Amide inequivalence in the fibrillar assembly of islet amyloid polypeptide
Department of Molecular Biophysics and Biochemistry, Yale University, 260 Whitney Avenue, New Haven, CT 06520-8114, USA
1 To whom correspondence should be addressed. E-mail: andrew.miranker{at}yale.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionFundingSupplementary dataAcknowledgementsReferences |
|---|
Amyloid fibers are aggregated, yet highly ordered, β-sheet-rich assemblies of misfolded proteins. Order is established in such systems following profiles indicative of nucleation-dependent assembly. Nucleation dependence suggests that specific interactions, such as long-range contacts and/or strand registration, are critical to establishing initial fiber structure. Here, we show that amino acids at selected positions participate in key interactions that modulate the pathway of amyloid fiber formation by the hormone, islet amyloid polypeptide (IAPP). Specifically, we investigated the role of amide side-chain interactions in the process of IAPP assembly. We mutated five of the asparagine side chains in IAPP and assessed their effects on the kinetics of assembly. We find that the asparagine amide side chains strongly dictate the ability of IAPP to form fibers. In particular, the elimination of two specific asparagines results in near and total loss of amyloid, respectively. Interestingly, the two asparagines are located in a recently identified domain with
-helical bias. These sensitivities are unusual for IAPP, as IAPP is generally tolerant to mutation. Here, we demonstrate this mutational tolerance by assessing 10 alterations at five distinct sites. In all cases, the constructs form fibers on timescales perturbed by less than a factor of two compared with wild-type protein. These findings indicate the presence of key specific interactions that are the determinants of IAPP amyloid formation.
Keywords: amylin/amyloid/IAPP/islet amyloid/type II diabetes
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionFundingSupplementary dataAcknowledgementsReferences |
|---|
The abnormal self-association of soluble protein into fibrillar aggregates termed amyloid is a common feature of a number of diseases such as Alzheimer's, type II diabetes, and Parkinson's disease (Westermark, 2005
; Chiti and Dobson, 2006). Amyloids are distinct from other protein aggregates in that they exhibit a high degree of structural order. Indeed, amyloid fiber structure is defined by its characteristic features. In particular, amyloids are fibrous and consist of β-sheets with strands oriented perpendicular to the long axis of the fiber and backbone hydrogen bonding occurring in parallel (Chiti and Dobson, 2006; Nelson and Eisenberg, 2006; Makin et al., 2006). This common core structure arises regardless of the conformation of the precursor. It is therefore likely that the mechanism of fiber assembly is similar in all amyloid systems.
The kinetics of different amyloid systems share common features. Typically, a protein placed under amyloidogenic solution conditions will initially remain soluble. This first period of time is termed the lag phase. Over time, precursor proteins undergo conformational change and self-association to transiently sample high energy states. The highest energy state is termed the nucleus. Addition of protein to the nucleus and/or conformational change results in stabilization and allows subsequent cooperative assembly into the fiber state (Ferrone, 1999). Amyloidogenesis can therefore be described as a nucleation-dependent polymerization process, akin to crystallization (Chiti and Dobson, 2006). As with crystallization, the addition of exogenous pre-formed fiber results in the bypass of nucleation and results in an exponential growth by elongation of seed material. Recently, intermediate states in amyloidogenesis have been demonstrated to be cytotoxic to cultured cells (Janson et al., 1999; Bucciantini et al., 2002; Caughey and Lansbury, 2003). Therefore, elucidating the common features in the mechanism of fiber assembly also contributes to our understanding of the pathology of amyloid-associated diseases.
Amyloid and amorphous aggregates represent distinct phases accessible to proteins including -synuclein (Hoyer et al., 2002; Uversky, 2003), Aβ (Gorman et al., 2003), β-2-microglobulin (Morgan et al., 2001) and immunoglobulin light chain (Bellotti et al., 2000; Khurana et al., 2001). The near crystalline nature of the former is a distinguishing feature which requires specific interactions for both stability and formation. Accordingly, the stability of amyloid is often exceptional, requiring harsh solution conditions for re-solubilization. The origin of this stability has enjoyed a number of recent insights. For example, atomic structures of short model peptide amyloid fibers have revealed an unusually high degree of complementarity and solvent exclusion at sheet–sheet interfaces (Nelson et al., 2005; Sawaya et al., 2007). Furthermore, parallel association of strands enable repeated stacking interactions to occur between identical, in register, side chains. In cases of amide containing side chains, this results in interactions comparable with those of the main-chain amides. These studies are consistent with statistical assessments of >4000 parallel β-sheet structures (Fooks et al., 2006). Stacking interactions between asparagine side chains are the most strongly biased of all self-pairs occurring four times more often than what would be expected by random chance.
