THE JOURNAL OF BIOLOGICAL CHEMISTRY
© 2001 by The American Society for Biochemistry and Molecular Biology, Inc.
Vol. 276, No. 20, Issue of May 18, pp. 17052–17057, 2001
Printed in U.S.A.
The Mannose/N-Acetylgalactosamine-4-SO4 Receptor Displays
Greater Specificity for Multivalent than Monovalent Ligands*
Received for publication, February 2, 2001, and in revised form, March 5, 2001
Published, JBC Papers in Press, March 6, 2001, DOI 10.1074/jbc.M101027200
Daniel S. Roseman and Jacques U. Baenziger‡
From the Department of Pathology, Washington University School of Medicine, St. Louis, Missouri 63110
Recognition of carbohydrates on glycosylated molecules typically requires multivalent interactions with
receptors. Monovalent forms of terminal saccharides engaged by the receptor binding sites typically display
weak affinities in the mM range and poor specificity. In
contrast, multivalent forms of the same saccharides are
bound with strong affinity (10ⴚ7–10ⴚ9 M) and significantly greater specificity. Although multivalency can
readily account for increased affinity, the molecular basis for enhanced specificity is not well understood. We
have examined the specificity of the cysteine-rich domain of the mannose/GalNAc-4-SO4 receptor using
monovalent and multivalent forms of the trisaccharide
GalNAc␤1,4GlcNAc␤1,2Man␣ (GGnM) sulfated at either
the C4 (S4GGnM) or C3 (S3GGnM) hydroxyl of the terminal GalNAc. Monovalent S4GGnM and S3GGnM have
Ki values of 25.8 and 16.2 ␮M, respectively. Multivalent
conjugates of the same GalNAc-4-SO4- and GalNAc-3SO4-bearing trisaccharides (6.7 mol of trisaccharide/mol
of bovine serum albumin) have Ki values of 0.013 and
0.170 ␮M, respectively. The 2000-fold versus 95-fold
change in affinity seen for the multivalent forms of these
4-sulfated and 3-sulfated trisaccharides reflects a difference in the impact of conformational entropy. A large
fraction of the SO4-3-GalNAc structures exists in a form
that is not favorable for binding to the Cys-rich domain.
This reduces the effective concentration of SO4-3-GalNAc as compared with SO4-4-GalNAc under the same
conditions and results in a markedly lower association
rate. This difference in association rate accounts for the
12-fold difference in the rate of clearance from the blood
seen with S4GGnM-BSA and S3GGnM-BSA in vivo.
We previously demonstrated that glycoproteins such as LH1
and thyrotropin bearing N-linked oligosaccharides terminating
* This work was supported by National Institutes of Health Grants
R37-CA21923 and R01-DK41738. The costs of publication of this article
were defrayed in part by the payment of page charges. This article must
therefore be hereby marked “advertisement” in accordance with 18
U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed. Tel.: 314-362-8730;
Fax: 314-362-8888; E-mail: [email protected].
1
The abbreviations used are: LH, lutropin; Man, mannose; S4GGnM,
SO4-4-GalNAc␤1,4GlcNAc␤1,2Man␣; S3GGnM, SO4-3-GalNAc␤1,4Glc
NAc␤1,2Man␣; BSA, bovine serum albumin; MCO, O-(CH2)8OOCH3;
3⬘-SO4-LewisX, 3⬘-O-SO3Gal␤1,4(Fuc␣1,3)GlcNAc; Mu0, Man/GalNAc4-SO4-Fc(Mu0), transmembrane and cytosolic domains of Man/GalNAc4-SO4 receptor replaced with the Fc refion of human IgG1; Mu10,
Man/GalNAc-4-SO4Fc(Mu10), fibronectin type II repeat and carbohydrate recognition domains 1–3 deleted from Man/GalNAc-4SO4Fc(Mu0); Mu11, Man/GalNAc-4-SO4-Fc(Mu11), fibronectin type II
repeat and carbohydrate recognition domains 1– 8 deleted from Man/
GalNAc-4-SO4-Fc(Mu0); SPR, surface plasmon resonance; APTS,
9-aminopyrene-1,4,6-trisulfonic acid; MOPS, 3-[N-morpholino]propane
sulfonic acid.
with ␤1,4-linked GalNAc-4-SO4 are recognized by a receptor
located in hepatic endothelial cells, the Man/GalNAc-4-SO4
receptor, and are rapidly removed from the circulation (1– 4).
Rapid clearance in conjunction with stimulated release from
dense core storage granules in LH-producing cells located in
the anterior lobe of the pituitary produce the episodic rise and
fall in serum LH levels that is important for the expression of
LH bioactivity in vivo. A novel feature of the Man/GalNAc-4SO4 receptor is its capacity to bind carbohydrate moieties terminating with Man and GalNAc-4-SO4 at physically distinct
domains that are not structurally related. The binding site for
GalNAc-4-SO4 is located in the cysteine-rich domain found at
the N terminus of the Man/GalNAc-4-SO4 receptor (5). The
Cys-rich domain is a member of the ␤-trefoil family of proteins
that includes proteins such as acidic fibroblast growth factor.
