(Received for publication, August 1, 1996, and in revised form, November 5, 1996)
From the Structural Biology Unit, National Institute of Immunology, Aruna Asaf Ali Marg, New Delhi 110 067, India
The shared surface topology of two chemically
dissimilar but functionally equivalent molecular structures has been
analyzed. A carbohydrate moiety (-D-mannopyranoside) and
a peptide molecule (DVFYPYPYASGS) bind to concanavalin A at a common
binding site. The cross-reactivity of the polyclonal antibodies (pAbs)
was used for understanding the topological relationship between these
two independent ligands. The anti-
-D-mannopyranoside
pAbs recognized various peptide ligands of concanavalin A, and the
anti-DVFYPYPYASGS pAbs recognized the carbohydrate ligands, providing
direct evidence of molecular mimicry. On the basis of differential
binding of various rationally designed peptide analogs to the
anti-
-D-mannopyranoside pAbs, it was possible to
identify different peptide residues critical for the mimicry. The
comparison of circular dichroism profiles of the designed analogs
suggests that the carbohydrate mimicking conformation of the peptide
ligand incorporates a polyproline type II structural fold. The
concanavalin A binding activity of these analogs was found to have a
direct correlation with the topological relationship between peptide
and carbohydrate ligands.
The functional mimicry involving unrelated molecules is often encountered. Many times it occurs by design and is used as an effective control during various regulatory mechanisms. However, sometimes the accidental structural resemblances lead to aberrations, which are expressed in terms of clinical manifestations. For example, it has been suggested that the mimicry between microbial or viral peptides and the self-peptides presented inappropriately on a target tissue could initiate an autoimmune attack (1). On the other hand, many enzymatically regulated events are controlled by intervention of proteinaceous inhibitors, which mimic substrate binding. The complementarity between serine proteases and the corresponding substrates is matched by aprotinin, the chymotrypsin inhibitor, and many other such inhibitors (2, 3). In addition, molecular mimicry has implications for rational drug design applications. Considerable efforts have been invested in the development of peptidomimetic drugs using nonpeptidyl surrogates (4-7).
Systematic approaches involving computational as well as experimental
tools have been used to analyze and exploit topological similarity
between dissimilar molecules (8-12). It is apparent that the
structural rules governing molecular mimicry are required to be defined
for its successful exploitation. Concanavalin A (ConA),1 a lectin known to be specific for
binding to certain mannose-containing carbohydrates, provides an
appropriate model system for this purpose. Peptidyl ligands have been
characterized that bind to ConA with affinities comparable with those
of the carbohydrate ligand methyl -D-mannopyranoside
(13, 14). We have shown that the peptide and the carbohydrate ligands
of ConA are true topological mimics. Cross-reactivity of polyclonal
antibodies (pAbs) against an assortment of designed peptide analogs is
used here to delineate the specific residues that may be contributing
to the mimicry of carbohydrate structure by the peptide ligand.
Conjugation of
mannose with bovine serum albumin (BSA; Sigma) or
diphtheria toxoid (DT; Serum Institute) was achieved by a two step
reaction. In the first step,
p-aminophenyl--D-mannopyranoside (Sigma) was
activated by an equimolar amount of glutaraldehyde (Sigma) in 0.1 M sodium carbonate buffer, pH 9.0, for 30 min at 20 °C
and then mixed with BSA or DT in the same buffer. Synthesized conjugate
was extensively dialyzed against normal saline at 4 °C. The
carbohydrate content of the conjugate was determined by the phenol-sulfuric acid assay (15), and its protein concentration was
measured by the Bradford assay (16) using BSA as a standard. Mannopyranoside-BSA or -DT conjugate was prepared using a molar ratio
of 50 mol of sugar:1 mol of BSA or DT.
Assembly of
the protected peptide chains was carried out on an automated peptide
synthesizer, (431A, Applied Biosystems Inc.) using solid phase
9-fluorenylmethyloxycarbonyl chemistry on a p-hydroxymethyl
phenoxymethyl polystyrene resin (Nova Biochem). All amino acids (Nova)
were protected at the -amino position with the
9-fluorenylmethyloxycarbonyl group. Cleavage was performed by using
trifluoroacetic acid (Sigma). The peptides were purified on a Delta Pak
C18-100Å column (19 mm × 30 cm, 15 µm, spherical; Waters)
using a linear gradient of trifluoroacetic acid and acetonitrile. The
absorption was monitored at 214 nm. Lysine was introduced at the
carboxyl terminus of the dodecapeptide (DVFYPYPYASGS) for its
conjugation to BSA or DT using glutaraldehyde. The glutaraldehyde solution in 0.1 M phosphate buffer, pH 7.1, containing 150 mM NaCl was slowly added to the cold mixture of peptide and
BSA or DT with the ratio of 100:1 in the same buffer. The final
concentration of glutaraldehyde was 0.1% in the reaction mixture. The
reaction mixture was incubated for 20 h at 4 °C. Synthesized
conjugate was exhaustively dialyzed against normal saline at
4 °C.
