Identification of a Heparin Binding Domain in the N-terminal Cleavage Site of Pro-islet Amyloid Polypeptide IMPLICATIONS FOR ISLET AMYLOID FORMATION*

Islet amyloid deposits are a characteristic pathologic lesion of the pancreas in type 2 diabetes and are com-posed primarily of the islet beta cell peptide islet amyloid polypeptide (IAPP or amylin) as well as the basement membrane heparan sulfate proteoglycan perlecan. Impaired processing of the IAPP precursor has been impli-cated in the mechanism of islet amyloid formation. The N-and C-terminal cleavage sites where pro-IAPP is processed by prohormone convertases contain a series of basic amino acid residues that we hypothesized may interact with heparan sulfate proteoglycans. This possibility was tested using affinity chromatography by applying synthetic fragments of pro-IAPP to heparin-agarose and heparan sulfate-Sepharose. An N-terminal human pro-IAPP fragment (residues 1–30) was retained by both hep-arin-agarose and heparan sulfate-Sepharose, eluting at 0.18 M NaCl at pH 7.5. Substitution of alanine residues for two basic residues in the N-terminal cleavage site abol-ished heparin and heparan sulfate binding activity. At pH 5.5, the affinity of the wild-type peptide for heparin/ heparan sulfate was increased, implying a role for histidine residues at positions 6 and 28 of pro-IAPP. A C-terminal pro-IAPP fragment (residues 41–67) had no specific affinity for either heparin or domain prior to exocyto-sis from the beta cell. Interestingly, the heparin binding activity of pro-IAPP we identified was in the N-terminal the C-terminal region of the peptide. Immuno-

Islet amyloid deposits are a characteristic pathologic lesion of the pancreas in type 2 diabetes and are composed primarily of the islet beta cell peptide islet amyloid polypeptide (IAPP or amylin) as well as the basement membrane heparan sulfate proteoglycan perlecan. Impaired processing of the IAPP precursor has been implicated in the mechanism of islet amyloid formation. The Nand C-terminal cleavage sites where pro-IAPP is processed by prohormone convertases contain a series of basic amino acid residues that we hypothesized may interact with heparan sulfate proteoglycans. This possibility was tested using affinity chromatography by applying synthetic fragments of pro-IAPP to heparin-agarose and heparan sulfate-Sepharose. An N-terminal human pro-IAPP fragment (residues 1-30) was retained by both heparin-agarose and heparan sulfate-Sepharose, eluting at 0.18 M NaCl at pH 7.5. Substitution of alanine residues for two basic residues in the N-terminal cleavage site abolished heparin and heparan sulfate binding activity. At pH 5.5, the affinity of the wild-type peptide for heparin/ heparan sulfate was increased, implying a role for histidine residues at positions 6 and 28 of pro-IAPP. A Cterminal pro-IAPP fragment (residues 41-67) had no specific affinity for either heparin or heparan sulfate, and the N-or C-terminal fragments had only weak affinity for chondroitin sulfate. These data suggest that monomeric N-terminal human pro-IAPP contains a heparin binding domain that is lost during normal processing of pro-IAPP.
Type 2 diabetes is characterized by peripheral insulin resistance (1) coupled with a progressive loss of insulin secretion (2) that is associated with a decrease in pancreatic islet beta cell mass and the deposition of amyloid in the pancreatic islets (3,4). The principal component of islet amyloid is a 37-amino acid peptide called islet amyloid polypeptide (IAPP 1 or amylin) (5,6). IAPP is a normal product of the islet beta cell (7,8) and is cosecreted from beta cells along with insulin (9 -12). In type 2 diabetes, IAPP aggregates to form amyloid fibrils within the islet that are thought to be toxic to islet beta cells (13). Other components of islet amyloid that are common to all amyloids include apolipoprotein E (3,14) and the heparan sulfate proteoglycan, perlecan (3,15).