The binding site is a neutral pocket that accommodates the
sulfate group, which accounts for the major interactions with
the protein (6). Inhibition and modeling studies with monovalent
ligands have indicated that the binding site in the Cys-rich domain can also accommodate saccharides with terminal Gal or
GalNAc when the sulfate is located at C3 rather than C4 (6, 7).
The latter observations were unexpected because we had
reported that bovine serum albumin substituted with 6 – 8 molecules of SO4-4-GalNAc␤1,4GlcNAc␤1,2Man␣ (S4GGnM-BSA)
is removed from the circulation at a rate that is at least 12-fold
greater than that seen for bovine serum albumin substituted
with 6 – 8 molecules of SO4-3-GalNAc␤1,4GlcNAc␤1,2Man␣
(S3GGnM-BSA). Furthermore, isolated hepatic endothelial
cells do not bind and internalize S3GGnM-BSA, and the purified Man/GalNAc-4-SO4 receptor does not display significant
binding of S3GGnM-BSA in precipitation assays (2, 3). We have
recently determined that the Man/GalNAc-4-SO4 receptor
must be dimeric and engage at least two terminal GalNAc-4SO4 moieties on separate oligosaccharides to mediate uptake
by hepatic endothelial cells with a Kd of 1.63 ⫻ 10⫺7 M (8).
These observations raise the possibility that the Man/GalNAc4-SO4 receptor may display significantly greater specificity
when binding ligands with multiple terminal sulfated saccharide moieties than when binding monovalent forms of the same
ligands. We have examined this possibility using well characterized ligands of known valency and structure. Our studies
demonstrate that the Man/GalNAc-4-SO4 receptor displays
greater specificity for multivalent forms of the same structures
than for monovalent forms and that this can be explained best
by the presence of favorable and unfavorable conformations of
the trisaccharides for binding to the Cys-rich domain. Remarkably, inhibition constants obtained with the monovalent forms
of these trisaccharides are not predictive of the properties of
multivalent forms of the same structures.
EXPERIMENTAL PROCEDURES
Materials—Man/GalNAc-4-SO4-Fc(Mu0) (the transmembrane and
cytosolic domains are replaced by the hinge, CH2, and CH3 domains of
17052
This paper is available on line at http://www.jbc.org
Man/GalNAc-4-SO4 Receptor Displays Multivalent Ligand Specificity
human IgG1), Man/GalNAc-4-SO4-Fc(Mu10) (deletion of the fibronectin
type II repeat and carbohydrate recognition domains (CRDs) 1–3 from
Mu0), and Man/GalNAc-4-SO4-Fc(Mu11) (deletion of the fibronectin
type II repeat and CRDs 1– 8 from Mu0) were prepared as described (5).
3⬘-SO4-LewisX (3⬘-O-SO3Gal␤1,4(Fuc␣1,3)GlcNAc) was from Calbiochem. S4GGnM-MCO, S3GGnM-MCO, and their BSA conjugates,
S4GGnM-BSA and S3GGnM-BSA, were prepared as described (9). The
LH␣ subunit was supplied by the National Hormone and Pituitary
Program (NIDDK, National Institutes of Health) and A. F. Parlow.
Surface Plasmon Resonance Analysis—Surface plasmon resonance
(SPR) analyses were performed on an Amersham Pharmacia Biotech
BIACORE 2000 instrument as described previously (8). For studies
using immobilized the Cys-rich domain, a predetermined amount of
Man/GalNAc-4-SO4-Fc(Mu11) was allowed to bind to protein A (Pierce),
which had been covalently attached to a CM5 sensor chip. The amount
of Mu11 bound was monitored by SPR. No detectable dissociation of
Mu11 from protein A was observed under the conditions of the assay.
For inhibition studies the bovine LH␣ subunit (685 nM) was passed over
the chip at a flow rate of 5 ␮l/min for 300 s in the presence of increasing
concentrations of saccharide inhibitors. The amount of LH␣ bound at
equilibrium was used to generate inhibition curves that were analyzed
by nonlinear regression using PRISM software (version 2.0). Ki values
were calculated from the concentration of inhibitor that effected 50%
inhibition. For glycoconjugate binding studies, differing concentrations
of S4GGnM-BSA and S3GGnM-BSA were passed over immobilized
Mu11 at a flow rate of 5 ␮l/min for 10 min. Dissociation was initiated at
10 min by elution with TBS buffer (20 mM Tris䡠HCl, pH 7.4, 150 mM
NaCl) ⫹ 0.005% (w/v) surfactant P20 or TBS buffer ⫹ 0.005% surfactant P20 containing 1 mM GalNAc-4-SO4.
Binding of Mu11 to immobilized S4GGnM-BSA and S3GGnM-BSA
was also monitored using SPR. Similar amounts of S4GGnM-BSA and
S3GGnM-BSA were attached covalently to a BIACORE CM5 sensor
chip through primary amine groups using the amine coupling kit provided by the manufacturer. The amount of S4GGnM-BSA and
S3GGnM-BSA conjugated was determined from the increase in response units. Mu11 was passed over immobilized BSA conjugates at a
flow rate of 5 ␮l/min for 10 min. The amount of Mu11 bound at equilibrium was used to generate saturation curves that were analyzed by
nonlinear regression using PRISM software (version 2.0).
