(Received for publication, December 12, 1996, and in revised form, February 26, 1997)
From the Kimmel Cancer Institute, Jefferson Medical College, Thomas Jefferson University, Philadelphia, Pennsylvania 19107
The interaction between CD4 and major
histocompatibility complex class II proteins provides a critical
co-receptor function for the activation of CD4+ T
cells implicated in the pathogenesis of a number of autoimmune diseases
and transplantation responses. A small synthetic cyclic heptapeptide
was designed and shown by high resolution NMR spectroscopy to closely
mimic the CD4 domain 1 CC surface loop. This peptide effectively
blocked stable CD4-major histocompatibility complex class II
interaction, possessed significant immunosuppressive activity in
vitro and in vivo, and strongly resisted proteolytic degradation. These results demonstrate the therapeutic potential of
this peptide as a novel immunosuppressive agent and suggest a general
strategy of drug design by using small conformationally constrained
peptide mimics of protein surface epitopes to inhibit protein
interactions and biological functions.
Protein-protein interactions play an important role in a wide range of physiological and pathological processes. The interactions between proteins generally involve large interfaces with many intermolecular contacts (1). As such, the rational design of small molecular inhibitors of these surfaces has long been considered a formidable challenge. Despite this commonly held view, recent studies indicate that proteins may actually interact through small surface binding epitopes, as in the human growth hormone-receptor complex (2) and the erythropoietin-receptor complex (3). These findings raise an intriguing possibility that mimics of such small binding epitopes may be sufficient for blockade of a large protein-protein interface. However, the general applicability of this hypothesis and its implications for rational drug design remain to be tested and demonstrated in different biological systems. Undoubtedly, the development of a general approach to inhibit protein-protein interactions will have a tremendous impact on understanding the structural basis of these interactions and in developing new therapeutic strategies for many human diseases.
CD4 is a glycoprotein consisting of four Ig-like extracellular domains (D1-D4)1 and is expressed on the surface of helper T cells (4). Major histocompatibility complex (MHC) class II is a heterodimeric glycoprotein expressed on the surface of antigen-presenting cells and binds antigenic peptides for recognition by the T cell receptor. The interaction between CD4 and non-polymorphic regions of the MHC class II molecule is critical for optimal CD4+ T cell activation, with CD4 serving as a co-receptor for T cell receptor-antigen engagement (5). Numerous mutation studies have been performed to determine the regions of CD4 involved in MHC class II binding, and like many protein-protein complexes, the interface is generally believed to involve a large surface area of both D1 and D2, with many contact sites (6-9).
CD4+ T cells participate in the pathogenesis of a number of immune-based human conditions, including autoimmune diseases, allogenic organ transplant rejection, and graft versus host disease (GVHD) following allogenic bone marrow transplantation. Small molecular inhibitors of the CD4-MHC class II interaction could potentially block the undesirable activation of CD4+ T cells and could thus serve as effective immunosuppressive agents.
As described previously (10), the
peptides were prepared by solid-phase synthesis using Fmoc strategy on
a Model 430A peptide synthesizer (Applied Biosystems, Inc., Foster
City, CA) and a Model 9050 Pepsynthesizer Plus (Perseptive Biosystems,
Cambridge, MA). A 4-fold excess of
N-Fmoc-amino acid,
2-(1H-benzotriale-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate,
and 1-hydroxybenzotriazole and a 10-fold excess of
diisopropylethylamine were used in every coupling reaction step.
Removal of the N-terminal Fmoc group was accomplished by 20%
piperidine in dimethylformamide. The cleavage of a peptide from the
resin was carried out with reagent K (11) for 2 h at room
temperature with gentle stirring. The crude peptide was then precipitated in ice-cold methyl-t-butyl ether, centrifuged,
and lyophilized. The cyclization of disulfide cyclic peptides was achieved by using a modified procedure of Misicka et al.
(12). The crude peptide was then purified by preparative reverse-phase HPLC using a Dynamax®-300Å C18 column (25 cm × 21.4 mm, inner diameter) with a flow rate of 9 ml/min and two solvent
systems of 0.1% trifluoroacetic acid/H2O and 0.1%
trifluoroacetic acid/acetonitrile. The fractions containing the
peptide were pooled together and lyophilized. The purity of the final
products was assessed by analytical reverse-phase HPLC, capillary
electrophoresis, and matrix-assisted laser desorption/ionization
time-of-flight mass spectrometry.
