Platform Synthetic Lectins for Divalent Carbohydrate Recognition in Water

Abstract Biomimetic carbohydrate receptors (“synthetic lectins”) have potential as agents for biological research and medicine. However, although effective strategies are available for “all‐equatorial” carbohydrates (glucose, etc.), the recognition of other types of saccharide under natural (aqueous) conditions is less well developed. Herein we report a new approach based on a pyrene platform with polar arches extending from aryl substituents. The receptors are compatible with axially substituted carbohydrates, and also feature two identical binding sites, thus mimicking the multivalency observed for natural lectins. A variant with negative charges forms 1:2 host/guest complexes with aminosugars, with K 1>3000 m −1 for axially substituted mannosamine, whereas a positively charged version binds the important α‐sialyl unit with K 1≈1300 m −1.

Abstract: Biomimetic carbohydrate receptors ("synthetic lectins") have potential as agents for biological researcha nd medicine.H owever,a lthough effective strategies are available for "all-equatorial" carbohydrates (glucose,e tc.), the recognition of other types of saccharide under natural (aqueous) conditions is less well developed. Herein we report an ew approach based on ap yrene platform with polar arches extending from aryl substituents.The receptors are compatible with axially substituted carbohydrates,a nd also feature two identical binding sites,t hus mimicking the multivalency observed for natural lectins.Avariant with negative charges forms 1:2h ost/guest complexes with aminosugars,w ith K 1 > 3000 m À1 for axially substituted mannosamine,w hereas ap ositively charged version binds the important a-sialyl unit with K 1 % 1300 m À1 .
Carbohydrate recognition is ac entral biological phenomenon that mediates ar ange of cellular processes. [1] Carbohydrate-binding molecules are important as research tools for investigating these processes,a nd potentially as diagnostic and therapeutic agents in medicine. [1][2][3] Studies in this area most commonly use lectins,t he major class of saccharidebinding proteins,but lectins often lack the desired selectivities and tend to show low affinities (generally 10 3 -10 4 m À1 for monosaccharides). [4] Moreover,a sp roteins,t heir therapeutic potential is limited by issues such as immunogenicity. [3] There is consequently much interest in small-molecule receptors, which could complement lectins and perhaps be developed for new types of application. [1,3,5] However,the design of such molecules has proved difficult, especially for biomimetic systems based on noncovalent bonding. [6] Although avariety of structures have been shown to be active in organic solvents, [7] there are few which can operate in the natural but challenging environment of water. [8] We have approached this problem by constructing symmetrical cavities with an aromatic "roof" and "floor" separated by polar spacers (e.g. 1,F igure 1a). [9] Thed esigns are complementary to saccharides with all-equatorial substitution patterns (e.g. 2), and have yielded encouraging results.S electivities are good, and some affinities are above 10 4 m À1 ,e ven for uncharged substrates. [9a,b] However,t he selectivity for all-equatorial carbohydrates is ac onstraint on potential applications,a smany substrates of interest do not belong to this family.H erein we report an alternative design strategy which rationally targets carbohydrates with axial substituents and which, for the first time,m imics the multivalencye xhibited by many lectins. [1] While all-equatorial saccharides possess two roughly similar hydrophobic patches,o ther carbohydrates tend to be [9a]). The symmetrical cavity matches the polar (red) and apolar (blue) groups in the substrate. b) As trategy for binding saccharides with axial substituents, illustrated for the b-galactosyl group 3 as the substrate. c) General design of the pyrene-based receptors described herein.