Mole ratios were determined by dividing the change in response unit
values obtained for immobilized or bound glycoproteins by their respective molecular weights. The molecular weights utilized were: LH␣,
14,600; Mu11, 98,100; S4GGnM-BSA, 73,100; and S3GGnM-BSA,
73,100. The values calculated for S4GGnM-BSA and S3GGnM-BSA
reflect the presence of 7 mol of sulfated trisaccharide conjugated/mol of
BSA and are consistent with their mobility when examined by SDSpolyacrylamide gel electrophoresis (data not shown).
Equilibrium Dialysis Studies—3⬘-SO4-LewisX and a glucose pentasaccharide (Glc5) prepared from dextran by mild acid hydrolysis (10)
were derivatized with 9-aminopyrene-1,4,6-trisulfonic acid (APTS) by
reductive amination as described (11). Binding was performed in a
two-chamber Teflon Micro-Equilibrium dialyzer (Amika, Inc.) utilizing
ultrathin 10,000-dalton cut-off membranes (The Nest Group, Inc.). One
chamber contained either 10 ␮M Mu11 (based on Mr ⫽ 98,100) or BSA
(1 mg/ml) in 25 ␮l of TBS buffer. The second chamber contained 0.1–10
nmol of 3⬘-SO4-LewisX-APTS and Glc5-APTS in 25 ␮l of TBS buffer.
Dialysis was allowed to proceed for 96 h at 4 °C. Thereafter, 2-␮l
aliquots were removed from each side of the membrane, and the
amounts of 3⬘-SO4-LewisX-APTS and Glc5-APTS were quantitated by
capillary electrophoresis using an N-CHO-coated capillary column on a
Beckman P/ACE 5000 and laser-induced fluorescence for detection as
described by the manufacturer. The Glc5-APTS was used to normalize
the amount of 3⬘-SO4-LewisX-APTS because it did not demonstrate any
binding to Mu11 or BSA.
Monosaccharide Composition Analysis of BSA Conjugates—Quantitative analysis of the sugar constituents of the sulfated glycoconjugates
was performed according to the method developed by Chen et al. (12).
S4GGnM-BSA and S3GGnM-BSA were hydrolyzed with 2.0 N trifluoroacetic acid to release their constituent sugars. After amino sugar Nacetylation with acetic anhydride in sodium carbonate, the released
monosaccharides were derivatized with APTS by reductive amination.
Capillary electrophoresis was performed using a fused silica column in
120 mM MOPS buffer, pH 7.0, on a Beckman P/ACE 5000 (13). The
APTS monosaccharides were identified and quantitated using laserinduced fluorescence and comparison with monosaccharide standards.
Solution Binding Assays—Affinity-purified Man/GalNAc-4-SO4Fc(Mu10) was incubated with 125I-labeled S4GGnM-BSA (1 nM) in 150
␮l of TBS buffer containing 1% (w/v) Triton X-100 and 0.1% (w/v) bovine
17053
FIG. 1. Inhibition of LH␣ binding to immobilized Man/GalNAc4-SO4-Fc(Mu11) by monovalent sulfated saccharides. LH␣ was
passed over immobilized Mu11 in the presence of the indicated concentrations of SO4-3-GalNAc␤1,4GlcNAc␤1,2Man␣-MCO (●), SO4-4GalNAc␤1,4GlcNAc␤1,2Man␣-MCO (E), GalNAc-4-SO4 (䡺), or 3⬘-SO4LewisX (f). The amount of LH␣ bound at equilibrium in the presence of
inhibitor was used to generate the curves shown. Ki values were calculated using the inhibitor concentration that effected 50% inhibition.
IgG in the presence of increasing concentrations of unlabeled S4GGnMBSA or S3GGnM-BSA. Mu10-S4GGnM-125I-BSA complexes were precipitated by the addition of 1.5 ml of ice-cold 10% polyethylene glycol-8000
(Sigma) and collected on Whatman GF/C filter discs as described (3).
RESULTS
Specificity of the Cys-rich Domain for Monovalent Ligands—We
have characterized the affinity and stoichiometry of binding for
ligands bearing different numbers and/or configurations of terminal ␤1,4-linked GalNAc-4-SO4 moieties utilizing SPR. The competitive inhibition studies shown in Fig. 1 were performed by passing
the LH␣ subunit over Man/GalNAc-4-SO4-Fc(Mu11) bound to immobilized protein A in the presence of increasing amounts of monovalent inhibitors. GalNAc-4-SO4, 3⬘-SO4-LewisX, S4GGnM-MCO,
and S3GGnM-MCO have Ki values of 28.2, 29.7, 25.8, and 16.2 ␮M,
respectively (Table I). The similar Ki values obtained for GalNAc4-SO4 and S4GGnM indicate that recognition is directed almost
exclusively at the terminal sulfated monosaccharide. The Ki value
determined for S3GGnM is half that obtained for S4GGnM. Thus,
for the monovalent trisaccharide SO4-GalNAc␤1,4GlcNAc␤1,
2Man␣-MCO location of the sulfate at the C3 hydroxyl as compared
with the C4 hydroxyl of the terminal, GalNAc enhances interaction
with the binding site in the Cys-rich domain.
