction of R5â1 positive tumor cells
with wild-type fibronectin in vivo, thus expanding the
potential for integrin antagonists.37 The results pre-
sented here provide still further evidence that blocking
integrin-mediated cell adhesion interactions, specifically
R4â1/VCAM, may lead to new classes of small molecule
anti-inflammatory agents.
Although 17 does inhibit R5â1/fibronectin, it is over
1000 times more potent as an inhibitor of R4â1/VCAM
(IC50 ) 2 nM) demonstrating a remarkable selectivity
of this peptide for R4â1. Highly selective antagonists
are desirable to maximize anti-inflammatory activity
while minimizing potential side effects due to unwanted
inhibition of other protein interactions.
In Vivo Lym p h ocyte Hom in g Activity. Lympho-
cyte homing to Peyer’s patch high-endothelial venules
(HEVs) is known to be dependent upon the interaction
of cell adhesion molecules expressed on HEVs and R4
integrins expressed on lymphocytes.5,33,35 The accumu-
lation of radiolabeled lymphocytes in Peyer’s patches
can be inhibited by treatment with anti-R4 antibodies.36
In order to assess the activity of a representative cyclic
peptide in vivo, mice were treated intravenously with
analog 17 (0.5 mg/mouse) prior to injection with 51Cr-
labeled mesenteric lymph node cells (see Experimental
Methods). As negative controls, mice were treated with
buffer or the linear peptide YASPA (75) which does not
inhibit the R4â1/VCAM-1 interaction. An anti-R4 IgG
(PS/2) was used as a positive control. The data shown
in Figure 4 demonstrate that pretreatment of mice with
17 significantly inhibits lymphocyte accumulation in
Peyer’s patches while similar treatment with the control
peptide 75 does not. Importantly, 17 did not inhibit
migration to other tissues such as the inguinal lymph
nodes where VCAM-1 is not expressed in high levels
(Figure 4). These results suggest that peptide or small
molecule antagonists similar to 17 may be useful for
treating R4â1-associated inflammation in vivo.
Exp er im en ta l Meth od s
P ep tid e Syn th esis. All peptides were synthesized manu-
ally using standard solid phase peptide chemistry38 with Fmoc-
protected amino acids39 on a p-alkoxybenzyl alcohol resin.40
All amino acids were purchased from Bachem (CA), Advanced
ChemTech U.S.A., or Calbiochem Corp. (CA). Couplings were
performed with 2-4 equiv of HBTU-activated amino acid and
4 equiv of N-methylmorpholine. Fmoc groups were removed
with 20% piperidine in DMA. Cleavage and deprotection with
TFA containing 5% triethylsilane afforded crude linear pep-
tides after washing the resin with diethyl ether to remove
protecting groups. The crude peptides were then extracted
from the resin by stirring with 100 mL of 2:1 H2O/CH3CN for
5 min followed by filtration. Disulfide oxidation was carried
out at 25 °C via dropwise addition of a saturated solution of
iodine in acetic acid to the crude extracts with vigorous stirring
until a slight yellow color persisted. The crude oxidized
peptides were lyophilized and purified by preparative reverse
phase C18 HPLC (CH3CN/H2O gradient, 0.1% TFA) to afford
purified material. Pure fractions (>98% pure by analytical
HPLC) were characterized by electrospray ionization mass
spectrometry (Sciex API100) and lyophilized to dryness. The
purified peptides were dissolved in DMSO at 10 mM just prior
to assay. Serial dilutions of peptide starting at 0.5 mM were
titrated into the ELISA assay, and the IC50 for each peptide
was determined.
Con clu sion s
We have demonstrated that cyclic peptides can com-
petitively inhibit the interaction of the integrin R4â1
with VCAM-1 at subnanomolar concentrations. The
peptides bind R4â1 with affinities several orders of
magnitude tighter than the native protein ligands

3366 J ournal of Medicinal Chemistry, 1997, Vol. 40, No. 21
J ackson et al.
N-Acetyl-3-(4-h yd r oxyp h en yl)p r olin e (B). The amino
acid analog B (Table 5) was synthesized (see Scheme 1) using
a modified procedure of Chung et al.32 and incorporated into
peptides 57 and 58 (Table 5) as described.
