Overall Sulfation of Heparan Sulfate from Pancreatic Islet β-TC3 Cells Increases Maximal Fibril Formation but Does Not Determine Binding to the Amyloidogenic Peptide Islet Amyloid Polypeptide*

Background: Stimulation of amyloid fibril formation by heparan sulfate is likely mediated by the extent of sulfation. Results: Islet β-cell heparan sulfate is poorly sulfated but still binds islet amyloid polypeptide (IAPP) and accelerates fibril formation. Conclusion: The degree of sulfation does not determine all aspects of heparan sulfate-mediated amyloid fibril formation. Significance: This information will be important for designing amyloid inhibitors targeting peptide-heparan sulfate interactions. Islet amyloid, a pathologic feature of type 2 diabetes, contains the islet β-cell peptide islet amyloid polypeptide (IAPP) as its unique amyloidogenic component. Islet amyloid also contains heparan sulfate proteoglycans (HSPGs) that may contribute to amyloid formation by binding IAPP via their heparan sulfate (HS) chains. We hypothesized that β-cells produce HS that bind IAPP via regions of highly sulfated disaccharides. Unexpectedly, HS from the β-cell line β-TC3 contained fewer regions of highly sulfated disaccharides compared with control normal murine mammary gland (NMuMG) cells. The proportion of HS that bound IAPP was similar in both cell lines (∼65%). The sulfation pattern of IAPP-bound versus non-bound HS from β-TC3 cells was similar. In contrast, IAPP-bound HS from NMuMG cells contained frequent highly sulfated regions, whereas the non-bound material demonstrated fewer sulfated regions. Fibril formation from IAPP was stimulated equally by IAPP-bound β-TC3 HS, non-bound β-TC3 HS, and non-bound NMuMG HS but was stimulated to a greater extent by the highly sulfated IAPP-bound NMuMG HS. Desulfation of HS decreased the ability of both β-TC3 and NMuMG HS to stimulate IAPP maximal fibril formation, but desulfated HS from both cell types still accelerated fibril formation relative to IAPP alone. In summary, neither binding to nor acceleration of fibril formation from the amyloidogenic peptide IAPP is dependent on overall sulfation in HS synthesized by β-TC3 cells. This information will be important in determining approaches to reduce HS-IAPP interactions and ultimately prevent islet amyloid formation and its toxic effects in type 2 diabetes.

Islet amyloid, a pathologic feature of type 2 diabetes, contains the islet ␤-cell peptide islet amyloid polypeptide (IAPP) as its unique amyloidogenic component. Islet amyloid also contains heparan sulfate proteoglycans (HSPGs) that may contribute to amyloid formation by binding IAPP via their heparan sulfate (HS) chains. We hypothesized that ␤-cells produce HS that bind IAPP via regions of highly sulfated disaccharides. Unexpectedly, HS from the ␤-cell line ␤-TC3 contained fewer regions of highly sulfated disaccharides compared with control normal murine mammary gland (NMuMG) cells. The proportion of HS that bound IAPP was similar in both cell lines (ϳ65%). The sulfation pattern of IAPP-bound versus non-bound HS from ␤-TC3 cells was similar. In contrast, IAPP-bound HS from NMuMG cells contained frequent highly sulfated regions, whereas the nonbound material demonstrated fewer sulfated regions. Fibril formation from IAPP was stimulated equally by IAPP-bound ␤-TC3 HS, non-bound ␤-TC3 HS, and non-bound NMuMG HS but was stimulated to a greater extent by the highly sulfated IAPP-bound NMuMG HS. Desulfation of HS decreased the ability of both ␤-TC3 and NMuMG HS to stimulate IAPP maximal fibril formation, but desulfated HS from both cell types still accelerated fibril formation relative to IAPP alone. In summary, neither binding to nor acceleration of fibril formation from the amyloidogenic peptide IAPP is dependent on overall sulfation in HS synthesized by ␤-TC3 cells. This information will be important in determining approaches to reduce HS-IAPP interactions and ultimately prevent islet amyloid formation and its toxic effects in type 2 diabetes.
Islet amyloid deposition is a pathological hallmark of the pancreatic islet in type 2 diabetes (1). Aggregation of islet amyloid polypeptide (IAPP), 2 a normal peptide product of the islet ␤-cell, underlies the deposition of islet amyloid, a process that contributes to the decreased ␤-cell volume that characterizes type 2 diabetes (2)(3)(4). The mechanism(s) that govern the aggregation of this normally soluble polypeptide are poorly understood. In addition to amyloidogenic IAPP, several other components of islet amyloid have been identified, including apolipoprotein E (5, 6), serum amyloid P component (7), and the heparan sulfate proteoglycan perlecan (6,8).
