Biosynthetic Oligosaccharide Libraries for Identification of Protein-Binding Heparan Sulfate Motifs* Exploring the structural diversity by screening for FGF1 and FGF2 binding

Heparan sulfate is crucial for vital reactions in the body because of its ability to bind various proteins. The identification of protein-binding heparan sulfate sequences is essential to our understanding of heparan sulfate biology and raises the possibility to develop drugs against diseases such as cancer and inflammatory conditions. We present proof-of-principle that in vitro generated heparan sulfate oligosaccharide libraries can be used to explore interactions between heparan sulfate and proteins, and that the libraries expand the available heparan sulfate sequence space. Oligosaccharide libraries mimicking highly 6-O-sulfated domains of heparan sulfate were constructed by enzymatic O-sulfation of O-desulfated, end-group (3)H-labeled heparin octasaccharides. Acceptor oligosaccharides that were 6-O-desulfated but only partially 2-O-desulfated yielded oligosaccharide arrays with increased ratio of iduronyl 2-O-sulfate/glucosaminyl 6-O-sulfate. The products were probed by affinity chromatography on immobilized growth factors, fibroblast growth factor-1 (FGF1) and FGF2, followed by sequence analysis of trapped oligosaccharides. An N-sulfated octasaccharide, devoid of 2-O-sulfate but with three 6-O-sulfate groups, was unexpectedly found to bind FGF1 as well as FGF2 at physiological ionic strength. However, a single 2-O-sulfate group in the absence of 6-O-sulfation gave higher affinity for FGF2. FGF1 binding was also augmented by 2-O-sulfation, preferentially in combination with an adjacent upstream 6-O-sulfate group. These results demonstrate the potential of the enzymatically generated oligosaccharide libraries.


SUMMARY
Heparan sulfate is crucial for vital reactions in the body due to its ability to bind various proteins. The identification of protein-binding heparan sulfate sequences is essential to our understanding of heparan sulfate biology, and raises the possibility to develop drugs against diseases such as cancer and inflammatory conditions.
We present proof-of-principle that in vitrogenerated heparan sulfate oligosaccharide libraries can be used to explore interactions between heparan sulfate and proteins, and that the libraries expand the available heparan sulfate sequence space. Oligosaccharide libraries mimicking highly The HS carbohydrate backbone consists of alternating hexuronate (HexUA) and α-D-glucosamine monosaccharide units (2,5). The HexUA can be either β-D-glucuronate (GlcUA) or α-L-iduronate (IdoUA). The amino group of the α-D-glucosamine is usually acetylated (GlcNAc) or sulfated (GlcNS), but it may also be unsubstituted. Within and adjacent to N-sulfated domains, 2-hydroxyl groups of IdoUA (IdoUA2S) or, less commonly, GlcUA, and 6-hydroxyl groups of GlcNS (GlcNS6S) or GlcNAc may be sulfated. Also, 3-O-sulfate groups can occur on GlcNS units. These modifications give rise to a large variety of different motifs along the HS chain, which usually consists of 50-400 monosaccharide units. Considering the heterogeneity of naturally occurring HS, the possibility of different proteins interacting with distinct HS epitopes is apparent. Indeed, following the development of HS sequencing methods (6, 7) protein-specific HS motifs have been elucidated (8,9).
Attempts to identify protein-binding HS motifs have so far relied on HS isolated from tissues (10). Owing to the structural variability of HS molecules from different cells (5,11,12) there is an impending risk that a HS epitope assigned to interaction with a given protein in vivo may be poorly represented in a tissue-derived HS pool used for screening in vitro.
We have therefore taken a novel approach, generating HS libraries in vitro, that are based on enzymatic O-sulfation of N-sulfated precursors. The potential of the library saccharides to functionally mimic HS domains was verified by their selective interactions with fibroblast growth factor-1 (FGF1) and FGF2. heparin pyridinium salt with 2.5 mL dimethylsulfoxide (DMSO) containing 10 % (v/v) methanol at 100 °C for 16 h (13). Part of the product (2 mg), was completely 2-O-desulfated by dissolution in 10 mL 30 mM NaOH, pH 12.5 followed by lyophilization (14). The N/O-desulfated heparins were re-Nsulfated by incubation with 4.5 mL 80 mM Na 2 CO 3 , 50 mM sulfur trioxide Affinity chromatography-Preparations of recombinant FGFs were as described (21). Affinity columns were made by coupling primary amines in the proteins (1 mg) to activated esters in Activated CH Sepharose 4B gel matrix (1 g, dry weight), in a total volume of 4 mL according to the instructions of the manufacturer (Amersham Biosciences). To protect lysine residues involved in binding of HS, FGF-Sepharose conjugates were prepared in the presence of heparin (5 mg), that had been treated with HNO 2 at pH 3.9 to eliminate Nunsubstituted glucosamine residues (17). The FGF columns were equilibrated with 50 mM Tris/HCl, pH 7.4 and the respective octasaccharide library (~2×10 6 cpm) was loaded. After washing the columns with buffer, bound octasaccharides were eluted using a stepwise NaCl gradient in 50 mM Tris/HCl, pH 7.4.
Characterization of affinity-fractionated oligosaccharides-Octasaccharides were separated further by anion-exchange HPLC on the ProPac PA1 column. Molecular weights of oligosaccharides were determined by MALDI-MS as described (22,23) with slight modifications. Oligosaccharides (10 pmol) were directly dissolved in 4 µL of the matrix solution (12 mg caffeic acid/mL in 50% aqueous acetonitrile) containing 20 pmol/mL of the basic peptide (Arg-Gly) 19 -Arg (Sigma-Genosys Ltd.). The peptideoligosaccharide sample (1 µL) was then deposited on the stainless-steel MALDI plate and dried at room temperature. Further sample purification was performed in situ on the MALDI target as described (24). Mass spectrometry measurements were performed on a Bruker Biflex III MALDI TOF instrument (delayed extraction, mass gate set to 2000 Da) calibrated with cytocrome C and adrenocorticotropic hormone.
The deamination and enzyme digestion products were analyzed on the ProPac PA1 column. the different ionic strengths used to develop the column (Fig. 3) [Fig. 3