5 mL of aqueous 4% sodium bicarbonate. Incubation overnight
at room temperature afforded 9 mg of the N-labeled product
BCK(N-fluorescene thiourea)PC (77) (IC50 )10 nM after
purification by preparative C18 HPLC. The product 77 was
characterized by HPLC and mass spectrometry as above and
gave a negative ninhydrin test confirming that labeling
occurred on the lysine side chain.
(1) To a solution of diethyl acetamidomalonate (73.7 g) in
600 mL of anhydrous ethanol was added 20 mL of sodium
ethoxide (1 M in ethanol), the solution was stirred at room
temperature for 15 min, 4-methoxycinnamaldehyde (50 g) was
added in one portion, and stirring was continued overnight.
The solvent was evaporated, and the crude residue was
chromatographed on silica gel (4:1 ethyl acetate/hexane).
Evaporation of solvent afforded 106 g (91%) of pure diethyl
1-acetyl-5-hydroxy-3-(4-methoxyphenyl)pyrrolidine-2,2-dicar-
boxylate: 1H NMR (300 MHz, CDCl3) ∂ 7.11 (d, 2H, J ) 9 Hz,
aromatic H), 6.86 (d, 2H, J ) 9 Hz, aromatic H), 4.29 (q, J )
7.5 Hz, 2H), 3.89 (m, 3H), 3.81 (s, 3H), 3.75 (m, 2H), 2.61 (m,
1H), 2.26 (m, 1H), 2.15 (s, 3H), 1.3 (t, J ) 7.5 Hz, 3H), .86 (t,
J ) 7.5 Hz, 3H); HRMS calcd for C19H25NO6 363.1682, found
363.1684.
(2a) To 100 g of diethyl 1-acetyl-5-hydroxy-3-(4-methoxy-
phenyl)pyrrolidine-2,2-dicarboxylate (from step 1) in 500 mL
of CHCl3 was added 50 g of triethylsilane followed by 150 mL
of trifluoroacetic acid dropwise. The solution was stirred
overnight at room temperature and evaporated to dryness to
afford 92 g of diethyl 1-acetyl-3-(4-methoxyphenyl)pyrrolidine-
2,2-dicarboxylate which was used directly in the next step
without further isolation or purification.
(2b) Diethyl 1-acetyl-3-(4-methoxyphenyl)pyrrolidine-2,2-
dicarboxylate (90 g, from above) was dissolved in dioxane (500
mL), a solution of sodium hydroxide (30 g) in water (250 mL)
was added, and the mixture was heated at reflux for 48 h.
After cooling the solution was filtered and the filtrate acidified
with concentrated HCl. The acidified solution was concen-
trated under vacuum to remove dioxane/water and resus-
pended in 300 mL of water. The product was extracted into
ethyl acetate (3 × 300 mL), the organics were concentrated,
and the product was purified on silica gel (eluted with ethyl
acetate) to afford 54 g of 1-acetyl-3-(4-methoxyphenyl)pyrro-
lidine-2-carboxylic acid (77%, two steps) after evaporation of
solvent: HRMS calcd for C14H17NO4 263.1157, found 263.1159;
1H NMR (300 MHz, CDCl3) ∂ 7.12 (d, 2H, J ) 9 Hz, aromatic
H), 6.87 (d, 2H, J ) 9 Hz, aromatic H), 4.61 (d, 1H, J ) 5 Hz,
CH), 3.87 (m, 1H, CH), 3.79 (s, 3H, OCH3), 3.65 (m, 2H, CH2),
2.40 (m, 1H, CH), 2.10 (d, 3H, J ) 9.6 Hz, acetyl), 2.03 (m,
1H, CH).
