Biochemical Journal

Research article

The chaperone proteins HSP70, HSP40/DnaJ and GRP78/BiP suppress misfolding and formation of β-sheet-containing aggregates by human amylin: a potential role for defective chaperone biology in Type 2 diabetes

Vita Chien , Jacqueline F. Aitken , Shaoping Zhang , Christina M. Buchanan , Anthony Hickey , Thomas Brittain , Garth J. S. Cooper , Kerry M. Loomes

Abstract

Misfolding of the islet β-cell peptide hA (human amylin) into β-sheet-containing oligomers is linked to β-cell apoptosis and the pathogenesis of T2DM (Type 2 diabetes mellitus). In the present study, we have investigated the possible effects on hA misfolding of the chaperones HSP (heat-shock protein) 70, GRP78/BiP (glucose-regulated protein of 78 kDa/immunoglobulin heavy-chain-binding protein) and HSP40/DnaJ. We demonstrate that hA underwent spontaneous time-dependent β-sheet formation and aggregation by thioflavin-T fluorescence in solution, whereas rA (rat amylin) did not. HSP70, GRP78/BiP and HSP40/DnaJ each independently suppressed hA misfolding. Maximal molar protein/hA ratios at which chaperone activity was detected were 1:200 (HSP70, HSP40/DnaJ and GRP78/BiP). By contrast, none of the chaperones modified the secondary structure of rA. hA, but not rA, was co-precipitated independently with HSP70 and GRP78/BiP by anti-amylin antibodies. As these effects occur at molar ratios consistent with chaperone binding to relatively rare misfolded hA species, we conclude that HSP70 and GRP78/BiP can detect and bind misfolded hA oligomers, thereby effectively protecting hA against bulk misfolding and irreversible aggregation. Defective β-cell chaperone biology could contribute to hA misfolding and initiation of apoptosis in T2DM.

  • aggregation
  • amylin
  • chaperone
  • glucoseregulated protein of 78 kDa/immunoglobulin heavy-chain-binding protein (GRP78/BiP)
  • heat-shock protein 70 (HSP70)
  • misfolding
  • Type 2 diabetes mellitus

INTRODUCTION

T2DM (Type 2 diabetes mellitus) is characterized by pancreatic islet β-cell dysfunction [1] and loss of β-cell mass [2,3], which leads to progressive impairment of insulin secretion and, ultimately, to overt diabetes. Amylin, also known as IAPP (islet amyloid polypeptide), is a 37-amino-acid polypeptide [46] secreted from pancreatic islet β-cells [7,8] and is a physiological component of insulin granules [9], wherein it is secreted through the cell membrane via the regulated secretory pathway. Amylin self-associates to form aggregates in solution [10,11] and pancreatic islet amyloid [12,13] in humans with T2DM [4]. Islet amyloid has been reported in 40–90% of patients with T2DM studied post-mortem [4,12,14,15], and has been linked to decreased β-cell mass, β-cell dysfunction and the aetiopathogenesis of T2DM [14,16].

In aqueous solution, hA (human amylin) spontaneously forms β-sheets and aggregates, whereas rA (rat amylin) does not [10,11,17,18], and hA aggregation has been linked to β-cell degeneration in T2DM [16,1921]. Mature amyloid fibrils, however, may not contribute directly to its cytotoxicity, since there is no strong correlation between the extent of amyloid deposits and disease severity [19,22], and mature fibrils are not cytotoxic when assessed by in vitro assays [18]. Misfolding of hA into β-sheet-containing oligomers evokes apoptosis originating on the surface and/or within cells, as shown by studies using cultured β-cells or transgenic rodent models susceptible to hA-evoked diabetes [19,20,22].

Cytotoxic hA aggregates have been reported in both the intracellular and extracellular spaces in islets [20,23,24], but the detailed molecular mechanisms by which such misfolding might trigger apoptosis remain to be fully elucidated. Previous studies have indicated that exposure of cultured islets or β-cells to extracellular fibrillogenic hA can result in cell death and suggested that direct contact of hA aggregates with cell membranes may be required to elicit apoptosis [20,2528]. Amylin-evoked membrane instability and leakage-induction via cell membrane interactions are considered to provide one potential cytotoxic mechanism [29,30]. Other mechanisms include increased cellular pro-oxidant responses [18], LDL (low-density lipoprotein) uptake evoked by hA aggregate–cell interactions [31] and activation of the ER (endoplasmic reticulum) stress response [32,33].