We also determined Kd and Bmax values for 3⬘-SO4-LewisX
binding to Mu11 using equilibrium dialysis. Preliminary experiments established both the conditions and the time required
for equilibrium to be reached (data not shown). Glc5-APTS did
not bind to Mu11, indicating no interaction between APTS and
the Cys-rich domain (data not shown). 3⬘-SO4-LewisX-APTS
bound to Mu11 with a Kd of 18.5 ␮M and a Bmax of 1.02 mol of
3⬘-SO4-LewisX-APTS bound/mol of Cys-rich domain (Fig. 2).
The Kd obtained by equilibrium dialysis is in good agreement
with the Ki of 29.7 ␮M obtained by inhibition studies on the
BIACORE as described above (Table I). Furthermore, in agreement with our previous studies (8) and the crystallographic
analysis (6), each Cys-rich domain has a single binding site for
a sulfated saccharide.
Specificity of the Cys-rich Domain for Multivalent Ligands—We previously determined the dissociation constant for
S4GGnM-BSA is 0.013 ␮M for Man/GalNAc-4-SO4-Fc(Mu0) and
Man/GalNAc-4-SO4-Fc(Mu11) (5). The assay utilized for these
determinations relied on polyethylene glycol-mediated precipitation of S4GGnM-125I-BSA complexed to Mu0 or Mu11. We
determined inhibition constants for S4GGnM-BSA and
S3GGnM-BSA using S4GGnM-125I-BSA and Man/GalNAc-4SO4-Fc(Mu10) (Fig. 3). Mu10 was used because it is secreted
more efficiently after transfection than Mu0 and is more efficiently precipitated by 10% polyethylene glycol-8000 than
Mu11. S4GGnM-BSA has a Ki of 0.013 ␮M, whereas S3GGnM-
17054
Man/GalNAc-4-SO4 Receptor Displays Multivalent Ligand Specificity
TABLE I
Binding constants for monovalent and polyvalent ligands sulfated at
either the C3 or C4 hydroxyl of terminal GalNAc or Gal
GalNAc-4-SO4a
3⬘-SO4-LewisXa
S3GGnMa
S4GGnMa
3⬘-SO4-LewisXb
S3GGnM-BSAc
S4GGnM-BSAc
Kd
Kid
Bmax
␮M
␮M
mol/mol
28.2 (⫾3.8)
29.7 (⫾1.6)
16.2 (⫾2.7)
25.8 (⫾4.2)
18.5
1.02
0.170 (⫾0.013)
0.013 (⫾0.007)
a
Competitive inhibition of LH␣ binding to immobilized Man/GalNAc4-SO4-Fc(Mu11) on the BIACORE.
b
Equilibrium dialysis studies.
c
Competitive inhibition of S4GGnM-[125I]BSA binding to Man/GalNAc-4-SO4-Fc(Mu10) (Mu10: fibronectin type II domains and carbohydrate recognition domains 1–3 have been deleted) in the precipitation
(PEG) assays.
d
Mean (⫾ standard deviation) for replicate experiments.
FIG. 2. Determination of the stoichiometry and dissociation
constant for 3ⴕ-SO4-LewisX binding to the Cys-rich domain of the
Man/GalNAc-4-SO4 receptor. Increasing amounts of 3⬘-SO4-LewisXAPTS and Glc5-APTS were dialyzed against 10 ␮M Man/GalNAc-4-SO4Fc(Mu11) for 96 h at 4 °C. The amount of 3⬘-SO4-LewisX-APTS and
Glc5-APTS in each 25-␮l chamber was determined by capillary electrophoresis. A Scatchard analysis of the results is shown and yields a Kd of
18.5 ␮M and a Bmax of 1.02 mol of 3⬘-SO4-LewisX-APTS bound/mol of
Cys-rich domain (N) at saturation. B/F, bound/free ligand.
FIG. 3. S4GGnM-BSA and S3GGnM-BSA have different Ki values. S4GGnM-125I-BSA was incubated with Man/GalNAc-4-SO4Fc(Mu10), and complexes were precipitated by the addition of 10%
polyethylene glycol-8000. Increasing amounts of either S3GGnM-BSA
(●) or S4GGnM-BSA (E) were added to generate the inhibition curves
shown. Ki values were determined as summarized in Table I.
BSA has a Ki of 0.170 ␮M (Table I). This results in a 2000-fold
change in the affinity for S4GGnM-BSA as compared with
monovalent S4GGnM (0.013 versus 25.8 ␮M) and a 95-fold
change in affinity for S3GGnM-BSA as compared with the
monovalent S3GGnM (0.170 versus 16.2 ␮M). Even though the
Ki for S4GGnM is 1.6-fold greater than the Ki for S3GGnM,
the Ki for S3GGnM-BSA is 13-fold greater than the Ki for
S4GGnM-BSA. Thus, the Man/GalNAc-4-SO4 receptor displays
significantly greater specificity for multivalent than for monovalent forms of the same ligand.