BALB/c mice were immunized
intraperitoneally with -D-mannopyranoside-BSA or peptide
(12 mer)-DT in alum-adsorbed 100 µg of protein. Mice received three
injections at intervals of 2 weeks and were bled from the retro-orbital
venous plexus 15 days after the last injection. The normal mice sera
were obtained from mice before the first injection. Collected sera were
stored in aliquots at
20 °C.
To carry out ELISA, the corresponding conjugate of peptide (1 µg/well) or sugar (5 µg/well) with DT or BSA was absorbed on Nunc (Roskilde, Denmark) immunoplates. After coating, wells were blocked with 1% BSA or gelatin to prevent nonspecific binding, before adding serial dilutions of mouse pAbs. The polyclonal antisera were used as such for the ELISA experiments. Horseradish peroxidase-labeled goat anti-mouse Abs (NII reagent bank) were used as second antibodies. The peroxidase substrate was o-phenylenediamine (Sigma). The color was developed using H2O2, and absorbance was recorded at 490 nm.
The general protocol of the competitive ELISA included incubation of constant amounts of pAbs with equal volumes of increasing concentrations of peptides. Binding of free pAbs to absorbed antigen was measured by ELISA. The levels of inhibition were calculated by comparison of the absorbance at 490 nm of the wells reacted with pAbs and soluble inhibitor and that of the control wells, which did not contain the inhibitor. The concentrations of pAbs used in all the inhibition assays were such that the absorbance values at 490 nm were in the range of 0.8 and 1.0 for the wells without inhibitor.
Concanavalin A Binding AssayPlates with preabsorbed antigen were incubated with biotinylated ConA (Sigma) in 50 mM phosphate, pH 7.1, containing 150 mM NaCl, 10 nM MnCl2, 100 nM CaCl2, and 0.1% Tween 20 for 1.5 h at 37 °C. The binding of ConA was detected by streptavidin-peroxidase (Vector Laboratories). As controls, wells coated with the same antigen and incubated with streptavidin-peroxidase without lectin were used. The procedure of competitive lectin binding assay was identical to the competitive ELISA.
Circular DichroismThe circular dichroism (CD) experiments were carried out on a JASCO 710 spectropolarimeter with a 2.0-nm bandwidth, 1-nm resolution, and 1-s response time. A 10-mm-path length cell was used. Typically, 10 scans at a scan speed of 50 nm/min were added and averaged. Peptide concentrations were 50 µM in H2O. Results are expressed as molar ellipticity in deg cm2/dmol.
ConA binds to the dodecapeptide DVFYPYPYASGS (12-mer), as
shown in Fig. 1A. A glycan ligand of ConA
(17), methyl -D-mannopyranoside, competitively inhibited
this binding (Fig. 1B). This confirmed that methyl
-D-mannopyranoside sterically interferes in the ConA peptide binding in a dose-dependent manner as suggested
earlier (13, 14). Thus, the carbohydrate binding site and the peptide binding site are overlapping.
The murine pAbs were raised against -D-mannopyranoside
using the glycan moiety conjugated to BSA. The reactivity of these pAbs
to
-D-mannopyranoside-DT conjugate in the ELISA
confirmed that the antisera had
-D-mannopyranoside-specific antibodies (Fig.
2A). The control wells coated with DT showed
a negligible signal, suggesting that there was no nonspecific binding.
Similarly, the pAbs raised against 12-mer DT showed binding to 12-mer
BSA but not to BSA, confirming that the antisera had 12-mer
peptide-specific antibodies (Fig. 2B). The control sera
obtained from unimmunized mice did not show any binding to either the
carbohydrate ligand
-D-mannopyranoside (Fig.
2A) or the ligand 12-mer peptide (Fig. 2B). The
cross-reactivities of the sugar and peptide pAbs were analyzed (Fig.
2C). ELISA carried out using various dilutions of the
anti-12-mer pAbs showed that the antibodies recognize the glycan
moiety,
-D-mannopyranoside. Conversely, the ELISA
carried out using similar dilutions of the
anti-
-D-mannopyranoside antibodies showed
activity against the 12-mer peptide.
