Heparan Sulfate Proteoglycans Are Important for Islet Amyloid Formation and Islet Amyloid Polypeptide-induced Apoptosis*

Background: Islet amyloid causes β cell death in T2D. Results: Overexpression of heparanase reduced islet amyloid formation in cultured islets, and cells lacking surface associated HS were protected against IAPP-mediated toxicity. Conclusion: IAPP fibrillation requires HSPG interaction for induction of apoptosis. Significance: Inhibition of the HS and hIAPP interaction poses a potential intervention target to prevent β cell death in diabetes. Deposition of β cell toxic islet amyloid is a cardinal finding in type 2 diabetes. In addition to the main amyloid component islet amyloid polypeptide (IAPP), heparan sulfate proteoglycan is constantly present in the amyloid deposit. Heparan sulfate (HS) side chains bind to IAPP, inducing conformational changes of the IAPP structure and an acceleration of fibril formation. We generated a double-transgenic mouse strain (hpa-hIAPP) that overexpresses human heparanase and human IAPP but is deficient of endogenous mouse IAPP. Culture of hpa-hIAPP islets in 20 mm glucose resulted in less amyloid formation compared with the amyloid load developed in cultured islets isolated from littermates expressing human IAPP only. A similar reduction of amyloid was achieved when human islets were cultured in the presence of heparin fragments. Furthermore, we used CHO cells and the mutant CHO pgsD-677 cell line (deficient in HS synthesis) to explore the effect of cellular HS on IAPP-induced cytotoxicity. Seeding of IAPP aggregation on CHO cells resulted in caspase-3 activation and apoptosis that could be prevented by inhibition of caspase-8. No IAPP-induced apoptosis was seen in HS-deficient CHO pgsD-677 cells. These results suggest that β cell death caused by extracellular IAPP requires membrane-bound HS. The interaction between HS and IAPP or the subsequent effects represent a possible therapeutic target whose blockage can lead to a prolonged survival of β cells.

peptide (IAPP), 3 the main amyloid component, forms intraand extracellular aggregates, with an implication for the progressive ␤ cell loss seen in these conditions, as reviewed in Ref.
1. The precise mechanism for amyloid development still needs to be clarified, but accumulating data suggest a common mechanism for amyloid toxicity (2) and that smaller aggregates (oligomers) formed during the earlier stage of amyloidogenesis constitute the principal toxic entity (2,3). Mirzabekov et al. (4) have shown early that human IAPP oligomers can form ion-leaking pores, and fibril formation is accelerated dramatically along anionic lipid membranes (5). A cholesterol-dependent internalization of IAPP oligomers into ␤ cells results in neutralization of cytotoxicity (6). Also, monomeric human IAPP can lead to increased fluidity and destabilization of the plasma membrane (7). Although five different hormone-producing cell types can be identified in the islets of Langerhans, extracellular IAPP deposits only affect ␤ cells without damaging other cell types (8). This suggests a cellular property unique to ␤ cells.
Several amyloid diseases exist, and they are classified on the basis of the specific protein that makes up the amyloid fibril. So far, more than 28 proteins have been identified to be able to form local or systemic amyloidosis in human (9). Beside the amyloid-specific protein, other components, such as serum amyloid P and proteoglycans, are always present in amyloid deposits, where both glycosaminoglycans (GAGs) (10,11) and core proteins (12) have been identified. Heparan sulfate (HS) is found on cell membrane-associated syndecan and glypican and on perlecan and agrin present in the extracellular matrix (13), and HS dominates as the most frequently encountered GAG in amyloid deposits. The function of HS in amyloidogenesis is not clear, but accumulated information points to an important role during initiation of amyloid formation. Human IAPP, but not the non-amyloid forming rat IAPP (rIAPP) binds to perlecan isolated from Engelbreth-Holm-Swarm tumors (14). Also, isolated ␤ cell-associated heparan sulfate proteoglycan binds human IAPP, and no interaction occurs with rIAPP (15). A specific binding site for HS has been identified within the N-terminal processing site of human proIAPP (16), and binding of HS to monomeric proIAPP1-30 stimulates amyloid formation from this otherwise non-amyloid-forming peptide (17). Although binding of heparan sulfate proteoglycan to IAPP is mainly found with the monomeric form of IAPP, Watson et al. (18) showed that binding of heparin to IAPP or amyloid ␤ depends on aggregation status and that binding requires mature fibrils. Also, chondroitin sulfate and keratan sulfate enhanced IAPP fibrillation (14), but with a significantly lower efficiency in comparison with HS.
Heparanase is a mammalian endoglycosidase that specifically cleaves HS chains (19), leading to reduced length of cell surface-bound and extracellular matrix-associated HS. Our earlier study showed that transgenic mice overexpressing human heparanase attenuated inflammatory induced AA amyloidosis (20). In the mouse, an organ-specific difference in human heparanase overexpression coincided with development of amyloid. Livers and kidneys with high levels of heparanase overexpression showed little or no amyloid depositions, whereas spleens without heparanase expression displayed extensive deposits.
