Early Events in the Fibrillation of Monomeric Insulin* Received for publication, April 20, 2005, and in revised form, October 4, 2005 Published, JBC Papers in Press, October 24, 2005, DOI 10.1074/jbc.M504298200 Atta Ahmad, Vladimir N. Uversky, Dongpyo Hong, and Anthony L. Fink1 From the Department of Chemistry and Biochemistry, University of California, Santa Cruz, California 95064 Insulin has a largely -helical structure and exists as a mixture of hexameric, dimeric, and monomeric states in solution, depending on the conditions: the protein is monomeric in 20% acetic acid. Insulin forms amyloid-like fibrils under a variety of conditions, especially at low pH. In this study we investigated the fibrillation of monomeric human insulin by monitoring changes in CD, attenu- ated total reflectance-Fourier transform infrared spectroscopy, 8-anilinonaphthalenesulfonic acid fluorescence, thioflavin T fluo- rescence, dynamic light scattering, and H/D exchange during the initial stages of the fibrillation process to provide insight into early events involving the monomer. The results demonstrate the exist- ence of structural changes occurring before the onset of fibril for- mation, which are detectable by multiple probes. The data indicate at least two major populations of oligomeric intermediates between the native monomer and fibrils. Both have significantly non-native conformations, and indicate that fibrillation occurs from a beta- rich structure significantly distinct from the native fold. A number of human diseases are caused by the pathogenic deposition of proteins in the form of amyloid-like fibrils (1���7). Several non-patho- genic proteins and peptides also undergo amyloid like fibril formation on destabilization of their native state (7���10). The fact that structurally and sequentially non-homologous proteins are able to self-assemble into fibrils possessing similar morphology (e.g. 10���18 nm width, bire- fringence to polarized light, and cross- structure) suggests a common molecular mechanism in the fibrillation pathways. A variety of hypoth- eses for the mechanism of fibril formation have been proposed. Insulin is a 51-residue hormone with a largely -helical structure. It exists as a mixture of hexameric, dimeric, and monomeric states in solution, with the relative population of different oligomeric species being strongly dependent on the environmental conditions: the protein is predominantly monomeric in 20% acetic acid, dimeric in 20 mM HCl, and hexameric at pH 7.5 in the presence of zinc. Insulin forms amyloid- like fibrils under a variety of conditions (11���13), with various overall morphologies depending on the arrangement of constituent protofila- ments (14, 15). Insulin fibrils pose a variety of problems in biomedical and biotechnological applications. Amyloid deposits of insulin have been observed in patients with diabetes after repeated injection and in normal aging, as well as after subcutaneous insulin infusion (16, 17). In our previous work we have shown the important role of partially folded intermediates in insulin fibrillation in vitro (13, 18���20). In fact, our studies showed that insulin, which is a hexamer at physiological pH, undergoes rapid fibrillation starting at relatively low concentrations of guanidine hydrochloride and urea. The predominant species character- ized by various biophysical techniques under these conditions was shown to be a partially folded, expanded monomer, which is present in solutions containing up to 6 M guanidine hydrochloride or 8 M urea (18, 19). Furthermore, the dissociation of the hexamer under these condi- tions was found to be the major rate-limiting step. Interestingly, in the presence of 20% acetic acid i.e. under the conditions where insulin exists as a native monomer, the addition of guanidine hydrochloride was shown to result in the formation of a partially folded expanded confor- mation with high propensity to fibrillate (18). However, the addition of urea under the same conditions slowed down the rate of fibrillation ( 5-fold at 0.75 M urea), which has been explained by the appearance of non-native conformation with significantly increased -helical content compared with the native form of the protein (19). It was very interest- ing to observe the effective formation of fibrils even at high denaturant concentration where the protein exists in an extensively unfolded con- formation (18, 19). All these observations led us to assume that the pathway for insulin fibrillation can be described by a simple model hex- amer 3 modified monomer 3 fibrils (13, 18���20). The present report, in conjunction with earlier studies on insulin fibrillation (11���13, 18���20), provides further evidence for the relationship between the con- formational changes and the propensity