Amyloid deposition is a pathological feature not only observed in T2DM, but also in several other diseases such as Alzheimer's disease, Parkinson's disease, Huntington's disease (and related polyglutamine disorders), transmissible spongiform encephalopathy and amyotrophic lateral sclerosis [3436]. Although unrelated proteins are involved in each case, the resulting amyloid structures share morphological and biochemical commonalities, such as the generation of β-fibril structures and their high resistance to proteolytic degradation [34,35].

Similarly, an increasing body of evidence suggests that endogenous chaperones play important protective roles in the pathophysiology of amyloid deposition in several diseases such as Alzheimer's disease, Parkinson's disease, Huntington's disease (and related polyglutamine disorders), transmissible spongiform encephalopathy and amyorophic lateral sclerosis [3436]. These cellular components can recognize non-native proteins and assist protein folding or remove the proteins by proteolytic machineries, such as the ubiquitin/proteasome system, and thus prevent cellular damage caused by protein aggregation. Several studies have now shown that chaperones inhibit amyloid fibril formation derived from α-synuclein [37,38], APP (amyloid precursor proteins) [39,40], polyQ (polyglutamine) [4143] and immunoglobulin light chain [44], which are primary peptides of amyloid associated with Parkinson's disease, Alzheimer's disease, Huntington's disease and systemic/modular amyloidosis respectively.

At present, the cellular processes that lead to or cause hA to misfold and form cytotoxic aggregates remain poorly understood. Clearly, there is a pressing need to better understand the molecular interactions between hA itself and components of the cellular machinery that control and regulate protein folding. Prominent among these processes could be those regulated by molecular chaperones, which bind non-native states of other proteins and assist them to achieve physiological conformations, and stabilize them against irreversible multimeric aggregation [45,46].

Previous studies with pancreatic islet sections have provided evidence for an increased expression of chaperone proteins in diabetes [47] and that overexpression of the HSP (heat-shock protein) HSP70 can attenuate ER stress in cultured islets [48]. Both amylin and insulin are co-localized within the human pancreatic β-cell granule [49] and co-purify in purified granule preparations from cultured insulinoma β-cells [50]. In addition, multiple chaperones including HSP70 were identified within highly purified β-cell granules derived from INS-1E islet β-cells [51]. These secretory granules also contained insulin [51] and amylin (C. Buchanan and A. Hickey, unpublished work), supporting the co-localization of all of these components to the same organelle.

As hA misfolding can induce ER stress and β-cell apoptosis [52,53] and therefore probably encounter chaperones along the regulated secretory pathway, we therefore decided to probe their intrinsic functional relationships. We analysed solution-based hA–chaperone interactions using a time-dependent CD assay, which measures hA β-sheet formation, and a ThT (thioflavin-T) assay that quantifies hA aggregation [18], both key features of hA misfolding. We now report that the ER-resident chaperone GRP78 (glucose-regulated protein of 78 kDa), the cytosolic chaperone HSP70 and the co-chaperone HSP40 separately suppressed misfolding and aggregation of hA. These findings demonstrate the existence of direct functional interactions by which GRP78, HSP70 and HSP40 can suppress hA misfolding. We conclude that these chaperones could well exert the physiologically significant regulation of hA during the processes by which it is synthesized and folded.

EXPERIMENTAL

Materials

Experiments were performed in water (18.2 MΩ·cm resistivity; Millipore). All reagents were of analytical grade, unless otherwise stated. Synthetic hA and rA were from Bachem. hA and rA were column-purified to give monomeric amylin as described previously [54]. Peptides thus treated are designated ‘purified’. Recombinant HSP70, HSP40 and GRP78 were from Stressgen. Recombinant human HSP70 and human HSP40 were in 1 × DPBS [2.7 mM KCl, 1.5 mM KH2PO4, 137 mM NaCl, 8.1 mM Na2HPO4 (pH 7.2–7.4)], and recombinant hamster GRP78/BiP (immunoglobulin heavy-chain-binding protein) in 37 mM Tris/HCl (pH 7.5) and 37 mM NaCl. BSA (low endotoxin; ICPbio) was dissolved in water (1 mg/ml stock).

Methods

CD

CD spectra were collected (π*−180; Applied Photophysics) as described previously [18]. Chaperone effects on amylin folding in solution (by measuring the rate of β-sheet formation at 217 nm/s for 1 h) were performed in buffer [100 mM KCl and 50 mM KH2PO4 (pH 7.4)] with purified hA or rA, relevant controls at the stated concentrations, molar ratios and other conditions, and without or with Mg3ATP (Sigma), as described in the Figure legends. Kinetic experiments were initiated by the addition of purified amylin (6.4 μM) and 2.5% (v/v) HFIP (hexafluoroisopropanol) (final concentration) to the buffer containing chaperones and Mg3ATP, where indicated. Chaperones were used at concentrations at which they did not contribute to any observed CD.