P ep t id e Ma ss Sp ect r a l Da t a , P ep t id e (ca lcd ; fou n d
MH+): 1 (590.67; 591.2 MH+), 2 (634.22; 635.2 MH+), 3 (562.20;
563.0 MH+), 4 (816.41; 818.2 MH+), 5 (703.29; 704.2 MH+), 6
(737.28; 738.5 MH+), 7 (776.29; 776.5 MH+), 8 (753.27; 754.1
MH+), 9 (725.27; 726.2 MH+), 10 (668.73; 669.5 MH+), 11
(597.65; 598.2 MH+), 12 (767.29; 767.7 MH+), 13 (752.29; 753.2
MH+), 14 (766.30; 767.2 MH+), 15 (595.64; 596.5 MH+), 16
(639.69; 640.5 MH+), 17 (611.68; 612.4 MH+), 18 (625.20; 626.0
MH+), 19 (645.21; 646.2 MH+), 20 (595.19; 596.0 MH+), 21
(555.16; 556.2 MH+), 22 (569.18; 570.0 MH+), 23 (609.71; 610.2
MH+), 24 (631.71; 632.2 MH+), 25 (553.65; 554.2 MH+), 26
(569.21; 570.2 MH+), 27 (569.21; 570.2 MH+), 28 (603.20; 604.0
MH+), 29 (607.13; 608.2 MH+), 30 (569.17; 570.2 MH+), 31
(659.23; 660.2 MH+), 32 (595.17; 596.2 MH+), 33 (581.32; 582.0
MH+), 34 (616.72; 617.5 MH+), 35 (589.23; 590.2 MH+), 36
(782.88; 783.2 MH+), 37 (781.91; 782.5 MH+), 38 (504.29; 505.2
MH+), 39 (503.31; 504.5 MH+), 40 (594.19; 595.2 MH+), 41
(569.18; 570.0 MH+), 42 (553.18; 554.2 MH+), 43 (510.21; 511.0
MH+), 44 (517.12; 518.0 MH+), 45 (555.12; 556.0 MH+), 46
(599.31; 600.2 MH+), 47 (575.11; 576.0 MH+), 48 (575.11; 576.0
MH+), 49 (569.18; 570.2 MH+), 50 (695.07; 696.0 MH+), 51
(587.11; 588.0 MH+), 52 (603.11; 604.0 MH+), 53 (602.67; 603.5
MH+), 54 (641.11; 642.0 MH+), 55 (578.11; 579.2 MH+), 56
(631.11; 632.2 MH+), 57 (637.11; 638.2 MH+), 58 (637.11; 638.2
MH+), 59 (553.64; 554.5 MH+), 60 (442.50; 443.5 MH+), 61
(456.13; 457.2 MH+), 62 (456.13; 457.2 MH+), 63 (470.21; 470.8
MH+), 64 (470.21; 471.0 MH+), 65 (496.51; 497.5 MH+), 66
(510.51; 511.5 MH+), 67 (504.29; 505.2 MH+), 68 (504.29; 505.2
MH+), 69 (518.59; 519.2 MH+), 70 (518.59; 519.2 MH+), 71
(524.09; 525.0 MH+), 72 (510.28; 511.2 MH+), 73 (524.28; 525.2
MH+), 74 (651.11; 652.0 MH+), 75 (507.61; 508.5 MH+), 76
(678.82; 679.9 MH+), 77 (1068.19; 1069.3 MH+).
Recep tor Bin d in g ELISA. Inhibitor concentrations af-
fording 50% inhibition (IC50s) of R4â1 binding to VCAM-1 was
determined by a protein-based receptor binding ELISA. The
R4â1 was extracted from Ramos cell membranes using wheat
germ lectin Sepharose chromatography followed by gel filtra-
tion in 1% octyl glucoside. MnCl2 (1 mM) was included in all
buffers during purification procedures and assays to maintain
the active binding state. Recombinant soluble human VCAM-1
(55 kDa fragment composed of the first five N-terminal Ig-
like domains) was purified from Chinese hamster ovary (CHO)
cell culture media. Nunc Maxisorp 96-well plates were coated
with 4 µg/mL VCAM-1 in phosphate-buffered saline. The wells
were blocked with 1% bovine serum albumin in phosphate-
buffered saline. Diluted samples of test inhibitor molecules
were evaluated for blocking of R4â1 binding to VCAM-1 by
addition of 50 µL of each inhibitor to the VCAM-1-coated wells
prior to the addition of 50 µL of optimally diluted R4â1. The
assay buffer was composed of 50 mM Tris-HCl (pH 7.4), 100
mM NaCl, 1 mM MnCl2, and 0.05% Tween-20. The sample
and R4â1 were allowed to incubate in the wells for 2 h at 37
°C. The bound R4â1 was detected with a nonblocking mouse
anti-human â1 integrin monoclonal antibody (clone 2D4.6;
Genentech, Inc.) followed by goat anti-mouse horseradish
peroxidase (BioSource International, Camarillo, CA). Peroxi-
dase activity was detected with TMB microwell peroxidase
substrate (Kirkegaard and Perry Laboratories, Inc., Gaithers-
burg, MD). Reactions were stopped with 1 M phosphoric acid,
and absorbance was measured at 450 nm. Results were
plotted as absorbance vs concentration, and the concentration
of peptide at the half-maximal absorbance value is reported
(3) 1-Acetyl-3-(4-methoxyphenyl)pyrrolidine-2-carboxylic acid
(50 g, from step 2b) was dissolved in CH2Cl2 (700 mL) and
cooled to -78 °C under N2. Boron tribromide (1 M in CH2Cl2,
380 mL) was added dropwise with stirring, and the solution
was allowed to warm to room temperature. Stirring was
continued overnight; the reaction mixture was poured into 500
g of ice and neutralized via addition of solid sodium bicarbon-
ate. The aqueous layer was separated and acidified with
concentrated HCl. The product was extracted into ethyl
acetate (3 × 300 mL) and purified on silica gel (9:1 ethyl
acetate/acetic acid) to afford 41 g (87%) of N-acetyl-3-(4-
hydroxyphenyl)proline (B) after evaporation of solvent. The
cis/ trans-isomers were separated by preparative C18 HPLC.