of insulin to fibrillate. To this end the changes in biophysical properties (detected by CD, ATR-FTIR,2 dynamic light scattering, ANS fluorescence, and H/D exchange mass spectrometry) accompanying the fibrillation of monomeric insulin (monitored by ThT fluorescence) have been analyzed as a function of time to provide insight into the very early events taking place in the monomer before it undergoes fibril formation. The results demonstrate the existence of significant structural changes occurring in monomeric insulin before the onset of fibril formation, which correspond to at least two major populations of intermediates. Both have significantly non- native conformations and indicate that fibrillation occurs from a -rich structure significantly distinct from the native fold. EXPERIMENTAL PROCEDURES Materials���Human insulin with zinc was kindly provided by Novo Nordisk, Copenhagen, Denmark. All the chemicals were analytical grade, and were from EM Sciences or Sigma. Thioflavin T was obtained from Fluka. Preparation of Samples���Solutions of monomeric human insulin, 2 and 4 mg/ml, were freshly prepared in 20% acetic acid. The concentra- tion of insulin was determined using an extinction coefficient of 1.0 for 1 mg/ml at 276 nm (14). Stock (1 mM) solutions of Thioflavin T were prepared by dissolving ThT in double-distilled water, and the concen- tration was determined using a molar extinction coefficient of 24,420 M 1 cm 1 at 420 nm. ThT was stored at 4 ��C, protected from light. Incubation of Insulin for Fibrillation���500 l of insulin, 2 or 4 mg/ml in 20% acetic acid, was incubated at 37 ��C in a glass vial on a stirrer with a small magnetic bead spinning at the bottom of the vial at 600 rpm. Aliquots from this solution were taken at desired time intervals. * This work was supported by a grant from UC BioSTAR and Novo Nordisk A/S (to A. L. F.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked ���advertisement��� in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 To whom correspondence should be addressed. Tel.: 831-459-2744 Fax: 831-459-2935 E-mail address: enzyme@ucsc.edu. 2 The abbreviations used are: ATR-FTIR, attenuated total reflectance-Fourier transform infrared spectroscopy ThT, thioflavin T ANS, 8-anilinonaphthalenesulfonic acid DLS, dynamic light scattering H/D exchange, hydrogen/deuterium exchange. THE JOURNAL OF BIOLOGICAL CHEMISTRY VOL. 280, NO. 52, pp. 42669���42675, December 30, 2005 �� 2005 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in the U.S.A. DECEMBER 30, 2005��� VOLUME 280���NUMBER 52 JOURNAL OF BIOLOGICAL CHEMISTRY 42669

ThT Assay for Determining Kinetics of Fibrillation���Free ThT has excitation and emission maxima at 350 and 450 nm, respectively. However, upon binding to fibrils the excitation and emission max change to 450 and 485 nm, respectively (21, 22). 5- l aliquots of sample were added to solution containing 20 M ThT in 20 mM Tris- HCl buffer, pH 7.4, shaken a few times before measuring the fluores- cence emission on a FluoroMax-3 spectrofluorometer from Instru- ments S.A., Inc. Jobin Yvon-Spex, at room temperature. A background fluorescence spectrum obtained by running a blank buffer was sub- tracted from each sample fluorescence spectrum. The excitation wave- length was 444 nm, and the emission was recorded at 482 nm. Fluores- cence intensity at 482 nm was plotted against time, and the kinetic profiles were analyzed by curve fitting, using SigmaPlot software (18). ANS Fluorescence���5- l aliquots from the incubated mixture were added to solutions containing 5 M ANS in 20 mM Tris-HCl buffer, pH 7.4, and the fluorescence was measured with a FluoroMax-3 spectroflu- orometer at room temperature. The excitation wavelength was 350 nm, and the emission was measured at 460 nm. The values of fluorescence intensity at 462 nm and the values of maximal wavelength of the emis- sion spectra were plotted against time. The profiles were fitted using curve fitting with SigmaPlot. Circular Dichroism���40 l of the aliquot were placed in a cuvette with 0.1 mm path length, and CD spectra were collected on an AVIV 60DS CD spectrophotometer (Aviv Associates, Lakewood, NJ) at 25 ��C. A CD spectrum of the buffer was subtracted from the sample spectra for background correction. Spectra were recorded using a step size of 0.5 nm and a bandwidth of 1.5 nm. Phase diagrams were plotted as previ- ously described (18, 19). ATR-FTIR���FTIR spectra were recorded on a ThermoNicolet Nexus 670 FTIR spectrophotometer from 4000 to 400 cm 1 using a resolution of 2 cm 1 and an accumulation of 256 scans. 