(Received for publication, September 11, 1996, and in revised form, December 18, 1996)
From the Departments of Molecular Pharmacology, and
Microbiology and Immunology, Albert Einstein College of Medicine,
Bronx, New York 10461 and the § Department of Organic
Chemistry, Stockholm University, Stockholm, Sweden
The trisaccharide
3,6-di-O-(-D-mannopyranosyl)-D-mannose,
which is present in all asparagine-linked carbohydrates, was previously shown by titration microcalorimetry to bind to the lectin concanavalin A (ConA) with nearly
6 kcal mol
1 greater enthalpy
change and 60-fold higher affinity than
methyl-
-D-mannopyranoside (Mandal, D. K., Kishore, N.,
and Brewer, C. F. (1994) Biochemistry 33, 1149-1156).
Similar studies of the binding of a series of monodeoxy derivatives of
the
(1-3) residue of the trimannoside showed that this arm was
required for high affinity binding (Mandal, D. K., Bhattacharyya, L.,
Koenig, S. H., Brown, R. D., III, Oscarson, S., and Brewer, C. F. (1994) Biochemistry 33, 1157-1162). In the present paper,
a series of monodeoxy derivatives of the
(1-6) arm and "core"
Man residue of the trimannoside as well as dideoxy and trideoxy analogs
were synthesized. Isothermal titration microcalorimetry experiments
establish that the 3-, 4-, and 6-hydroxyl groups of the
(1-6)Man
residue of the trimannoside binds to the lectin, along with the 2- and
4-hydroxyl groups of the core Man residue and the 3- and 4-hydroxyl
groups of the
(1-3)Man residue. Dideoxy analogs and trideoxy
analogs showed losses of affinities and enthalpy values consistent with
losses in binding of specific hydroxyl groups of the trimannoside. The
free energy and enthalpy contributions to binding of individual
hydroxyl groups of the trimannoside determined from the corresponding
monodeoxy analogs are observed to be nonlinear, indicating differential
contributions of the solvent and protein to the thermodynamics of
binding of the analogs. The thermodynamic solution data agree well with
the recent x-ray crystal structure of ConA complexed with the
trimannoside (Naismith, J. H., and Field, R. A. (1996) J. Biol. Chem. 271, 972-976).
The ability of concanavalin A (ConA)1 to bind with high affinity to certain N-linked carbohydrates has made it a valuable tool to investigate the carbohydrates of normal and transformed cells, as well as to isolate carbohydrates, glycoconjugates, and cells on ConA-affinity matrixes (1, 2). Thus, it is important to establish the nature of the molecular interactions between N-linked carbohydrates and ConA in order to understand the specificity of the lectin for cellular carbohydrates.
ConA is a tetramer above pH 7 and a dimer below pH 6, with each monomer
(Mr = 25, 600) possessing one saccharide-binding
site as well as a transition metal ion site (S1) (typically
Mn2+) and a Ca2+ site (S2) (3-5). The
three-dimensional structure of the lectin at 1.75-Å resolution has
been determined by x-ray diffraction analysis (6), and a complex with
Me--Man to 2.9-Å resolution (7). Recently, the x-ray crystal
structure of ConA bound with methyl-3,6-di-O-(
-D-mannopyranosyl)-
-D-mannopyranoside
to 2.3 Å has been reported (8).
Early studies by Goldstein and co-workers (9) established that ConA has
specificity for -pyranose forms of Glc and Man, which contain
similar hydroxyl group configurations at the 3-, 4-, and 6-positions.
Monodeoxy derivatives of the monosaccharides at these positions showed
essentially complete loss of binding to the lectin, thus establishing
the specificity of the so-called "monosaccharide-binding site" in
ConA. Later studies demonstrated that certain N-linked
oligomannose and complex-type oligosaccharides possessed much higher
affinities (~50-fold or greater) than the monosaccharide
methyl-
-D-mannopyranoside (Me-
-Man) (10), thus suggesting extended site binding interactions with the lectin. The
trisaccharide moiety
3,6-di-O-(
-D-mannopyranosyl)-D-mannose, which is part of all N-linked oligosaccharides, was shown to
bind with nearly 100-fold higher affinity than Me-
-Man, and to
induce conformational changes in ConA similar to those of the larger N-linked carbohydrates (10, 11). These results indicated
that the trimannosyl moiety in N-linked carbohydrates was
primarily responsible for their high affinity binding to ConA.