Heparan sulfate proteoglycans (HSPGs) are a heterogeneous population of proteoglycans involved in a diverse range of cellular processes ranging from vascular development to cell signaling (9). HSPGs are components of amyloid deposits that form in a number of diseases, including type 2 diabetes (6,8) and Alzheimer disease (10). HSPGs may play a role in stimulating amyloid deposition via a direct interaction with amyloidogenic peptides. We and others have shown that HSPGs are capable of binding the amyloidogenic peptide IAPP (11,12), its precursor pro-IAPP (13) and other amyloidogenic peptides, including amyloid-␤ (14), and serum amyloid A (15), the unique amyloidogenic peptides from Alzheimer disease-related amyloid and inflammation-associated AA amyloidosis, respectively. Upon binding amyloidogenic peptides, HSPGs or their oligosaccharide heparan sulfate (HS) glycosaminoglycan chains have been shown to induce structural changes in the peptide to the ␤-sheet structure required for amyloid fibril formation (16) and to increase fibril formation from the amyloidogenic peptides (11,14). These findings suggest that HSPGs are important for amyloid formation.
Heparan sulfate undergoes a complex set of modifications following its synthesis in the endoplasmic reticulum. Many of these modifications are sulfation reactions, catalyzed by a diverse array of enzymes including N-deacetylase/N-sulfotransferase and 2-, 3-, or 6-O-sulfotransferases (17). These modifications typically result in the formation of domains containing either non-sulfated (acetylated) or highly sulfated oligosaccharides, with the size, arrangement, and composition of these sulfated domains determining the ability of HS to bind a diverse set of protein ligands (18). The composition of HS attached to a given core protein has been shown to vary widely among tissue types, suggesting a complex level of tissue specificity in HS synthesis and modification (17). For example, HS composition, particularly sulfation, of perlecan (19) and syndecan (20) can differ among cell types, and this has been implicated in altered affinity for the perlecan ligand fibroblast growth factor 2 and fibroblast growth factor receptors (19).
Islet amyloid is a form of localized amyloidosis, with its deposition occurring in close proximity to the islet ␤-cell, the source of the amyloidogenic peptide IAPP. We previously showed that the immortalized murine pancreatic islet ␤-cell line ␤-TC3 synthesizes and secretes several proteoglycans, predominantly HSPGs (12,21). Furthermore, we also demonstrated that these ␤-TC3 cell proteoglycans are capable of binding amyloidogenic human IAPP (12), suggesting that these locally produced proteoglycans may be important in IAPP fibril formation and the deposition of islet amyloid in type 2 diabetes. We therefore hypothesized that the degree of sulfation of ␤-TC3 cell HS is an important determinant of IAPP binding and thereby IAPP fibril formation. Here, we report the composition, IAPP-binding, and fibril-enhancing ability of HS synthesized by ␤-TC3 cells.

EXPERIMENTAL PROCEDURES
Cell Culture and Metabolic Labeling-The murine insulinoma cell line, ␤-TC3 (passage 55-57), was utilized for this study, with normal murine mammary gland epithelial (NMuMG, passage 4 -7) cells as a control cell line. The latter were obtained from ATCC. Cells were seeded at 1 ϫ 10 5 cells per ml and cultured for up to 4 days in DMEM containing 10% fetal bovine serum and 5.5 mM or 25 mM glucose for ␤-TC3 or NMuMG cells, respectively, followed by metabolic labeling for 48 h in DMEM containing 10% FBS and 16.7 or 25 mM glucose. 5.5 mM glucose was selected for the 4-day culture of ␤-TC3 cells to avoid glucotoxic effects of long term culture that have been well documented in islets/␤-cells. For IAPP affinity column and solid-phase binding studies with intact proteoglycans, ␤-TC3 cells were labeled for 48 h with 50 or 5 Ci/ml Na 2 [  Isolation of Proteoglycans from ␤-TC3 Conditioned Medium-Culture medium was collected in the presence of protease inhibitors (5 mM benzamidine, 100 mM 6-aminohexanoic acid, and 50 mM phenylmethylsulfonyl fluoride), and proteoglycans were extracted by ion exchange chromatography using DEAE Sephacel in 8 M urea buffer (8 M urea, 2 mM EDTA, 0.25 M NaCl, 50 mM Tris-HCl, and 0.5% Triton X-100, pH 7.4) and eluted with 4 M guanidine buffer (4 M guanidine HCl, 2.5 mM EDTA, 100 mM Na 2 SO 4 , 100 mM Tris, 0.5% Triton X-100 detergent, pH 7.4). Samples were either used as total proteoglycan or subjected to size fractionation via Sepharose CL-4B to yield three pools of proteoglycans based on size (denoted M1, M2, and M3, described in detail in Ref. 12).
Isolation of Heparan Sulfate from ␤-TC3 and NMuMG Cells-Cell layers were extracted in PBS containing 1% (v/v) Triton X-100 and pooled with conditioned medium. Samples were then treated with Pronase (0.1 mg/ml in PBS, pH 7.2; Sigma) for 16 h at 37°C to digest core proteins and then subjected to DEAE ion exchange chromatography with the DEAE equilibrated in PBS containing 0.15 M NaCl, pH 7.2, and samples eluted in PBS containing 3 M NaCl. Glycosaminoglycans were liberated from residual core protein fragments by alkaline elimination and borohydride reduction; samples were treated with 1 M NaBH 4 in 0.025 M NaOH for 24 h at 45°C, followed by neutralization with acetic acid. Samples were then desalted by precipitation (three times) in 95% ethanol/1.3% CH 3 CO 2 K followed by resuspension in deionized water. Chondroitin/dermatan sulfate glycosaminoglycans were removed by digestion with chondroitinase ABC (0.5 units/ml, Seikagaku, Tokyo, Japan; 16 h at 37°C). Samples were then dialyzed over 24 -48 h against at least four changes of deionized water, and the presence of HS/lack of contaminating chondroitin or dermatan sulfate in the sample was verified by size exclusion chromatography Ϯ heparinase I, II, and III digestion.