Ratio of isomers ) 3:1 trans/ cis by analysis: C18 HPLC (25
cm × 1.5 mm, 1.5 mL/min, 20-50% CH3CN/H2O in 15 min) tR
) 7.21 min (trans), 7.63 min (cis); HRMS calcd for C13H15NO4
249.1001, found 249.1004; 1H NMR for (()-trans-B (300 MHz,
CD3OD) ∂ 7.10 (d, 2H, J ) 8.7 Hz, aromatic), 6.75 (d, 2H, J )
8.7 Hz, aromatic), 4.31 (d, 1H, J ) 6.6 Hz, CHR), 3.79 (m, 1H,
CH), 3.75 (m, 1H, CH), 3.40 (m, 1H, CH), 2.35 (m, 1H, CH),
1
2.11 (s, 3H, acetyl), 2.08 (m, 1H, CH); H NMR for (()-cis-B
(300 MHz, CD3OD) ∂ 7.10 (d, 2H, J ) 8.7 Hz, aromatic), 6.75
(d, 2H, J ) 8.7 Hz, aromatic), 4.40 (d, 1H, J ) 5.4 Hz, CHR),
3.79 (m, 1H, CH), 3.75 (m, 1H, CH), 3.40 (m, 1H, CH), 2.35
(m, 1H, CH), 2.11 (s, 3H, acetyl), 2.08 (m, 1H, CH). Cis/ trans
assignments were made by analogy of CHR coupling constants
with N-acetyl-3-phenylproline from ref 32.
as the IC50
.
Cell Ad h esion Bin d in g Assa ys. As a secondary assay,
the ability of test molecules to inhibit binding of Ramos cells
(American Type Culture Collection, Bethesda, MD) to VCAM-1
was assessed. Ramos cells were labeled with calcein (Molec-
ular Probes, Eugene, OR). Nunc Maxisorp 96-well plates were
coated with 2 µg/mL rabbit anti-human IgG (Fc specific;
Syn th esis of BCK(N-flu or escen e th iou r ea )P C (77).
The peptide BCKPC (76) (10 mg, ELISA IC50 ) 2 nM) was
synthesized using standard methods as descibed and added
to a solution of fluorescene isothiocyanate (10 mg; Sigma) in

R4â1 Peptide Antagonists as Anti-Inflammatory Agents
J ournal of Medicinal Chemistry, 1997, Vol. 40, No. 21 3367
J ackson Immunoresearch Laboratories, West Grove, PA). The
wells were blocked with 1% bovine serum albumin in phosphate-
buffered saline. Human VCAM-1-IgG chimera (domains 1
and 2 of human VCAM-1 linked to the CH2 and CH3 regions
of human IgG1) was captured onto the anti-IgG-coated sur-
faces during a 1 h incubation. Diluted test compounds (10 µL)
were added to 40 µL of binding buffer (4 mM CaCl2, 4 mM
MgCl2 in 50 mM Tris-HCl, pH 7.5) in the washed assay wells.
This was followed by 2 × 105 cells in 50 µL of 1% bovine serum
albumin in phosphate-buffered saline. Cells were allowed to
bind for 1 h at room temperature. Unbound cells were
removed by gentle washing. Bound cells were lysed with 1%
SDS in Tris-HCl at pH 8.0. The fluorescence was recorded on
a Cytofluor 2350 plate reader (Millipore, Bedford, MA) with
filters set at 485 nm excitation and 530 nm emission.