Comparison between Different Carbohydrate Moieties
The
specificity of pAbs against the sugar moiety,
-D-mannopyranoside, was further confirmed by competitive
ELISA against various sugars, as shown in Fig.
3A. At a 500 mM concentration, methyl
-D-glucopyranoside showed less than half the
activity of the specific antigen methyl
-D-mannopyranoside. No cross-reactivity of these pAbs
was observed against
-lactose. Thus, the antibodies raised against
-D-mannopyranoside showed structurally correlated differential binding to different sugar moieties.
Similarly, the polyclonal antisera against the peptide molecule also
distinguished finer details of the topological relationship between
different carbohydrate molecules. It is known that ConA recognizes
methyl -D-glucopyranoside with an affinity lower than that for methyl
-D-mannopyranoside (18). The comparative
binding of antipeptide pAbs to methyl
-D-glucopyranoside, methyl
-D-mannopyranoside, and
-lactose was analyzed. It was
found that anti-12-mer pAbs exhibited binding to methyl
-D-mannopyranoside better than to methyl
-D-glucopyranoside and did not recognize
-lactose
(Fig. 3B). About a four times lower concentration of the
methyl
-D-mannopyranoside was required than that of
methyl
-D-glucopyranoside for achieving 30% inhibition.
The difference between the two sugars in their antipeptide pAb binding
profile is comparable with the difference in their ConA binding.
Different
peptide ligands of ConA were analyzed using polyclonal antibodies
against the carbohydrate ligand -D-mannopyranoside. Besides the 12-mer peptide, two other peptides, MYWYPY (6- mer) and
RVWYPYGSYLTASGS (15 mer), have also been identified to bind to ConA
with affinities comparable with that of the carbohydrate ligand (13,
14). The binding affinity of antisugar pAbs to these peptides was
analyzed (Fig. 4A). It was observed that 12- and 15-mer peptides inhibited the reaction of
anti-
-D-mannopyranoside pAbs with the solid phase 12-mer
peptide in a dose-dependent manner, but the 6-mer peptide
did not bind to the pAbs in the concentration range of up to 2 mM. Despite the fact that
anti-
-D-mannopyranoside pAbs reacted with the 12-mer
peptide attached to solid support (Fig. 2C), its inhibition
by peptide in solution (Fig. 4A) was more in the case of the
15-mer peptide than in the case of the 12-mer peptide. At a 2 mM peptide concentration, about 45% inhibition was
achieved by the 15-mer peptide compared with about 11.5% by the 12-mer
peptide. This inhibition pattern probably reflects higher binding
affinity of these pAbs to the 15-mer peptide than to the 12-mer
peptide.
The circular dichroism spectra (Fig. 4B) of the three
peptides reflect the profile of binding to
anti--D-mannopyranoside pAbs. All three peptides showed
a positive molar ellipticity at about 230 nm, generally attributed to
the aromatic residues (19). This was expected, since these peptides are
rich in tyrosine and tryptophan residues. The 12- and 15-mer peptides
showed two distinct negative peaks, one at 207 nm and another at 212 nm. However, the 6-mer peptide showed a negative peak at 212 nm and a
positive peak at 202 nm. Thus, the CD profile of the 6-mer peptide was markedly different from those of the other two peptides.
The CD spectra
suggested that the 12- and 15-mer peptides are structurally similar to
each other. However, the CD spectrum of the 6-mer peptide was very
different from those of the other two peptides. This difference was
also reflected in the antibody reactivities. Comparison of the
sequences of the three peptides indicated that the carboxyl-terminal
ASGS sequence is the most common feature in the 12- and 15-mer
peptides. In previous studies (13), the sequence ASGS was incorporated
to improve the solubilities of the ConA binding peptides. This sequence
is absent in the 6-mer peptide. It could, therefore, be inferred that
this region of the sequence may be functionally important. A peptide
with the sequence MYWYPYASGS (10-mer) was synthesized by incorporating ASGS at the carboxyl terminus of the 6-mer peptide. Fig.
5A shows comparison between the 10- and 6-mer
peptides in antisugar pAbs binding. It is clear that the decapeptide
inhibits binding of pAbs to the absorbed 12-mer conjugate in a
dose-dependent manner. The control peptide (6- mer) shows
no such competitive inhibition in the same concentration range. These
data suggest that ASGS in the peptide sequence has direct contribution
to the binding with pAbs.