IAPP Affinity Column-Human IAPP (200 g; Bachem, Torrance, CA) was covalently linked to AminoLink Plus Coupling Gel (Pierce, Thermo Fisher Scientific, Rockford, IL) according to the manufacturer's instructions. Human IAPP was used for this and subsequent experiments as this form of IAPP is amyloidogenic, whereas mouse and rat IAPP are not (22). Furthermore, we have shown that ␤-TC3 proteoglycans bind human but not rodent IAPP (12). Samples were applied to the IAPP affinity column equilibrated in 50 mM Tris-HCl, 0.15 M NaCl, pH 7.4 (Tris-buffered saline, TBS). For intact [ 35 S]sulfate-labeled proteoglycans, 200,000 dpm were loaded, eluted over a 0.15-3 M NaCl gradient in TBS, and finally washed in 8 M urea buffer containing 3 M NaCl. For double-labeled HS samples 100,000 dpm (based on 3 H incorporation to yield equal quantities of HS) were loaded, the column washed in TBS containing 0.15 M NaCl and bound material eluted in TBS containing 3 M NaCl, followed by a final wash in 8 M urea buffer containing 3 M NaCl. Eluted material was analyzed by SDS-PAGE (for intact proteoglycans; 20,000 dpm per lane was loaded) or was subjected to heparinase I or heparinase III enzyme digestion (for HS samples; see details below) followed by Bio-gel P10 size exclusion chromatography.
Solid Phase IAPP Binding Assay-␤-TC3 proteoglycans were subjected to Sepharose CL-4B size exclusion chromatography and subdivided into three fractions based on molecular weight. IAPP binding of proteoglycans by solid phase assay was assessed in fractions M1 and M2 only, as these fractions but not M3 were shown to bind amyloidogenic human IAPP by affinity chromatography. Proteoglycans were concentrated by precipitation in 95% ethanol containing 1.3% potassium acetate, reconstituted in transfer buffer (12.5 mM Tris, 96 mM glycine, 10% (v/v) methanol), and 2,000 dpm per sample was applied to a prewet nitrocellulose membrane via a Bio-Dot microfiltration system (Bio-Rad). Heparan sulfate or chondroitin/dermatan sulfate glycosaminoglycans were selectively removed from the immobilized proteoglycans by heparinase digestion (2.6 units/ml heparinase I, 1.3 units/ml heparinase II ϩ 2.6 units/ml heparinase III, Sigma) or chondroitin ABC lyase digestion (0.1 units/ml), respectively, for 2 h at 37°C. Control (undigested) samples were incubated in the absence of enzyme and negative (no proteoglycan) controls consisted of digestion buffer (0.1 M Tris, 10 mM calcium acetate, 18 mM sodium acetate, pH 7.0) alone. Following digestion, proteoglycans were incubated with 4% (w/v) nonfat dry milk in TBS (100 mM Tris, pH 7.5, 150 mM NaCl, 0.02% NaN 3 ) containing 0.1% (v/v) Tween 20) for 3 h at room temperature followed by overnight incubation at 4°C with human IAPP (10 g/ml in TBS ϩ 0.1% (v/v) Tween 20). Bound IAPP was detected using the monoclonal anti-IAPP antibody F055 (kind gift from Amylin Corp., San Diego, CA; 1:100,000) and visualized using alkaline phosphatase-conjugated anti-mouse immunoglobulins and enhanced chemiluminescence. Presence of residual proteoglycans on the nitrocellulose membrane following heparinase digestion was achieved using the heparan sulfate stubs antibody 3G10 (Seikagaku; 1:1000) and visualized as for anti-IAPP antibody.
Compositional Analyses of HS-Heparan sulfate from ␤-TC3 and NMuMG were analyzed in total HS preparations and in IAPP-bound and non-bound fractions following human IAPP affinity chromatography. Samples (10,000 dpm) were subjected to no treatment (intact HS), low pH (1.5) nitrous acid treatment (23), and heparinase I or heparinase III digestion to assess the proportion of N-sulfated highly sulfated or poorly sulfated heparan sulfate disaccharides, respectively (2.6 units/ml for each enzyme, digested for 2 h at 37°C). Reaction products were separated by size exclusion chromatography on a Bio-gel P10 column equilibrated in 0.5 M NH 4 HCO 3 and/or Sepharose CL-6B equilibrated in TBS, pH 7.0, and the content of [ 3 H]glucosamine and [ 35 S]sulfate-labeled HS per fraction was determined by liquid scintillation counting. Undigested HS elutes in the void volume (V 0 ) of the Bio-gel P10 column (24) (data not shown) and elutes in the included volume of the Sepharose CL-6B column (12).