NMR Str u ctu r a l An a lysis. Peptide 5 (3.5 mg) was dis-
solved in 500 µL of 90% H2O/10% D2O/25 mM sodium acetate-
d3, pH 3.9. All NMR spectra were acquired at 7 °C on a Bruker
AMX-500 spectrometer. In addition to a high-resolution one-
dimensional 1H NMR spectrum, the following homonuclear
two-dimensional (2D) NMR spectra were recorded using
standard pulse sequences and phase cycling: COSY, TOCSY,
NOESY, and ROESY.42 All 2D spectra were acquired in the
phase-sensitive mode using time-proportional phase incre-
mentation for quadrature detection in the t1 dimension.43 The
solvent signal was suppressed by low-power phase-locked
presaturation for 1.5 s. The TOCSY spectrum was acquired
using a 96 ms “clean” DIPSI-2rc isotropic mixing sequence.44
The NOESY spectrum was acquired with irradiation of solvent
during the 350 ms mixing time. The ROESY spectrum was
acquired with a mixing time of 175 ms and a spin-lock field
strength of 4 kHz. Spectra were processed and analyzed on
Silicon Graphics workstations using the program FELIX
(Biosym Technologies). Resonance assignments were obtained
using standard methods utilizing the 2D spectra discussed
above.45
NOE cross-peaks were characterized according to their
integrated volumes as strong, medium, or weak, corresponding
to upper bound distance restraints of 2.7, 3.5, and 4.5 Å,
respectively; pseudoatom corrections were added when neces-
sary.46 Dihedral angle restraints for φ and ø1 and stereospecific
resonance assignments for the â-methylenes of Asp4, Pro5, and
Cys6 were derived from coupling constants measured from 1D
spectra and ROE data. Distance geometry calculations were
carried out using DGII (Biosym Technologies). Fifty structures
were calculated, and the 30 with the lowest total restraint
violations were further refined with restrained molecular
dynamics and minimization using the program DISCOVER
(Biosym Technologies). The all-atom AMBER force field47,48
was used with a 15.0 Å cutoff for nonbonded interactions and
a distance-dependent dielectric constant (ꢀ ) 4r) to compensate
for the lack of explicit solvent. Side chain and terminal
charges were reduced to give total charges of (0.2 in order to
reduce artifacts due to charge-charge interactions.
this project and for supplying us with the human VCAM
used in the protein assay.
Refer en ces
(1) Hynes, R. O. Integrins: versatility, modulation, and signaling
in cell adhesion. Cell 1992, 69, 11-25.
(2) Hemler, M. E. VLA proteins in the integrin family: structures,
functions, and their role on leukocytes. Annu. Rev. Immunol.
1990, 8, 365-368.
(3) Sharar, S. R.; Winn, R. K.; Harlan, J . M. The adhesion cascade
and anti-adhesion therapy: an overview. Springer Semin. Im-
munopathol. 1995, 16, 359-378.
(4) Nakajima, H.; Sano, H.; Nishimura, T.; Yoshida, S.; Iwamoto,
I. Role of VCAM-1/VLA-4 and ICAM-1/LFA-1 interactions in
antigen-induced eosinophil and T cell recruitment into the tissue.
J . Exp. Med. 1994, 179, 1145-1154.
(5) Butcher, E. C.; Picker, L. J . Lymphocyte homing and homeo-
stasis. Science 1996, 272, 60-66.
(6) Elices, M. J .; Osborn, L.; Takada, Y.; Crouse, C.; Luhowsky, S.;
Hemler, M. E.; Lobb, R. R. VCAM-1 on activated endothelium
interact with the leukocyte integrin VLA-4 at a site distinct from
the VLA-4/fibronectin binding site. Cell 1990, 60, 577-584.
(7) Makarem, R.; Newham, P.; Askari, J . A.; Green, L. J .; Clements,
J .; Edwards, M.; Humphries, M. J .; Mould, A. P. Competitive
binding of VCAM-1 and the HepII/IIICS domain of fibronectin
to the integrin R4â1. J . Biol. Chem. 1994, 269, 4005-4011.
(8) Chuluyan, H. E.; Issekutz, A. C. R4 Integrin-dependent emigra-
tion of monocytes. Springer Semin. Immunopathol. 1995, 16,
391-404.