ThT assay

ThT fluorescence intensity correlates directly with hA aggregates in solution [18]. End point fluorescence signals were collected (Spectra-MAX Gemini-XS; Molecular Devices) as described previously [55]. Chaperone/hA molar ratios were constituted by adjusting chaperone concentrations, and duplicate group measurements were made.

Chaperone binding to amylin by co-immunoprecipitation

Purified amylin was diluted in buffer [100 mM KCl and 50 mM KH2PO4 (pH 7.4) per 1500 μl] and then hA or rA (6.4 μM, final concentration) and 2.5% (v/v) HFIP (final concentration) was added, followed by HSP70 or GRP78/BiP (0.032 μM, final concentration), incubated for 1 h at ambient temperature before the addition of a rabbit anti-amylin antibody (1:1000 dilution; in-house preparation). Following incubation for 24 h at 4 °C with gentle rotation, 50 μl of Protein G–Sepharose™ fast-flow (GE Healthcare) was added and a second 24-h incubation was performed (under the same conditions). Amylin–chaperone complexes were collected by centrifugation (12000 g for 20 s). Pellets were washed three times in 1 ml of lysis buffer [500 mM NaCl, 1% NP-40 (Nonidet P40), 50 mM Tris/HCl (pH 8.0) and 1 mM PMSF] then once with 50 mM Tris/HCl (pH 8.0), and retrieved by centrifugation (12000 g for 20 s). Pellets were suspended in 20 μl of reducing buffer [1% (w/v) SDS, 100 mM DTT (dithiothreitol) and 50 mM Tris/HCl (pH 7.5)], heated (95 °C for 3 min) and the mixture was then centrifuged (12000 g for 20 s).

Western blotting

Immunoprecipitates (10 μl) were mixed with NuPAGE® LDS sample agent 10 × (2 μl) (Invitrogen) and 1 M DTT (1 μl). Samples and SeeBlue® (5 μl) plus pre-stained standards (Invitrogen) were loaded on to a 4–12% Bis-Tris gradient gel (NuPAGE® Novex; Invitrogen) in 1 × Mes buffer (Invitrogen) and developed (120 V/70 mA for 35 min). Proteins were transferred {30 V/110 mA in transfer buffer [100 mM Tris, 80 mM glycine and 20% (v/v) methanol] for 1.25 h} on to a pre-activated (100% methanol) PVDF membrane (GE Healthcare). Membranes were blocked in PBS containing 0.05% Tween 20 and 5% (w/v) non-fat milk for 2 h at ambient temperature and washed three times for 10 min each in PBS containing 0.05% Tween 20. A goat anti-HSP70 antibody (sc-24, 1:200 dilution; Santa Cruz Biotechnology) and a goat anti-GRP78 antibody (sc-1050, 1:200 dilution; Santa Cruz Biotechnology) were diluted in PBS containing 0.05% Tween 20 and 1% (w/v) non-fat milk and incubated with the membrane for 2 h at an ambient temperature. The washing steps described above were then repeated. Membranes were then incubated with an HRP (horseradish peroxidase)-conjugated donkey anti-(goat IgG) antibody (sc-2033, 1:2500 dilution; Santa Cruz Biotechnology in PBS containing 0.05% Tween 20 and 1% (w/v) non-fat milk for 2 h at an ambient temperature. Membranes were then washed three times for 10 min each as above. ECL (enhanced chemiluminescence) detection reagent (GE Healthcare) was added, membranes stood for 1 min and signals were detected (lumi-film; Roche).

Curve fitting and statistical analysis

Kinetic analysis on CD spectra were performed using four-parameter logistic curve-fits (Prism 4; GraphPad). Aggregation rates in the ThT assay were determined from the slopes fitted during the exponential phase. Statistical analysis (Prism 4) was performed according to tests indicated in the Figure legends, and P values <0.05 were considered significant.

RESULTS AND DISCUSSION

The presence of misfolded hA is a common observation in β-cell pathobiology, particularly in T2DM. Although pure hA is well known to spontaneously misfold and aggregate in solution, it does not do so in healthy humans. As the aggregation process is known to be extremely sensitive to amyloid concentration, it is possible that the pathobiology associated with hA aggregation is itself associated with hA overexpression [56].