All NMR experiments were
performed on a Bruker AMX 600 spectrometer equipped with a 5-mm
broad-band inverse probe or a 5-mm triple resonance
(13C,1H,15N) z-gradient
probe, at a proton frequency of 600.13 MHz, using the XWIN-NMR Version
1.0 software package run on a Silicon Graphics INDY workstation. The
NMR sample of peptide IV (final concentration 5 mM) was prepared by dissolving the lyophilized powder in
0.4 ml of Me2SO-d6 (Wilmad). Spectra
were acquired locked and at 305 K, unless otherwise specified.
Intraresidue connectivities and spin systems were deduced by means of
double quantum filtered COSY and total correlation spectroscopy
(
m = 55 and 110 ms) (13) experiments. Gradient-assisted
13C-1H heteronuclear single quantum correlation
spectroscopy (coherence pathway rejection) (14) was employed to verify
the proton resonance assignments obtained as described above. All
two-dimensional experiments were acquired phase-sensitive via time
proportional phase increments (15) and processed with shifted
(45-90°) sine-bell functions, followed by automatic base-line
correction in both dimensions. NOE intensities classified as strong,
medium, and weak were obtained from a NOE correlation spectroscopy
experiment with a
m of 200 ms; the NOE build-up was found to
be linear up to this value, translating into minimal spin diffusion for
mixing times up to
200 ms. 1H chemical shifts reported
here are referenced to the residual Me2SO signal (
= 2.525 ppm at 305 K). The observation of a single set of resonances in
the 1H and 13C-1H spectra is
consistent with the presence of one major form of the peptide and the
absence of impurities as well as multiple conformations in slow
exchange with each other on the NMR time scale. The absence of such
exchange is also corroborated by rotating frame nuclear Overhauser
effect spectroscopy experiments (data not shown).
NMR-based structure determination of peptide IV was carried out within
the QUANTA® Version 4.1 molecular modeling program (Molecular
Simulations Inc.) using the XPLOR Version 3.1 package and the CHARMm
(Version 23.0) force field in tandem with the NMRCompassTM Version 2.5 package used for the complete analysis and bookkeeping of NOE
correlation spectroscopy data. A set of 61 NOE-based distance restraints (29 intraresidue, 25 sequential, and 7 long-range) involving
the amide, -, and
-protons within the molecule was used in the
conformational search according to the following protocol. First, a set
of 75 initial structures was generated from a covalent template of the
cyclic peptide by a round of distance geometry, followed by simulated
annealing (10-ps time steps at 1000 K, with subsequent 2-ps cooling to
a final temperature of 300 K) and finally 5 ps of refinement dynamics
(0.001-ps steps, 100 K final temperature). After this stage of the
protocol, it was apparent that the
angle for Ser-3 in a majority of
the structures generated fell in the vicinity of one of the allowed
solutions based on the spin-spin coupling data (i.e.
60°). Hence, this angle was restrained initially to the range
80
to 40° and subsequently to
60 ± 20°. Using this augmented
set of restraints, two rounds of refinement dynamics (10 ps) followed
by Powell energy minimization (500 steps) were conducted. In the fourth
and final stage of structure calculations, the distance between the
carbonyl oxygen of Asn-2 and the amide hydrogen of Gln-5 was
constrained to the range 1.9-2.5 Å, consistent with a
-turn in the
region of the sequence encompassing these residues, and in accordance
with the spin-spin coupling and amide temperature coefficient data (see
Table I). Using this final set of restraints, another 10 ps of
refinement dynamics plus 5000 energy minimization steps were performed.
Of the 75 structures refined in this fashion, the best 10, which
satisfied all dihedral angle restraints (± an additional 10°) and
exhibited no upper bound distance violations in excess of 0.3 Å, were
selected.