facially amphiphilic.T he inversion of as tereocenter,a sinbgalactosyl 3,adds to the polarity of one face while decreasing the polarity of the other. Ac omplementary binding site should therefore contain just one extended apolar surface,the remainder being mainly polar. [10] An approach to such structures might involve an aromatic platform with polar substituents that can arch over ab ound carbohydrate (e.g. Figure 1b). When considering options for realizing this architecture,w en oted the potential of 1,3,6,8-tetraarylpyrenes ( Figure 1c). These compounds are readily prepared from tetrabromopyrene (8)b yS uzuki-Miyaura methodology,a nd are forced to adopt nonplanar conformations owing to interactions between aryl and peri-H groups. [11] The meta positions on the aryl groups could provide anchor points for the polar "arches". Stereoisomers would be possible in the general case,b ut could be avoided by using symmetrically substituted aryl groups,a ss hown in Figure 1c.I nterestingly, this arrangement would generate two equivalent binding sites, thus mimicking the multivalencycommon in lectins. [1] Groups Zc ould be used to confer water-solubility as well as to provide polar interactions.F or example,p olyionic dendrimers,w hich are highly water solubilizing and capable of hydrogen bonding to polar carbohydrate substituents,c ould be readily installed. [9c] As ap rototype for this design, we chose the tetracosacarboxylate 9.M odeling [12] showed that the dendrimers in 9, though relatively small, possessed sufficient reach to interact with axial groups on as ubstrate.P rotonated mannosamine 10·H + ,w ith an axial NH 3 + group,w as found to be an especially promising substrate ( Figure 2). Receptor 9 was prepared in 23 %y ield over four steps from diacid 4,a mine 5, [13] and 1,3,6,8-tetrabromopyrene (8; [14] Scheme 1). [12] The anionic receptor 9 dissolved freely in water to give wellresolved 1 HNMR spectra, which were concentration-independent below 1.2 mm,t hus implying am onomeric species. Fluorescence spectra showed an emission maximum at 425 nm (excitation wavelength:3 80 nm), detectable down to nanomolar concentrations and with intensity directly proportional to concentration.
Carbohydrate recognition was studied by fluorescence and 1 HNMR titrations of 9 with ar ange of sugars. [12] Solutions were adjusted to pH 7, and the pH value was confirmed to be unchanged after each experiment. Titrations with aminosugars 10-12 (protonated at pH 7) yielded clear  Figure 3a). AJ ob plot [15] based on the receptor 4-H signal displacements gave amaximum at 9/10 = 1:2, thus confirming divalency (see Figure S14 in the Supporting Information). Analysis of the receptor signals,a ssuming the 1:2b inding model, gave stepwise binding constants K 1 = 3120 m À1 and K 2 = 540 m À1 (see Figure S9). [16] As K 1 = 4 K 2 for noncooperative two-site binding, [17] it seems the two associations are almost independent. Theresults were supported by fluorescence titrations of 9 with 10,w hich showed large increases in receptor emission intensity and could be analysed to give almost identical binding constants (Figure 4a nd Table 1).
Changes in the positions of carbohydrate signals during the 1 HNMR titration were also informative ( Figure 3b). All the signals from 10 moved downfield during the titration implying that, as expected, these protons are shielded in the complex. [18] Thes pectra provide separate information for a and b anomers,p resent in the ratio 1:1.9. [19] Signals for hydrogen atoms on the a face of the b anomer [for example, H1(b), H5(b)] showed especially large movements,consistent with the modeled structure ( Figure 2). From the a anomer, the proton H5(a)s ignal also moved substantially.N OE data (see Figure S15) confirmed that 9 and the carbohydrate are closely associated. Binding could not be quantified reliably because of the complexity of the system (two substrates,both forming 1:1a nd 1:2c omplexes). However,t he signal movements were consistent with K 1 % 3000 m À1 for both anomers.
Titrations with galactosamine (11)a nd glucosamine (12) gave similar changes in both 1 HNMR and fluorescence spectra. Analysis of the data gave the binding constants shown in Table 1. Affinities were somewhat lower, suggesting that the binding sites of 9 favor axial NH 3 + substitution. Uncharged monosaccharides did not appear to bind, but titrations with the disaccharides cellobiose,l actose,a nd maltose,a nd also methylamine,g ave evidence of weakcomplex formation (K 1 16 m À1 ). [12] It thus seems that the high affinities for aminosugars result from combining carbohydrate-specific interactions (hydrophobic,C H-p,h ydrogen bonding) with electrostatic attraction.