(9) Simmons, P. J .; Masinovsky, B.; Longenecker, B. M.; Berenson,
R.; Torok-Storb, B.; Gallatin, W. M. VCAM-1 expressed by bone
marrow stromal cells mediates the binding of hematopoietic
progenitor cells. Blood 1992, 80, 388-395.
(10) J uliano, R. L.; Varner, J . A. Adhesion molecules in cancer: the
role of integrins. Curr. Opin. Cell Biol. 1993, 5, 812-818.
(11) Laberge, S.; Rabb, H.; Issekutz, T. B.; Martin, J . G. Role of VLA-4
and LFA-1 in allergen-induced airway hyperresponsiveness and
lung inflammation in the rat. Am. J . Respir. Crit Care Med.
1995, 151, 822-829.
(12) Barbadillo, C.; G-Arroyo, A.; Salas, C.; Mulero, J .; Sanchez-
Madrid, F.; Andreu, J . L. Anti-integrin immunotherapy in RA:
protective effect of anti-R4 antibody in adjuvant arthritis.
Springer Semin. Immunopathol. 1995, 16, 375-379.
(13) Podalski, D. K. Inflammatory bowell disease. N. Engl. J . Med.
1991, 325, 928-937.
(14) Powrie, F.; Leach, M. W. Genetic and spontaneous models of
inflammatory bowel disease in rodents: evidence for abnormali-
ties in mucosal immune regulation. Ther. Immunol. 1995, 2,
115-123.
(15) Papayannopoulou, T.; Craddock, C.; Nakamoto, B.; Priestley, G.
V.; Wolf, N. S. The VLA-4/VCAM-1 adhesion pathway defines
contrasting mechanisms of lodgement of transplanted murine
hemopoietic progenitors between bone marrow and spleen. Proc.
Natl. Acad. Sci. U.S.A. 1995, 92, 9647-9651.
(16) Papayannopoulou, T.; Nakamoto, B. Peripheralization of he-
mopoietic progenitors in primates treated with anti-VLA-4
integrin. Proc. Natl. Acad. Sci. U.S.A. 1993, 90, 9374-9378.
(17) Barker, P. L.; Bullens, S.; Bunting, S.; Burdick, D. J .; Chan, K.
S.; Deisher, T.; Eigenbrot, C.; Gadek, T. R.; Gantzos, R.; Lipari,
M. T.; Muir, C. D.; Napier, M.; Pitti, R. M.; Padua, A.; Quan, C.;
Stanley, M.; Struble, M.; Tom, J .; Burnier, J . P. Cyclic RGD
peptide analogs as antiplatelet antithrombotics. J . Med. Chem.
1992, 35, 2040-2048.
Lym p h ocyte Hom in g Assa y. The in vivo activity of
inhibitor 17 was evaluated in a lymphocyte homing assay
using 4 × 106 51Cr-labeled syngeneic mesenteric lymph node
cells administered to normal 6-8 week old female BALB/c mice
(J ackson Laboratories, Bar Harbor, ME) according to a modi-
fied procedure described by Hamann et al.41 Cells were labeled
with 150 µCi of 51Cr (Amersham, Arlington Heights, IL) for
25 min at 37 °C. Mice were injected intravenously with
radiolabeled cells in the presence of 0.5 mg of test peptide at
time 0. The animals were sacrificed 1 h after receiving the
radiolabeled cells. The Peyer’s patches and inguinal lymph
nodes were collected and counted for accumulated radioactiv-
ity. The PS/2 monoclonal antibody against the R4 subunit
(American Type Culture Collection) was used as a positive
control at 0.67 µM given intravenously at time 0. The number
of lymph node cells based upon accumulated radioactivity in
various tissues is reported.
(18) McDowell, R. S.; Blackburn, B. K.; Gadek, T. R.; McGee, L. R.;
Rawson, T.; Reynolds, M. E.; Robarge, K. D.; Somers, T. C.;
Thorsett, E. D.; Tichler, M.; Webb, R.; Venuti, M. C. From
peptide to non-peptide. The de novo design of potent, non-
peptidal inhibitors of platelet aggregation based on a benzodi-
azepinedione scaffold. J . Am. Chem. Soc. 1994, 116, 5077-5083.