Following a modified procedure of Moebius et al. (16), 5 × 104 COS-7 cells/well of a six-well plate were transfected with 2 µg of T4-pcDNA3 (Invitrogen) and 6 µg of DOSPER (Boehringer Mannheim) according to the supplier's protocol for the DOSPER liposomal transfection reagent. Normally 30~40% of transfected COS-7 cells express human CD4 as determined by immunofluorescence. MHC class II-expressing Raji Burkitt's lymphoma cells (107) were added to CD4-transfected COS-7 cells (48 h post-transfection) for 1 h at 37 °C (in 1 ml of RPMI 1640 medium containing 10% fetal calf serum and 200 mM glutamine) in the presence of peptides at appropriate concentrations. The inhibition of rosette formation by peptides was determined by the number of rosettes obtained in the presence of the peptides relative to the number of rosettes in the positive control. COS-7 cells transfected with pcDNA3 vector alone served as negative controls, which showed no rosette formation. Other studies have demonstrated that the enumeration of rosetting as performed here correlates well with quantitative cell binding assays employing radiolabeled Raji cells (16).
Mixed-lymphocyte Reaction (MLR) AssayAs described previously (17), MLR assays were established by co-culturing 1 × 105 responder peripheral blood lymphocytes with 2 × 105 irradiated (30 Gy) stimulator peripheral blood lymphocytes in quadruplicate wells of a 96-well microtiter plate. Peptide analogs were added at a final concentration of 100 µM immediately after the cells were plated. For measuring proliferation, 1 µCi of [3H]dThd was added to the wells for the final 6 h of day 6. Cells were harvested onto glass-fiber filters and counted in a Model 1205 Beta-Plate reader (Wallac, Gaithersburg, MD). The mean thymidine incorporation was calculated, and results are expressed as mean percent inhibition by peptide analogs relative to control (untreated) T cell proliferation.
In Vivo Animal StudyFemale SJL/J H2s, male
C57BL6/J (B6) H2b, and male MHC class II mutant
B6.C-H2bm12 (bm12) mice were purchased from Jackson
Laboratory (Bar Harbor, ME). All animals used were between 7 and 9 weeks of age. The assay protocols for experimental allergic
encephalomyelitis (EAE) and skin allograft rejection have been
described previously (18). For GVHD in the (B6xDBA/2)F1 (B6xCBA)F1 (950 cGy) strain combination, recipient mice
were irradiated (950 cGy) and transplanted 6 h later with 2 × 106 T cell-depleted donor bone marrow cells alone or in
combination with 3 × 106 CD4+ T cells.
Peptide (0.5 mg in 0.2 ml of buffered saline solution) was administered
intravenously on days 0, 3, and 6. Statistical significance was
determined by Wilcoxin rank analysis.
To search for potential CD4 functional epitopes that could be
targeted for the design of new inhibitors, a computer analysis was
conducted for CD4 D1 in conjunction with synthetic peptide mapping
using a procedure published previously (10). This led to the
identification of a surface pocket potentially involved in the CD4-MHC
class II interaction (Fig. 1) (18). This CD4 surface
pocket is formed by the FG loop (also known as the third complementarity-determining region or CDR3) and the CC loop. While
CDR3 has long been proposed to be involved in MHC class II interaction
(6) and recent studies with peptide mimics have confirmed their
involvement (17, 19, 20), our analysis suggested that the highly
protruded CC
loop may also serve as a surface epitope critical for CD4
function.
A series of linear peptides based upon the CC loop region were
synthesized to test whether this site was involved in CD4-MHC class II
interaction. Peptide I (KNSNQIK) was derived from the entire CC
loop
(amino acids 29-35) of human CD4 D1. Peptides II (KNSNQ) and III
(SNQIK) were truncated analogs of peptide I corresponding to the C and
N termini, respectively. In assays for T cell proliferation as measured
by MLRs, peptide I exhibited an inhibitory effect of 21% at 100 µM, and the N- and C-terminal truncated analogs II and
III exhibited reduced activities of 11 and 15%, respectively, at the
same concentration. This finding indicates that the bioactive region of
the CD4 D1 CC
loop lies on the sequence of NSNQI, which adopts a type
I
-turn, as observed in the crystal structure of CD4 D1 and D2 (9).