Detailed insights into the specificity of carbohydrate-protein
interactions requires not only relative binding affinity data, but also
thermodynamic data to establish whether extended binding site
interactions occur. We recently described titration microcalorimetry studies of the binding of
methyl-3,6-di-O-(-D-mannopyranosyl)-
-D-mannopyranoside (1) (Fig. 1) to ConA (12). The results showed
that 1 possesses nearly 6 kcal mol
1 greater
change in binding enthalpy (
H) than Me-
-Man, thus providing direct evidence for extended recognition site by the lectin
of the trimannoside epitope. Similar studies with deoxy analogs of the
(1-3)Man residue of the trimannoside established that the 3-OH on
this arm is required for high affinity binding (13).
In the present study a series of monodeoxy derivatives of the (1-6)
arm and "core" Man residue of the trimannoside, as well as dideoxy
and trideoxy analogs of 1 (Fig. 1), were synthesized and
their binding to ConA investigated by isothermal titration microcalorimetry measurements. In light of the recently reported x-ray
crystal structure of ConA and the trimannoside (8), the present results
provide information on the energetics of binding of the trimannoside to
ConA, and, in turn, on the structure of the complex in solution.
ConA was prepared from Jack bean (Canavalia
ensiformis) seeds (Sigma) according to the method of Agrawal and
Goldstein (14). The concentration of ConA was determined
spectrophotometrically at 280 nm using
A1 cm1% = 13.7 at pH 7.2 (9) and
expressed in terms of monomer (Mr = 25, 600).
Me--Man and 1 were purchased from the Sigma. Synthesis of
trimannoside derivatives 2-13 in Fig. 1 has been
reported (15). The synthesis of 14 will be described
elsewhere. The concentrations of carbohydrates were determined by
modification of the Dubois phenol-sulfuric acid method (16, 17) using
appropriate monosaccharides (Man, 2-deoxymannose, 3-deoxymannose,
4-deoxymannose, and 6-deoxymannose) as standards.
Isothermal titration
microcalorimetry was performed using an OMEGA Microcalorimeter from
Microcal, Inc. (Northampton, MA). In individual titrations, injections
of 4 µl of carbohydrate were added from the computer-controlled
250-µl microsyringe at an interval of 4 min into the lectin solution
(cell volume = 1.3424 ml) dissolved in the same buffer as the
saccharide, while stirring at 350 rpm. An example of an isothermal
titration microcalorimetry experiment is shown in Fig. 2
for trisaccharide 13 with ConA at 27 °C. Control
experiments performed by making identical injections of saccharide into
a cell containing buffer with no protein showed insignificant heats of
dilution. The experimental data were fitted to a theoretical titration
curve using software supplied by Microcal, with H
(enthalpy change in kcal mol
1), Ka
(association constant in M
1) and n
(number of binding sites per monomer), as adjustable parameters. The
quantity, c = Ka
Mt(0), where Mt(0) is the initial
macromolecule concentration, is of importance in titration
microcalorimetry (18). All experiments were performed with c
values 1 < c < 200. The instrument was
calibrated using the calibration kit containing ribonuclease A (RNase
A) and cytidine 2
-monophosphate (2
-CMP) supplied by the manufacturer.
Thermodynamic parameters were calculated from,
![]() |
(Eq. 1) |
Our previous titration calorimetry studies showed that ConA binds
to trimannoside 1 which is present in all
N-linked carbohydrates with nearly 6 kcal
mol
1 greater enthalpy change (
H) and
60-fold higher affinity than Me-
-Man (Table I) (12).