Disaccharide Analysis-Non-radiolabeled HS was quantified by 1,9-dimethylmethylene blue assay, with glycosaminoglycans of known concentration as standards (25). Analysis of heparan sulfate disaccharide abundance was performed by the Glycotechnology Core Resource at the University of California, San Diego. The limit of detection for heparan sulfate disaccharides is 0.5 pmol/liter. Chemical Desulfation-Pyridine salts of HS samples (10 g for each) were generated by ion exchange on Dowex 50 (H ϩ ) under acidic conditions followed by neutralization with pyridine. These samples were then incubated in 10% (v/v) methanol in dimethyl sulfoxide for 18 h at 80°C, adjusted to pH 9 with NaOH, dialyzed against several changes of deionized water and lyophilized (26,27).
Human IAPP Fibril Formation Assay-For these studies, IAPP was obtained commercially (Bachem) or was kindly provided by Dr. Daniel Raleigh (State University of New York, Stony Brook, NY). IAPP fibril formation alone and in the presence of HS samples did not differ by source of IAPP. For the end point fibril formation assay, HS samples (12.5 l in duplicate) in water were added to an equal volume of synthetic human IAPP peptide to give final concentrations of 25 g/ml for HS and 100 g/ml (25 M) IAPP, yielding a 1:4 (wt/wt) HS:IAPP ratio. For the kinetic fibril formation assay, this HS:IAPP ratio was titrated to ensure good resolution of lag phases and maximal fibril formation within the 132-h time frame of the assay; a 1:32 HS:IAPP ratio was used for these studies. IAPP stock solution (1.6 mM) was filtered (0.2 m), lyophilized, and reconstituted (final concentration 48 g/ml) immediately prior to use to ensure no aggregates were present at base line. Sulfated or desulfated HS samples used at a final concentration of 1.5 g/ml. For both studies, control samples were IAPP alone (positive control for fibril formation) and HS samples with no IAPP (negative controls). Thioflavin T (4 M) was added to samples, and fibril formation assessed by fluorescence at 480 nm/530 nm. For the end point assay, fibril formation was assessed over a period of 6 h, and data are reported after 30 min, which is representative of the whole time course. For the kinetics assay, data were collected every 10 min over 132 h, and data are shown for the whole period.
Data Analysis-Data shown are representative for the following sample sizes. For intact proteoglycan studies, IAPP affinity chromatography was performed twice, and solid phase binding assay was performed at least three times on two independent proteoglycan preparations. For HS analyses, IAPP affinity columns for radiolabeled samples were performed on three independent preparations; radiolabeled composition studies (total and after IAPP column) were performed on the same preparations, with each analysis being performed at least twice. Disaccharide analysis was performed on two independent HS samples. IAPP affinity columns for non-radiolabeled HS samples were performed on two independent preparations, one of which was used for disaccharide analysis, and both of which were used for fibril formation assays. Chemical desulfation was performed on two independent preparations of HS, both of which were used to verify desulfation by disaccharide analysis, and one of which was used for fibril formation assay.

RESULTS
Binding of ␤-Cell Proteoglycans to Amyloidogenic Human IAPP Requires HS-We have previously shown that ␤-TC3 proteoglycans from medium and cell layer bind human IAPP equivalently (12); therefore, we only analyzed the binding ability of proteoglycans from medium for the first portion of the study. [ 35 S]sulfate-labeled total medium proteoglycans were ␤-TC3 Heparan Sulfate and IAPP Binding applied to an IAPP affinity column and eluted over a 0.25-3 M NaCl gradient (Fig. 1A). The six fractions demarcated in Fig. 1A were collected, and the nature of the proteoglycans in these fractions was then analyzed by SDS-PAGE (Fig. 1B). Proteoglycans in the low molecular weight M3 pool (eluted in fraction 1), which we have previously shown to be comprised of chondroitin/dermatan-free glycosaminoglycans (12), did not bind IAPP, demonstrated by elution at low NaCl. Material from pool M2 contains proteoglycans that bound IAPP with intermediate (fractions 2-4) and high affinity (fractions 5 and 6), whereas proteoglycans from M1 bound IAPP with high affinity (eluted in fractions 5 and 6).
IAPP binding of the proteoglycans in pools M1 and M2 was further investigated by solid phase binding assay. As expected, proteoglycans from both M1 and M2 bound IAPP (Fig. 1C, upper row). Binding of proteoglycans from both M1 and M2 was markedly reduced following treatment with heparinases I, II, and III (Fig. 1C, middle row), suggesting that the binding of IAPP by ␤-TC3 proteoglycans requires HS. Chondroitinase ABC treatment was not associated with a reduction in IAPP binding (Fig. 1C, lower row). If anything, chondroitinase treatment resulted in an increase in IAPP binding, suggesting both that chondroitin/dermatan sulfate are not important for the binding of IAPP and may even impede the binding of IAPP by ␤-TC3 HSPGs. Thus, HS from ␤-TC3 proteoglycans in both pools M1 and M2 appeared to mediate IAPP binding.
Composition of HS from ␤-TC3 and NMuMG Cells-For compositional analysis of HS, we used total HS pooled from both medium and cell layer. As described above, we have previously shown that glycosaminoglycans from ␤-TC3 medium versus cell layer are very similar with respect to molecular weight, heparinase/chondroitinase sensitivity, and IAPP binding capacity (12). Furthermore, ␤-TC3 HS from both M1 and M2 fractions were capable of binding human IAPP (Fig. 1). Finally, HS from fractions M1 and M2 had very similar composition based on sensitivity to nitrous acid at pH 1.5, and to heparinase I or heparinase III (data not shown). HS from NMuMG cells served as a control for all subsequent analyses.