Heparan sulfate proteoglycans, and in particular the basement membrane proteoglycan perlecan, have been proposed to play an important role in amyloid deposition in Alzheimer's disease, familial amyloidoses, prion diseases, and type 2 diabetes (16,17). Heparan sulfate proteoglycans are widely expressed and are an important component of extracellular matrix and basement membranes, where amyloid tends to accumulate. Many proteins, for example lipoprotein lipase (18) and apolipoprotein E (19), interact with heparan sulfate proteoglycans via electrostatic interaction of heparin binding domains (clusters of basic residues) with the dense negative charge on the highly sulfated glycosaminoglycan chains. In many cases, these interactions serve important physiological roles, but interaction of amyloidogenic precursors with heparan sulfate proteoglycans may also play an important initiating role in amyloid formation (17). Most amyloidogenic precursor proteins, including the Alzheimer's precursor protein (20,21) and serum amyloid A (SAA) (22), have been shown to have heparin binding domains that facilitate their interaction with glycosaminoglycans. These heparin binding domains usually, but not always, follow the consensus sequence for heparin binding proposed by Cardin and Weintraub (23) as follows: XBBBXXBX or XBBXBX, where B is a basic amino acid and X is a nonbasic amino acid. Although a heparin binding domain in IAPP has not yet been identified, fibrils derived by in vitro aggregation of synthetic human IAPP have been shown to bind to both heparin (24) and perlecan (25).
Although the mechanism of islet amyloid formation is unknown, one potential cause has been proposed to be alterations in the processing of the IAPP precursor molecule, pro-IAPP, by the islet beta cell (3,26). Pro-IAPP is processed to mature IAPP in beta cell secretory granules by the action of prohormone convertase enzymes at the C-terminal side of dibasic residues. The prohormone convertases PC2 and PC3 (also called PC1) are likely responsible for pro-IAPP processing, since they are present in beta cell granules, are known to similarly process proinsulin at the C-terminal side of dibasic residues (27,28), and appear to be capable of fully processing pro-IAPP in vitro (29). Furthermore, we have recently shown that PC2 is essential for processing of pro-IAPP at its N-terminal cleavage site (30). In type 2 diabetes, processing of proinsulin by beta cells is defective, resulting in elevated release of proinsulin relative to insulin (31,32). Since proinsulin and pro-IAPP are processed in parallel in beta cells, it seems likely that pro-IAPP processing may also be defective in this disease, leading to the hypersecretion of unprocessed or partially processed forms of pro-IAPP. Interestingly, immunoreactivity for the N-terminal (but not the C-terminal) region of pro-IAPP has been found to be present in islet amyloid deposits (33) suggesting that partially processed, N-terminally extended pro-IAPP might be an important molecule in islet amyloid formation.
We hypothesized that unprocessed or partially processed pro-IAPP might interact with basement heparan sulfate proteoglycans. To test this hypothesis, we synthesized fragments of pro-IAPP containing the N-and C-terminal cleavage sites, and we assessed their ability to bind to heparin and heparan sulfate.

EXPERIMENTAL PROCEDURES
Materials-Heparin, heparan sulfate, chondroitin sulfate A, bovine serum albumin (BSA), Sepharose CL-4B, heparin-agarose CL-4B, and thioflavin T were purchased from Sigma. CNBr-activated Sepharose 4B was purchased from Amersham Pharmacia Biotech. Wild-type and mutant pro-IAPP fragments were synthesized at the Nucleic Acid Protein Synthesis unit at the University of British Columbia. All synthetic peptides were high performance liquid chromatography-purified, and the N-terminal peptides containing cysteine residues at positions 6 and 9 of IAPP were cyclized. Rat and human IAPP were purchased from Bachem (Torrance, CA).
