Bioconjugate Chem. 2004, 15, 87−98
87
Persubstituted Cyclodextrin-Based Glycoclusters as Inhibitors of Protein-Carbohydrate Recognition Using Purified Plant and Mammalian Lectins and Wild-Type and Lectin-Gene-Transfected Tumor Cells as Targets† Sabine Andre´,*,‡ Herbert Kaltner,‡ Tetsuya Furuike,§ Shin-Ichiro Nishimura,*,| and Hans-Joachim Gabius‡ Institute of Physiological Chemistry, Faculty of Veterinary Medicine, Ludwig-Maximilians-University Munich, Veterina¨rstr. 13, D-80539 Munich, Germany, and Laboratory for Glycocluster Project, Japan Bioindustry Association, Graduate School of Science, Hokkaido University, Sapporo 060-0810, Japan, and Division of Biological Sciences, Graduate School of Science, Hokkaido University, Kita-10 Nishi-8, Kitaku, Sapporo 060-0810, Japan. Received May 1, 2003; Revised Manuscript Received October 15, 2003
Multivalent glycoclusters have the potential to become pharmaceuticals by virtue of their target specificity toward clinically relevant sugar receptors. Their application can also provide fundamental insights into the impact of two spatial factors on binding, i.e., topologies of ligand (branching mode, cluster presentation) and carbohydrate recognition domains in lectins. Persubstituted macrocycles derived from nucleophilic substitution of iodide from heptakis 6-deoxy-6-iodo-β-cyclodextrin by the unprotected sodium thiolate of 3-(3-thioacetyl propionamido)propyl glycosides (galactose, lactose and N-acetyllactosamine) were prepared. The produced glycoclusters were first tested as competitive inhibitors in solid-phase assays. A plant toxin from mistletoe and an immunoglobulin G fraction from human serum were markedly susceptible. A nearly 400-fold increase in inhibitory potency of each galactose moiety in the heptavalent form relative to free lactose (217-fold relative to free galactose) was detected. Thus, these glycoclusters can efficiently interfere, for example, with xenoantigendependent hyperacute rejection. Among the tested galectins selected from this family of adhesionand growth-regulatory endogenous lectins, the substituted β-cyclodextrins acted as sensors to delineate topological differences between the two dimeric prototype proteins. The relatively strong reactivity with chimera-type galectin-3, a mediator of tumor metastasis, disclosed selectivity for glycocluster binding among galectins. Equally important, the geometry of ligand display (maxiclusters, bi- or triantennary N-glycans) made its mark on the inhibitory potency. To further determine the sensitivity of a distinct galectin presented on the cell surface and not in solution, we established a stably transfected tumor cell clone. We detected a significant response to presence of the multivalent inhibitor. This type of chemical scaffold with favorable pharmacologic properties might thus be exploited for the design of galectin- and ligand-type-selective glycoclusters.
INTRODUCTION
The emergence of the concept of the sugar code entails an increasing interest in the activities of endogenous lectins (Winterburn and Phelps, 1972; Laine, 1997; Reuter and Gabius, 1999; Ru¨diger et al., 2000; Gabius et al., 2002; Kilpatrick, 2002). On account of cell surface presentation and capacity for endocytosis, medical applications are being examined for several C-type lectins (Gabius, 1991, 1997; Rice, 1997; Yamazaki et al., 2000, 2001; Davis and Robinson, 2002), by exploiting glycoligands as a postal code in vectorized drug delivery. Acting as role models for this purpose, the hepatic and macrophage asialoglycoprotein receptors and the tandemrepeat mannose macrophage receptor have already served * To whom correspondence should be addressed. S.A.: Fax: +49-89-2180-2508; e-mail:
[email protected]. S.N.: Fax: +81-11-706-3435; e-mail:
[email protected]. † Dedicated in respect and thankful commemoration to Prof. Dr. F. Cramer who recently deceased three months before his 80th birthday. ‡ Ludwig-Maximilians-University Munich. § Laboratory for Glycocluster Project, Hokkaido University. | Division of Biological Sciences, Hokkaido University.