Kinetics for Binding S4GGnM-BSA and S3GGnM-BSA—
The above inhibition studies alone did not provide insight to
the molecular basis for the difference in specificity for monovalent and multivalent forms of the same trisaccharides. We
therefore used SPR to examine the kinetics of S4GGnM-BSA
(Fig. 4A) and S3GGnM-BSA (Fig. 4B) binding to immobilized
Mu11. At a concentration of 400 nM, the binding of S4GGnMBSA rapidly reaches saturation. Elution with buffer devoid of
S4GGnM-BSA beginning at 700 s results in little or no dissociation. The highest amount of S4GGnM-BSA bound yields a
mole ratio of 0.34 mol S4GGnM-BSA/mol of Mu11 or 0.17 mol
S4GGnM-BSA/mol of Cys-rich domain. This is equal to a Cysrich domain/S4GGnM-BSA mole ratio of 5.9, which is a value
close to the theoretical maximum of 6.7 based on the number of
trisaccharides conjugated to the BSA (see below) and a single
binding site for terminal GalNAc-4-SO4/Cys-rich domain. Further, this indicates that the slow dissociation rate is a result of
multivalent binding. At a concentration of 8 nM S4GGnM-BSA
(Fig. 4A) the amount of S4GGnM-BSA bound increases
throughout the binding phase of the experiment. Because of the
lack of significant dissociation of the bound ligand, if given
sufficient time the amount of S4GGnM-BSA bound to Mu11
would attain the same maximum as was seen with 400 nM
S4GGnM-BSA. Further evidence for the contribution of multivalency to the slow dissociation rate was obtained by co-injection of 1 mM GalNAc-4-SO4 during the dissociation phase (Fig.
4C). The S4GGnM-BSA was rapidly and completely dissociated, indicating that the equilibrium at individual GalNAc-4SO4 binding sites is rapid. The dissociation rate in the presence
of 1 mM GalNAc-4-SO4 is so rapid that the dissociation rate
could not be determined by SPR.
S3GGnM-BSA was also bound by immobilized Mu11 (Fig.
4B). The amount of S3GGnM-BSA bound was less than the
amount of S4GGnM-BSA bound at the same concentration
(compare 400-nM concentrations in Fig. 4, A and B). However,
the amount of S3GGnM-BSA bound to Mu11 continues to
increase with time. Furthermore, little or no dissociation
of S3GGnM-BSA is observed when elution with buffer free of
S3GGnM-BSA is initiated at 700 s. Thus, as was seen for
S4GGnM-BSA, complexes formed with S3GGnM-BSA are multivalent and highly stable. Furthermore, bound S3GGnM-BSA
is rapidly dissociated from Mu11 in the presence of 1 mM
GalNAc-4-SO4 (Fig. 4C). Given sufficient time and/or increased
concentrations of S3GGnM-BSA, maximal amounts of binding
equal to those obtained with S4GGnM-BSA can be attained
(data not shown).
Because neither S4GGnM-BSA nor S3GGnM-BSA dissociates at a significant rate from the immobilized Mu11, the
results shown in Fig. 4, A and B, indicate that at identical
concentrations the association rate for S3GGnM-BSA is considerably slower than for S4GGnM-BSA. A difference in the ratio
of trisaccharide conjugation to BSA was excluded as an explanation for the different rates by determining the mole ratio of
trisaccharide to BSA for each conjugate. After acid hydrolysis,
the released monosaccharides were derivatized with APTS and
quantitated by capillary electrophoresis using a laser-induced
fluorescence detector (data not shown). The mole ratio of trisaccharide to BSA was 6.73 for S4GGnM-BSA and 6.75 for
S3GGnM-BSA, which is in agreement with the analyses performed at the time of synthesis. Thus, the different characteristics seen for S4GGnM-BSA and S3GGnM-BSA binding to
immobilized Mu11 reflect a significantly slower association
rate for S3GGnM-BSA than for S4GGnM-BSA.
The multivalent nature of the binding of S4GGnM-BSA and
Man/GalNAc-4-SO4 Receptor Displays Multivalent Ligand Specificity
17055
FIG. 4. S4GGnM-BSA and S3GGnMBSA display markedly different rates
of binding to immobilized Man/GalNAc-4-SO4-Fc(Mu11). Mu11, 4180 response units, was bound to the protein A
that had been coupled to a CM5 biosensor
chip. Concentrations of 400, 80, and 8 nM
S4GGM-BSA (A) and S3GGM-BSA (B)
were injected at a flow rate of 5 ␮l/min,
and the amount of binding was monitored
by SPR. After 700 s, buffer without
S4GGnM-BSA or S3GGnM-BSA was
passed over the chip to monitor dissociation. C, S4GGnM-BSA (80 nM) and
S3GGnM-BSA (400 nM) were injected as
above except that 1 mM GalNAc-4-SO4
was co-injected during the dissociation
phase.
S3GGnM-BSA to immobilized Mu11 and the consequent lack of
dissociation prevent kinetic analyses. We therefore used SPR to
characterize the binding of Mu11 to identical concentrations of
immobilized S4GGnM-BSA and S3GGnM-BSA. Although the
immobilized BSA conjugates are multivalent, the Man/GalNAc4-SO4-Fc(Mu11) chimera is bivalent and can only engage two
terminal sulfated GalNAc moieties. The results of these analyses
are shown in Figs. 5 and 6 and summarized in Table II. Mu11
reaches equilibrium rapidly when binding to either immobilized
S4GGnM-BSA or S3GGnM-BSA (Fig. 5). Mu11 bound to immobilized S4GGnM-BSA with a Kd of 3.90 ␮M and bound to immobilized S3GGnM-BSA with a Kd of 3.93 ␮M. However, Mu11
bound to S4GGnM-BSA with a Bmax of 0.74 mol/mol and bound to
S3GGnM-BSA with a Bmax of 0.09 mol/mol (see Fig. 6 and Table
II). Thus, even though Mu11 binds to both conjugates with a
similar affinity under these conditions, there is an 8-fold difference in the amount of Mu11 bound at saturation. This is true
even though the number and concentration of terminal sulfated
saccharides are essentially identical.