These results led to the design of the cyclic peptide IV
(cyclo(CNSNQIC)), incorporating a disulfide bridge to enhance the
structural stability of the
-turn around NSNQ. Molecular modeling
studies predicted that the cyclic heptapeptide IV closely mimics the
conformational feature of the
-turn of the native CC
loop (data not
shown).
To confirm our design principle, the structure of the cyclic peptide IV
was determined by high resolution two-dimensional NMR spectroscopy.
Table I summarizes the data from NMR experiments. Several lines of evidence are consistent with the existence of a type I
-turn spanning NSNQ and closely resembling the native CC
loop. (i)
There are a number of weak NOE contacts between Asn-2 and Gln-5 (the
residue numbering follows the sequence of the peptide) as well as
between Ser-3 and Gln-5. (ii) There is a strong dNN
interaction between Asn-4 and Gln-5. (iii) The coupling constant data
conform closest to those expected for a type I
-turn (i.e.
3JNH-
H values of 4 and 9 Hz are
expected for the i+1 and i+2 turn residues,
respectively (21)). (iv) The low amide temperature coefficients for
Gln-5 and, to a lesser extent, Asn-4 are highly diagnostic of their
participation in intramolecular hydrogen bonding, whereas the remaining
amides are largely solvent-exposed. This finding is consistent with the
expected hydrogen bond between the carbonyl of residue i and
the NH of residue i+3 in the
-turn (22).
Structure determination based on the obtained NMR data was carried out
for the cyclic peptide IV. Ten structures of the peptide that best fit
the NOE and dihedral angle data were selected from the extensive
conformational search (Fig. 2). The conformational search of the peptide was performed without restricting its structure to a specific type of turn (i.e. type I) a
priori. As shown in Fig. 2, the cyclic peptide IV adopted well
defined conformation around the sequence NSNQ, which approximately
resembled the type I -turn structure of the CC
loop in the native
CD4 protein. The
(i+1) and
(i+2) angles
deviate from ideality and push the backbone amide of Asn-4 in toward
the carbonyl of Asn-2. This results in an appreciably shorter
interatomic distance than that observed for the
i,i+3 putative hydrogen bond (
2.1
versus 2.5 Å) in these structures. As such, the structure
of the peptide deviated slightly from the ideal type I
-turn;
nevertheless, the overall backbone and side chain topologies of the
functionally important region of the peptide closely mimic those of the
native CD4 CC
loop. In contrast to the bioactive region, the disulfide bridge region was somewhat ill defined, probably due to the low number
of observed NOE restraints in this region.
If the CC loop of CD4 D1 is a critical epitope for MHC class II
binding, as predicted by theoretical analysis, the structural mimicry
of the native CD4 D1 CC
loop region by peptide IV suggests that the
peptide might block CD4-MHC class II interaction. CD4-MHC class II
binding studies, using a cell rosetting assay (5, 16), were performed
to test this hypothesis. Peptide IV inhibited rosette formation by as
much as 50% in a concentration-dependent manner (Fig.
3A). The linear peptide I exhibited flexible
conformations as suggested by modeling studies and consequently
displayed decreased potency in comparison with the constrained cyclic
peptide IV. This result strongly suggests that the stable
-turn
conformation mimicking the native protein surface region is important
for inhibitory activity of the peptide. The inhibitory effect of
peptide IV was sequence-specific, as demonstrated by the lack of
activity of a scrambled peptide (peptide IV-scr; identical amino acid
composition, but a fixed randomized sequence). The selective effect of
peptide IV on CD4-MHC class II interaction was also indicated by its
inability to inhibit cell rosetting mediated by CD8-MHC class I
interaction (data not shown). These results strongly support the notion
that the CC
loop of CD4 D1 is a critical functional epitope for MHC class II binding and that small peptide mimics such as peptide IV are
sufficient to block this interaction.
The ability of peptide IV to inhibit CD4-MHC class II interaction suggested that it could interrupt the activation of CD4+ T cells. In this regard, the peptide was tested in MLRs and found to inhibit proliferation by at least 40% of the control response at 100 µM (Fig. 3B). In addition, peptide IV exhibited significantly higher activity than the other linear peptides derived from the same region, again supporting the conformational dependence of the peptide inhibition.