Affinity Chromatography-Heparin-agarose affinity chromatography was performed as per Ancsin and Kisilevsky (22). In brief, an 8-ml column of heparin-agarose was equilibrated with 20 mM Tris-HCl at either pH 7.5 or 5.5, as indicated. Peptide (300 g) was dissolved in buffer and loaded onto the column, washed with four column volumes of buffer, and then developed at a flow rate of 0.5 ml/min with a 0 -1 M NaCl gradient over five column volumes using a GM-1 gradient mixer (Amersham Pharmacia Biotech). Fractions (0.5 ml) were collected, and absorbance was measured at 214 nm. Salt concentrations in fractions were directly measured using a hand-held conductivity meter. Heparan sulfate and chondroitin sulfate affinity chromatography were performed in a similar manner, except the columns were made by coupling free heparan sulfate or chondroitin sulfate to CNBr-activated Sepharose 4B beads following the protocol provided by the manufacturer. Glycosaminoglycan coupling efficiency to the Sepharose beads, assessed by the toluidine blue method (34) was found to be 0.25 mg/ml gel for chondroitin sulfate and 0.1 mg/ml gel for heparan sulfate. To minimize cost for preparation of heparan sulfate-Sepharose the coupling protocol was scaled down, and a 3-ml column was used.
Fluorometry-The degree of peptide aggregation to form amyloid fibrils was determined by thioflavin T fluorescence using an assay adapted from Naiki et al. (35) and Kudva et al. (36) for a 96-well plate format. All peptides were dissolved in 100% dimethyl sulfoxide (Me 2 SO), aliquoted, and kept frozen as 250 M stock solutions until thawing immediately prior to use. Peptide was added to wells containing 10 M thioflavin T in 10 mM Tris-HCl, pH 7.4, 100 mM NaCl, and 0.1% Triton X-100 to a final peptide concentration of 12.5 M and final Me 2 SO concentration of 5%. The plate was sealed with parafilm and fluorescence measured at 37°C using a Fluoroskan (Labsystems, Vista, CA) fluorometer with filters set at 444 (excitation) and 485 (emission) nm and bandwidth slits of 12 and 14 nm, respectively. When bound to amyloid fibrils, thioflavin T fluoresces with excitation and emission maxima of 450 and 482 nm, respectively (35), and fluorescence under these conditions correlates with the degree of amyloid fibril formation as assessed by electron microscopy (36). Triplicate measurements for each peptide were made every minute from 0 to 100 min, every 10 min from 100 to 150 min, and finally every 60 min from 150 min to 15 h, allowing kinetic assessment of IAPP fibrillogenesis. The mean data were plotted and analyzed using Kaleidagraph TM software (Synergy Software, Reading, PA). Since amyloid fibril elongation is thought to follow first-order rate kinetics (37,38), the data were fit using nonlinear least squares to the equation: where F t is the fluorescence at time t; F ϱ is the steady state fluorescence, and k is the initial rate constant.

Affinity of Human Pro-IAPP Fragments for Heparin-To test
whether the cluster of basic amino acids in the regions of the Nand C-terminal cleavage sites in pro-IAPP might constitute heparin binding domains, we synthesized peptides corresponding to the 30 N-terminal or 27 C-terminal amino acids of human pro-IAPP (see Fig. 1), containing these domains, and applied these peptides to a heparin-agarose column. Since fibril formation might enhance heparin binding (24), a critical amyloidogenic region in IAPP (amino acids 31-40 of pro-IAPP) was omitted from these synthetic peptides to minimize the likelihood of protein aggregation. The N-terminal pro-IAPP fragment was retained by the heparin column, eluting at 0.18 M NaCl on a 0 -1 M NaCl gradient developed over 80 min ( Fig. 2A). In contrast, BSA, a protein known to not bind to heparin (22), eluted in the void volume. When applied to a column containing uncoupled Sepharose, the N-terminal pro-IAPP fragment eluted in the void volume ( Fig. 2A), indicating that its retention on heparin-agarose was not due to a nonspecific interaction between the peptide and the column matrix. The C-terminal region of pro-IAPP contains two pairs of basic residues, including one at the C-terminal cleavage site involved in processing of pro-IAPP to IAPP-(1-37). Unlike the Nterminal pro-IAPP fragment, the C-terminal pro-IAPP peptide interacted only weakly with heparin, eluting in wash fractions prior to commencement of the NaCl gradient (Fig. 2B).