In this study, we aimed to investigate the effect of heparanase overexpression on IAPP aggregation and islet amyloid formation. A double-transgenic mouse overexpressing both human heparanase and human IAPP (hpa-hIAPP) was generated. Expression of heparanase did not alter glucose-stimulated insulin or IAPP release but led to a significant reduction of IAPP amyloid in islets cultured in 20 mM glucose. Culture of human islets in the presence of heparin fragments reduced islet amyloid, and this effect was influenced by fragment size.

Experimental Procedures
Animals-Double transgenic (tg) mice overexpressing human heparanase and hpa-hIAPP were generated by crossing human heparanase C57BL (21) with hIAPP FVB/N mice (22). Littermates expressing only hIAPP without concomitant expression of human heparanase were used as controls (hIAPP). Both hpa-hIAPP and hIAPP mice lack the gene for endogenous mouse IAPP shown previously to interfere in IAPP fibril formation (22). Animals were maintained at the animal facility at the Biomedical Centre, Uppsala University, and experiments were approved by the regional Animal Ethics Committee in Uppsala, Sweden.
Islets-Mice (9 -13 weeks old) were sacrificed by cervical dislocation. The pancreas was excised, and islets were isolated by collagenase digestion (Clostridium histolyticum, Sigma-Aldrich, St. Louis, MO) as described previously (23). Mouse islets were cultured free-floating in RPMI 1640 culture medium containing 11 mM glucose (Sigma-Aldrich) supplemented with 10% (v/v) FCS, 100 IU/ml penicillin, 100 g/ml streptomycin, and 45 M ␤-mercaptoethanol at 37°C in 5% CO 2 and allowed to recover for at least 48 h post-isolation before being used in any experiment. Isolated human islets from brain-dead donors with beating hearts were provided by the Nordic Network for Clinical Islet Transplantation, Uppsala University. Human islets were cultured in CMRL 1066 medium containing 5.5 mM glucose supplemented with 10% (v/v) FCS, 100 IU/ml penicillin, 100 g/ml streptomycin, and 2 mM L-glutamine.
Islet Culture-Islets were cultured for 19 days in culture media supplemented with glucose to yield a final concentration of 20 mM (high-glucose (HG)) to stimulate IAPP secretion and subsequent amyloid deposition. To study the effect of heparin chain length on amyloid load, mouse islets were cultured in HG medium with 4-, 8-, 12-, or 18-mers or full-length heparin corresponding to ϳ50-mers (150 nM if nothing else stated). Isolated mouse islets were cultured in 11 mM glucose, and human islets were cultured in 5.5 mM glucose (normal glucose) for 19 days to determine amyloid development under unstimulated conditions. The medium was changed every second day. To confirm the diffusion of heparin into islets, they were cultured with 1 M biotinylated full-length heparin for 3 h, followed by fixation in 4% paraformaldehyde and overnight incubation with Alexa Fluor 488-labeled streptavidin (Molecular Probes) diluted 1:250 in TBS at 4°C.
Quantification of Islet Amyloid Load and Apoptosis-Islets were fixed in 4% paraformaldehyde and stained for amyloid with thioflavin S (0.125%, Sigma-Aldrich). Optical sections of islets were acquired with an EZ-C1 digital camera connected to a Nikon eclipse E600 microscope (Nikon, Kawasaki, Japan) with a Nikon C1 confocal unit using an argon 488-nm laser (Nikon). The volume of thioflavin S-stained regions and islet volume were determined with image analysis software (Imaris 7.6, Bitplane). Islets were restained with Mayer's hematoxylin, and the number of apoptotic nuclei per islet area was determined (ImageJ software, http://imagej.nih.gov/ij/). All quantifications were made in a blinded manner.
Islet Hormone Release-Analysis of hormone release was performed on islets 3 days after isolation. Three samples of 10 islets per mouse were incubated in Krebs-Ringer bicarbonate buffer (114 mM NaCl, 4.7 mM KCl, 1.2 mM MgSO 4 , 1.2 mM KH 2 PO 4 , 25 mM NaHCO 3 , 4.3 mM NaOH, 2.5 mM CaCl 2 ) supplemented with 10 mM HEPES and 2 mg/ml bovine serum albumin (BSA) (KRBH-BSA) containing 1.7 mM glucose for the first hour followed by an incubation in 16.7 mM glucose for a second hour. Next, the islet triplicates were pooled, homogenized, and mixed with acid ethanol (180 mM HCl in 95% ethanol) to extract total islet insulin and IAPP content. Insulin was determined with rat insulin ELISA (Mercodia, Uppsala, Sweden) and IAPP with human amylin ELISA (Merck Millipore, Darmstadt, Germany).
Immunoelectron Microscopy-Islets were fixed in 2% paraformaldehyde with 0.25% glutaraldehyde and processed for immunoelectron microscopy. Ultrathin sections were immunolabeled as described previously (25), with rabbit antiserum A110 raised against IAPP1-37 and contrasted with 5% uranyl acetate in water and Reynold's lead solution. For morphological analysis of aggregates formed in vitro, samples were diluted 1:30 in double-distilled H 2 O, adhered on formvar-coated copper grids and negatively contrasted with 2.5% uranyl acetate in 50% ethanol. Samples were studied at 75 kV in a Hitachi H-7100 transmission electron microscope (Hitachi, Tokyo, Japan), and images were obtained with Gatan 832 Orius SC1000 (Gatan).