Since a major limitation of peptide-based therapeutics is their susceptibility to proteolytic degradation, the synthetic CD4 peptides were tested for their proteolytic stability by incubation in 90% human serum.2 The linear peptide I was highly susceptible to serum proteases and was completely degraded after 24 h, whereas peptide IV exhibited significant proteolytic resistance, with nearly 75% of the peptide remaining intact after 72 h. This resistance is likely due to the small size and cyclic nature of peptide IV, so the molecule is constrained into conformations that are not favorable for proteolytic recognition.
The in vivo immunosuppressive activity of peptide IV was
tested in three different CD4+ T cell-dependent
murine models: EAE, skin allograft rejection across a MHC class II
antigen difference, and GVHD across a haplomismatch MHC difference.
Following induction of EAE in SJL mice, the maximum mean disease
severity level reached by the untreated control group was 1.8 (Fig.
4A), whereas mice treated with a single dose
of peptide IV on day 12 (0.5 mg intravenously) attained a significantly lower maximum of 0.4 (p < 0.01 on days 18-22). In the
skin allograft model, a single dose of peptide IV (0.5 mg; intravenous)
3 h before transplantation significantly (p < 0.01) prolonged the median survival time to 20 days, compared with 15 days for the untreated control (Fig. 4B). Similarly, in the
murine GVHD model, when donor CD4+ T cells were
transplanted into irradiated recipient mice of the untreated control
group, 50% succumbed to GVHD with a median survival time of 36 days
(Fig. 4C). In contrast, mice treated with peptide IV
exhibited 90% survival (median survival time > 60 days,
p < 0.04), not significantly different from those
transplanted with only anti-T cell antibody-treated bone marrow cells
(p > 0.60). The effectiveness of the peptide in these
different animal models for autoimmune disease and transplantation
reactions demonstrated the in vivo immunosuppressive
activity of this small constrained peptide and its potential as a novel
therapeutic agent.
Current immunotherapeutic strategies include the use of monoclonal antibodies, such as monoclonal anti-CD4 antibody, to block T cell activation (23). However, these broad-based monoclonal antibody approaches can result in pan-T cell depletion, and in addition, their value as an effective treatment have been reduced by their inherent immunogenicity. In comparison, small peptide-based therapeutics are less immunogenic and can therefore be used over longer periods of time. In regard to pan-T cell depletion, spleen and lymph nodes from mice treated 48 h earlier with peptide IV have normal cellularity and T and B cell subset composition (data not shown). The combined results of the above studies clearly indicate the potential of synthetic chemically modified peptides as an alternative therapeutic approach.
The present study of small peptide mimics of the CD4 surface may
suggest a starting point for developing a general approach to inhibit
other Ig-related protein structures and interactions. As members of the
Ig superfamily have a conserved backbone-folding pattern, it is likely
that this generic structure provides some common scaffolds for
efficient protein-protein interactions. For example, a surface pocket
consisting of the FG and CC loops, analogous to that seen here in CD4,
is found in the following: CD8, which mediates dimerization (24); the
IgE high affinity receptor, which binds IgE (25); CD2, which binds
LFA-3 (26); and CD28, which binds CD80/86 (27).
In summary, we have proposed that the CC loop is an important
functional epitope on the CD4 surface for intermolecular binding and
found that small peptide mimics of this epitope are sufficient to
interrupt a larger protein-protein interface. In particular, a
synthetic cyclic heptapeptide (peptide IV) has been shown to closely
mimic the CC
surface epitope, to effectively block CD4-MHC class
II-dependent cell rosetting, to possess significant
immunosuppressive activity in vitro and in vivo,
and to strongly resist proteolytic degradation. These findings have
demonstrated a general approach of bioactive peptide design by the
functional mimicry of protein surface epitopes to generate potential
novel therapeutic agents.
We thank Zhengdong Wu, Simei Shan, and Zhixian Lu for peptide synthesis; Anna Wiaderkiewicz for MLR assays; the Scientific Support Group at Molecular Simulations, Inc. (San Diego, CA) for technical assistance pertaining to the NMR structure determination package; and also Dr. J. M. Varnum (Thomas Jefferson University) for helpful discussion concerning the structural work.