The observations based on antisugar antibody reactivity seem consistent
with the ConA binding activities (Fig. 5B) and CD profiles
of the peptides (Fig. 5C). Unlike in the case of the 6-mer
peptide, the CD profile of the 10-mer peptide has a negative molar
ellipticity at 207 nm. In fact, the CD spectrum of the 10-mer peptide
(Fig. 5C) is similar to that of the 12- and 15-mer peptides (Fig. 4B). Thus, addition of ASGS in the hexapeptide appears
to have led to a definite structural fold similar to that of the other
nonhomologous peptide ligands of ConA. Furthermore, a competitive ConA
binding assay was carried out to investigate whether the designed
decapeptide can also bind to ConA. The 6-mer peptide was used again as
the control. The results indicated that the 10-mer peptide binds to
ConA in a dose-dependent manner, but the 6-mer peptide does
not (Fig. 5B), corroborating that the sequence ASGS also
participates in the -D-mannopyranoside mimicking by the
peptide.
The comparison of binding properties of the 12-, 15-, and 10-mer
peptides (Figs. 4 and 5), and their sequences suggested that the 15- and 10-mer peptides have comparable antisugar pAbs binding activity.
However, the 12-mer peptide showed significantly lower activity against
these antibodies. This difference could be correlated with their
sequences. There is Phe at position 3 in the 12-mer peptide compared
with Trp in the other two peptides. Therefore, an analog of the 12-mer
peptide was synthesized by changing Phe3 to Trp, leading to
the sequence DVWYPYASGS (12-mer), and assayed for the reactivity to
antisugar pAbs. The substituted peptide showed significantly higher
binding to pAbs (Fig. 6A). The ConA binding
of the peptide was also analyzed (Fig. 6B). The binding curve of the 12-mer peptide to ConA showed a sharper increase with
respect to peptide concentration compared with that of the 12
-mer
peptide. However, there is no significant change in the CD spectrum of
the 12-mer peptide after mutation (data not shown). The ConA binding of
the 12
-mer peptide was similar to that of the 10-mer peptide (Fig.
5B). A nonspecific peptide of comparable size
(PTPSPPMSPLRPG) was used as a negative control for antibody binding
(Fig. 6A) as well as ConA binding (Fig. 6B),
which showed no significant binding in either case.
The complementarity of geometry and charge, expressed in terms of
the weak noncovalent forces at the interface, is responsible for
mediating molecular recognition. This property is common to diverse
molecular interactions, which include enzyme-substrate, protein-nucleic
acid, and antigen-antibody. Functional equivalence of the chemically
dissimilar molecules sharing common surface topology is, therefore,
anticipated. Many such equivalences involving proteinaceous molecules
and glycan moieties have been identified. Tendamistat, a regulatory
protein inhibitor of -amylase, binds to the carbohydrate binding
site of the enzyme through a tripeptide epitope (20, 21). It has been
suggested that streptococcal group A carbohydrate induces antibodies
cross-reacting with cytokeratin peptides, leading to the clinical
manifestations of autoimmune responses (22). The analysis of finer
details of the mimicry between peptide and carbohydrate ligands of ConA
is relevant for defining structural correlations corresponding to such
functionally important protein-carbohydrate mimics.
The peptide ligand (12-mer) not only binds to ConA but can also be
displaced by methyl -D-mannopyranoside in a
dose-dependent manner. However, this alone is not
sufficient to suggest that the two are structurally similar. There are
many examples in which an organic molecule can bind to a receptor as
strongly as the normal ligand without being structurally similar (23).
But the anti-12-mer peptide-specific polyclonal antibodies cross-react with
-D-mannopyranoside, and the
anti-
-D-mannopyranoside-specific polyclonal antibodies
cross-react with the peptide, suggesting that the peptide and the
carbohydrate ligands are topological equivalents. A truly polyclonal
antibody population maps the entire surface of the antigen through the
distribution of specificities. It essentially represents a topological
map of the antigen. Such an antibody population can also provide a
measure of similarity of the reciprocally related ligands. As seen in
Fig. 3, the antipeptide antibodies distinguish high and low affinity
ligands of ConA in a way comparable with the antisugar antibodies. It
is known that methyl
-D-mannopyranoside binds to ConA
more strongly than methyl
-D-glucopyranoside. The
antipeptide pAbs bind to methyl
-D-mannopyranoside four
times as efficiently as to methyl
-D-glucopyranoside.