DISCUSSION
Our analyses demonstrate that inhibition constants obtained
with monovalent ligands do not necessarily predict the properties of multivalent forms of the same ligands. This study provides new insights into the basis for the remarkable difference
in specificity displayed by the Man/GalNAc-4-SO4 receptor for
multivalent as compared with monovalent ligands. The clearance studies we performed in the rat (1), as well as subsequent
analyses with the isolated hepatic endothelial cells (2) and the
purified Man/GalNAc-4-SO4 receptor (3), indicated that this
receptor displays specificity for the location of the sulfate on
terminal ␤1,4-linked GalNAc in the context of the sequence
GalNAc␤1,4GlcNAc␤1,2Man␣. The recent report (7) that
monovalent saccharides with a sulfate located at the C3 hydroxyl of Gal have Ki values similar to those of monovalent
structures with the sulfate located at the C4 hydroxyl of Gal or
GalNAc seemed to contradict these observations. The studies
we have presented here demonstrate that the Man/GalNAc-4-
17056
Man/GalNAc-4-SO4 Receptor Displays Multivalent Ligand Specificity
TABLE II
Binding of increasing concentrations of Mu11 to equal amounts
of immobilized S3GGnM-BSA and S4GGnM-BSA was monitored
by SPR
The amount bound at equilibrium for each concentration in Fig. 5 was
used to generate the curves shown in Fig. 6. The Kd and Bmax were
determined by nonlinear regression analysis.
Kda
S3GGnM-BSA
S4GGnM-BSA
a
FIG. 5. Analysis of Man/GalNAc-4-SO4-Fc(Mu11) binding to immobilized S4GGnM-BSA and S3GGnM-BSA using surface plasmon resonance. Increasing concentrations of Mu11, ranging from 2 to
500 ␮g/ml (0.02–5 ␮M), were injected over a surface with immobilized
S4GGnM-BSA (A) or S3GGnM-BSA (B) for 10 min. at a flow rate of 5
␮l/min. The surfaces contained 6267 and 5769 response units of immobilized S4GGnM-BSA and S3GGnM-BSA, respectively. No significant
change in the refractive index was observed in the presence of 1 mM
GalNAc-4-SO4 or when Mu11 was injected over a surface with immobilized Man-BSA (data not shown).
FIG. 6. Man/GalNAc-4-SO4-Fc(Mu11) binds to immobilized
S4GGnM-BSA and S3GGnM-BSA with the same Kd but a different Bmax. The equilibrium binding data from the experiments shown in
Fig. 5 were used to generate saturation curves for Mu11 binding to
immobilized S4GGnM-BSA (E) and S3GGnM-BSA (●). The results of
regression analyses are summarized in Table II. RU, response units
SO4 receptor displays far greater specificity for multivalent
ligands than would be predicted on the basis of binding constants obtained for monovalent forms of the same ligands.
Each Cys-rich domain is able to engage a single terminal
sulfated monosaccharide. In agreement with the studies of
Leteux et al. (7) and Liu et al. (6, 14), monovalent forms of the
trisaccharide GalNAc␤1,4GlcNAc␤1,2Man␣-MCO sulfated at
either the C3 or C4 hydroxyl of the terminal GalNAc are bound
Bmaxa
␮M
mol/mol
3.93 (⫾0.38)
3.90 (⫾0.15)
0.091 (⫾0.007)
0.735 (⫾0.021)
Mean (⫾ standard deviation) for replicate experiments.
with similar affinities of 16.2 and 25.8 ␮M, respectively. How
can we, therefore, account for the dissociation constants of
0.170 and 0.013 ␮M obtained for S3GGnM-BSA and S4GGnMBSA, respectively? Investigators have previously observed
greater specificity on the part of lectins for polyvalent ligands
as compared with their monovalent counterparts (15–22). The
specificity for multivalent ligands has been attributed to a
requirement for unique spacing patterns (15–18), secondary
protein-protein and protein-carbohydrate interactions (19, 20),
and amplification in the polyvalent state of small differences in
the affinities between monovalent forms (21). One prediction of
a model in which small differences in affinity of monovalent
ligands are amplified in the polyvalent state is that the affinities obtained for Mu11 binding to immobilized S4GGnM-BSA
and S3GGnM-BSA would differ but that the Bmax obtained at
saturation would be the same. This is not the case because
Mu11 binds to immobilized S4GGnM-BSA and S3GGnM-BSA
with the same affinity but with an 8-fold difference in Bmax at
saturation. The behavior seen in Fig. 6 is explicable if the
trisaccharide S3GGnM exists in two distinct conformations,
only one of which is favorable for binding (see Fig. 7A). The
crystal structures of the Cys-rich domain binding 3⬘-SO4Lewisa (3⬘-(SO4-Gal␤1,3(Fuc␣1,4) GlcNAc)) and 3⬘-SO4-LewisX
indeed display significant differences in the structure of the
carbohydrate, suggesting that one or both of the 3⬘-sulfated
oligosaccharides undergo a conformational change when binding to the Cys-rich domain (14). The N-acetyl present on the
GalNAc-3-SO4 may require an even greater change in conformation to allow binding. Formation of stable complexes requires the simultaneous engagement of two terminal GalNAc3-SO4 moieties by the bivalent Mu11. If only a small portion of
the S3GGnM is in a form favorable for binding at any point in
time, and all of the S4GGnM is in a favorable form for binding,
the effective concentration of S3GGnM-BSA would appear to be
a fraction of that of S4GGnM-BSA. However, when binding
does occur to the favorable form of S3GGnM, it would have the
same properties as the binding to S4GGnM-BSA. Hence the
two would have the same Kd but different Bmax values.