Isolation and Characterization of HS from Pancreas-One heparanase overexpressing mouse and one C57BL mouse were injected intraperitoneally with 100 mCi of Na 35 SO 4 (catalog no. NEX041H, PerkinElmer LifeSciences). After 1 h, the mice were sacrificed, and the whole pancreata were dissected. Following homogenization, the lysates were incubated with 0.5 mg of protease at 55°C overnight to digest proteins, followed by incubation with benzonase (12.5 units) to degrade DNA. After 10-min centrifugation at 8000 ϫ g, the supernatants were collected and applied to a DEAE-Sephacel to purify 35 S-labled glycosaminoglycans (heparan sulfate and chondroitin sulfate). The yielded products were treated with chondroitinase ABC (#100330; Serkagaku, Tokyo, Japan) and re-applied to DEAE-Sephacel column to remove degraded chondroitin sulfate. The purified 35 S-labeled heparan sulfate was analyzed by Superose-12 column (GE Healthcare) connected to a HPLC system. The purity of 35 S-labeled heparan sulfate was confirmed by complete heparinase/heparitinase digestion.
Heparin Labeling-The 3 H-labeled heparin was prepared by N-deacetylation followed by re-N-acetylation with N-[ 3 H]acetic anhydride (26). The heparin fragments were generated by partial deaminative cleavage of the 3 H-labeled heparin, followed by reduction with NaBH 4 . The fragmented heparin samples were separated on a Bio-gel P-10 column (Bio-Rad) as described previously (27).
Thioflavin T Assay-A stock solution of hIAPP1-37 dissolved in DMSO was diluted to a final concentration of 1.4 M (0.5% DMSO) in PBS (pH 7.4) with 10 M thioflavin T (ThT, Sigma). Fibrillization of hIAPP was measured by monitoring emission at 480 nm with 440-nm excitation on a FLUOstar Omega microplate reader (BMC Labtech, Stockholm, Sweden) at room temperature. All reactions were carried out in Sigmacote-treated (Sigma-Aldrich) black 96-well plates (Nunc). Fibrillation studies were performed in the absence or presence of equimolar full-length heparin or 12-mer heparin fragments. Also, an aliquot of the samples used for the FRET cell assay was incubated with 10 M ThT in PBS, and aggregation was monitored as described above but at 37°C.
FRET Assay-Cells were seeded into black 96-well optical bottom plates (Nunc) and cultured in the presence of 4% DMSO for 48 h to increase vector expression because longterm culture is known to inactivate the CMV promotor (30). The assay was performed in Krebs-Ringer solution (120 mM NaCl, 4.7 mM KCl, 2.5 mM CaCl 2 , 1.2 mM MgSO 4 , and 0.5 mM KH 2 PO 4 (pH 7.4)) supplemented with 20 mM HEPES and 2 mM glucose (KRHG), shown previously to give low autofluo-rescence at 480 nm and 540 nm (31). Cell culture medium was replaced with KRHG, and the experiment was initiated by addition of a mixture of 50 M hIAPP1-37 and sonicated preformed fibrils (seeds) corresponding to 125 nM monomeric IAPP. Monomeric IAPP and seeds were diluted in KRHG, and final DMSO concentration was 1.5%. IAPP fibril seeds were added to ensure the consistency of IAPP fibrillation kinetics between experiments. To analyze the effect of heparin and 12-mer heparin fragments on IAPP-induced cytotoxicity, an equimolar concentration (50 M) was mixed with the seeded IAPP solution just before addition to the cells. For caspase inhibitor analysis, cells were preincubated for 30 min with 200 M caspase-8 inhibitor (Z-IETD-fmk) or caspase-9 inhibitor (Z-LEHD-fmk) (R&D Systems) and present at 100 M during the analysis. As positive control for induction of apoptosis, the cells were incubated with 2 M staurosporine (Sigma-Aldrich) or 4 g/ml mouse Fas Ligand His 6 with 10 g/ml polyhistidine antibody (R&D Systems). Cells incubated with only seeds or polyhistidine antibody served as negative controls. FRET was measured by monitoring emission at 480 and 540 nm with 440-nm excitation once per hour in a FLUOstar Omega microplate reader at 37°C.
Quantitative Real-time PCR-CHO WT -DEVD and pgsD-677-DEVD cells incubated for 48 h in 4% DMSO were lysed in Qiazol lysis reagent (Qiagen), and total RNA was purified using RNeasy MiniElute cleanup columns (Qiagen). 5 g of RNA, quantified using Nanodrop 2000c (Thermo Scientific), was used for first-strand cDNA synthesis with oligo(dT) primer according to the instructions of the manufacturer (catalog no. K1632, Thermo Scientific). Reactions (10 l) containing 10 ng of cellular cDNA, 400 nM primer, and FastStart Universal SYBR Green Master (ROX) (Roche Diagnostics) were used for realtime quantitative PCR on a 7900HT real-time PCR system (Applied Biosystems) at 95°C for 5 min, followed by 50 cycles of 95°C for 30 s and 54°C for 30 s. The following primer sequences were used: Fas receptor (XM 003505961), 5Ј CACG-GAAGGGAAGGAATACA and 3Ј CAGTGTTCGCAGACA-GAAGC; housekeeping gene GAPDH, 5Ј AACTTTGGCATT-GTGGAAGG and 3Ј CACATTGGGGGTAGGAACAC. All primers were designed with Primer3 and produced at TAG (Copenhagen, Denmark). PCR products were confirmed by agarose gel electrophoresis and melting curve analysis.