This implies that the surface similarity of the peptide ligand is more to the methyl
-D-mannopyranoside than to methyl
-D-glucopyranoside. As expected, both the anti-12-mer
peptide and the anti-
-D-mannopyranoside pAbs do not
recognize
-lactose, which is structurally very different from
-D-mannopyranoside. Thus, the antibody population not
only provides a topological description of a molecular mimic but can also detect the extent of mimicry.
The three peptide ligands have varying cross-reactivities to the
antisugar pAbs. The 6-mer peptide is apparently a weak ligand of ConA.
Within the concentration range used in the ELISA, no binding of this
peptide to the antibodies was detected. However, 12- and 15-mer
peptides show binding to these antibodies. The structural properties of
the peptides, analyzed by CD spectroscopy, are consistent with these
observations. All the peptide analogs exhibited characteristic
structural features in their CD spectra. Short peptides containing a
sequence motif in which proline is sandwiched between two aromatic
residues is known to have definitive structure in solution (24). The
feature corresponding to the -sheet secondary structure with a
negative ellipticity at 212 nm is shared by all three peptides.
However, a characteristic polyproline type II feature detected in the
12- and 15-mer peptides corresponding to negative ellipticity at 207 nm
is absent in the 6-mer peptide. Thus, the 12- and 15-mer peptides are
structurally closer to each other compared with the 6-mer peptide.
The relative differences in binding of different peptides to antisugar
pAbs implies that it is possible to delineate the residues responsible
for the topological similarity with the sugar. It had been suggested
that the consensus sequence YPY is critical for binding to ConA (13,
14). However, the residues away from YPY also appear to participate in
binding. The carboxyl-terminal four residues (ASGS) of these peptides
significantly contribute to the topological resemblance to the sugar
moiety and the ConA binding activity (Fig. 5). A decapeptide designed
by incorporating these four residues at the carboxyl terminus of the
6-mer peptide not only shows better binding to antisugar antibodies but
has structural similarity with the 12- and 15-mer peptides. In
addition, it binds to ConA, unlike the 6-mer peptide. The hydrogen
bonds involving hydroxyl groups from the sugar moieties critically
contribute to the ConA-carbohydrate interaction (17, 25). The serine side chains in the ASGS sequence can also be attractive substitutes as
hydrogen bonding donors in place of sugar hydroxyl groups. The point
mutation at an amino-terminal residue Phe3 Trp in the
12-mer peptide (called 12
-mer) improves its binding to antisugar pAbs
without significantly affecting its structure. Although the binding
affinities of the two analogs, 12-mer and 12
-mer, to ConA appears to
be comparable, the mutation seems to have brought about a transition in
the nature of binding. The 12-mer analog shows a much sharper increase
in the ConA binding compared with that of the 12
-mer analog over the
same concentration range of the peptide. ConA is known to exhibit
structural and thermodynamic diversity in the mode of interaction with
different carbohydrate moieties (25-28). The differences in the nature
of binding by various peptide analogs may also reflect similar
diversity.
The carboxyl-terminal extension consisting of ASGS and the
amino-terminal region incorporating WYPY are common in all peptide analogs sharing a strong topological relationship with the carbohydrate moiety. The three peptides, 10-, 12-, and 15- mer, which share these
two sequence motifs exhibit similar ConA binding activities. The
anticarbohydrate pAb binding of these peptides is comparatively high.
Also, the circular dichroism spectra of these peptides show significant
resemblance with each other. However, the region sandwiched between
these two motifs is highly diverse in terms of length and the nature of
the residues in the three peptides. Whereas the 15-mer peptide has five
residues sandwiched between WYPY and ASGS, the 12
-mer peptide has two,
and the 10-mer peptide has none. The two residues in the 12
-mer
peptide are not related to the corresponding sandwiched sequence in the
15-mer peptide. It may, therefore, be inferred that this part of the
molecule does not significantly contribute to either structure,
carbohydrate mimicry or the ConA binding activity of the peptide.
Thus, the analysis involving mutual pAb cross-reactivity of a peptide and a carbohydrate ligand has provided a detailed description of the topological similarity between these two essentially dissimilar molecules. The structural features and the amino acid residues of the peptide ligand, critical for carbohydrate mimicry, were identified. A direct correlation was observed between the extent of carbohydrate mimicry and the ConA binding activities of various peptide analogs. The immunological probe involving pAbs used here in analyzing the topological relationship between peptide and carbohydrate ligands of ConA can also be effective in designing peptidomimetic drugs.
We thank Drs. A. Surolia, S. Rath, and J. K. Batra for useful discussions and H. S. Sarna for technical assistance.