To determine whether the Lys 10 -Arg 11 -Lys 12 sequence in the N-terminal pro-IAPP cleavage site is essential for heparin binding, we synthesized a peptide in which alanine residues were substituted for the two basic residues (lysine and arginine) that compose the cleavage site recognized by prohormone convertases during normal pro-IAPP processing (see Fig. 1). The K10A/R11A mutant N-terminal pro-IAPP peptide fragment had no affinity for heparin (Fig. 2C), eluting in the void volume when applied to the heparin-agarose column. The complete loss of heparin binding in the K10A/R11A mutant peptide indicates that one or both of the basic residues in the Nterminal cleavage site of pro-IAPP are essential for heparin binding.
Affinity of Human Pro-IAPP Fragments for Heparan Sulfate-To determine whether the N-terminal region of human

FIG. 1. Amino acid sequence of fulllength human pro-IAPP (A) and the synthetic N-and C-terminal pro-IAPP peptides used in these studies (B).
Arrows denote sites of pro-IAPP cleavage during normal processing by prohormone convertases PC2 and PC3 to produce mature IAPP (in bold). Boxed amino acids in A represent sequence of synthetic peptides shown in B. ϩ designates basic amino acid; underlined residues designate alanine substitutions.
pro-IAPP has the ability to bind to the heparan sulfate chains of heparan sulfate proteoglycans, as it does heparin, we applied the synthetic N-terminal pro-IAPP fragment to a column in which heparan sulfate was coupled to Sepharose. As shown in Fig. 3A, the N-terminal pro-IAPP peptide bound to the heparan sulfate column, eluting at a NaCl concentration (0.17 M) almost identical to that observed when the peptide was applied to heparin-agarose. The C-terminal pro-IAPP fragment did not, by contrast, bind to the heparan sulfate, eluting in wash fractions (Fig. 3A). Similarly, the K10A/R11A mutant N-terminal pro-IAPP fragment eluted in the void volume when applied to the heparan sulfate-Sepharose column (Fig. 3A). Thus, the cluster of basic residues in the N-terminal cleavage site of pro-IAPP appears to confer affinity of this molecule for both heparin and heparan sulfate.
To elucidate which of the basic residues in the N-terminal cleavage site of pro-IAPP may be critical for heparan sulfate binding, we next synthesized two mutant peptides, in which an alanine residue was substituted for either the lysine at position 10 (K10A) or the arginine at position 11 (R11A) of pro-IAPP. When applied to the heparan sulfate column, both the K10A and the R11A mutant peptides eluted in the void volume (Fig. 3B), indicating that both of these basic residues are critical for interaction of N-terminal pro-IAPP with heparan sulfate.
Affinity of Human Pro-IAPP Fragments for Chondroitin Sulfate-When applied to a chondroitin sulfate-Sepharose column, both the N-terminal and C-terminal pro-IAPP peptide fragments appeared to have weak interaction with this glycosaminoglycan, eluting after the void volume but prior to commencement of the salt gradient (Fig. 4). Thus, the affinity of the N-terminal pro-IAPP peptide appears to be much stronger for heparan sulfate than for chondroitin sulfate. Nonetheless, the weak binding of the N-terminal pro-IAPP peptide to chondroitin sulfate does seem to be dependent on the presence of the basic residues in the N-terminal cleavage site, since the K10A/ R11A mutant N-terminal pro-IAPP peptide did not bind to chondroitin sulfate, eluting in the void volume (Fig. 4).