(19) Nowlin, D. M.; Gorcsan, F.; Moscinski, M.; Chiang, S.; Lobl, T.
J .; Cardarelli, P. M. A novel cyclic pentapeptide inhibits R4â1
and R5â1 integrin-mediated cell adhesion. J . Biol. Chem. 1993,
268, 20352-20359.
(20) Kieffer, N.; Phillips, D. R. Platelet membrane glycoproteins:
functions in cellular interactions. Annu. Rev. Cell Biol. 1990, 6,
329-357.
(21) Pytela, R.; Pierschbacher, M. D.; Ruoslahti, E.; A 125/115-kDa
cell surface receptor specific for vitronectin interacts with the
RGD adhesion sequence derived from fibronectin. Proc. Natl.
Acad. Sci.U.S.A. 1985, 82, 5766-5770.
(22) Cardarelli, P. M.; Yamagata, S.; Taguchi, I.; Gorscan, F.; Chiang,
S.; Lobl, T. The collagen receptor R2â1, from MG-63 and HT1080
cells, interactes with a cyclic RGD peptide. J . Biol. Chem. 1992,
267, 23159-23164.
(23) Osborn, L.; Vassallo, C.; Benjamin, C. D. Activated endothelium
binds lymphocytes through a novel binding site in the alterna-
tively spliced domain of VCAM-1. J . Exp. Med. 1992, 176, 99-
107.
Ack n ow led gm en t. We would like to thank our col-
laborators J eff Tilley, Barry Wolitzky, Kuo-Sen Huang,
and Shirley Li at Roche (Nutley, NJ ) for their input on

3368 J ournal of Medicinal Chemistry, 1997, Vol. 40, No. 21
J ackson et al.
(24) J ones, E. Y.; Harlos, K.; Bottomley, M. J .; Robinson, R. C.;
Driscoll, P. C.; Edwards, R. M.; Clements, J . M.; Dungeon, T.
J .; Stuart, D. I. Crystal structure of an integrin-binding fragment
of vascular cell adhesion molecule-1 at 1.8A resolution. Nature
1995, 373, 539-544.
either of two integrin â chains, one of which is novel. EMBO
1989, 8, 1735-1741.
(36) Holzmann, B.; McIntyre, B. W.; Weismann, I. L. Identification
of a murine Peyer’s patch-specific lymphocyte homing receptor
as an integrin molecule with an a chain homologous to human
VLA-4R. Cell 1989, 56, 37-46.
(37) Pasqualini, R.; Bourdoulous, S.; Koivunen, E., J r.; Ruosllahti,
E. A polymeric form of fibronectin has antimetastatic effects
against multiple tumor types. Nature Medicine 1996, 2, 1197-
1203.
(38) Merrifield, R. B. Solid phase peptide synthesis 1. The synthesis
of a tetrapeptide. J . Am. Chem. Soc. 1963, 85, 2149-2154.
(39) Carpino, L. A.; Han, G. Y.; Fluorenylmethyl carbamate protection
for amino acids. J . Org. Chem. 1972, 37, 3404-3409.
(40) Wang, S. S.; Kulesha, I. D.; Winter, D. P.; Makofske, R.; Kutney,
R.; Meinhofer, J . Preparation of protected peptide hydrazides
from the acids and hydrazine by dicyclohexylcarbodiimide-
hydroxybenzotriazole coupling. Int. J . Pept. Protein Res. 1978,
11 (4), 297-299.
(41) Hamann, A.; Andrew, D. P.; J ablonski-Westrich, D.; Holzmann,
B.; Butcher, E. C. Role of R4-integrins in lymphocyte homing to
mucosal tissues in vivo. J . Immunol. 1994, 152, 3282-3285.
(42) Cavanagh, J .; Fairbrother, W. J .; Palmer, A. G.; Skelton, N. J .
Protein NMR Spectroscopy: Principles and Practice; Academic
Press: San Diego, CA, 1995.
(43) Marion, D.; Wuthrich, K. Application of phase sensitive two-
dimensional correlated spectroscopy (COSY) for measuring 1H-
1H spin-spin coupling constants in proteins. Biochem. Biophys.
Res. Commun. 1983, 113, 967-974.
(44) Cavanagh, J .; Rance, M. Suppression of cross-relaxation effects
in TOCSY spectra via a modified DIPSI-2 mixing sequence. J .
Magn. Reson. 1992, 96, 670-678.