An understanding of the final conformation of a protein does not, however, inform on the pathway to that state. The crystalline repeat of a fiber requires that the pathway include energetically distinct interactions for nucleation and fiber growth. For larger proteins such as islet amyloid polypeptide (IAPP), this includes intra- and inter-molecular interactions giving rise to core fiber structure as well as turns and other features. IAPP (also known as amylin) is a peptide hormone (Marzban et al., 2003) normally co-secreted with insulin by the β cells of the pancreas. In patients who develop type II diabetes, IAPP deposits as amyloid in the extracellular spaces of the pancreas (Kahn et al., 1999; Höppener and Lips, 2006). IAPP is a C-terminally amidated 37 residue peptide with a disulfide bond bridging residues 2 and 7 (Fig. 1a). Human IAPP forms fibers on the minutes to hours timescale upon dilution into a physiological buffer solution. In addition, we have shown that IAPP nucleates by both fiber-independent and fiber-dependent pathways (Padrick and Miranker, 2002; Ruschak and Miranker, 2007).
|
We and others have previously reported on the properties of a construct of IAPP in which the constrained N-terminal disulfide (Cys2–Cys7) is removed (Goldsbury et al., 2000
; Abedini and Raleigh, 2005; Koo and Miranker, 2005; Abedini and Raleigh, 2006), i.e. IAPP8–37. IAPP8–37 has been found to be an effective model of the full length protein. For example, we have shown that the elimination of the first seven residues results only in the modest reduction of fiber-dependent nucleation processes (Koo and Miranker, 2005). In this work, we aim to elucidate the principles of molecular organization that govern IAPP fiber assembly using IAPP8–37 as the background construct. We assess the effects of mutating native amide side chains on fiber formation kinetics. We also compare these effects to the mutation of other non-amide side chains in IAPP. Our mutagenic studies highlight position-specific side-chain interactions affecting the fiber assembly pathway of IAPP. |
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionFundingSupplementary dataAcknowledgementsReferences |
|---|
Materials
Buffers and salts were purchased from J.T.Baker. Thioflavin T (ThT) was purchased from Acros. N-ethylmaleimide (NEM) was purchased from Pierce. (1,1,1,3,3,3)-hexafluoroisopropanol (HFIP) was purchased from Sigma-Aldrich and re-purified by fractional distillation. All peptides used in this study were synthesized by the W.M.Keck facility (New Haven, CT, USA) using standard Fmoc methods and purified in house by reverse-phase HPLC (RP-HPLC). Protein concentrations were determined by absorbance measurements at 280 nm (
= 1280 cm–1 M–1).
For cysteine mutants, peptide stocks were prepared by solubilizing freeze-dried peptide in 7 M guanidine hydrochloride (GuHCl), 10% dimethylsulfoxide (DMSO), 0.1% trifluoroacetic acid (TFA), filtering through 0.2 µm filters (Pall Corporation), then loading onto Vydac C-18 microspin columns (Amika Bioscience). Column was washed with 10% acetonitrile, 0.1% TFA followed by MilliQ water, followed by elution with HFIP.
NEM-blocked constructs were prepared by equilibrating cysteine mutants with 500 µM TCEP in 7 M GuHCl, 100 mM potassium phosphate, pH 7.0 for 30 min at room temperature under nitrogen. Ten-fold molar excess NEM from DMSO was then added and allowed to incubate for 2 h. Reaction was quenched with 80 mM 2-mercaptoethanol and purified by RP-HPLC.
All de novo fiber formation reactions were performed at a concentration of 25 µM protein in aqueous buffer, 100 mM KCl, pH 7.4 and 25°C. For N#X mutants, buffer was 50 mM potassium phosphate, whereas for cysteine constructs, buffer was 100 mM potassium phosphate. For N#X constructs, lyophilized protein was solubilized with water followed by addition of 2x buffer to initiate reaction. All mutant reactions were paired with a WT8–37 reaction conducted on the same day under exactly matched conditions. For reactions monitored by ThT, buffer also contained 25 µM ThT. For seeding reactions, fiber seeds were first generated by incubating 50 µM protein in 50 mM potassium phosphate, 100 mM KCl, pH 7.4 for 12 h at room temperature. Seeded reactions were conducted at 25 µM precursor, 2.5 µM seed (i.e. 10% seed), 50 mM potassium phosphate, 100 mM KCl, pH 7.4 and 25°C.