Statistical Analysis-All statistics were carried out using GraphPad Prism version 6.01 (GraphPad Software, La Jolla, CA). p Ͻ 0.05 was considered statistically significant.

Results
Heparanase Overexpression in hpa-hIAPP Islets-Heparanase degrades cell surface-associated and extracellular heparan sulfate glycosaminoglycan chains, and, to study the effect of heparanase overexpression in IAPP amyloidogenesis, a doubletransgenic mouse expressing both human heparanase and human IAPP (hpa-hIAPP) was generated. Littermates expressing hIAPP were used as controls (hIAPP). Animals with double transgene expression did not develop any apparent phenotypic alterations, and the pancreas showed a normal histological picture, in line with observations from the single-transgenic mice (21,22). Overexpression of heparanase in islets from hpa-hIAPP mice was verified using an anti-heparanase antibody (733) that detected the cleaved 50-kDa form (Fig. 1A). In pancreas sections, most heparanase reactivity was detected throughout the islets, where it colocalized with insulin reactivity, and less heparanase reactivity was found in the exocrine pancreas tissue (Fig. 1B). The lack of reactivity in hIAPP mice indicates low levels of endogenous mouse heparanase in islets.
Heparanase overexpression resulted in reduced length of HS chains in the pancreas, as demonstrated by gel chromatography of HS extracted from whole pancreata (Fig. 1C). Because HS and heparanase expression is pronounced in islets and low in exocrine tissue (33), the shift in HS size observed in whole pancreata from hpa mice should reflect islet-associated HS.

FIGURE 1. Heparanase overexpression and HS fragmentation in hpa-hIAPP islets.
A, extracts of isolated islets from hIAPP and hpa-hIAPP mice were analyzed by Western blot using antiserum 733 reactive against the active 50-kDa subunit of heparanase. A single band corresponding to 50 kDa was detected in hpa-hIAPP islets, whereas no reactivity was detected in hIAPP islets. Bottom blot, insulin reactivity in both islet extracts. Molecular weight is shown in kilodaltons. B, pancreas sections were double-immunostained for heparanase (733, red) and insulin (green). Bottom panels and inset, merged images demonstrating the colocalization of heparanase (733) and insulin reactivity in hpa-hIAPP islet ␤ cells. hIAPP islets were only positive for insulin. Nuclei stained with DAPI are shown in blue. Scale bars ϭ 30 and 3 m (inset). C, gel chromatography analysis of 35 S-labeled HS isolated from the pancreata of a heparanase overexpressing mouse (hpa-tg, red) and a wild-type C57BL mouse (blue). The arrow indicates the expected elution position of heparin (14 kDa). A right shift of the elution profile from the hpa-tg mouse relative to the wild-type mouse reflects a significant reduction in HS chain length.
Heparin Binds to the N-terminal Region of IAPP in a Size-dependent Manner-The interaction between heparin and IAPP peptides was analyzed with a nitrocellulose filter binding assay using 3 H-labeled heparin. Heparin is used as a model for HS and only differs in its higher sulfate content and more uniform sulfation pattern. Heparin bound hIAPP1-37 as well as rat IAPP1-37 and the IAPP related peptide hCGRP1-37, whereas no binding was seen with a peptide spanning the N-terminal cleavage site of hproIAPP (hproIAPP6 -15) or with N-terminally truncated hIAPP (hIAPP8 -37) (Fig. 4A). Rat IAPP differs from the human variant at six positions and has an N-terminal region identical to hIAPP. Human CGRP shares almost 50% overall sequence identity with hIAPP, with the greatest homology found in the N-terminal region, where identity increases up to more than 70% (residues 1-12). No interaction of heparin with the other components of the secretory granule, insulin or   JUNE 12, 2015 • VOLUME 290 • NUMBER 24 JOURNAL OF BIOLOGICAL CHEMISTRY 15125 C-peptide, was observed. The binding between heparin and hIAPP1-37 was affected by the size of heparin, and heparin fragments Յ8-mers displayed little or no binding with hIAPP, whereas the longer chains of 12-, 16-, and 18-mers showed progressively increased binding to hIAPP (Fig. 4B).