These results indicate that the high affinity of the the trimannoside
moiety is due to extended site interactions. In order to examine the
nature of these extended site interactions, a complete set of monodeoxy
analogs of 1 as well as two dideoxy and a trideoxy analog
were synthesized and their thermodynamics of binding determined by
titration microcalorimetry. The results have provided structural
information on the complex in solution which has been compared with the
recently reported x-ray crystal structure of the complex (8).
|
Our previous
studies examined the thermodynamics of binding of a series of monodeoxy
analogs of 1 possessing substitutions on the (1-3) arm
(13). The results, which are shown in Table I for comparison, indicate
that the 2- (2) and 6-deoxy (5) analogs (Fig. 1)
possess similar affinities,
H and entropy
(T
S) values as that of the parent
trimannoside. However, the 3-deoxy analog (3) showed a
nearly 10-fold decrease in affinity and a 3.4 kcal mol
1
decrease in
H with respect to 1, indicating
its involvement in binding. Although we previously reported no change
in the thermodynamics of binding of the 4-deoxy analog (4)
relative to 1, a reinvestigation shows that 4 binds to ConA with a 5-fold decrease in affinity and a 2.1 kcal
mol
1 decrease in
H with respect to
1 (Table I). Thus, the 3- and 4-hydroxyl groups on the
(1-3) arm of 1 bind to ConA.
Our previous studies also revealed that analogs of 1 with a
Glc or Gal residue substituted on the (1-6) arm possessed decreased
affinities and
H values relative to 1 (13). These results provided evidence that the
(1-6)Man residue binds to
the so-called "monosaccharide"-binding site of ConA, by analogy to
the requirements of monosaccharide binding to ConA (9). However, in
order to directly determine the interactions of the
(1-6) arm of
1 with ConA, the 2- (6), 3- (7), 4- (8), and 6-deoxy (9) derivatives of the
(1-6)Man arm of 1 were synthesized. The thermodynamic
binding data for these derivatives are shown in Table I. The binding
parameters of 6 are nearly the same as that of 1,
indicating no binding of the the 2-OH group of the
(1-6) arm. On
the other hand, the Ka and
H values
for 7, 8, and 9 are significantly
lower than that of 1, indicating the involvement of 3-, 4-, and 6-OH of the
(1-6)Man residue of the trimannoside in binding
(19). The magnitude of the reductions in Ka and
H for 7-9 are similar to those observed for 3 and 4 above (Table I). Since the 3-, 4-, and 6-OH groups of the monosaccharides Man and Glc are required
for binding to ConA (9), the data are consistent with binding of the
(1-6)Man of 1 to the "monosaccharide site" of the
lectin.
Titration microcalorimetry data for the 2-deoxy (10) and
4-deoxy (11) derivatives of the central Man residue of
1 are shown in Table I. The results show reductions in
Ka and H for both analogs,
indicating the involvement of the 2- and 4-OH of the central Man
residue of 1 in binding to ConA.
The 2,3- (12) and 4,3- (13) dideoxy analogs of
1, and trideoxy analog 14 (Fig. 1) were also
synthesized. Table I shows that the 12 binds about 14-fold
more weakly than 1 and possesses a H of
10.6
kcal mol
1 which is a loss of
3.8 kcal
mol
1 compared to 1. This reflects the combined
loss in
H by the 2-deoxy analog 10 and 3-deoxy
derivative 3 relative to 1. Table I shows that
13 binds about 25-fold more weakly than 1, and
possesses a
H of
9.7 kcal mol
1 which is
4.7 kcal mol
1 less favorable than that of 1.
This reflects the combined loss in
H of the 3-deoxy
derivative 3 and 4-deoxy derivative 11 relative
to 1.
Trideoxy analog 14, which possesses deoxy substitutions at
the 3-OH group on the (1-3) arm and at the 2- and 4-OH groups of
the central Man residue of 1, exhibits a
Ka value about 50-fold lower than that of
1 (Table I). The
H of 14 is
8.7
kcal mol
1 which is 5.7 kcal mol
1 less than
that of 1. This reflects losses in
H of the corresponding monodeoxy analogs 3, 10, and
11, relative to 1.
The x-ray structure of the complex formed by the free sugar
of 1 with ConA has recently been reported (8). A view of the
H-bonding interactions between 1 and the binding site of the
lectin is shown in Fig. 3, with the individual hydrogen bonds and their distances listed in Table II. The
crystal structure indicates that the (1-6)Man residue of
1 binds via its 3-, 4-, and 6-hydroxyl groups in the same
manner as Me-
-Man in its crystalline complex with the lectin (7).