Molecular chaperones were originally identified by their increased abundance following heat shock [45,46] and bind non-native states of other proteins to assist them to achieve native conformations. They act by recognizing exposed hydrophobic surfaces on misfolded proteins, which will ultimately be buried in the correctly folded state, and form non-covalent interactions with them. This stabilizes misfolded proteins and prevents irreversible multimeric aggregation [46]. Polypeptides are then released and permit subsequent steps of physiological polypeptide folding or modification to occur [46]. As islet amyloid formed from hA comprises irreversible multimeric aggregates of misfolded hA [4,57], we hypothesized that chaperones might play some role in modulating its misfolding. We chose to study chaperones that are known to be expressed in different organelles in islet β-cells, where they have evident biological roles.

hA misfolding and aggregation in solution

Consistent with previous reports, purified hA misfolded in aqueous solution, as demonstrated by the time-dependent increases in the amounts of β-sheet formation, as measured by CD (Figures 1A and 1B). The aggregation kinetics are dependent on HFIP concentration [54], and we observed that appreciable aggregation only occurred in 2.5% (w/v) HFIP (final concentration) (Figures 1A and 1B). Non-fibrillogenic rA was used as a control in all experiments and has a random coil structure, which is the structure adopted by hA in its native conformation (results not shown), indicating that HFIP at all of the concentrations employed had no effect on the observed data.

Figure 1 hA underwent time-dependent misfolding and aggregation in aqueous solution

Purified hA was diluted in buffer [100 mM KCl and 50 mM KH2PO4 (pH 7.4)] to 6.4 μM (final concentration) with 2.5% (v/v) (black line), 4.0% (v/v) (dashed) or 8.0% (v/v) (dotted) HFIP. (A) Molar ellipticity was measured by CD at 217 nm every 1 s for 1 h. (B) Corresponding CD spectra after completion of 1 h of kinetic measurements. Curves in (A) and (B) are representative of three independent experiments. (C) ThT assay for time-dependent hA aggregation. Fluorescence was measured in solutions containing 2.5% (v/v) HFIP, 6.4 μM hA and 10 μM ThT (final concentrations), and results are expressed as the means±S.E.M. of three independent experiments.

Amylin aggregation was also measured using the complementary ThT assay, where hA preparation and experimental conditions were identical with those used in the CD experiments. Although there was some variability in the extent of the ‘lag phase’, hA nevertheless aggregated progressively during the 1 h observation period with a generally pronounced ‘lag phase’, lasting for typically 1200 s, followed by an ‘exponential phase’ (Figure 1C). Under these experimental conditions, both the CD and ThT assays were mutually consistent in that the progressive aggregation indicated in the ThT assay began only after a completed β-sheet spectrum was observed by CD (Figures 1A and 1C).

Prior to these experiments, purified synthetic hA (Bachem AG) was stabilized as a stock solution in 100% (w/v) HFIP to prevent self-aggregation and then this solution was diluted into the reaction buffer, which did not contain HFIP. By adopting this preparation method [54], larger oligomeric seeds could be eliminated and the rate of hA β-sheet aggregate formation standardized between experiments [54]. rA was prepared in an identical fashion. Our results are comparable with those described previously [54] in that column-purified hA can progressively and reproducibly self-aggregate into β-sheet-containing structures within a 1-h time frame when dissolved in phosphate buffer. This spontaneous time-dependent β-sheet formation and aggregation by purified hA was not observed for its non-fibrillogenic rodent homologue rA. Minor CD spectral deviations between hA and rA, such as seen in Figure 5(B), point to subtly altered hA conformations [17,18].

HSP70 and HSP40 suppressed hA β-sheet formation and independently cause concentration-dependent suppression of aggregation in solution

HSP70 assists in the stabilization and folding of many substrates in an ATP-dependent fashion. We found that HSP70 effectively abolished β-sheet formation at an HSP70/hA molar ratio of 1:200 (Figures 2A–2C, red traces and bar), but not at a chaperone/hA molar ratio of 1:2000 (Figures 2A–2C, blue traces and bar). Furthermore, the inhibition observed at the 1:200 ratio occurred in the absence of ATP, showing that a nucleotide-dependent folding cycle was not intrinsically required. In support of this conclusion, we found no evidence for ATP turnover with HSP70 chaperone/hA molar ratios ranging from 1:1 to 1:100 using a luminescent kinase assay (kinase-Glo; Promega) (results not shown). HSP70 did not modify the CD spectrum of rA solutions (Figure 2B, grey and purple traces).