HS Proteoglycans Are Important for Islet Amyloid Formation
The 12-mer Heparin Fragment Attenuated Amyloid Load in Human Islets-Next we determined the potential for extracellular heparin and heparin fragments to influence amyloid load in cultured islets. Isolated islets from hIAPP mice (hIAPP FVB/N, the same strain used to generate hpa-hIAPP mice) were cultured in 20 mM glucose in the absence or presence of heparin fragments; 4-, 8-, 12-, or 18-mer heparin fragments; or fulllength heparin. Of the five fragments tested, only 12-mers had a tendency to lower islet amyloid load (Fig. 5A). Therefore, the culture of transgenic mouse islets in 20 mM glucose and 20 mM glucose with a 12-mer heparin fragment was repeated (in total three times). The amyloid load in percent was determined, and, for comparison, the value was first normalized against the HG group for analysis (HG, 100% Ϯ 0% versus 12-mer, 71.63% Ϯ 2.8%; p Ͻ 0.05, ANOVA, Bonferroni correction) (Fig. 5A). The failure of 4-and 8-mer heparin fragments to interfere with IAPP amyloid formation is in accordance with the result from the binding assay (Fig. 4B), and the reason for the lack of effect by the longer heparin chains (Ͼ12-mer) is unclear. It was not due to their inability to diffuse into the islets because biotinylated full-length heparin was detected throughout the islets, including the islet core, already after 3 h of incubation. In a single experiment, hIAPP transgenic mouse islets isolated from three mice were cultured in 20 mM glucose and 20 mM glucose supplemented with 150 nM or 300 nM of 12-mer heparin fragments for 19 days. Irrespective of the 12-mer heparin fragment concentration used for incubation, there was a tendency for lowering of islet amyloid load (HG, 1.85% Ϯ 0.9%; 12-mer, 150 nM, 1.28% Ϯ 0.8%; 12-mer, 300 nM, 1.14% Ϯ 0.9%). To test whether the effect of heparin fragments also applies to human tissues, isolated human islets from three different donors were cultured in 20 mM glucose with or without addition of 12-mer heparin fragments. Culture in the presence of 150 nM 12-mer heparin fragments in the culture media reduced the amyloid load in human islets compared with islets cultured in 20 mM glucose alone (HG, 0.30% Ϯ 0.06% versus 12-mer, 0.10% Ϯ 0.02%; p Ͻ 0.05; ANOVA; Bonferroni correction) (Fig. 5B). A 2-fold increase of the 12-mer heparin fragment concentration to 300 nM resulted in reduction of the amyloid load; however, it did not reach statistical significance (12-mer, 300 nM, 0.16% Ϯ 0.04%).
The Heparin Fragment Promotes in Vitro IAPP Fibril Growth-To determine how heparin chain size affects IAPP fibril formation, aggregation kinetics in the presence of fulllength heparin or 12-mer heparin fragments were analyzed. We used a ThT assay where fibril formation is monitored as an increase in fluorescence intensity at 480 nm with excitation at 440 nm. The lag phase is the period before any rise in fluores-  . Islets were stained with thioflavin S, and the amyloid load was determined (islet amyloid volume/total islet volume). Islets cultured in HG with 12-mers had a reduced amyloid load compared with islets cultured in HG only (*, p Ͻ 0.05 versus HG, ****, p Ͻ 0.0001 versus HG; ANOVA; Bonferroni correction). The numbers in parenthesis define the number of experiments (n ϭ 1, performed on islets pooled from Ն6 mice). B, islets isolated from human donors (n ϭ 3) were cultured for 19 days in 5.5 mM glucose (normal glucose, gray column) or 20 mM glucose (HG, black columns) alone or supplemented with 12-mer heparin fragments (150 or 300 nM). The amyloid load was determined by thioflavin S staining. A significant decrease in amyloid load was seen in islets cultured in HG with 150 nM 12-mer compared with islets cultured in HG only (*, p Ͻ 0.05 versus HG; ANOVA; Bonferroni correction; mean Ϯ S.E.).
cence has occurred and is defined in this study as the period from the incubation start point until fluorescent light intensity at 480 nm reaches Ն 3ϫ baseline value. Monomeric hIAPP alone formed fibrils after a 6.5-h lag phase (Fig. 6A). Addition of an equimolar concentration of heparin dramatically decreased the lag phase to 2.1 h, whereas the addition of equimolar 12-mer heparin fragments had a more modest effect and reduced the lag phase to 4.6 h.
HS-deficient pgsD-677 Cells Are Protected from hIAPP-induced Apoptosis-Because our observations demonstrate that HS affects the deposition of islet amyloid, we continued to investigate the role of HS in IAPP-induced apoptosis. This was investigated using the mutant CHO cell line pgsD-677, defective in HS biosynthesis and, therefore, lacking HS. CHO WT cells were used as a control. Confirmation of HS deficiency in pgsD-677 cells has been carried out elsewhere (34). To monitor caspase-3 activation and subsequent induction of apoptosis in real time, cells were stably transfected with a vector producing EYFP and ECFP, connected via a short linker containing the caspase-3 cleavage sequence DEVD. The short linker allows FRET to occur between the two fluorophores. The DEVD sequence is recognized and cleaved by active caspase-3 during apoptosis, resulting in loss of FRET, measured as a reduction in the 540/480 nm emission ratio at 440-nm excitation.