Micrographs of negatively stained samples were taken on a Phillips Tecnai 12 transmission electron microscope at 120 kV accelerating voltages. Grids were prepared by applying sample in aqueous buffer to a copper mesh grid freshly coated with carbon and glow discharged at 25 mA for 30 s. Grids were incubated for 1 min and stained with 1% (w/v) phosphotungstic acid (PTA) at pH 7.0. Images were acquired using a 1000 x 1000 pixel Gatan 794 slow-scan CCD at a magnification of x11 000 and between –10 and –20 µm underfocus. Image analysis was performed using Gatan Digital Micrograph software. All conclusions drawn from images in this work include at least one repeat in which the sample identity was withheld from the investigator preparing and analyzing images.
Circular dichroism measurements were made at 25 µM protein concentration on an Aviv 215 spectrometer using 1 mm path length cuvettes. Spectra shown have had buffer spectrum subtracted. Spectra were collected at 0.5 or 1 nm intervals from 195 to 260 nm with 5 s averaging time.
Kinetic measurements of fluorescence were performed as previously described (Koo and Miranker, 2005) using a two-channel fluorimeter (QuantaMaster C-61 fluorescence spectrophotometer, PTI, London, ON, Canada). For anisotropy measurements, instrument was fitted with linear polarizers and excitation was performed at 278 nm wavelength with emission detected at 303 nm. For Thioflavin T, fluorescence was monitored with excitation at 440 nm and emission at 492 nm. Sample holder was temperature regulated to 25°C.
Kinetic data were fit as previously described (Padrick and Miranker, 2002) to the following sigmoid model:
|
))–1, r1 and r2 are lower and upper baselines, respectively, m1 and m2 are the slopes of the corresponding baselines. Data were fit using the Nonlinear Regress function in Mathematica 4.2 (Wolfram Research). All reported data in this work are expressed as ±1 SEM for 
3 repeats. |
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionFundingSupplementary dataAcknowledgementsReferences |
|---|
WT8–37 (i.e. IAPP8–37) contains five native asparagines residues (Fig. 1a). To assess the relative contributions of these asparagines to the mechanism of IAPP fiber formation, we systematically mutated these to leucine. By mutating the native asparagines to leucines, we were able to eliminate the hydrogen-bonding capacity of the amide side chain, without significantly disrupting the steric volume. Furthermore, although leucine has a stronger helical propensity than asparagine, this substitution maintains the calculated overall helical bias (not shown) of the recently identified helical domain of IAPP spanning residues 5–22 (Williamson and Miranker, 2006
). The assembly kinetics of each mutant was then evaluated using intrinsic fluorescence anisotropy which reflects the rigidity of the C-terminal tyrosine (Padrick and Miranker, 2001), and by the amyloid-specific fluorescence enhancement of the dye, Thioflavin T (ThT) (LeVine, 1993).
The asparagine residues of IAPP are kinetically inequivalent for fibrillar assembly. Half of WT8–37 converts to amyloid in 11 000 ± 400 s (i.e. t50) as monitored using the extrinsic dye, ThT (Fig. 1b). Under matched reaction conditions, the mutant N22L converts on a similar timescale, 11 000 ± 600 s. We present these t50 as ratios of the mutant t50 divided by the t50 of WT8–37 protein conducted on the same day with identical buffers (Fig. 1c). On the renormalized timescale, N22L has a value of 1.0 (Fig. 1c). The confidence intervals quoted above on the seconds timescale are unrealistically small. We have instead estimated a much larger confidence interval which when projected onto the renormalized timescale is ±0.25. This is based on the variance of the t50 of WT8–37 protein determined from different synthetic lots on different days. As with N22L, the mutations N31L and N35L result in only modest delay of t50 to within a factor of 3 of WT8–37. In marked contrast, ThT fluorescence of N14L is never observed to increase even after 4 days (Fig. 1b). Similarly, the mutation N21L gives no response to ThT on the timescale of WT8–37. A small increase (
10% that of WT8–37) is observed for N21 mutants on timescales 3 times that of a concomitant WT8–37 reaction. These results are paralleled for the mutants N14A, N14S and N21S (Fig. 1b, inset). This indicates that the effects are a consequence of elimination of the asparagine side chain, and not introduction of the more hydrophobic leucine. Clearly, the interactions of asparagines in establishing fiber structure are different at these five positions.