First, the proportion of N-sulfated disaccharides was assessed by cleavage with nitrous acid at pH 1.5. Surprisingly, ␤-TC3 cell HS was relatively insensitive to nitrous acid treatment as demonstrated by the persistence of larger molecular weight [ 3 H]glucosamine-labeled oligosaccharides for ␤-TC3 ( Fig. 2A, solid circles) compared with NMuMG HS (Fig. 2B,  solid circles). As expected, almost all [ 35 S]sulfate-labeled material (open circles) for both ␤-TC3 and NMuMG cells eluted at the total column permeation volume (V t ) following nitrous acid treatment (pH 1.5), consistent with liberation of N-sulfates  from glucosamine residues as free sulfate (Fig. 2, A and B). Of note, however, a small peak (ϳ10% of total counts) of sulfated material remained at the V 0 in ␤-TC3 cell HS following nitrous acid treatment, suggesting that ␤-TC3 HS contained a minor component of O-sulfated residues in the absence of N-sulfation in the same or neighboring disaccharide. Second, to determine the presence of highly sulfated disaccharides, ␤-TC3 and NMuMG HS were digested with heparinase I. ␤-TC3 HS was relatively insensitive to heparinase I treatment (Fig. 3A), with only 29% of labeled HS showing heparinase I sensitivity, compared with NMuMG HS where virtually all HS was sensitive to heparinase I (Fig. 3B). In addition, this analysis using Sepharose CL-6B showed that HS from NMuMG cells was larger than that from ␤-TC3 cells. Third, to determine the proportion of poorly sulfated disaccharides, ␤-TC3 and NMuMG HS were digested with heparinase III. In this case, ␤-TC3 HS was more sensitive to heparinase III digestion (Fig. 4A) than NMuMG HS (Fig. 4B), with the majority of [ 3 H]glucosamine-labeled material eluting at or near the column V t , suggesting digestion to disaccharides. In contrast, heparinase III digestion of NMuMG HS resulted in the presence of larger oligosaccharides, compatible with a more even distribution of poorly sulfated oligosaccharides separated by regions of highly sulfated (heparinase III insensitive) disaccharides. Together, these data suggest that ␤-TC3 HS contains far fewer regions of highly sulfated disaccharides compared with NMuMG HS.
The findings for the radiolabeled enzyme digestion composition studies were confirmed by disaccharide analysis ( Table  1). The proportion of N-acetylated disaccharides was higher in ␤-TC3 HS than in NMuMG HS. Conversely, the proportion of N-sulfated disaccharides was ϳ60% lower in ␤-TC3 than NMuMG HS. Other sulfated disaccharides were more frequent in NMuMG HS, especially the trisulfated UA-2S-(1,4)-Gl-cNS-6S that was three times more abundant in NMuMG HS than in ␤-TC3 HS.
Composition of IAPP-bound and Non-bound HS-Given our unexpected finding that ␤-TC3 HS contains relatively few regions of highly sulfated disaccharides, we next sought to determine whether there was a difference in the proportion of HS from ␤-TC3 and NMuMG cells that was capable of binding IAPP and whether the sulfation of each bound and non-bound fraction differed. Surprisingly, the proportion of heparan sulfate glycosaminoglycans that bound human IAPP was similar between ␤-TC3 cells and NMuMG cells (63 Ϯ 15% and 68 Ϯ 19%, respectively, data are mean Ϯ S.D.). The sulfation patterns of IAPP-bound and non-bound HS from each cell type was assessed by heparinase I or III digestion. IAPP bound HS from ␤-TC3 cells was relatively insensitive to heparinase I ( Fig. 5A;

Double-labeled (closed symbols, [ 3 H]glucosamine; open symbols, [ 35 S]sulfate) HS from ␤-TC3 cells (A) and NMuMG cells (B) was subjected to
Sepharose CL-6B size exclusion chromatography prior to (circles) or following heparinase I digestion (triangles), which cleaves highly sulfated regions of heparan sulfate. Intact ␤-TC3 heparan sulfate eluted at K av ϭ 0.6, whereas two HS peaks of approximately equal abundance were observed after heparinase I treatment: at K av ϭ 0.58 and 0.8. Thus, ␤-TC3 HS was only partially sensitive to heparinase I digestion, demonstrating poor overall sulfation. Intact NMuMG HS eluted at K av ϭ 0.45, whereas following heparinase I digestion, a larger peak was present at K av ϭ 0.7 and a smaller peak at K av ϭ 0.9. Thus, NMuMG HS was more sensitive to heparinase I digestion, demonstrating higher overall sulfation in this HS preparation.