Human N-and C-terminal Pro-IAPP Fragments Are Non-Fibrillogenic-Amyloid fibrils formed by mature human IAPP (pro-IAPP- ) and other amyloidogenic peptides including the amyloid-␤ protein of Alzheimer's disease are known to bind to heparin (24). This affinity is thought to be dependent on the aggregation state of the peptide, since human but not nonfibrillogenic rodent IAPP has been found to bind to both heparin (24) and perlecan (25). To rule out the possibility that the heparin binding activity of our synthetic N-terminal pro-IAPP peptide was simply due to its aggregation into fibrils that subsequently bound heparin, we measured fibril formation using thioflavin T fluorescence. As expected, thioflavin T fluorescence rapidly increased in the presence of human IAPP but was unchanged in the presence of rat IAPP, demonstrating the known fibrillogenic properties of human but not rat IAPP (Fig. 5). Neither the N-nor the C-terminal pro-IAPP fragments formed fibrils as measured by thioflavin T fluorescence (Fig. 5), producing identical traces to nonfibrillar rat IAPP. These data indicate that both N-and Cterminal pro-IAPP fragments were not fibrillar over the time course of these experiments and therefore that the affinity for heparin and heparan sulfate demonstrated by N-terminal pro-IAPP was not due to its prior aggregation.
Involvement of Histidine Residues in Heparin Binding at Acidic pH-By having determined that both the lysine and arginine residues at positions 10 and 11 of N-terminal pro-IAPP are crucial for its ability to bind heparan sulfate at pH 7.5 (Fig. 3B), we next investigated the involvement of histidine residues, which would be protonated in the acidic milieu of beta cell secretory granules. There are two histidine residues present in human pro-IAPP, one in the N-terminal flanking region (position 6 of pro-IAPP) and one at amino acid 18 of mature IAPP (position 29 of pro-IAPP) (see Fig. 1). At pH 5.5, the heparin affinity of the N-terminal human pro-IAPP fragment was increased, with the peptide eluting at 0.28 M NaCl (versus 0.18 M at pH 7.5) (Fig. 6A). The affinity of the peptide for heparan sulfate was similarly increased at acidic pH (Fig. 6B). However, the K10A/R11A mutant had little affinity for heparin (Fig. 6A) or heparan sulfate (Fig. 6B) under these conditions. Thus, although either or both of the two histidine residues present in pro-IAPP may increase its affinity for heparin/heparan sulfate at acidic pH, they cannot substitute for the Lys 10 -Arg 11 -Lys 12 sequence in the N-terminal cleavage site, which is still essential for heparin binding at either acidic or neutral pH. DISCUSSION In addition to IAPP, islet amyloid deposits have been shown to contain immunoreactivity for the N-terminal region of pro-IAPP (the IAPP precursor) (33) and for basement membrane heparan sulfate proteoglycan (15). The presence of these molecules in islet amyloid deposits suggests their involvement in the mechanism of islet amyloid deposition, yet it is unknown what role they might play. In the present study, we demonstrate that a synthetic fragment of the N-terminal region of pro-IAPP binds to heparin and heparan sulfate. We further demonstrate that the heparin binding domain in N-terminally extended pro-IAPP requires the presence of basic residues in the N-terminal cleavage site at which one step in pro-IAPP processing occurs and, therefore, that normal processing of pro-IAPP would be predicted to destroy this heparin binding domain. The C-terminal region of pro-IAPP showed no affinity for heparin or heparan sulfate, despite the presence of two pairs of basic residues in this region. Our findings raise the possibility that if secretion of unprocessed pro-IAPP (or partially processed, N-terminally extended pro-IAPP) from the beta cell is increased in type 2 diabetes, it might bind to the  6. Affinity of N-terminal pro-IAPP fragment for heparin and heparan sulfate is increased at pH 5.5. Synthetic peptides corresponding to human pro-IAPP amino acids 1-30 (N-terminal) or 1-30 with alanine substitutions (K10A/R11A) were applied to a heparin-agarose (A) or heparan sulfate-Sepharose column (B), and affinity chromatography was performed as in Fig. 2, except the Tris-HCl buffer was pH 5.5, as indicated. Arrows denote elution time of N-terminal pro-IAPP peptide in Tris-HCl at pH 7.5. sulfated glycosaminoglycan side chains of heparan sulfate proteoglycans, creating a nidus for amyloidogenesis within the pancreatic islet. We speculate that this mechanism may be an important initiating step in islet amyloid formation in type 2 diabetes.