Figure 2 HSP70 with or without HSP40 suppressed hA β-sheet formation

(A) Recombinant HSP70±HSP40 were mixed in buffer [100 mM KCl and 50 mM KH2PO4 (pH 7.4)] with purified hA (6.4 μM, final concentration) or rA (6.4 μM) at molar concentration ratios to hA as indicated below±ATP (25.6 μM). β-Sheet formation was measured at 217 nm over 1 h. (B) CD spectra obtained at the end of 1 h of kinetic measurements. Curves shown in (A) and (B) are each representative of three independent experiments. (C) Rates of β-sheet formation. If the curves had a visible ‘lag phase’, then the slopes of the ‘lag phase’ were calculated from best-fit curves between t=0 and t=1000 s and the late phase between t=1000 s and t=3000 s. If there was no ‘lag phase’, then best-fit lines were used to measure rates of β-sheet formation in each experiment between t=0 and t=1800 s. Experimental groups are: hA, black; HSP70/hA (1:200), red; HSP70/hA (1:2000), blue; HSP70/HSP40/hA (1:1:2000), green; HSP70/HSP40/hA (1:1:2000) + ATP, orange; rA, grey; and HSP70/rA (1:200), purple. rA data are only shown in (B). Bars are means±S.E.M. (n=4); **P<0.01 (unpaired two-tailed Student's t test).

HSP40 acts as a co-chaperone to HSP70, where it assists the rate-limited ATP hydrolysis step of HSP70. To determine whether HSP40 could act synergistically with HSP70, it was added to reaction mixtures containing HSP70 and hA. In the absence of ATP and under conditions where there was no effect on β-sheet formation by HSP70 (Figures 2A–2C, blue traces and bar), the addition of HSP40 (total chaperone/hA ratio of 1:1000) resulted in a reproducible ‘lag phase’ in the kinetic time course (Figures 2A and 2C, green trace and bar). After that, the rate of β-sheet formation was not significantly different from those in samples that did not contain chaperones (Figure 2C). A CD spectrum confirmed the presence of β-sheet structure at the end of the time course study (Figure 2B, green trace).

These results showed that HSP70 and HSP40 together attenuated hA β-sheet formation by introducing a ‘lag phase’, which occurred in the absence of ATP. In addition, HSP40 could function intrinsically as a chaperone in its own right, as it inhibited hA β-sheet formation in solution at a 1:200 molar ratio (Figures 3B and 3C, blue trace and bar). This inhibition was still evident after 16.5 h (Figure 3B, orange trace) and was concentration-dependent, as it was attenuated when the HSP40/hA ratio was increased to 1:2000 (Figures 3A–3C, green traces and bar). We found no evidence for the involvement of an ATP-dependent cycle, as there were no significant changes in the kinetic traces in the presence of ATP (Figures 2A and 2C, orange trace and bar). These findings show that the modulating effects of HSP70 and HSP40 on β-sheet formation occur intrinsically through a chaperone function independent of nucleotide turnover.

Figure 3 HSP40 alone inhibited β-sheet formation by hA in solution

(A) Recombinant HSP40 protein was mixed with purified hA (6.4 μM, final concentration) in buffer [100 mM KCl and 50 mM KH2PO4 (pH 7.4)] at the chaperone/hA molar concentration ratios as indicated and molar ellipticity was measured at 217 nm. (B) Corresponding CD spectra were obtained after 1 h of kinetic measurements, unless otherwise indicated. Curves shown in (A) and (B) are representative of three independent experiments. (C) Rates of β-sheet formation. Best-fit lines between t=0 s and t=1800 s were used to measure rates at 217 nm. Experimental groups are: hA, black; HSP40/hA (1:2000), green; HSP40/hA (1:200), blue (B and C); and HSP40/hA (1:200) at 16.5 h after reaction initiation, orange (B). Bars are means±S.E.M. (n=4); **P<0.01 (unpaired two-tailed Student's t test).

Although either HSP70 or HSP40 completely suppressed β-sheet formation at molar ratios to hA of 1:200, when hA aggregation was measured using the complementary ThT assay the rate of fluorescence enhancement over 1 h was relatively unaffected by either chaperone at this molar ratio (Figures 4A and 4B). However, HSP70 but not HSP40 exerted significant suppression at a 1:100 ratio (Figures 4C and 4D), whereas HSP40 was only effective at a 1:50 ratio (Figures 4E and 4F). At these higher molar ratios, it is possible that the chaperones were attenuating the ‘lag phase’. There was no additional suppression of hA aggregation when both chaperones were used at the indicated chaperone/hA ratios.