These results agree with the thermodynamic data for the 7,
8, and 9 in Table I. The x-ray data also shows
binding of the 3-OH of the
(1-3) Man residue to the N-H and side
chain O of Thr-15, and the 4-OH of the
(1-3)Man residue to the side
chain -OH of Thr-15 (8) (Fig. 3; Table II). These results agree with the thermodynamic data for 3 and 4 in Table I. The x-ray data also shows binding of the 2-OH of the central Man residue of 1 to a water molecule which, in turn, is bound to
the protein, as shown in Fig. 3, and the 4-OH of the central Man
residue to the aromatic 4-OH of Tyr-12 (8). These results agree with
the thermodynamic data for 10 and 11 in Table I.
Thus, the titration calorimetry data in Table I which provides
structural information on the solution complex of 1 with
ConA agrees with the x-ray structure of the complex (8).
|
The present thermodynamic data
indicate that the H values for the monodeoxy analogs
in Table I are nonlinear. For example, the sum of the
H values for monodeoxy analogs 7,
8, and 9 is ~
10.5 kcal mol
1
(Table I). To a first approximation, this represents the combined
H contribution of the 3-, 4-, and 6-OH of the
(1-6)Man residue of 1 at the monosaccharide site of the
lectin. This value can be compared to the
H for
Me-
-Man binding of
8.2 kcal mol
1 to the same site.
Furthermore, the value of
10.5 kcal mol
1 does not take
into account the
H contribution of the ring oxygen of the
(1-6)Man residue of 1 (Fig. 3) which would make this
value even greater and hence the difference with the
H of Me-
-Man larger. Similarly, the combined
H values
for the 3-OH and 4-OH of the
(1-3)Man residue of 1 obtained from 3 and 4, respectively, and the 2-OH
and 4-OH of the central Man residue obtained from 10 and
11, respectively, is ~
8.8 kcal mol
1. This
can be compared to the difference in
H between
1 and Me-
-Man of
6.2 kcal mol
1 which
reflects binding of the
(1-3)Man and the central Man residues of
1. Furthermore, the sum of the
H values for
the 3-, 4- and 6-OH of the
(1-6)Man residue (7,
8, and 9), the 3- and 4-OH of the
(1-3)Man
residue (3 and 4), and the 2- and 4-OH of the
central Man residue (10 and 11) is
19.3 kcal
mol
1 which is greater than the
H for
1 of
14.4 kcal mol
1. Thus, the sum of the
H values for the hydroxyl groups of 1 obtained from the monodeoxy analogs in Table I does not correspond to
the measured
H of 1. In all of the above
cases, the sum of the
H values for specific hydroxyl
groups on certain Man residues of 1 obtained from the
corresponding monodeoxy analogs is greater than the measured
H for that residue(s).
This nonlinear relationship in H is also present in
the di- and trideoxy analogs of 1.
H for
dideoxy analog 12 is
3.8 kcal mol
1 as
compared to the sum of the
H values for corresponding
monodeoxy analogs 3 and 10 of
4.4 kcal
mol
1. Likewise,
H for dideoxy analog
13 is
4.7 kcal mol
1 as compared to the sum
of the
H values for corresponding monodeoxy analogs
3 and 11 of
5.7 kcal mol
1. In the
case of trideoxy analog 14, its
H value is
5.7 kcal mol
1 as compared to
6.7 kcal
mol
1 for 3, 10, and 11.
Thus, the
H values for the monodeoxy analogs are
nonlinear in terms of their contributions to the
H
values for the two dideoxy analogs and the trideoxy analog.
The same nonlinearity is also present in the G values
of the monodeoxy analogs. For example, the sum of the
G values for 3, 4,
10, and 11 is
4.7 kcal mol
1,
however, the difference in
G between 1 and
Me-
-Man is
2.5 kcal mol
1 (Table I). In addition, the
sum of the
G values for 3 and 10 is
2.3 kcal mol
1 while the
G for
12, the corresponding dideoxy analog, is
1.6 kcal
mol
1. The sum of the
G values for
3 and 11 is
2.8 kcal mol
1 while
the
G for 13, the corresponding dideoxy
analog, is
1.9 kcal mol
1. The sum of the
G values for 3, 10, and
11 is
3.7 kcal mol
1 while the
G value for 14, the corresponding trideoxy analog, is
2.3 kcal mol
1. Thus, the
G
values for the monodeoxy analogs are nonlinear in terms of their
contributions to the
G values for the two dideoxy
analogs and the trideoxy analog.