(45) Wuthrich, K. NMR of Proteins and Nucleic Acids; J ohn Wiley
& Sons, Inc.: New York, 1986.
(46) Wuthrich, K.; Billeter, M.; Braun, W. Pseudostructures for the
20 common amino acids for use in studies of protein conforma-
tion by measurements of intramolecular proton-proton distance
constraints with nuclear magnetic resonance. J . Mol. Biol. 1983,
169, 949-961.
(47) Weiner, S. J .; Kollman, P. A.; Case, D. A.; Singh, U. C.; Ghio,
C.; Alagona, G. S.; Profeta, J .; Weiner, P. A new force field for
molecular mechanical simulation of nucleic acids and proteins.
J . Am. Chem. Soc. 1984, 106, 765-784.
(48) Weiner, S. J .; Kollman, P. A.; Nguyen, D. T.; Case, D. A. An
all-atom force field for simulations of proteins and nucleic acids.
J . Comput. Chem 1986, 7, 230-252.
(25) Wang, J .; Pepinsky, R. B.; Stehle, T.; Liu, J .; Karpusas, M.;
Browning, B.; Osborn, L. The crystal structure of an N-terminal
two-domain fragment of vascular cell adhesion molecule
1
(VCAM-1): a cyclic peptide based on the domain 1 C-D loop can
inhibit VCAM-1-R4 integrin interaction. Proc. Natl. Acad. Sci.
U.S.A. 1995, 92, 5714-5718.
(26) Renz, M. E.; Chiu, H. H.; J ones, S.; Fox, J .; Kim, K. J .; Presta,
L. G.; Fong, S. Structural requirements for adhesion of soluble
recombinant murine VCAM-1 to R4â1. J . Cell. Biol. 1994, 125,
1395-1406.
(27) Quan, C.; Skelton, N.; Clark, K.; J ackson, D. Y.; Renz, M. E.;
Chui, H. H.; Keating, S. M.; Beresini, M.; Fong, S.; Artis, D. R.
Structure-based design of small, potent cyclic peptides which
mimic the binding epitope of VCAM. Submitted for publication.
(28) Chiu, H. H.; Crowe, D. T.; Renz, M. E.; Presta, L. G.; J ones, S.;
Weismann, I. L.; Fong, S. Similar but nonidentical amino acid
residues on VCAM-1 are involved in the interaction with R4â1
and R4â7 under different activity states. J . Immunol. 1995, 155,
5257-5267.
(29) Pierschbacher, M. D.; Ruoslahti, E. Influence of stereochemistry
of the sequence Arg-Gly-Asp-Xaa on binding specificity in cell
adhesion. J . Biol. Chem. 1987, 262, 17294-17298.
(30) Lobl, T. J .; Chiang, S.; Cardarelli, P. M. Cyclic cell adhesion
modulation compounds. World Patent 92/00995, 1992.
(31) Komoriya, A.; Green, L. J .; Mervic, M.; Yamada, S. S.; Yamada,
K. M.; Humphries, M. J . The minimal essential sequence for a
major cell type-specific adhesion site (CS1) within the alterna-
tively spliced type III connecting segment domain of fibronectin
is leucine-aspartic acid-valine. J . Biol. Chem. 1991, 266, 15075-
15079.
(32) Chung, J . Y. L.; Wasicak, J . T.; Arnold, W. A.; May, C. S.;
Nadzan, A. M.; Holiday, M. W. Conformationally constrained
amino acids. Synthesis and optical resolution of 3-substituted
proline derivatives. J . Org. Chem. 1990, 55, 270-275.
(33) Viney, J . L.; J ones, S.; Chiu, H. H.; Lagrimas, B.; Renz, M. E.;
Presta, L. G.; J ackson, D. Y.; Hillan, K. J .; Lew, S.; Fong, S.
Mucosal addressin cell adhesion molecule 1: a structural and
functional analysis demarcates the integrin binding motif. J .
Immunol. 1996, 157, 2488-2497.
(34) Koivunen, E.; Gay, D. A.; Ruoslahti, E. Selection of peptides
binding to the R5â1 integrin from phage display library. J . Biol.
Chem. 1993, 266, 20205-20210.
(35) Holzmann, B.; Weismann, I. L. Peyer’s patch-specific lymphocyte
homing receptors consist of a VLA-4-like R chain associated with
J M970175S