The formation of rigid structure need not be coincident with amyloid formation. We have previously shown that Y37 incorporates into the interior of IAPP amyloid structure resulting in an increase in fluorescence anisotropy (Padrick and Miranker, 2001, 2002). This property also enables monitoring of fiber formation and, to date, has always yielded kinetic profiles comparable with those collected using the extrinsic fluorophore, ThT. Surprisingly, mutations at two of the five positions, N14 and N21, do not yield the same kinetic profile when monitored by these two alternative methods. Dramatically, the N14L mutant reaches the limiting anisotropy of 0.28 (Lakowicz and Maliwal, 1983), in the experimental dead-time (<30 s) (Fig. 2) yet never gives rise to a ThT-positive species (Fig. 1b). Similarly, the N21L construct forms a weakly (10% that of WT8–37) ThT-positive species in 6 h (Fig. 1b), whereas an increase in tyrosine anisotropy occurs after only 25 min (Fig. 2). ThT is not inducing these effects as our observations hold regardless of whether ThT is introduced before or after formation of the high-anisotropy state (data not shown). Anisotropy also increases markedly for alternative mutations at N14 and N21, namely N14A and N14S, and to a somewhat lesser extent with N21S (Fig. 2). Clearly, residues 14 and 21 mediate specific interactions that are central to the formation of ordered fibrous states of IAPP.
|
The N14 and N21 mutant constructs form poorly ordered aggregates. Approximately 65% of N14L and N21L can be pelleted by centrifugation (18 000g, 10 min) at time points after formation of the high-anisotropy state. In addition, these constructs can be visualized by electron microscopy (Fig. 3). Image evaluations performed and documented with investigators blinded to sample identity show ultrastructural morphologies which are overtly distinct from those of WT8–37. In all cases, the aggregates are dense structures that display either no evidence of fibrillar nature or else dense networks of highly curved filamentous assemblies. For the latter cases, we note that ThT fluorescence intensities of N14 and N21 mutants are significantly less (7–18%) than that of WT8–37 (Fig. 3, insets). The EM images may therefore represent a subpopulation of the total aggregate species. Alternatively, aggregated species may also occlude ThT from contact with fibrillar structures. Far-ultraviolet circular dichroism spectroscopy (far-UV CD) indicates that aggregated N14 (N14ag) and N21 (N21ag) samples are predominantly unstructured. Indeed, the CD spectra of N14Aag, N14Lag and N14Sag indicate weak β-sheet content that is overall more comparable with the soluble lag phase of WT8–37 under matched conditions (Fig. 4). Thus, although it is true that N14 and N21 mutant constructs form aggregates on a minutes timescale, they are not amyloid.
|
|
The altered capacity of N14 and, in part, N21 mutants to form amyloid cannot be rescued by addition of WT8–37 fibers as seed. Preformed fibers can act as seed to template fiber formation either through elongation or by the walls of fibers serving to stabilize nucleating species (Ruschak and Miranker, 2007
). We compared the kinetics of fiber formation for mutant constructs in the presence and absence of a 10% addition of preformed WT8–37 fibers. For N21L, N21S, N22L, N31L and N35L, fiber formation was accelerated by 5-fold compared with a matching de novo reaction (Fig. 5a and c). In marked contrast, N14 mutants (i.e. N14L, N14A and N14S) are wholly unaffected by the presence of WT8–37 fibers (Fig. 5a and c). Conversely, we also tested the ability of mutant fibers and aggregates to accelerate WT8–37 assembly. Those constructs that adopt structures that yield both elevated anisotropy and ThT fluorescence (N21L, N21S, N22L, N31L and N35L) are able to template WT8–37 fiber formation. In contrast, aggregated samples that show only weak ThT enhancement have little or no effect on WT8–37 assembly kinetics (Fig. 5b and d). These are N14L, N14A, N14S as well as early time points of N21L. We further note that the aggregated state of N14L shows no decrease in tyrosine anisotropy when diluted in the absence of WT8–37 (data not shown). Thus, the failure of N14Lag to act as seed is not a consequence of dissociation of the aggregated state. Evidently, N14 mutations result in protein structure that is wholly incompatible with the amyloidogenic structure of WT8–37.