Figure 4 HSP70 and HSP40 independently inhibited hA aggregation as determined by ThT fluorescence

Solutions were constituted to contain purified hA (6.4 μM, final concentration) and ThT (10 μM) with recombinant HSP70 (Δ), HSP40 (∇) or both chaperones (○) at the stated chaperone/hA molar concentration ratios, or neither chaperone (■) in buffer [100 mM KCl and 50 mM KH2PO4 (pH 7.4)], and fluorescence was recorded every 10 min for 1 h. (A, C and E) Single representative curves done in triplicate for the stated chaperone/hA molar ratios. (B, D and F) Corresponding composite data comprising best-fit slopes of the ‘exponential phases’ as estimates of rates of hA aggregation. Chaperone/hA ratios were 1:200 in (A and B) (n=3); 1:100 in (C and D) (n=4); and 1:50 in (E and F) (n=4). Bars are means±S.E.M; *P<0.05 and **P<0.01 (unpaired two-tailed Student's t tests).

GRP78 concentration-dependent suppression of hA misfolding and aggregation in solution

Addition of GRP78 to hA in solution at a molar ratio of 1:200 completely suppressed β-sheet formation by hA (Figure 5B, red trace), similar to the actions of HSP70 and HSP40. In the absence of ATP and at lower GRP78 concentrations (GRP78/hA molar ratio of 1:2000), there was a reproducible ‘lag phase’, followed by a late ‘exponential phase’ (Figures 5A–5C, blue traces and bar). There was no additional attenuating effect following the addition of ATP (Figures 5A–5C, orange traces and bar). Interestingly, the CD spectrum of GRP78-treated hA was not identical with that of equivalently treated rA over the region spanning 200–240 nm. rA either in the absence (Figure 5B, grey trace) or presence (Figure 5B, purple trace) of GRP78 had a random coil conformation, while the GRP78–hA spectrum (Figure 5B, red trace) indicated a residual secondary structure, suggesting that GRP78 did not return hA completely to its native random coil conformation.

Figure 5 GRP78 caused ATP-independent abrogation of β-sheet formation by aqueous hA

(A) Recombinant GRP78 was combined with purified amylin preparations (6.4 μM, final concentration) at the stated molar chaperone/hA ratios±ATP (25.6 μM, final concentration) in buffer [100 mM KCl and 50 mM KH2PO4 (pH 7.4)], and molar ellipticity was measured over 1 h at 217 nm. (B) Corresponding CD spectra. Curves in (A) and (B) are each representative of three independent experiments. (C) Rates of β-sheet formation (n=3). If the curves had a visible ‘lag phase’ then the slopes of the ‘lag phase’ were calculated from best-fit curves between t=0 and t=600 s and the late phase between t=600 s and t=3000 s. If there was no ‘lag phase’, then best-fit lines were used to measure rates of β-sheet formation in each experiment between t=0 s and t=1800 s. Experimental groups are: hA, black line; GRP78/hA (1:200), red (B); GRP78/hA (1:2000), blue; GRP78/hA (1:2000) + ATP, orange; rA, grey; and GRP78/rA (1:200), purple [rA data are only shown in (B)]. Bars are means±S.E.M; ***P<0.001 (unpaired two-tailed Student's t test).

Similar to the findings with HSP70, GRP78 suppressed hA aggregation as measured by the ThT assay at a molar ratio of 1:100, but not 1:200 (Figure 6). However, GRP78 may manifest a slightly more effective chaperone function on hA than HSP70, as it completely suppressed hA aggregation in the ThT assay at a molar ratio of 1:100. None of the chaperones used in the present study exerted any direct effects on ThT and so did not influence the fluorescence derived from hA aggregation in these assays (results not shown).

Figure 6 GRP78 inhibited hA aggregation as determined by ThT fluorescence

Purified hA (6.4 μM, final concentration) was dissolved in buffer [100 mM KCl and 50 mM KH2PO4 (pH 7.4)] with ThT (10 μM), and with GRP78 at molar chaperone/hA ratios of 1:200 (Δ) or 1:100 (∇), or without GRP78 (■) (n=6/condition), and fluorescence signals were recorded every 10 min for 1 h. (A) A single representative experiment done in triplicate. (B) Rates of hA aggregation were determined by ThT fluorescence. Slopes from best-fit lines were determined to measure aggregation in replicates of hA, n=13; GRP78 (1:200), n=8; and GRP78 (1:100), n=5. Bars are means±S.E.M.; **P<0.01 (unpaired two-tailed Student's t test).