Differences in the contribution of conformational entropy
(23), defined as the contribution of entropy arising from a
difference in the number of conformations available before
complexation and after complexation, are the most likely basis
for the 13-fold difference in the Ki values obtained for
S4GGnM-BSA and S3GGnM-BSA. When S3GGnM-BSA and
S4GGnM-BSA bind to immobilized Mu11, this difference is
manifested as a slow association rate for S3GGnM-BSA as
compared with S4GGnM-BSA (Fig. 4). Once a sufficient number of terminal sulfated saccharides are engaged, the multivalent interaction results in a negligible dissociation rate for both
S3GGnM-BSA and S4GGnM-BSA (see Fig. 7B). As a result, if
given sufficient time S3GGnM-BSA will reach the same maximal level of binding as S4GGnM-BSA when examined on the
BIACORE. Equilibration at individual binding sites is rapid,
however, because both S3GGnM-BSA and S4GGnM-BSA dis-
Man/GalNAc-4-SO4 Receptor Displays Multivalent Ligand Specificity
FIG. 7. A model for how the Cys-rich domain of the Man/GalNAc-4-SO4 receptor interacts differently with S4GGnM-BSA and
S3GGnM-BSA. A, bivalent Mu11 (tulip shape) binds to both immobilized S4GGnM-BSA and S3GGnM-BSA. Equilibration between conformations favorable and unfavorable for Mu11 binding reduces the effective concentration of S3GGnM termini as compared with S4GGnM
termini. B, Mu11 immobilized on a surface (dark line) will bind multivalent forms of S4GGnM-BSA (filled triangles, GalNAc-4-SO4) and
S3GGnM-BSA (open triangles, GalNAc-3-SO4). Because S3GGnM exists in both favorable and unfavorable forms for binding to Mu11, it
takes longer to form multiple interactions with the immobilized Mu11.
S4GGnM and S3GGnM are both in rapid equilibrium at individual sites
but do not dissociate because of the inability to simultaneously release
all of the bound termini. In the presence of free GalNAc-4-SO4 dissociation is rapid because no rebinding can occur.
sociate from the immobilized Mu11 in the presence of 1 mM
GalNAc-4-SO4 at a rate that is too rapid to determine the koff
(Figs. 4 and 7B). Rapid equilibration at individual binding sites
in conjunction with rapid equilibration between conformations
that are favorable and nonfavorable for binding may account
for the similar Ki values obtained for monovalent forms of these
sulfated trisaccharides.
The different rates of binding to immobilized Mu11 and the
different Ki values obtained for S4GGnM-BSA and S3GGnMBSA mirror the behavior seen in vivo (2). Radiolabeled
S4GGnM-BSA is rapidly removed from the blood (t1⁄2 ⫽ 1.3
min), whereas S3GGnM-BSA is removed slowly (t1⁄2 ⫽ 15.3
min). Thus, the 12-fold difference in clearance rate correlates
well with the 13-fold difference in Ki. Furthermore, S4GGnMBSA but not S3GGnM-BSA is bound and internalized by isolated hepatic endothelial cells (2). Because the GalNAc-4-SO4specific form of the Man/GalNAc-4-SO4 receptor in hepatic
endothelial cells exists as a dimer (8), binding two terminal
GalNAc-3-SO4-moieties is not likely to produce a sufficiently
stable complex to mediate internalization. In addition, the interaction with a greater number of terminal GalNAc-3-SO4moieties by bridging between dimeric GalNAc-4-SO4 receptors
is likely a slow process relative to the rate of receptor internalization. As a result ligands terminating with Gal or GalNAc
sulfated at the C3 hydroxyl are not likely to represent ligands
for the Man/GalNac-4-SO4 receptor in vivo.
Soluble forms of the Man/GalNAc-4-SO4 receptor have been
reported in the circulation and are thought to arise as a result of
proteolytic digestion (24). These soluble forms of the Man/
GalNAc-4-SO4 receptor would not be likely to bind terminal
GalNAc-4-SO4, GalNAc-3-SO4, or Gal-3-SO4 unless they remain
17057
dimeric or are in some manner multimerized. The amount of
GalNAc-3-SO4 or Gal-3-SO4 that would be required would be at
least 8-fold higher than GalNAc-4-SO4. Whether the soluble
forms of the Man/GalNAc-4-SO4 receptor in serum exist as dimeric or monomeric molecules remains to be addressed.