|
The unusual behavior of N14L could not be rescued by a potentially compensatory mutation. The sequence of IAPP fortuitously includes leucine at positions i–2 and i+2 relative to N14 (Fig. 1a). Double mutants were therefore prepared in an effort to rescue the WT8–37 behavior from the effects of N14L. As with N14L, neither L12N N14L nor N14L L16N showed positive signal in kinetic assays by ThT fluorescence (Fig. 6b). In addition, whereas N14L reached limiting anisotropy in <30 s (Fig. 2), N14L L16N took
3 h (Fig. 6a) and L12N N14L was never observed to give an elevated measure. Correspondingly, no aggregates were evident by EM after 30 min of incubation. Secondary structure measured by far-UV CD shows both L12N N14L and N14L L16N to have an increase in random coil character compared with N14L (Fig. 4). Indeed, L12N N14L is more random coil-like than the lag-phase conformation of WT8–37. The only significant difference we observe between L12N N14L and N14L L16N is in its capacity to be templated into fibers. As with N14L, no increase in ThT fluorescence is observed when L12N N14L is incubated with 10% WT8–37 fiber seeds. In contrast, N14L L16N is readily templated giving exponential kinetics to a state that has ThT fluorescence enhancement comparable with WT8–37 (Fig. 6b). Overall, it is plain that the introduction of asparagine at residue 16 or 12 on an N14L background grossly affects structure formation and solubility properties without rescue of wild-type kinetic behavior. This further demonstrates the strong positional dependence of asparagine's effects on IAPP assembly.
|
N mutations at position 12 or 16 on an N14L background. (A) Representative plots of de novo fiber formation by L12N N14L and N14L L16N, as monitored by tyrosine anisotropy. Inset shows statistical assessment of the time required for 50% of the total change in fluorescence anisotropy. For L12N N14L, tyrosine anisotropy change was never observed (ND). (B) Representative plots of 25 µM assembly reactions either de novo, or in the presence of 2.5 µM WT8–37 performed fibers. Kinetics are monitored by fluorescence enhancement of ThT.The IAPP sequence shows a high tolerance to residue alteration at a variety of positions. In order to determine a point of reference for the magnitude of kinetic effects of mutagenesis on IAPP fiber formation, we mutated an array of five additional residues to cysteine (Fig. 7a). In parallel, we modified these cysteines with NEM, deliberately introducing steric bulk at the same positions. These mutants and their NEM derivatives displayed only modest, albeit position-dependent effects on the kinetics of assembly. Indeed, we find that all of these constructs readily form amyloid fibers with minimal alteration in t50. For two of the mutants, V17C and T36C, the timescales of assembly are, respectively, faster than or equivalent to that of WT8–37 (Fig. 7b). The t50 of A13C, A25C and T30C are delayed by a factor of
2 (Fig. 7c). The addition of a bulky NEM group results in a 2-fold kinetic delay at positions 17 and 36, 2-fold acceleration at residues 25 and 30, and no effect at residue 13 (Fig. 7d). Morphological assessment by EM reveals nothing to distinguish these fibers from WT8–37 (data not shown). In no instance do we detect the rapid disordered aggregation as seen for the N14 or N21 constructs above. Evidently, modifications at these sites affect the stability of the nucleating species without fundamentally altering the assembly pathway.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionFundingSupplementary dataAcknowledgementsReferences |
|---|
Determining the molecular detail of rate-limiting steps is essential to understanding the mechanism of amyloid fiber formation. The highly ordered nature of fibrillar aggregation suggests that specific structures are important to the rate-limiting processes. Here, we have studied the effects of perturbing putative structural elements of the nucleating species in IAPP fiber assembly. Specifically, we find that the five native asparagines in WT8–37 make unequal contributions to the structure of IAPP fibers and the kinetics of their assembly. Mutation at two sites, 14 and 21, grossly affect assembly in a manner that is independent of the particular choice of substitution. In contrast, three other amide side-chain sites, as well as 10 other mutant constructs at five unrelated sites show only modest alteration in overall kinetics. The amyloid nucleus of WT and WT8–37 constructs are generally reported to be tolerant of mutation. For example, in studies of chimeric constructs between human IAPP and a non-amyloidogenic sequence variant from rat, numerous mutations were shown to be simultaneously tolerated by the human sequence (Green et al., 2003
). Most notably, fibers formed even for a triple proline mutant (A25P S28P S29P). To our knowledge, only the simultaneous introduction of three proline mutations, V17P, S19P and T30P, has been shown to wholly abolish fiber formation (Abedini and Raleigh, 2006). In this work, a comparable result was obtained with conservative point mutations, at two different sites. Clearly, positions 14 and 21 appear critical to amyloid formation.