Pro-IAPP is thought to be processed to mature IAPP in beta cell secretory granules by the action of PC2 and/or PC3 (PC1) cleaving on the C-terminal side of pairs of basic residues (in both cases lysine-arginine), followed by trimming of these basic residues by carboxypeptidase E (30). As a result, normal pro-IAPP processing results in the removal of the basic residues that we have shown are critical for heparin binding of the N-terminal region of pro-IAPP. Interestingly, the Lys 10 -Arg 11 -Lys 12 cluster of basic residues in the N-terminal pro-IAPP fragment does not represent a classic linear heparin binding domain as proposed by Cardin and Weintraub (23); however, many proteins that bind heparin do not possess these sequences (39,40). One model has suggested that a spacing of ϳ20 Å between two basic amino acids is a critical determinant of heparin binding ability (41). Such spacing can be achieved by a peptide in ␣-helical conformation by basic amino acids spaced 13 residues apart or, in ␤-strand conformation, 7 residues apart. Although the basic amino acids close to the N-terminal Lys 10 -Arg 11 cleavage site (His 6 , Arg 22 , and His 29 ) are not appropriately spaced according to this model, it is still conceivable that they contribute to the binding of N-terminal pro-IAPP to heparin/heparan sulfate. Indeed, the increased affinity for heparin and heparan sulfate of the N-terminal pro-IAPP peptide at pH 5.5 implies a possible role for the histidine residues at positions 6 and 29. Also, the disulfide bond between the cysteines at positions 13 and 18 of pro-IAPP might be predicted to bring the arginine at position 22 in closer proximity to the Lys 10 -Arg 11 -Lys 12 sequence. However, the uncyclized form of the N-terminal pro-IAPP peptide binds as well to heparin as the cyclized form. 2 The heparin binding activity of the N-terminal pro-IAPP peptide therefore does not depend upon any conformational change induced by formation of the disulfide bond. Thus, while the Lys 10 -Arg 11 -Lys 12 sequence at the Nterminal cleavage site of pro-IAPP is clearly essential for heparin binding, whether this sequence is in itself sufficient or whether the basic residues outside of this sequence are also critical will require further study using additional mutant peptides.
Heparin binding domains may also be created by protein aggregation, as illustrated by the binding to heparin (24) and perlecan (25) of aggregated human IAPP. Since our synthetic N-and C-terminal pro-IAPP fragments remained soluble and did not form fibrils as assessed by thioflavin T fluorescence, our findings cannot be explained by prior aggregation of the Nterminal pro-IAPP peptide to form a heparin binding domain. It is possible, however, that binding of soluble pro-IAPP to heparin/heparan sulfate might induce conformational changes in the protein that would enhance fibrillogenesis. Indeed, binding of soluble amyloid-␤ protein to glycosaminoglycans is known to stimulate ␤-sheet conformation and aggregation (42), and perlecan has been shown to stimulate fibril formation from mature human IAPP (25).
Our finding that heparin/heparan sulfate binding activity of the N-terminal pro-IAPP peptide is increased at pH 5.5 raises the possibility that alterations in the local pH, for example in acidic intracellular compartments, might impact pro-IAPP binding to heparan sulfate proteoglycans and subsequently amyloidogenesis. Alternatively, it cannot be ruled out that such interactions might play a normal physiological role, for example in (pro)IAPP trafficking and/or processing, although it is unknown whether heparan sulfate proteoglycans are a component of the acidic beta cell secretory granules in which (pro)IAPP resides. Heparan sulfate proteoglycans have been shown to be secreted via the constitutive pathway in other secretory cells (43), and we have recently found that immature (neonatal) rat beta cells secrete a significant proportion of (pro)IAPP immunoreactivity by the constitutive secretory pathway (12). The possible significance of the increased heparin binding of pro-IAPP at acidic pH in both amyloid pathogenesis and in normal physiology may therefore be worthy of further investigation.