␤-TC3 Heparan Sulfate and IAPP Binding
ϳ20% of [ 35 S]sulfate-labeled material eluted in included volume) and sensitive to heparinase III (Fig. 5B), similar to that seen in the whole preparation (Figs. 3A and 4A). Non-bound material from ␤-TC3 cells had similar sensitivity to heparinase I ( Fig. 5C; 25% of material eluted in the included volume) and heparinase III (Fig. 5D) compared with IAPP-bound material. To rule out the possibility that the non-bound fraction had not bound IAPP on the affinity column due to saturation of binding sites, non-bound ␤-TC3 HS was collected and reapplied to the IAPP affinity column. Only 9.0 Ϯ 0.8% of previously non-bound material bound to the IAPP affinity column (compared with 63% of the total HS sample). Thus, the non-bound material is truly incapable of binding IAPP, despite the fact that its composition does not appear to differ from that of the human IAPPbound ␤-TC3 HS.
In contrast to the findings for ␤-TC3 cell HS, IAPP-bound HS from NMuMG cells was extremely sensitive to heparinase I digestion (Fig. 6A) with heparinase III digestion yielding mainly large fragments eluting in or near the Bio-gel P10 V 0 (Fig. 6B). In marked contrast, non-bound heparan sulfate from NMuMG cells had a very different enzyme sensitivity profile, with heparinase I sensitive material eluting in or near the Bio-gel P10 V 0 (Fig. 6C), but showing greater sensitivity to heparinase III digestion (Fig. 6D). This demonstrates that regions of highly sulfated HS are important in determining the capability of NMuMG HS to bind IAPP. Furthermore, it raises the interesting observation that NMuMG-derived HS is heterogeneous and that two populations of HS that have different arrangement of highly sul-fated disaccharides can be separated from the same pool on the basis of IAPP binding ability.
Disaccharide analysis confirmed that the composition of IAPP-bound and non-bound HS from ␤-TC3 cells was similar ( Table 2), consistent with the size exclusion profiles shown in Fig. 5. Disaccharide composition of NMuMG HS also did not differ between IAPP-bound and non-bound material. This finding was unexpected, given the marked difference in enzyme sensitivity between the bound and non-bound fractions. However, these data are consistent with differences in the sequential arrangement of the variously sulfated disaccharides between the bound and non-bound fractions rather than a difference in disaccharide composition per se. These data suggest that regions of highly sulfated disaccharides occur more frequently in the IAPP-bound NMuMG HS, generating regions of heparinase I-sensitive material, whereas in the non-bound samples, these highly sulfated disaccharides may be distributed more evenly along the HS chain. HS size and the hypothesized arrangement of sulfated and non-sulfated regions for each sample is illustrated in Fig. 7.
Ability of HS Preparations from ␤-TC3 and NMuMG Cells to Stimulate IAPP Fibril Formation-To determine whether HS from ␤-TC3 or NMuMG cells that bound or did not bind IAPP had a differential effect to stimulate fibril formation, we performed thioflavin T assays. As demonstrated in Fig. 8, IAPP alone resulted in fibril formation as expected, and all HS samples had some effect to increase maximal fibril formation, as demonstrated by increased thioflavin T fluorescence. HS from ␤-TC3 cells had a moderate effect to stimulate IAPP fibril formation, and this effect was similar in HS populations regardless of whether or not they bound IAPP in the affinity column experiments. In contrast, the highly sulfated population of NMuMG HS that bound IAPP in the affinity column experiment was more effective at stimulating IAPP fibril formation compared with both the more poorly sulfated non-bound NMuMG HS and to either IAPP-bound or non-bound ␤-TC3 HS.
Finally, the requirement of HS sulfation for stimulation of fibril formation was tested. Chemical desulfation resulted in at

. Double-labeled (closed circles, [ 3 H]glucosamine; open circles, [ 35 S]sulfate) HS from ␤-TC3 cells (A) and NMuMG cells (B) was subjected to
heparinase III digestion and reaction products separated by Bio-gel P-10 size exclusion chromatography. ␤-TC3 HS (A) showed two peaks following heparinase III digestion, with the major peak eluting in the V t of the column, whereas the minor peak was highly sulfated and eluted at K av ϭ 0.3. NMuMG HS (B) had a different pattern of sensitivity to heparinase III treatment, with a larger distribution of peaks throughout the resolved molecular weight range of the Bio-gel P-10 column, suggesting more evenly distributed regions of non-sulfated disaccharides throughout the length of the HS chain.  least an 80% decrease in abundance of sulfated disaccharides in ␤-TC3 and NMuMG HS, respectively. Intact HS from both ␤-TC3 and NMuMG cells markedly accelerated fibril formation (Fig. 9), resulting in a decrease in lag time from 72 h with IAPP alone to essentially zero. (Fibrils were already detectable at the first measurement.) Intact HS from both cell types also increased maximal fibril formation, with NMuMG HS having a greater effect to do so. In contrast, desulfated HS from ␤-TC3  ) and B (heparinase III digested), whereas non-bound HS is shown in C (heparinase I digested) and D (heparinase III digested). ␤-TC3 HS showed similar composition regardless of IAPP binding ability, with both bound and non-bound material being relatively insensitive to heparinase I treatment but sensitive to heparinase III treatment. sulfate) NMuMG HS following application to and elution from an IAPP affinity column, following heparinase I or heparinase III digestion. Reaction products were separated by Bio-gel P-10 size exclusion chromatography. IAPP-bound HS is shown in A (heparinase I digested) and B (heparinase III digested), whereas non-bound HS is shown in C (heparinase I digested) and D (heparinase III digested). NMuMG HS that bound IAPP has a markedly different composition from non-bound material. IAPP-bound HS from NMuMG cells was extremely sensitive to heparinase I digestion, but resistant to heparinase III digestion, suggesting the frequent presence of regions of highly sulfated disaccharides. In contrast, non-bound HS from NMuMG cells was relatively resistant to heparinase I digestion but was sensitive to heparinase III digestion, suggesting that this population of HS contained fewer regions of highly sulfated disaccharides.