The H and
G values for each monodeoxy
analog of 1 also do not scale with the number of H-bonds at
each position as determined from x-ray crystallography (Fig. 3; Table
II). For example,
H for 7 is
3.2 kcal
mol
1, as compared to
2.7 and
2.8 kcal
mol
1 for 8 and 9, respectively.
However, the x-ray crystal structure shows only one H-bond from the
protein to the 3-OH of the
(1-6)Man residue and at least two strong
H-bonds each to the 4-OH and 6-OH of the
(1-6)Man residue. Thus,
the
H of
3.2 kcal mol
1 for
7 reflects the loss of a single H-bond with the protein, however, the
H values for 8 (
2.7 kcal
mol
1) and 9 (
2.8 kcal mol
1) do
not scale with the loss of two H-bonds. The type of H-bonds involved
(Table II) does not provide an explanation for this apparent discrepancy. Likewise, the
H for 3 (
3.4
kcal mol
1) is nearly the same as that for 7 (
3.2 kcal mol
1), even though in the latter case there
are two H-bonds to the 3-OH of the
(1-3)Man residue. Thus, the
H values for the monodeoxy analogs are not
proportional to the number or type of H-bonds involved at specific
hyroxyl groups of 1.
This same lack of scaling is also present in the G
values (Table I). This is of particular interest since it has been
suggested that the free energy associated with elimination of a H-bond
between an uncharged donor/acceptor pair is 0.5-1.5 kcal
mol
1, and between a neutral-charged pair 3.5-4.5 kcal
mol
1 (20). The data in Table I, however, indicate no such
relationship in the free energy difference (
G) of
monodeoxy analogs that represent the loss of one or more H-bonds such
as 7 versus 8 and 9.
The presence of nonlinear relationships in the H and
G values for the deoxy analogs in Table I indicates
other contributions to these terms such as solvent and protein effects.
Thus, the magnitude of the
H and
G
values represent not only the loss of the H-bond(s) involved, but also
differences in the solvent and protein contributions to binding of
1 and the deoxy analogs. Indeed, a recent study suggests a
substantial contribution of solvent to the
H of binding
of 1 to ConA (21). Thus, titration microcalorimetry
measurements of the binding of deoxy analogs of a substrate to a
macromolecule do not provide direct measurements of the free energy and
enthalpy of the H-bonding involved.
Enthalpy-entropy compensation
plots have previously been observed for carbohydrate interactions with
lectins (22, 23) and antibodies (24-26), and attributed to the unique
properties of water (23). A plot of the H versus
T
S values at 300 K for ConA binding to the
oligosaccharides in Table I shows that it is also compensatory (Fig.
4). The plot shows a linear relationship with a slope of
1.55 and the correlation coefficient to 0.94. The enthalpy-entropy plot
in Fig. 3 is similar to those reported earlier for other
lectin-carbohydrate interactions in that their slopes are greater than
unity (27, 28), in contrast to antibody-carbohydrate interactions where
the slope is often less than unity (25, 26). A slope greater than unity
means that the free energy of binding is predominantly driven by
enthalpy, while a slope less than unity indicates dominant entropy
contributions.
Summary
The present study provides a thermodynamic
description of the binding of trimannoside 1 and a series of
mono-, di-, and trideoxy analogs to ConA using titration
microcalorimetry. The results are consistent with binding of the 3-, 4-, and 6-hydroxyls of the (1-6)Man, the 2- and 4-OH groups of the
core Man, and the 3- and 4-OH on the
(1-3) an of 1 to
the lectin. These results agree with the recently described x-ray
crystal structure of the complex (8).
Nonlinear effects in the H and
G
values of the deoxy analogs indicate differential contributions of the
solvent and protein to their binding, and not direct measurements of
the loss in H-bonding interactions.