As a control, BSA was used with BSA/hA molar ratios of 1:200 (Figures 7A and 7B) and 1:50 (Figures 7C and 7D) to coincide with the chaperone/hA assays employed as described above. BSA, not known as a chaperone, neither modified the rates of hA aggregation as determined by CD spectroscopy (Figures 7A and 7B) nor influenced the rate of hA-mediated change in ThT fluorescence (Figures 7C and 7D). Likewise, it had no direct effect itself in either assay.

Figure 7 BSA did not modify misfolding or aggregation of hA

(A) Aqueous BSA was added to hA (6.4 μM, final concentration) in 2.5% (v/v) HFIP at the indicated molar concentrations and molar ellipticity was measured at 217 nm. hA, black line; BSA/hA (1:200), grey line. (B) Rates of β-sheet formation. Best-fit lines between t=0 and t=1800 s were calculated (n=3). (C) A representative experiment showing aggregation as determined by ThT fluorescence. Purified hA (6.4 μM, final concentration) was dissolved in 100 mM KCl and 50 mM KH2PO4 (pH 7.4) with ThT (10 μM) in the absence (■) or presence (Δ) of BSA at a BSA/hA molar ratio of 1:50 (n=6 individual wells for each data point). (D) Rates of hA aggregation as determined from three independent experiments. Slopes of the exponential phase of fluorescence–time curves were determined (best-fit lines between t=1000 s and t=3600 s). (A and C) Single representative experiments. Data in (B) and (D) are means±S.E.M. (differences were analysed with unpaired two-tailed Student's t tests).

Co-immunoprecipitation of hA with HSP70 or GRP78

We investigated whether either of these chaperones could directly bind to hA with the formation of stable complexes. Each chaperone was independently incubated with hA or rA as a control under conditions where chaperones inhibited hA misfolding (1:200 chaperone/amylin molar ratio in each case). Amylin–chaperone complexes were then immunoprecipitated with an anti-amylin antibody that cross-reacts with both amylin forms and immunoprecipitates were analysed by Western blotting with anti-GRP78 and anti-HSP70 antibodies (Figure 8). The overexposed blots show that HSP70 and GRP78 were detectable in hA immunoprecipitates, but not in those formed by the rA samples. Non-specific bands in lanes 3 and 4 could arise from other species in the purified chaperone preparations or from the rabbit in-house polyclonal antibody that was used for the immunoprecipitations. These findings show that chaperone-binding interactions must be due to the amyloidogenic properties of hA.

Figure 8 HSP70 and GRP78 were co-immunoprecipitated with hA

Purified hA or rA were diluted in buffer [100 mM KCl and 50 mM KH2PO4 (pH 7.4)] to final concentrations of 6.4 μM with 2.5% (v/v) HFIP and incubated with HSP70 or GRP78 at a 1:200 chaperone/amylin molar ratio. Co-immunoprecipitation and Western blotting were then performed, as described in the Experimental section. A single blot representative of three independent experiments is shown. Lane 1, hA + HSP70; lane 2, hA + GRP78; lane 3, rA + HSP70; and lane 4, rA + GRP78.

Fibril formation by hA is thought to occur through a self-assembled nucleation process involving small aggregates. These then attract free monomers and elongate into amyloid fibrils [58,59]. Our present findings show a suppression of β-sheet formation at sub-stoichiometric chaperone/hA molar ratios. Consequently, these chaperones must interact with small ‘on-pathway’ oligomers instead of monomers. This may suppress the initiation and progression of bulk hA aggregation into higher-order fibril structures. The ‘lag phases’ observed in the presence of limiting concentrations of chaperones were also consistent with the attenuation of these initial nucleation processes.

Immunoprecipitation experiments performed under the same conditions at which HSP70 and GRP78 suppressed β-sheet formation of hA peptides provided further evidence that this suppression was due to direct binding interactions. By contrast, none of these chaperones had observable effects on the secondary structure of rA, pointing to a fundamental difference between chaperone–rA and chaperone–hA interactions.

Interestingly, higher molar concentrations of chaperones were required to inhibit hA aggregation as measured by the ThT assay compared with β-sheet formation as measured by CD. ThT is a benzothiazole dye that binds to amyloid fibrils as micelles and leads to an enhancement in fluorescence emission [60]. Amylin aggregation may produce both ‘on-pathway’ and ‘off-pathway’ nucleation species, which lead to fibril elongation and other higher-order assemblies respectively [61]. The existence of multiple species has also been proposed for other amyloidogenic proteins [36]. ThT-enhanced fluorescence could therefore reflect the sum of all higher-order assemblies. It is possible that relatively higher concentrations of chaperones are required to bind and sequester ‘off-pathway’ species and therefore attenuate ThT-enhanced fluorescence.