40 l of the aliquot from the incubation mixture (4 mg/ml insulin at 37 ��C with continuous stir- ring) was withdrawn at various intervals of time and spread uniformly on the surface of a germanium crystal using nitrogen to form a hydrated thin film. The system was continuously purged with dry nitrogen. Back- ground and water vapor subtractions were performed until a straight baseline was obtained between 2000 and 1750 cm 1. Curve fitting of the amide I regions (raw spectra) was performed using GRAMS 32. Second derivative and Fourier self-deconvolved spectra were used as a peak position guide for the curve fitting procedure. The phase diagrams were plotted based on the extension of the concept described in the circular dichroism section above. H/D Exchange Electrospray Ionization Mass Spectrometry Measurements���The mass spectra were recorded on a Quattro II mass spectrometer (Micromass, Altrimcham, UK) equipped with an electro- spray ionization ion source. The sample solution was introduced into the ion source using a Harvard Apparatus model 11 syringe pump. To improve the sensitivity and stability of the ion beam, the source was operated at 80 ��C. The electrospray capillary was set at 3.5 kV, and the cone voltage was 20 V. Nitrogen was used as the nebulizing and drying gas at a flow rate of 15 and 300 liters/h, respectively. The spectra were acquired and processed using MassLynx software supplied with the instrument. The mass spectrum was scanned from m/z 500 to 2200 in 6 s, and the spectra presented are averages of five spectra. The charge states were determined by MassLynx software, and the deconvolution of the spectra was carried out using MaxEnt software. Stock insulin solution of 10 mg/ml was prepared in 20% acetic acid. Aliquots from this were added to obtain final concentration of 4 mg/ml insulin in 20% acetic acid with 5, 10, and 60% of D2O, respectively. The samples were incubated for 0, 6, 12, and 24 h and diluted 100 times before injecting into the mass spectrometer. Monitoring Monomeric and Oligomeric Insulin by Dynamic Light Scattering���DLS measurements were performed at 20 ��C using a DynaPro (Wyatt Technology) model 99-E-50 instrument. Aliquots of 10 l of the incubation solution (4 mg/ml insulin at 37 ��C) were centri- fuged before measuring DLS. 1 l of the supernatant was diluted with 10 l of filtered 20% acetic acid buffer in the DLS cuvette, and measure- ments were taken within 3���7 min. Scattering peaks of 0.1 nm radius and more than 1000 nm were ignored. RESULTS Changes in the conformation of insulin were monitored by a variety of probes during incubation of monomeric insulin under conditions leading to fibrillation. By starting with monomeric insulin complica- tions due to associated states of native insulin (such as the hexamer or dimer) were avoided. The apparent pH of a 20% acetic acid solution is 1.8. Such low pHs are used in the commercial production of recombi- nant human insulin. Unless otherwise noted the experiments involved incubation of 4 mg/ml insulin in 20% acetic acid at 37 ��C with continu- ous agitation. Fibril Formation Monitored by ThT Fluorescence���Thioflavin T is a fluorescent dye that is frequently used as a specific probe for fibril for- mation in vitro (21, 22). Changes in ThT emission at 482 nm as a func- tion of time of insulin incubation in 20% acetic acid at 37 ��C (i.e. condi- tions where insulin is known to be monomeric) with continuous shaking are shown in Fig. 1. The figure indicates that the increase in ThT fluorescence intensity follows a typical sigmoidal pattern both at 2.0 and 4.0 mg/ml insulin. The initial region with no significant changes in fluorescence corresponds to the nucleation phase, the region showing rapid increase in fluorescence is the elongation phase, and the final plateau region reflects the maturation phase of fibrillation. Analysis of the ThT curves showed that the fibrillation of protein at 4.0 and 2.0 mg/ml is characterized by lag times of 5.0 and 11.4 h, respectively. Thus at 2-fold higher protein concentrations the onset of fibrillation takes 2-fold shorter time than that at the lower concentration. The 2-fold higher fluorescence signal in the case of 4.0 mg/ml insulin solution also indicates a 2-fold higher quantity of fibrils. ANS Fluorescence���ANS is able to bind to solvent-exposed hydro- phobic regions of proteins and has been widely used for the character- ization of the partially folded intermediates (23���25). The excitation FIGURE 1. Fibrillogenesis of insulin as detected by changes in ThT fluorescence. Insulin, 4 mg/ml (filled circles) and 2 mg/ml (open circles), in 20% acetic acid was incu- bated at 37 ��C with continuous agitation, and aliquots were removed every 30 min, and assayed with ThT. Analysis of the kinetics gives a lag time of 11 h and 5 h for 2 and 4 mg/ml insulin, respectively. Early Stages of Insulin Fibrillation 42670 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 280���NUMBER 52��� DECEMBER 30, 2005