Incubation of CHO WT -DEVD cells with hIAPP and sonicated preformed IAPP fibrils (seeds) resulted in a progressive loss of FRET over time (p Ͻ 0.05, two-way ANOVA with Bonferroni correction) (Fig. 7A). In contrast, exposure of pgsD-677-DEVD to seeded hIAPP did not reduce FRET at any time point studied (Յ11 h) (Fig. 7B). Staurosporine was used as a positive control and reduced FRET in both CHO WT -DEVD and pgsD-677-DEVD after 4 -5 h (Fig. 7, A and B) but had no effect in CHO WT -KEAF or pgsD-677-KEAF carrying a sequence not recognized by caspase-3 (Fig. 7C). Aggregation kinetics of hIAPP solutions used in the FRET assay were monitored in parallel using a ThT assay. At time points corresponding to the reduction of FRET in CHO WT cells (i.e. in lag phase/early elongation phase), the ThT samples contained a mixture of aggregate species (Fig. 7D). Some of these early species had a typical fibril-like morphology when analyzed with transmission electron microscopy after negative staining, and fibrils increased with time (Fig. 7D). Incubation of CHO WT -DEVD with solution containing seeds only did not result in activation of caspase-3. At the end point of the FRET assay, after 12 h of incubation with seeded hIAPP, large aggregates of hIAPP were found mainly extracellularly and associated with the plasma membrane of both cell types (Fig. 7E). However, more extensive hIAPP reactivity was found associated with CHO WT -DEVD cells compared with pgsD-677-DEVD cells.
Fibril Formation Neutralizes Toxic Species-In the ThT assay, heparin and 12-mer heparin fragments shortened the lag phase for IAPP aggregation (Fig. 6A), and the 12-mer fragment attenuated the amyloid load in cultured islets (Fig. 5). When heparin and 12-mer fragments were added to seeded IAPP solution in the caspase-3 activation assay, it lead to a brief acceleration of apoptosis at 4 and 6 h, but, after 12 h, toxicity was attenuated significantly compared with seeded IAPP alone (p Ͻ 0.05, two-way ANOVA with Bonferroni correction) (Fig. 6B). The early caspase-3 activation at 4 and 6 h is in line with the ThT results and reinforces the assumption that rapid fibril formation neutralizes toxic species.
Human IAPP Induces Apoptosis in CHO WT via a Caspase-8dependent Pathway-The signaling pathway involved in hIAPPinduced apoptosis of CHO WT cells was determined using caspase-8 and caspase-9 inhibitors. Treatment of CHO WT -DEVD cells with caspase-8 inhibitor, but not caspase-9 inhibitor, effectively prevented the observed hIAPP-induced reduction of FRET (p Ͻ 0.05, one-and two-way ANOVA with Bonferroni correction) (Fig. 7, F and G). Caspase inhibitors had no effect on cells incubated with seeds alone, demonstrating that the effect of caspase-8 inhibition was specific to hIAPPinduced apoptosis. On the basis of these results, we analyzed whether a defect in the extrinsic pathway could explain the lack of IAPP-induced apoptosis in pgsD-677-DEVD cells. However, this seems unlikely because no difference in Fas receptor mRNA levels (Fig. 8A) or FasL-induced loss of FRET was  n ϭ 2). The presence of heparin promoted IAPP fibril formation, seen by the dramatically reduced lag phase compared with IAPP alone. 12-mers also promoted fibril formation, although less potently compared with full-length heparin. B, CHO WT cells were stably transfected with a vector encoding protein pairs for FRET (ECFP/EYFP) linked via residues DEVD. Activation of cellular caspase-3 during apoptosis cleaves the DEVD link, measured as loss of FRET signal. Incubation of CHO WT -DEVD cells with 50 M IAPP and sonicated seeds (corresponding to 125 nM monomeric IAPP) led to detectable caspase-3 activation at Ն6 h (green columns), reaching a maximal effect at ϳ18 h. The presence of equimolar full-length heparin (red columns) or 12-mers (blue columns) significantly reduced the activation of cellular caspase-3 induced by seeded IAPP (n ϭ 3; *, p Ͻ 0.05 versus IAPP; two-way ANOVA; Bonferroni correction). The addition of 12-mers alone was used as a negative control (gray columns). Data are presented relative to untreated cells (mean Ϯ S.E.). observed in CHO WT -DEVD and pgsD-677-DEVD cells (Fig. 8,  B and C).

Discussion
The codeposition of heparan sulfate proteoglycan with islet amyloid in patients with type 2 diabetes, A␤-amyloid in Alzheimer disease, and amyloid in all other amyloid-related diseases suggests that this ubiquitously expressed macromolecule plays an active role in amyloidogenesis (35,36). Overexpression of heparanase has been reported to reduce the amyloid load in animal models of other amyloid-related diseases, including AA amyloidosis (20) and Alzheimer disease (37). In this study, we show, for the first time, that overexpression of the HS-degrading enzyme heparanase in pancreatic islets reduces islet amyloid formation.