The molecular origins for disruption of fibrillogenesis are similar but not identical for N14 and N21 point mutations. Several characteristics of N14 and N21 mutants more closely resemble each other than they do WT8–37. For example, both give rise to aggregates on timescales that are short compared with WT8–37 (Fig. 2). Furthermore, neither of these aggregates gives significantly elevated fluorescence intensity by ThT (Fig. 1b). The most notable difference is the capacity of N21 mutants to be templated into amyloid by the addition of preformed seeds of WT8–37 protein (Fig. 5c). N14 mutants are wholly incapable of forming WT8–37 fibers even in response to seeding. This distinction may be related to the far more β-strand character of N21L (Fig. 4) compared with those of the N14 mutants (Fig. 4). In addition, we note that structural studies using EPR required placement of nitroxide spin labels at eight positions that included residue 21 (Jayasinghe and Langen, 2004). Although no kinetic assessments are presented, derivitization of N21C did not abrogate the ability of these authors to form or measure amyloid β-sheet structure. Thus, both residues 21 and 14 are disrupted by similar mechanisms, although residue 14 is plainly more potent.
Positional specificity is further illuminated by the effects of mutation on flanking residues. Whereas amyloid disruption occurs upon mutation of N14 to alanine, serine or leucine, mutation of neighboring A13 to cysteine results in a protein with characteristics that are closely similar to WT8–37. The kinetics of fiber formation are diminished, but by only 2-fold. Furthermore, blocking of this cysteine residue with NEM gives no further alteration in the assembly timescale (Fig. 7d). Likewise, mutation of F15 to a serine results in only a 2-fold delay in kinetics (data not shown). At N22, mutation to leucine has little effect on the kinetics of fiber formation, whereas mutation of N21 to leucine or serine causes marked alterations (Fig. 1b and c). These observations suggest the presence of highly localized contacts. Such contact might represent a side chain to main-chain intramolecular interaction that is within one or two residues, or may correspond to an intermolecular interface that is well ordered, albeit transient in nature.
The nucleating structure sampled near N14 is unlikely to resemble β-strand. IAPP fiber structure is principally composed of parallel β-strands (Jayasinghe and Langen, 2004). This is similar to studies on Aβ (Petkova et al., 2002), tau (Margittai and Langen, 2004), Sup35 (Shewmaker et al., 2006) and model peptide fibers (Sawaya et al., 2007) all of which show parallel stacking of β-strands. This form of β-sheet assembly results in stacking of amino acid side chains, diminishing the role of sequence order on structure formation by any given set of amino acids. For example, the yeast prion, Ure2, is tolerant to randomization of an 88 residue, asparagines and glutamine-rich sequence central to its amyloid assembly (Ross et al., 2005). The EPR study revealing parallel β-sheet structure in IAPP included a probe at residue 12 (Jayasinghe and Langen, 2004). Furthermore, recent NMR data are consistent with a continuous β-strand spanning residues 8–17 (Luca et al., 2007). In a β-strand, the side chains of residues 12, 14 and 16 would reside on the same face. The double mutants examined in this work amount to an amino acid content preserving exchange of residues between 12/14 and 14/16. These exchanges grossly perturb fiber formation in marked contrast to the effects of sequence randomizations on Ure2(35). Despite the gross changes in properties compared with WT8–37, both double mutants retain the capacity to form amyloid. In the case of N14L L16N, this is evident in its capacity to be seeded by WT8–37 (Fig. 6b). For L12N N14L, we note that our controls included assessment of the capacity of EM stain to induce fiber formation. For example, fibers can be visualized for L12N N14L if a wash step is omitted between application of sample and application of stain (not shown). Thus, we do not believe these double mutants negate the capacity of IAPP to form fibers. Rather, we propose that although residues 12, 14 and 16 are in a β-strand conformation in the fiber state (Luca et al., 2007), they are in an alternative conformation in the structure(s) responsible for nucleation.
The N-terminal 22 residues of IAPP sample partially, and/or transiently structured states in aqueous buffer. For residues 5–19, these states are predominantly -helical in character (Williamson and Miranker, 2006). In addition, we and others have demonstrated that IAPP fiber formation can be catalyzed by binding to lipid surfaces (Knight and Miranker, 2004; Jayasinghe and Langen, 2005). Moreover, lipid binding results in stabilization of -helical states of IAPP (Jayasinghe and Langen, 2005) and these states appear to form higher order -helical assemblies prior to fiber formation (Knight et al., 2006). We suggested that the transient -helical sampling observed in solution by NMR may correspond to the same structures stabilized on bilayers (Williamson and Miranker, 2006). Interestingly, all mutants that significantly disrupt IAPP amyloid formation lie within the putative helical region. This includes N14 and N21 mutants reported here. It is also consistent with the observation that V17P S19P T30P (Abedini and Raleigh, 2006) but not A25P S28P S29P (Green et al., 2003) is more effective at abrogating fiber formation. It is therefore plausible that mutations at and/or near N14 and N21 reflect the interactions of -helical and not β-strand precursor states.