␤-TC3 Heparan Sulfate and IAPP Binding
and NMuMG cells did not enhance maximal fibril formation over that seen with IAPP alone. However, desulfated HS from both cell types still accelerated fibril formation, albeit less effectively than intact HS, with lag times at ϳ30 h for IAPP ϩ desulfated ␤-TC3 HS and Ͻ20 h for IAPP ϩ desulfated NMuMG HS.

DISCUSSION
We have shown that ␤-TC3 cells synthesize HS with few regions of highly sulfated disaccharides, a pattern that differs from the more conventional arrangement of sulfated domains seen in the control NMuMG cells. Fig. 7 summarizes our data, presenting a likely arrangement of sulfated versus non-sulfated disaccharides in the different HS samples analyzed in the present study.
It is well known that HS sulfation can vary markedly among different cell types (17). Our study reports, for the first time, the sulfation of HS in ␤-cells, and our data suggest that HS from ␤-TC3 cells have a low degree of overall sulfation. Whether expression of different HS modification enzymes or sulfatases in ␤-TC3 versus NMuMG cells can explain the differences in sulfation is currently unknown. One possible mechanism by which the observed compositional differences could have occurred is altered availability of the sulfate donor 3-phosphoadenosine 5-phosphosulfate. In vitro studies showed that the presence or absence of phosphoadenosine 5-phosphosulfate resulted in marked differences in modification of a bacterial HS-like substrate (28). Specifically, in the absence of phosphoadenosine 5-phosphosulfate, N-deacetylase/N-sulfotransferase activity was uncoupled, leading to N-deacetylation at random intervals, even if phosphoadenosine 5-phosphosulfate was reintroduced. This contrasted with the coupled N-deacetylation/Nsulfotransferase activity in a sequential manner creating regions of consecutively sulfated disaccharides (28). Of note, these two different patterns of modification are similar to those we observed in the present study in ␤-TC3 and NMuMG HS samples, respectively.
Despite the low overall sulfation, HS from ␤-TC3 cells was capable of binding amyloidogenic human IAPP to the same extent as the much more highly sulfated HS synthesized by the control NMuMG cells. Interestingly, the composition of ␤-TC3 HS was very similar regardless of its ability to bind IAPP. Because this was an unexpected finding, we confirmed that the non-bound fraction was really incapable of binding IAPP by reapplying it to the IAPP column and noting that only 10% of the material bound. These data raise the possibility that there is   Open regions are non-or poorly sulfated; black bars are sulfated regions. Both the ratio of sulfated to non-sulfated disaccharides, and their arrangement is similar for human IAPP-bound or non-bound HS from ␤-TC3 cells. It is likely, however, that a small difference exists between bound and non-bound material with the sequence or clustering of sulfated disaccharides creating an additional highly sulfated region (demonstrated by third black bar) in IAPP-bound ␤-TC3 HS. ␤-TC3 HS is similar to non-bound HS from NMuMG cells, except for the difference in molecular weight. The overall ratio of sulfated to non-sulfated disaccharides in NMuMG HS is similar between IAPP-bound and non-bound material, but the arrangement is different, such that larger regions of highly sulfated disaccharides are present in the IAPP-bound material. Fibril formation from IAPP was stimulated by all HS samples, demonstrated by increased thioflavin T fluorescence above that seen with IAPP alone. Stimulation of IAPP fibril formation by ␤-TC3 HS was similar regardless of its IAPP-binding ability. IAPP fibril formation was stimulated to the greatest extent by NMuMG HS that bound IAPP in the affinity chromatography experiment. NMuMG HS that did not bind IAPP-stimulated IAPP fibril formation to a lesser extent than NMuMG HS that bound IAPP and was similar to both ␤-TC3 HS samples. Data are mean Ϯ S.D. for duplicates from one representative experiment.