as proof-of-principle. In fact, the design of multivalent ligands matching the rather rigid topological display of these lectins’ carbohydrate recognition domains was the basis for achieving astounding affinity enhancements of custom-made glycoclusters relative to their monovalent form, termed the glycoside cluster effect (Lee and Lee, 1994; Biessen et al., 1996; Grandjean et al., 2001; East and Isacke, 2002; Weigel and Yik, 2002). A similarly impressive example for how rational design of the ligand display accomplishes complementarity to fixed geometry of receptor sites is afforded by starfish and modularly prepared pentavalent inhibitors for bacterial AB5 toxins (Schengrund, 2003). Thus, there is already ample reason to conclude that the combination of synthetic chemistry with lectin research promises insights into receptor functionality and a clinical perspective (Lee and Lee, 1994; Bovin and Gabius, 1995; Roy, 1996, 2002; Kiessling 1 Abbreviations: ASF, asialofetuin; CD, cyclodextrin; Gal, galactose; IgG, immunoglobulin G; Lac, lactose; Lac-BSA, lactosylated bovine serum albumin; LacNAc, N-acetyllactosamine; SAP, serum amyloid P component; VAA, Viscum album L. agglutinin.
10.1021/bc0340666 CCC: $27.50 © 2004 American Chemical Society Published on Web 11/22/2003
88 Bioconjugate Chem., Vol. 15, No. 1, 2004
et al., 2000; Gabius, 2001a; Cloninger, 2002; Houseman and Mrksich, 2002). Presently, a major challenge in this area is to figure out how spatial factors and binding avidity toward soluble lectins are correlated. Model studies revealed that sugar surface density is a crucial factor for selectivity of binding, rate of cluster formation, and their stoichiometry (Horan et al., 1999; Cairo et al., 2002). However, increases in epitope density in neoglycoconjugates do not necessarily entail improved capacity as competitive inhibitors in model systems (Zanini and Roy, 1997a,b; Ashton et al., 1998; Andre´ et al., 1999a, 2001). Thus, systematic studies are warranted. After the chemistry to graft bioactive carbohydrates onto synthetic/natural scaffolds had been mastered, first studies on molecular interactions were nearly exclusively performed with plant lectins. Of note, the behavior of endogenous lectins cannot be reliably deduced on this experimental basis. As study objects, we consequently included mammalian proteins from a lectin family with adhesion/growth-regulatory activity on normal and malignant cells, i.e., the galectins (Gabius, 1997; Hirabayashi, 1997; Ohannesian and Lotan, 1997; Kaltner and Stierstorfer, 1998; Andre´ et al., 1999b; Brewer, 2002; Danguy et al., 2002; Rabinovich et al., 2002). For comparison, we also worked with a galactoside-specific plant lectin with potent signaling activity andsat increased concentrationstoxicity on mammalian cells (Endo, 1989; Timoshenko et al., 1999; Gabius, 2001b) and an affinity-purified human immunoglobulin G fraction, a model for auto- or xenoreactive carbohydrate-binding antibodies in serum. Similar to galectins, these two sugar receptors are also cross-linking modules (Gupta et al., 1996). Persubstituted β-cyclodextrins maintaining their Cn-symmetry were chosen as test substance owing to several favorable properties for potential applications. Cyclodextrins are nonimmunogenic natural macrocycles with inherently low pharmacological activity and high biocompatibility, the capacity to host pharmaceuticals in their truncated cone-shaped hydrophobic cavity coming along as added value. Functionalization with carbohydrate derivatives was accomplished with the longterm aim to devise an “intelligent” drug delivery system (Fulton and Stoddart, 2001a; Houseman and Mrksich, 2002; Ortiz Mellet et al., 2002). Having developed the facile synthesis of persubstituted