The affinity of the N-terminal pro-IAPP peptide for another highly sulfated glycosaminoglycan, chondroitin sulfate, was much less than that observed for heparan sulfate. The weak interaction between chondroitin sulfate and the N-terminal pro-IAPP peptide likely does, however, involve the Lys 10 -Arg 11 -Lys 12 sequence in the N-terminal cleavage site, since substitution of either the Lys 10 or Arg 11 residues with alanines resulted in total loss of binding. Chondroitin sulfate has not been shown to be a component of islet amyloid, although it may be a component of amyloid deposits in experimental murine inflammation-associated (AA) amyloidosis (44) and Alzheimer's disease (45). This glycosaminoglycan has also been shown to bind to human IAPP-derived fibrils in vitro, albeit less strongly than does heparan sulfate (25). The finding that the N-terminal pro-IAPP peptide binds to heparan sulfate more avidly than it does to chondroitin sulfate, despite the higher degree of sulfation of the latter, suggests that the degree of sulfation is not the most important determinant of sulfated glycosaminoglycan interaction with pro-IAPP. We speculate that the spacings of the sulfate groups in heparan sulfate and heparin are more appropriate than that of chondroitin sulfate for their interaction with pro-IAPP, as has been recently suggested for interaction of serum amyloid A, another amyloidogenic precursor, with glycosaminoglycans (22).
Altered proteolytic processing of precursors to produce a more amyloidogenic molecule may be a general mechanism underlying amyloid deposition in several different amyloidoses. This idea was first suggested by the work of Glenner et al. (46) on immunoglobulin light chain (AL) amyloidosis but has since been implicated in the mechanism of amyloid formation in Alzheimer's disease (17,47,48) and recently in familial British dementia (49). In the case of islet amyloid, we propose that impaired proteolytic processing of pro-IAPP by beta cells may result in disproportionate secretion of forms of (pro)IAPP with high affinity for heparan sulfate proteoglycans present on islet cell basement membranes. Once exocytosed from the beta cell, these molecules may bind to perlecan on the basement membranes of beta cells or islet vascular endothelial cells, preventing their entry into islet capillaries. Indeed, ultrastructural evidence from a transgenic mouse model of islet amyloid formation suggests that amyloid fibrils first accumulate extracellularly between islet beta cells and blood vessels (50). Interaction with the sulfated glycosaminoglycan side chains might then induce conformational changes in pro-IAPP that favor ␤-sheet formation, enhancing its tendency to aggregate. The local accumulation of pro-IAPP bound to perlecan might form a nidus for amyloid formation to which other amyloidogenic forms of IAPP including the major secreted form, IAPP, could be incorporated following their secretion from neighboring beta cells. In nondiabetic patients (in which islet amyloid is not usually observed) (51), pro-IAPP processing would be expected to be nearly complete (based on proinsulin (32)), resulting in loss of the pro-IAPP heparin binding domain prior to exocytosis from the beta cell. Interestingly, the heparin binding activity of pro-IAPP that we identified was observed only in the Nterminal and not the C-terminal region of the peptide. Immuno-reactivity for the N-terminal flanking region of pro-IAPP, but not the C-terminal region, is present in islet amyloid in type 2 diabetic human pancreas (33), raising the possibility that a partially processed, N-terminally extended pro-IAPP conversion intermediate may be an important molecule in islet amyloid formation. Whether partially processed pro-IAPP is secreted in excessive amounts in type 2 diabetes is unknown; however, beta cells of patients with type 2 diabetes have disproportionately elevated secretion of both proinsulin and a partially processed proinsulin conversion intermediate, des-31,32-proinsulin (32). Moreover, prolonged culture in high glucose causes marked accumulation of the N-terminally extended pro-IAPP conversion intermediate in human islets (52) as well as rapid islet amyloid formation in islets from transgenic mice expressing amyloidogenic human IAPP (53). If indeed hyperglycemia in type 2 diabetes is associated with excessive secretion of the N-terminal pro-IAPP conversion intermediate and its subsequent deposition as islet amyloid, we hypothesize that binding of N-terminal pro-IAPP to basement membrane heparan sulfate proteoglycans may be an important pathogenic event in this pathway.