In this work, we have demonstrated that specific residues are required to mediate fibrillar nucleation in IAPP. The central players in this process are amide containing side chains, as their alteration results in the formation of disordered and partially ordered aggregates. Just as backbone amides dictate the fundamental structures of proteins (i.e. -helix, β-sheet), the amides of side chains can dictate specific interactions with the same spacing as that of the backbone. This is evident in structural analyses of parallel β-sheets that reveal a particularly strong bias toward asparagine stacking (Fooks et al., 2006). In the atomic structure of the amyloid peptide GNNQQNY, both asparagine and glutamine side chains form a repeating hydrogen-bonded network running in parallel to the corresponding network of the backbone (Nelson et al., 2005). Asparagines and glutamines are enriched in the sequences of a number of amyloidogenic proteins. This is most obvious in polyglutamine expansion diseases such as Huntington's. In such systems, the number of glutamines directly correlates to the amyloidogenicity of the protein (Scherzinger et al., 1999). It is intriguing, therefore, that the amide side chains of IAPP are strongly inequivalent. We note that IAPP is a far more aggressive amyloid-forming protein than polyglutamine expansions of similar length. For example, a 28-residue polyglutamine repeat does not form fibers for hundreds of hours (Chen et al., 2002), whereas under similar conditions here, WT8–37 forms fibers in 4–6 hours. Polyglutamine self-assembly may experience frustration as its misregistered states will be more similar in energy than misregistered states of IAPP. Our studies therefore illuminate a role for specific and not necessarily β-sheet interactions in establishing the structures necessary for amyloid nucleation.
|
Funding |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionFundingSupplementary dataAcknowledgementsReferences |
|---|
National Institute of Health (DK54899).
|
Supplementary data |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionFundingSupplementary dataAcknowledgementsReferences |
|---|
Supplementary data (colour images for figures 1, 2, 4, 5, 6, and 7) are available at PEDS online.
|
Footnotes |
|---|
Edited by Dek Woolfson
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionFundingSupplementary dataAcknowledgementsReferences |
|---|
We thank Dr J.Knight for critical reading of the manuscript, Dr V.Unger for use of the electron microscope and Dr L.Regan for the use of the CD.
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionFundingSupplementary dataAcknowledgementsReferences |
|---|
Abedini A., Raleigh D.P. Biochemistry (2005) 44:16284–16291.[CrossRef][Medline]
Abedini A., Raleigh D.P. J. Mol. Biol. (2006) 355:274–281.[CrossRef][Medline]
Bellotti V., Mangione P., Merlini G. J. Struct. Biol. (2000) 130:280–289.[CrossRef][Web of Science][Medline]
Bucciantini M., Giannoni E., Chiti F., Baroni F., Formigli L., Zurdo J., Taddei N., Ramponi G., Dobson C.M., Stefani M. Nature (2002) 416:507–511.[CrossRef][Medline]
Caughey B., Lansbury P.T. Annu. Rev. Neurosci. (2003) 26:267–298.[CrossRef][Web of Science][Medline]
Chen S., Ferrone F.A., Wetzel R. Proc. Natl Acad. Sci. USA (2002) 99:11884–11889.
Chiti F., Dobson C.M. Annu. Rev. Biochem. (2006) 75:333–366.[CrossRef][Web of Science][Medline]
Ferrone F. Methods Enzymol. (1999) 309:256–274.[Web of Science][Medline]
Fooks H.M., Martin A.C., Woolfson D.N., Sessions R.B., Hutchinson E.G. J. Mol. Biol. (2006) 356:32–44.[CrossRef][Web of Science][Medline]
Goldsbury C., Goldie K., Pellaud J., Seelig J., Frey P., Müller S.A., Kistler J., Cooper G.J., Aebi U. J. Struct. Biol. (2000) 130:352–362.[CrossRef][Web of Science][Medline]
Gorman P.M., Yip C.M., Fraser P.E., Chakrabartty A. J. Mol. Biol. (2003) 325:743–757.[CrossRef][Web of Science][Medline]
Green J., Goldsbury C., Mini T., Sunderji S., Frey P., Kistler J., Cooper G., Aebi U. J. Mol. Biol. (2003) 326:1147–1156.[CrossRef][Web of Science][Medline]
Hoyer W., Antony T., Cherny D., Heim G., Jovin T.M., Subramaniam V. J. Mol. Biol. (2002) 322:383–393.[CrossRef][Web of Science][Medline]
Höppener J.W., Lips C.J. Int. J. Biochem. Cell Biol. (2006) 38:726–736.[CrossRef][Medline]
Janson J., Ashley R.H., Harrison D., McIntyre S., Butler P.C. Diabetes (1999) 48:491–498.[Abstract]
Jayasinghe S.A., Langen R. J. Biol. Chem. (2004) 279:48420–48425.