For the implementation of this work, we generated a double transgene hpa-hIAPP mouse and found active heparanase in islet ␤ cells. HS isolated from whole pancreata demonstrated decreased chain size, consistent with an overproduction of active heparanase, which must originate from islets because FIGURE 7. HS is required for hIAPP-induced caspase-3 activation. CHO WT cells and HS-deficient pgsD-677 cells were stably transfected with a vector encoding protein pairs for FRET (ECFP/EYFP) linked via residues DEVD (CHO WT -DEVD and pgsD-677-DEVD). Activated cellular caspase-3 during apoptosis cleaves the DEVD link, measured as loss of FRET signal, i.e. reduced 540/480 nm ratio. FRET analysis was performed with monomeric IAPP with the addition of sonicated IAPP fibrils (seed) to initiate the fibril propagation process. Incubation with seed only was used as a negative control (neg. ctrl). A, incubation of CHO WT -DEVD cells with 50 M IAPP and sonicated IAPP seeds (corresponding to 125 nM monomeric IAPP) resulted in a progressive loss of FRET over time (green; n ϭ 4; *, p Ͻ 0.05 versus negative control (seeds, gray); two-way ANOVA; Bonferroni correction). Stauro, staurosporine. B, in HS-deficient pgsD-677-DEVD cells, there was no difference in FRET signal between cells incubated with seeded 50 M IAPP (green) and the negative control (gray; n ϭ 4; p Ͼ 0.05 versus negative control (seeds, gray); two-way ANOVA; Bonferroni correction). Incubation with staurosporine was used as a positive control (blue). ns, not significant. C, cells expressing ECFP/EYFP linked via the KEAF residues, which is not recognized by caspase-3, displayed no loss of FRET during exposure to staurosporine (n ϭ 4). D, ThT assay demonstrates aggregation kinetics of the seeded IAPP used in A and B (representative of four individual experiments), and the insets show transmission electron microscopy images of negatively stained samples removed from the ThT assay at the indicated time points. Scale bars ϭ 100 nm. E, immunofluorescence using antiserum A133 specific for hIAPP of cells incubated with seeded IAPP for 12 h. IAPP reactivity (red) was mainly associated with cell membranes and was more abundant on CHO WT -DEVD cells compared with pgsD-677-DEVD cells. Cell cytoplasm with ECFP/EYFP is shown in green, and nuclei were stained with DAPI (blue). Scale bars ϭ 5 m. F, treatment of CHO WT -DEVD cells with caspase-8 inhibitor (purple) prevented IAPP-induced loss of FRET over time, whereas no effect was seen with caspase-9 inhibitor (black) (n ϭ 3; *, p Ͻ 0.05 versus IAPP; two-way ANOVA; Bonferroni correction). G, summary of the effect shown in F of caspase-8 and caspase-9 inhibitors on IAPP-induced apoptosis in CHO WT -DEVD cells at 12 h, including negative controls where IAPP was omitted (gray columns). *, p Ͼ 0.05 versus IAPP; two-way ANOVA; Bonferroni correction. Data are mean Ϯ S.E. very low levels of active heparanase were found in the exocrine pancreas.
Hull et al. (38) demonstrated that blocking GAG chain elongation in islets with an N-acetylglucosamine analogue significantly reduced both islet amyloid prevalence and severity, with an associated increase in ␤ cell area. However, the authors raised a concern regarding impaired islet viability and islet hormone secretion as a consequence of loss of GAGs. HS is ubiquitously expressed in virtually all tissues, including pancreatic islets (33) and ␤ cells (38), and HS has been shown to affect both ␤ cell survival (33) and insulin release (39). However, the decreased islet amyloid deposition in hpa-hIAPP mice was not due to reduced IAPP release because glucose-induced IAPP secretion remained unaltered. The preserved hormone secretion from ␤ cells in hpa-hIAPP mice may be explained by the specific degradation of HS, leaving the other GAGs intact, which may act as substitutes for HS. In addition, small HS fragments have been shown to retain the ability to perform some of the functions of full-length HS.
We hypothesize that the observed reduction in islet amyloid load by overexpression of heparanase results from the digestion of HS present on cell surface-associated syndecans and glypicans or in the extracellular matrix, reducing the availability of HS, which, otherwise, would act as a promoter of amyloid formation. A further effect of heparanase overexpression is the generation of soluble heparin fragments that compete with cellular and extracellular HS for binding of the released IAPP, thereby preventing IAPP from forming aggregates directly adjacent to the cell surface. However, the contribution of other effects of heparanase overexpression, in addition to increased enzymatic digestion of HS, such as increased sulfation degree of HS (40), altered gene expression (41), or increased release of extracellular matrix-bound factors (21), cannot be ruled out.
Several in vitro studies have demonstrated an accelerating effect of HS on fibril formation of amyloid peptides, including hIAPP and hproIAPP (14,17). When binding of heparin to IAPP was analyzed by nitrocellulose filter assay, we detected the same degree of binding to hIAPP1-37 as to rIAPP1-37, a result that differs from previous studies in which binding between proteoglycans and rIAPP was missing (14,15). The amino acid sequences of hIAPP and rIAPP are identical in the N-terminal region spanning residues 1-17, and this, together with the absence of heparin binding to hIAPP 8 -37, suggest that residues 1-7 comprise part of the binding site. We also found that heparin binds to hCGRP, an IAPP-related peptide that shares five of seven N-terminal residues with IAPP.
Amyloid fibrils consist of a ␤ sheet structure where strands are arranged perpendicularly to the fibril axis. There are several models describing IAPP aggregates (42)(43)(44)(45). Commonly, the seven to ten most N-terminal residues are not included in the fibril core. Biological activity of IAPP and CGRP requires a disulfide bond between cysteines at positions 2 and 7, which causes a restriction in mobility in this region. We used oxidized hIAPP, rIAPP, and CGRP in our study, and the observed binding between the N-terminal region and HS may lead to an orientation of bound peptides allowing intermolecular interactions and, in the case of hIAPP, aggregation. Such binding at the N terminus of the peptide may also prevent other interactions of IAPP regions important for ␤ sheet formation.