Our studies demonstrate that although the binding site of
the Cys-rich domain of the Man/GalNAc-4-SO4 receptor can
accommodate a number of different sulfated monosaccharides
as monovalent ligands, it displays much more highly restricted
specificity when simultaneously engaging two or more terminal
sulfated GalNAc or Gal moieties. That this specificity is attained through differences in conformational entropy is remarkable and may apply to other instances in which multivalent ligands are recognized by lectins with greater specificity
than the same monovalent ligands. The affinity and specificity
of the Man/GalNAc-4-SO4 receptor for oligosaccharides terminating with ␤1,4-linked GalNAc-4-SO4 are essential for regulating the half-life of LH in the circulation throughout the
ovulatory cycle including the preovulatory surge. The ability to
ignore other ligands whether monovalent or multivalent assures that the clearance of LH will be maintained even if other
sulfated ligands are present in the blood. Precisely how multivalent forms of S3GGnM are excluded from binding while
multivalent forms of S4GGnM are not excluded will require
real-time analysis of such ligands in the presence of dimeric
Cys-rich domains. Models of oligosaccharides terminating with
GalNAc-4-SO4 and GalNAc-3-SO4 reveal marked differences in
their three-dimensional structures that would potentially have
a major impact on access to the binding site on the Cys-rich
domain. This represents a highly novel mechanism to dictate
the specificity of carbohydrate receptors for complex multivalent saccharide structures in vivo.
Acknowledgment—We thank Nancy Baenziger for critical comments.
REFERENCES
1. Baenziger, J. U., Kumar, S., Brodbeck, R. M., Smith, P. L., and Beranek, M. C.
(1992) Proc. Natl. Acad. Sci. U. S. A. 89, 334 –338
2. Fiete, D., Srivastava, V., Hindsgaul, O., and Baenziger, J. U. (1991) Cell 67,
1103–1110
3. Fiete, D., and Baenziger, J. U. (1997) J. Biol. Chem. 272, 14629 –14637
4. Fiete, D., Beranek, M. C., and Baenziger, J. U. (1997) Proc. Natl. Acad. Sci.
U. S. A. 94, 11256 –11261
5. Fiete, D. J., Beranek, M. C., and Baenziger, J. U. (1998) Proc. Natl. Acad. Sci.
U. S. A. 95, 2089 –2093
6. Liu, Y., Chirino, A. J., Leteux, C., Feizi, T., Misulovin, Z., Nussenzweig, M. C.,
and Bjorkman, P. (2000) J. Exp. Med. 191, 1105–1115
7. Leteux, C., Chai, W., Loveless, R. W., Yuen, C.-T., Uhlin-Hansen, L.,
Combarnous, Y., Jankovic, M., Maric, S. C., Misulovin, Z., Nussenzweig,
M. C., and Feizi, T. (2000) J. Exp. Med. 191, 1117–1126
8. Roseman, D. S., and Baenziger, J. U. (2000) Proc. Natl. Acad. Sci. U. S. A. 97,
9949 –9954
9. Vandana, Hindsgaul, O., and Baenziger, J. U. (1987) Can. J. Chem. 65,
1645–1652
10. Mellis, S. J., and Baenziger, J. U. (1981) Anal. Biochem. 114, 276 –280
11. Guttman, A., Chen, F. T., Evangelista, R. A., and Cooke, N. (1996) Anal.
Biochem. 233, 234 –242
12. Chen, F. T., Dobashi, T. S., and Evangelista, R. A. (1998) Glycobiology 8,
1045–1052
13. Chen, F. T., and Evangelista, R. A. (1995) Anal. Biochem. 230, 273–280
14. Liu, Y., Misulovin, Z., and Bjorkman, P. J. (2001) J. Mol. Biol. 305, 481– 490
15. Baenziger, J. U., and Maynard, Y. (1980) J. Biol. Chem. 255, 4607– 4613
16. Baenziger, J. U., and Fiete, D. (1980) Cell 22, 611– 620
17. Lee, Y. C., and Lee, R. T. (1995) Acc. Chem. Res. 28, 321–327
18. Adler, P., Wood, S. J., Lee, Y. C., Lee, R. T., Petri, W. A., Jr., and Schnaar, R. L.
(1995) J. Biol. Chem. 270, 5164 –5171
19. Horan, N., Yan, L., Isobe, H., Whitesides, G. M., and Kahne, D. (1999) Proc.
Natl. Acad. Sci. U. S. A. 96, 11782–11786
20. Liang, R., Loebach, J., Horan, N., Ge, M., Thompson, C., Yan, L., and Kahne,
D. (1997) Proc. Natl. Acad. Sci. U. S. A. 94, 10554 –10559
21. Mortell, K. H., Weatherman, R. V., and Kiessling, L. L. (1996) J. Am. Chem.
Soc. 118, 2297–2298
22. Dam, T. K., Roy, R., Das, S. K., Oscarson, S., and Brewer, C. F. (2000) J. Biol.
Chem. 275, 14223–14230
23. Mammen, M., Choi, S.-K., and Whitesides, G. M. (1998) Angew. Chem. Int. Ed.
Engl. 37, 2754 –2794
24. Martinez-Pomares, L., Mahoney, J. A., Kaposzta, R., Linehan, S. A., Stahl,
P. D., and Gordon, S. (1998) J. Biol. Chem. 273, 23376 –23380