β-cyclodextrins with galactose (Gal), lactose (Lac), or N-acetyllactosamine (LacNAc) as headgroup (Furuike et al., 2000), these glycoclusters (for schematic illustration, please see Figure 1) were introduced to assays with galectins. These endogenous lectins harbor marked intrafamily complexity. It encompasses different modes of spatial orientations of the carbohydrate recognition domains and crosslinking capacity, leading to functional divergence in disease mechanisms such as tumor invasion and spread (Kopitz et al., 2001; Lahm et al., 2001; Brewer, 2002; Camby et al., 2002; Cooper, 2002; Nangia-Makker et al., 2002; Nagy et al., 2003). For this study, we selected two related prototype proteins (galectins-1 and -7) and the chimera-type galectin-3. Because sugar density and topology are emerging as crucial factors modulating lectin binding, we systematically tested the glycoclusters’ capacity to inhibit lectin binding to a panel of surfacepresented ligands. Toward this end, (neo)glycoproteins presenting either lactose maxiclusters, bi- or triantennary N-glycans, or a complex mixture of N- and O-glycans were used for establishing the model surface. Last but not least, our study design accounted for another parameter of importance: the potency to interfere with multi-
Andre´ et al.
Figure 1. Structural illustration of the perglycosylated β-cyclodextrin derived from coupling 3-(3-thioacetyl propionamido)propyl glycosides after de-O,S-acetylation to yield unprotected glycosides with a terminal sodium thiolate to heptakis 6-deoxy6-iodo-β-cyclodextrin.
valent interactions can vary with the nature of the test system (Mann et al., 1998). To address this issue properly, we performed solid-phase, haemagglutination, and cell binding assays. In these experimental settings the lectins were invariably present in solution. To be able to probe galectin behavior when presented on a cell surface, we generated a stable galectin-1-overexpressing colorectal cancer cell clone by transfection as a test model. The presented data reveal (a) efficiency of the substituted cyclodextrins to abolish receptor binding, (b) differential sensitivity of the tested galectins, (c) the improtance of ligand topology, and (d) of lectin presentation either in solution or on the cell surface. EXPERIMENTAL PROCEDURES
Reagents. The persubstituted glycoclusters with galactose, lactose, and N-acetyllactosamine as headgroup (please see Figure 1 for structural details) were prepared by conjugation of the respective sodium thiolate derived from treatment of 3-(3-thioacetyl propionamido)propyl sugars with NaOMe to heptakis 6-deoxy-6-iodo-β-cyclodextrin at 70 °C for 24 h, as described (Furuike et al., 2000). Following gel filtration on Sephadex G25, concentration of peak fractions, washing of the crude solid and lyophilization analytical procedures (optical rotation, 1H and 13C NMR spectroscopy, and matrix-assisted laser desorption ionization time-of-flight mass spectrometry) checked structure and purity of the products prior to the inhibition studies. Galectin-1 from bovine heart, murine galectin-3 (using the plasmid prCBP35s and E. coli JA 221 cells; a kind gift of Dr. J. L. Wang), and human galectin-7 (using the plasmid pQE-60/hGal-7 and E. coli M15[pREP4] cells; a kind gift of Dr. F.-T. Liu), the
Inhibition of Lectin Binding by Glycoclusters
Bioconjugate Chem., Vol. 15, No. 1, 2004 89
Table 1. Determination of the IC50 Values and the Inhibitory Capacity (relative potency, rel pot.) of Gal-, Lac-, and LacNAc-Containing β-Cyclodextrins Relative to the Univalent Inhibitor Lactose in a Solid-Phase Assay inhibitor
Lac-BSA sugar content/ molecule IC50 (µM) rel pot.
SAP IC50 (µM)
ASF rel pot.
IC50 (µM)
laminin rel pot.
IC50 (µM)
rel pot.
0.6 1 24.6 (3.5) 1304 (186)