In addition to HS effects on IAPP amyloid formation, we investigated its role in IAPP-induced cytotoxicity. It is still unknown what drives the conversion of native protein to amyloid fibrils, but it is clear that the process involves the formation of smaller aggregates, oligomers (46,47). Although the species responsible for toxicity is debated (48), ex vivo studies show that oligomers or seeded IAPP amyloid propagation induce apoptosis in various cell lines, whereas the addition of mature fibrils does not (49 -51). Molecular dynamic studies on IAPP aggregation show that hIAPP trimers form a stable core that can seed IAPP fibrillization, whereas rIAPP aggregates never adopt the stability required for fibril propagation (52). It is possible that HS creates a microenvironment on the cell surface that allows extracellular hIAPP seeds to form.
In CHO WT -DEVD cells, IAPP-induced toxicity corresponded to aggregated species early in the elongation phase of the ThT curve, although already with fibrillar morphology. It is likely that IAPP fibril formation is initiated much earlier in the presence of cells compared with what was seen in the ThT assay because cellular components such as GAGs and negatively charged lipids can promote the aggregation of IAPP (5).
Previously, B-TC6 cells transfected with the FRET-DEVD vector were used in a study in which we showed that there was no difference between the ability of hproIAPP and hIAPP to activate caspase-3, whereas incubation with monomeric hIAPP, mature hIAPP fibrils, or rIAPP failed to do so (31). B-TC6-DEVD and CHO WT -DEVD cells with HS on the cell surface are expected to bind both hIAPP and rIAPP, but only hIAPP activates caspase-3. Therefore, it seems unlikely that binding of hIAPP or rIAPP to HS is sufficient for induction of toxicity. Instead, further aggregation or fibril propagation in close association with the cell membrane is required. This is supported by the finding that immunolabeling of cells incubated with hIAPP and IAPP seeds showed an enrichment of IAPP-reactive material in close proximity to the CHO WT -DEVD cell membrane, different from the picture in pgsD-677-DEVD cells. Also, pgsD-677-DEVD cells lacking HS on their surface remain unaffected by exogenously added hIAPP.
In CHO WT -DEVD cells, incubation with aggregating hIAPP resulted in caspase-3 activation after 6 h, and this is in good agreement with the results from the ThT assay, where IAPP aggregates could be demonstrated after 6 h. When CHO WT -DEVD cells were incubated with heparin and IAPP or a 12-mer heparin fragment and IAPP, caspase-3 activation started at 4 and 6 h, respectively, but never reached the same level of cell toxicity as after incubation with IAPP alone. In the ThT assay, both heparin and 12-mer fragments triggered fibrillation. This supports the hypothesis that fibrils are less toxic and that rapid fibrillation of IAPP prevents cytotoxicity. A recent study showed that copper ions interfere with IAPP aggregation and can suppress fibril formation but potentiate the toxicity of formed aggregates (53). The presence of heparin and 12-mer heparin fragments in the extracellular medium also competes with IAPP binding to cell surface HS and redirects the fibrillation of IAPP away from the vicinity of the cell surface, discussed above as an important aspect in the induction of cytotoxicity. Culture of CHO WT cells in the presence of hIAPP and IAPP seeds induced apoptosis via the caspase-8-dependent extrinsic pathway, in agreement with earlier work with ␤ cells demonstrating the extrinsic pathway, involving Fas receptor signaling, as the primary pathway in hIAPP-induced apoptosis (8,54,55).
In a recent paper, Carufel et al. (56) demonstrated that the removal of cell surface GAGs had no effect on IAPP-induced cytotoxicity, measured 24 h after exposure to Ն25 M hIAPP. This is not unexpected because formation of fibrillar species and, therefore, cytotoxicity should occur at such supraphysiological concentrations of hIAPP well before 24 h, independent of the presence or absence of GAGs. In addition to type 2 diabetes, IAPP amyloid develops rapidly in human islet implanted under the kidney capsule of nude mice (57) and in allogenic islets transplanted to the liver as a treatment strategy for improvement of blood glucose regulation in type 1 diabetes (58). In the human situation, islets from more than one donor are required to achieve insulin independence. A vast mass of the implant is lost in connection with the transplantation and is ascribed to the process of instant blood-mediated inflammatory reaction. The addition of heparin has been shown to improve immediate islet survival (59). However, heparin increases the risk of bleeding and is therefore given as a bolus dose for peritransplantation treatment. In an attempt to improve long-term islet survival, functional heparin has been immobilized on the islet surface to create a shield that, after implantation, prevents coagulation and activation of the complement cascade (60). This heparin shield can create foci with high binding for IAPP released continuously from ␤ cells in the entrapped islet. Therefore, it is possible that IAPP aggregation is activated in islets pretreated with heparin or implanted into a heparin-rich environment.
In conclusion, we show, for the first time, that overexpression of the HS-degrading enzyme heparanase reduces islet amyloid formation and that the presence of small heparin fragments (12-mers) decreases the islet amyloid load in cultured human islets. Moreover, we show that cell-associated HS is important for IAPP-induced caspase-3 activation. This study provides proof of HS playing a causative role in islet amyloid formation and IAPP-induced toxicity in ␤ cells. HS may be a suitable target for the development of new approaches to preserve ␤ cell viability and function in type 2 diabetes and after islet transplantation.