here].
As seen for Library 1 with FGF1 as probe (Fig. 2), as well as previously for However, selected octasaccharides of lower charge density remained bound to the growth factor at relatively high ionic strength, e.g., fragments e and g in  Fig. 4 shows an example of such analysis, of the fragments making up peak g in Fig. 3 [Fig. 4 here].
Notably, the method enables concomitant analysis of mixtures of two (or even three) components, provided that their degree of sulfation is known (8). The occurrence of two components in fraction g was indicated by the identification of two distinct disaccharides, IdoUA-[ 3 H]aMan R and IdoUA2S-[ 3 H]aMan R , following exhaustive deaminative cleavage of the endgroup-labeled parent fraction (data not shown). Accordingly, exoenzyme digestions of the products obtained by partial deamination revealed two distinct tetrasaccharides and also two hexasaccharides (Fig. 4). All structures deduced are shown in Fig. 5.
No GlcUA residues were found, except at the nonreducing terminus (unit 1) of fragment c. The relative molecular masses of three octasaccharides were In HS biosynthesis 2-O-sulfation generally precedes 6-O-sulfation, and the distribution of the latter substituents appears to be quite strictly regulated (12,42). Thus, whereas excessively 6-O-sulfated sequences such as that displayed by fragment a are rarely seen in nonmanipulated systems, 2-O-and 6-O-sulfated structures of g and h type may be detected in authentic HS (8,42). The physiological significance of the high affinity for FGF1 (and other growth factors) expressed by these structures, as compared to weaker binders, remains unclear. Conceivably, a gradient of HS motifs with increasing affinities may serve to capture and direct a growth factor molecule towards the site of interaction with its tyrosine kinase receptor target at the cell surface (43). However, we cannot exclude that high-affinity binding may sequester growth factors at sites on HS chains out of reach for receptor 22 by guest on July 9, 2020 http://www.jbc.org/ Downloaded from interaction. Productive interaction appears to depend on binding of not only FGF but also its receptor to the same HS chain (25). It will therefore be important to isolate and structurally characterize HS domains capable of accommodating both proteins in a ternary complex.
In conclusion, the oligosaccharide libraries obtained in the current work present a novel tool to explore protein-HS interactions. Other potential applications include the analysis of substrate specificity for biosynthetic enzymes such as sulfotransferases. Finally, the technique may be useful in guiding future enzyme engineering toward catalysts that can generate tailored HS oligosaccharides, for use as reagents in research or even as drugs.