Jayasinghe S.A., Langen R. Biochemistry (2005) 44:12113–12119.[CrossRef][Web of Science][Medline]
Kahn S.E., Andrikopoulos S., Verchere C.B. Diabetes (1999) 48:241–253.[Abstract]
Khurana R., Gillespie J.R., Talapatra A., Minert L.J., Ionescu-Zanetti C., Millett I., Fink A.L. Biochemistry (2001) 40:3525–3535.[CrossRef][Medline]
Knight J.D., Miranker A.D. J. Mol. Biol. (2004) 341:1175–1187.[CrossRef][Web of Science][Medline]
Knight J.D., Hebda J.A., Miranker A.D. Biochemistry (2006) 45:9496–9508.[CrossRef][Web of Science][Medline]
Koo B.W., Miranker A.D. Protein Sci. (2005) 14:231–239.[CrossRef][Medline]
Lakowicz J.R., Maliwal B.P. J. Biol. Chem. (1983) 258:4794–4801.
LeVine H.D. Protein Sci. (1993) 2:404–410.[Web of Science][Medline]
Luca S., Yau W.M., Leapman R., Tycko R. Biochemistry (2007) 46:13505–13522.[CrossRef][Medline]
Makin O.S., Sikorski P., Serpell L.C. Curr. Opin. Chem. Biol. (2006) 10:417–422.[CrossRef][Medline]
Margittai M., Langen R. Proc. Natl Acad. Sci. USA (2004) 101:10278–10283.
Marzban L., Park K., Verchere C.B. Exp. Gerontol. (2003) 38:347–351.[CrossRef][Web of Science][Medline]
Morgan C.J., Gelfand M., Atreya C., Miranker A.D. J. Mol. Biol. (2001) 309:339–345.[CrossRef][Web of Science][Medline]
Nelson R., Eisenberg D. Curr. Opin. Struct. Biol. (2006) 16:260–265.[CrossRef][Web of Science][Medline]
Nelson R., Sawaya M.R., Balbirnie M., Madsen A.O., Riekel C., Grothe R., Eisenberg D. Nature (2005) 435:773–778.[CrossRef][Medline]
Padrick S.B., Miranker A.D. J. Mol. Biol. (2001) 308:783–794.[CrossRef][Medline]
Padrick S.B., Miranker A.D. Biochemistry (2002) 41:4694–4703.[CrossRef][Web of Science][Medline]
Petkova A.T., Ishii Y., Balbach J.J., Antzutkin O.N., Leapman R.D., Delaglio F., Tycko R. Proc. Natl Acad. Sci. USA (2002) 99:16742–16747.
Ross E.D., Edskes H.K., Terry M.J., Wickner R.B. Proc. Natl Acad. Sci. USA (2005) 102:12825–12830.
Ruschak A.M., Miranker A.D. Proc. Natl Acad. Sci. USA (2007) 104:12341–12346.
Sawaya M.R., et al. Nature (2007) 447:453–457.[CrossRef][Medline]
Scherzinger E., Sittler A., Schweiger K., Heiser V., Lurz R., Hasenbank R., Bates G.P., Lehrach H., Wanker E.E. Proc. Natl Acad. Sci. USA (1999) 96:4604–4609.
Shewmaker F., Wickner R.B., Tycko R. Proc. Natl Acad. Sci. USA (2006) 103:19754–19759.
Uversky V.N. J. Biomol. Struct. Dyn. (2003) 21:211–234.[Web of Science][Medline]
Westermark P. FEBS J. (2005) 272:5942–5949.[CrossRef][Medline]
Williamson J.A., Miranker A.D. Protein Sci. (2007) 16:110–117.[CrossRef][Medline]
Received November 16, 2007; revised November 16, 2007; accepted November 19, 2007.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  J. R. Cort, Z. Liu, G. M. Lee, K.N.L. Huggins, S. Janes, K. Prickett, and N. H. Andersen Solution state structures of human pancreatic amylin and pramlintide Protein Eng. Des. Sel., August 1, 2009; 22(8): 497 - 513. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||