There are parallels between our findings and other amyloid pathophysiologies. Dedmon et al. [37] demonstrated that α-synuclein aggregation, the primary constituent of amyloid plaque in Parkinson's disease, could also be attenuated solely by HSP70 alone at sub-stoichiometric molar concentrations. The inhibitory effects of HSP70 on amyloidogenicity of peptides associated with neurodegenerative diseases in the absence of both co-chaperone and ATP were also demonstrated by its homologues in Escherichia coli DnaK or HSC70 (heat-shock cognate 70 kDa protein). Here, there was an inhibition of insoluble aggregate formation of the huntingtin peptide HD53Q (containing 53 glutamine repeats) in vitro [42]. Although direct binding interactions were not shown, HD53Q was co-immunoprecipitated with the HSP70 homologue in Saccharomyces cerevisiae Ssa1 from a transgenic yeast model [42].

Within the context of other associated studies, direct binding of HSP70 or GRP78/BiP to other amyloidogenic peptides has been demonstrated, including Ssa3/α-synuclein in yeast [62], HSP70/human mutant SOD-1 (superoxide dismutase-1) in NIH-3T3 cells [63], either HSP70 or HSC70 and HD150Q in mouse Neuro2A cells [41], and GRP78/BiP and Aβ (amyloid β) [39]. All of these chaperones again attenuated the formation of amyloid-like aggregates. In addition, a mutant form of HSP70, which lacked the ATPase domain, also bound to the huntingtin peptide HD150Q (containing 150 glutamine repeats) and prevented aggregation [41], showing further that ATP hydrolysis was not required for this inhibition.

GRP78 is mainly ER-resident [64], and its ability to suppress hA misfolding points to the ER as one organelle where chaperone–amylin interactions could have physiological relevance. Interestingly, GRP78/BiP was reportedly elevated in the islet β-cells of an hA-transgenic mouse model of diabetes concomitantly with other ER-stress markers [52]. Taken together with our present results, that finding suggests that a possible compensatory increase in GRP78/BiP was occurring in this transgenic model, and points to a potential role for GRP78/BiP in the physiological regulation of hA folding.

The research hypothesis underlying the present study was that hA amyloidogenicity can be inhibited by HSP70 or GRP78/BiP. Molecular chaperones are able to eliminate misfolded/unfolded peptides and consequently prevent peptide aggregation [64,65]. Our results are therefore consistent with the hypothesis that a diminished availability and/or functionality of β-cell chaperones could be a factor contributing to hA amyloidogenesis, thereby contributing to the initiation of apoptosis, either through cell-membrane-mediated [21,66] or ER-mediated [52,53] mechanisms, or perhaps a combination of both. It is clearly crucial to understand hA misfolding in the context of the molecular and cellular regulation of β-cell chaperone biology in T2DM.

AUTHOR CONTRIBUTION

Vita Chien, Jacqueline Aitken, Shaoping Zhang, Christina Buchanan, Anthony Hickey, Thomas Brittain, Garth Cooper and Kerry Loomes designed the research, Vita Chien performed the research, and Vita Chien, Garth Cooper and Kerry Loomes wrote the paper.

FUNDING

This work was supported by the Endocore Research Trust [grant number 60187]; the Health Research Council of New Zealand [grant number HRC06–190]; the Foundation for Research, Science and Technology (FRST, New Zealand) [grant number PMIX0201]; the Maurice & Phyllis Paykel Trust (equipment); NZ Lottery Grants Board (equipment); and the University of Auckland (equipment) [grant number TRIF 60178].

Acknowledgments

We thank P.J. Scott for his ongoing encouragement and advice, M. Stojkovic for anti-amylin antibody production, and technical support from C. Tse.

Abbreviations: BiP, immunoglobulin heavy-chain-binding protein; DTT, dithiothreitol; ER, endoplasmic reticulum; GRP78, glucose-regulated protein of 78 kDa; hA, human amylin; HFIP, hexafluoroisopropanol; HSC70, heat-shock cognate 70 kDa protein; HSP, heat-shock protein; rA, rat amylin; T2DM, Type 2 diabetes mellitus; ThT, thioflavin-T

References

View Abstract