Whereas cationic sugars are relatively uncommon in nature,a nionic carbohydrates are widespread and play important roles.E specially significant is the a-linked Nacetylneuraminic acid (a-sialyl) unit 13.T his moiety commonly appears as at erminus of oligosaccharides,a ccessible for binding and therefore an important potential target. [20]

Communications
Thes ubstitution pattern in 13 is all-equatorial with an additional axial negatively charged substituent. This compound should be nicely complementary to our platform design (Figure 1b), provided that the polar arches are positively charged.
To test this concept, we prepared the cationic receptor 16, possessing 24 guanidinium units (see the Supporting Information). Guanidinium substituents were chosen to ensure that the receptor would be fully protonated at pH 7. [21] Spectroscopic studies implied that receptor 16 is monomeric in water below ac oncentration of 1.2 mm.A se xpected, the 1 HNMR spectrum of 16 was unaffected by the pH value in the range 6-8, confirming full protonation.
Studies of 16 as areceptor for the a-sialyl unit 13 required amodel substrate.The parent saccharide N-acetylneuraminic acid is readily available,b ut exists mainly (> 90 %) as the b anomer 15,a nd is thus unrepresentative of 13.S imple asialosides are not commercially available,a nd we therefore synthesized the methyl derivative 14 through av ariation of aliterature procedure. [12] Thebinding of 14 to 16 was studied by 1 HNMR and fluorescence spectroscopy.N OESY crosspeaks between substrate and receptor aromatic proton signals supported complex formation (see Figure S55). On the other hand, NMR titrations yielded relatively small changes in the signal positions,implying alooser geometry than for 9 + 10.A Jobp lot based on ar eceptor aromatic signal confirmed the expected 16/14 = 1:2b inding stoichiometry (see Figure S54). Forq uantitative analysis,w ee mployed at itration in which receptor 16 was added to substrate 14.T he carbohydrate signals moved upfield as expected, and several could be followed throughout (see Figure S57). Simultaneous analysis of four of these signals was consistent with three successive binding events,w ith K 1 , K 2 ,a nd K 3 = 1310, 570, and 30 m À1 , respectively.Given the high density of positive charge on 16, it is reasonable to suppose that athird (and possibly afourth) molecule of 14 might bind to the receptor.
Fluorescence titration of 16 with 14 yielded asurprisingly strong effect ( Figure 5);r eceptor emission was reduced almost to zero by addition of the carbohydrate.A nalysis of the changes assuming a1:2 binding model (see Figure 5) gave stepwise binding constants K 1 , K 2 = 1300, 790 m À1 ,c onsistent with the NMR data. Thed ecrease in emission was found to depend on the chloride counterions;w hen these were replaced with trifluoroacetate,f luorescence increased on binding.W ep resume that the addition of 14 to 16 causes ar earrangement of the counterions,t hus promoting fluorescence quenching.
Fluorescence titrations were also performed for receptor 16 with several uncharged monosaccharides. [12] Efficient quenching was again observed, although analysis suggested that binding was much weaker than for 14.Moderate affinities were estimated for methyl b-d-glucoside (K 1 = 43 m À1 ), methyl b-d-galactoside (K 1 = 46 m À1 ), and galactose (K 1 = 17 m À1 ). In these cases,NMR shifts were too small for analysis, but association between the receptor and the carbohydrate was confirmed by NOE enhancements. [22] Sodium acetate was also bound weakly (K 1 = 17 m À1 ). Theaffinity for glucose was too small to be quantified implying that, as intended, the platform design can reverse the selectivity shown by our earlier synthetic lectins.
In conclusion, we have reported ar ational design for synthetic lectins which mimic the multivalencys hown by natural lectins and, unlike earlier systems,can accommodate substrates with axial substituents.T here is scope for varying the system, for example,byaltering/extending the side chains or changing one aryl substituent. Thepotential for binding the b-sialyl group 13 is especially significant, and will be afocus of future efforts.