␤-TC3 Heparan Sulfate and IAPP Binding OCTOBER 26, 2012 • VOLUME 287 • NUMBER 44 a relatively short oligosaccharide binding sequence required for HS-IAPP interactions (as depicted in Fig. 7), such as has been classically described for HS ligands such as antithrombin (29) or basic fibroblast growth factor (30), which is not discernable using the enzyme digestion or disaccharide analysis techniques employed here. Whatever the arrangement of oligosaccharide structure in ␤-TC3 HS required to bind IAPP, it is not important for stimulation of fibril formation, as both IAPP bound and non-bound material stimulated fibril formation equivalently. This suggests that there is a divergence in the ability of HS to bind to IAPP, which may represent a physiological interaction unrelated to its aggregation, and its ability to stimulate fibril formation, which represents a pathological interaction. For the control NMuMG cells, there was a marked difference in regions of highly sulfated disaccharides in HS fractions that were and were not able to bind IAPP. However, the disaccharide analysis revealed that the overall abundance of acetylated and mono-, di-, and tri-sulfated disaccharides was equivalent between the IAPP-bound and non-bound fractions. This suggests that two distinct populations of HS, with similar disaccharide composition but that differ in the arrangement of sulfated domains, are synthesized by this cell type. Because we used total HS from NMuMG cells for these analyses, it is possible that these different HS populations are derived from different HSPGs (e.g. perlecan or a member of the syndecan family) and/or are medium or cell layer derived, respectively. Although in our ␤-TC3 cells, HS size and composition does not differ regardless of the core protein to which the HS is attached (12), 3 we have previously reported differences in the sulfation of chondroitin/dermatan glycosaminoglycans produced by monkey aortic smooth muscle cells (31,32).
Although the overall sulfation of HS was not important in determining the proportion of HS that bound IAPP by affinity chromatography (ϳ65% for both cell types), our data raised the question of whether distribution of highly sulfated disaccharides into regions that are susceptible to heparinase I digestion resulted in a greater effect to enhance IAPP fibril formation. Despite the low overall sulfation of ␤-TC3 HS, it was capable of increasing IAPP fibril formation, and this effect was similar regardless of whether or not ␤-TC3 HS bound IAPP during affinity chromatography experiments. HS from NMuMG cells that did not bind IAPP was also equally effective in stimulating fibril formation compared with both ␤-TC3 samples. However, highly sulfated HS (IAPP-bound material from NMuMG cells) had the greatest effect to stimulate maximal IAPP fibril formation. We directly assessed the importance of HS sulfation to the stimulation of IAPP fibril formation for each cell type by comparing the effect of untreated versus chemically desulfated HS in a fibril formation kinetics assay. Desulfation of HS from both cell types ablated their ability to enhance maximal fibril formation relative to IAPP alone, suggesting that the degree of sulfation is critical in determining the extent to which HS can enhance total fibril formation. This observation is consistent with the previous report by Castillo et al. (11) that sulfation of HS determines fibril formation from IAPP and that perlecan or perlecan-derived HS from Engelbreth-Holm-Swarm sarcoma cells were effective at stimulating IAPP fibril formation, an effect diminished upon chemical desulfation of HS. However, our data show that desulfated HS was still able to accelerate IAPP fibril formation, although to a lesser degree relative to fully sulfated HS, with desulfated NMuMG HS having a greater effect to accelerate fibril formation than desulfated ␤-TC3 HS. An additional explanation for the more effective acceleration in fibril formation with NMuMG HS even after desulfation may be its larger size relative to ␤-TC3 HS. Thus, while confirming the previously published requirement for the presence of highly sulfated HS to markedly stimulate IAPP fibril formation, we have made the novel and interesting observation that highly sulfated HS is not required for IAPP binding or acceleration of fibril formation per se.
The low overall HS sulfation in ␤-TC3 cells could be a result of the use of this isolated cell line, which is derived from a transgenic mouse with ␤-cell expression of SV40 and is therefore transformed. However, our previous studies suggest that the degree of sulfate incorporation into HS in primary mouse islets is similar to that of ␤-TC3 cells (21). The analysis of HS composition in primary islets is complicated, however, by the multiple cell types present in islets. Although insulin and IAPPproducing ␤ cells are the predominant cell type in islets, the islet has a complex cellular composition, making ␤-cell lines such as ␤-TC3 a more suitable model for analyses of ␤-cell HS such as those performed here. The effect of ␤-TC3 HS to increase maximal IAPP fibril formation is relatively small (ϳ1.5-fold). However, it is important to note that increases in amyloid formation of similar magnitude (1.5-fold or less) in cultured islets from transgenic mice expressing amyloidogenic human IAPP (a model of islet amyloid deposition) result in significant increases in ␤-cell apoptosis (33,34). Furthermore, because islet amyloid deposition is a cumulative process, with the extent of fibril formation being associated with increased ␤-cell loss and/or cell death in islet culture models (35), animal models (36,37), and in humans with type 2 diabetes (4), even a small acceleration and/or increase in maximal IAPP fibril formation will likely have a dramatic impact on ␤-cell function and viability over the long term.
In summary, we have determined that ␤-cells synthesize poorly sulfated HS that is capable of both binding human IAPP and stimulating fibril formation. Although the stimulation of fibril formation occurs to a lesser degree than the more highly sulfated HS synthesized by NMuMG cells, this effect does not rely solely on sulfation. Our data raise several possibilities; that cell types within the pancreatic islet other than the ␤-cell may synthesize more highly sulfated HS that could play a role in stimulation of islet amyloid formation in type 2 diabetes or that sulfation of ␤-cell HS could be altered under conditions of islet amyloid formation as has been described for hepatic and splenic amyloidosis (38). Finally, development of compounds designed to interrupt HS-IAPP interactions and the resultant fibril formation must take into account that this interaction may not always be determined by the degree of HS sulfation. Thus, these data provide new information that may aid approaches to reduce HS-IAPP interactions and ultimately prevent islet amyloid formation and its toxic effects in type 2 diabetes.