De novo Refolding and Aggregation of Insulin in a Nonaqueous

Jun 26, 2008 - β-pleated conformer precipitates from solutions of DMSO-denatured insulin upon dilution with chloroform. DMSO destroys hydrogen bond ...
0 downloads 0 Views 616KB Size
8744

J. Phys. Chem. B 2008, 112, 8744–8747

De novo Refolding and Aggregation of Insulin in a Nonaqueous Environment: An Inside out Protein Remake Aleksandra Fulara,†,‡ Sławomir Wojcik,†,‡ Anna Loksztejn,†,‡ and Wojciech Dzwolak*,†,‡ Institute of High Pressure Physics, Polish Academy of Sciences, Sokolowska 29/37, 01-142 Warsaw, Poland, and Department of Chemistry, UniVersity of Warsaw, Pasteura 1, 02-093 Warsaw, Poland ReceiVed: February 15, 2008

While thermodynamic penalties associated with protein-water interactions are the key driving force of folding, perturbed hydration of destabilized protein molecules may trigger aggregation, which in ViVo often causes cellular and histological damage. Here we show, that the denatured state of an R-helical protein, insulin, converts to a non-native β-sheet-rich structure upon de noVo “refolding” in an anhydrous environment. The β-pleated conformer precipitates from solutions of DMSO-denatured insulin upon dilution with chloroform. DMSO destroys hydrogen bond network of the native protein acting as a strong acceptor of main chain hydrogen bonds. Upon the addition of chloroform, which is a weak hydrogen bond donor per se, competitive hydrogen bonds between DMSO and chloroform are formed. This leads to the release of unfolded insulin molecules. In the absence of water, the imminent saturation of polypeptide’s dandling hydrogen bonds does not produce the native and predominantly R-helical state but a β-sheet-rich structure, which is morphologically and spectrally distinct from insulin amyloid fibrils. Unlike insulin fibrils, the β-sheet conformer is metastable and refolds spontaneously to the native form in an aqueous environment. This implies that “folding” in the absence of water results in inefficient burial of hydrophobic side-chains, and thermodynamic frustration at the water-protein interface. Introduction For free polypeptide chains placed in aqueous environment, folding and formation of singly dispersed compact molecules on the one hand, or aggregation of multiple chains into densely packed fibrils on the other hand, may be conceptualized as two competing strategies minimizing frustration stemming from dangling hydrogen bonds, and solvent-exposure of hydrophobic residues.1 Although unfavorable interactions between nonpolar amino acid side chains and water have been long recognized as the driving force of protein folding,2 recent studies also accentuate the role of perturbed solvation of a protein in its propensity to misfold and aggregate.1,3–5 Insulin has proved to be an excellent model amyloidogenic peptide for many physicochemical studies focused on the role of protein-solvent interactions in misfolding and aggregation.6–10 In this work, conformational fate of insulin random coil state during refolding in an entirely nonaqueous environment has been studied. Unfolding and refolding of insulin were carried out in two organic solvent: DMSO and chloroform, respectively. DMSO is a strong acceptor of hydrogen bonds, which dissolves and denatures proteins through disruption of intramolecular hydrogen bonds.11,12 Chloroform, on the other hand, is a hydrogen bond donor, too weak to interact strongly with the protein, yet readily forming HB-complexes with DMSO.13,14 The “refolding” process was induced by transferring DMSO-denatured insulin to chloroform, which acts as a competitive hydrogen bond donor eluting the denaturant from the protein while maintaining the nonaqueous environment. We are showing that this approach * Corresponding author. Phone: +48 22 888 0237. Fax: +48 22 632 4218. E-mail: [email protected]. † Polish Academy of Sciences. ‡ University of Warsaw.

enables accessing “nonaqueous” folding and aggregation pathways of proteins. Experimental Section For FT-IR measurements, bovine insulin from Sigma was dissolved at 1 wt % concentration in D2O/DMSO mixtures containing 100 mM NaCl and 15 mM DCl, which in DMSOfree samples corresponds to an uncorrected value of pD 1.9. DCl and D2O (either of 99.5% isotope grade) were from ARMAR Chemicals, and pure anhydrous DMSO was purchased from Aldrich. Insulin samples in pure DMSO did not contain NaCl nor DCl. For CD measurements, protein samples were diluted 20 times with appropriate buffers. Other details concerning routine FT-IR, CD, and thioflavin T fluorescence measurements have been described earlier.1,16 For the sake of uniformity of solvent conditions, thioflavin T stained samples of amyloid fibrils, native insulin, and its new conformational variant were suspended in acetone prior to fluorescence measurements. The β-pleated conformational variant of insulin was induced in 5 wt % solutions of bovine insulin in anhydrous DMSO through an abrupt dilution with a 30-fold excess of chloroform. Subsequently, filtered and centrifuged pellets of the precipitated protein were washed with portions of chloroform 5 times in order to remove traces of DMSO. Chloroform-suspended insulin precipitates were analyzed with transmission FT-IR spectroscopy, while vacuum-dried and carbon-sputtered protein films were prepared for SEM imaging on a GEMINI LEO 1530 microscope. For tapping-mode AFM imaging, a small droplet of 5 wt % bovine insulin in DMSO was diluted 2500 times with chloroform. A small amount of the liquid sample was deposited onto mica surface and left to dry up for several hours. Other details were described earlier.1

10.1021/jp8029727 CCC: $40.75  2008 American Chemical Society Published on Web 06/26/2008

Insulin in a Nonaqueous Environment

J. Phys. Chem. B, Vol. 112, No. 29, 2008 8745

Figure 2. The amide I band of the conformer obtained by elution of DMSO-dissolved insulin with chloroform (thick solid line). Peak-fitting enabled estimation of spectral components (shaded areas) assigned to parallel (A) and antiparallel (B, [ν˜ (ο,π)]; C, [ν˜ (π,ο)]) β-sheet structures. The band is overlapped by spectra of the native insulin (dotted line) and insulin amyloid fibrils (dashed line). Inset: Savitzky-Golay 2nd derivative spectrum of the DMSO/chloroform-induced insulin conformer (left panel).

Figure 1. (A) The influence of increasing concentration of DMSO on the infrared amide I band of bovine insulin (1 wt %, 25 °C) dissolved in D2O/DMSO solvent. The spectra are solvent-subtracted. (B) The corresponding peak-positions (diamonds) and HHBW (bars) of the amide I band.

Results and Discussion The spectral changes visible in the conformation-sensitive infrared amide I band region reflect the gradual denaturation of insulin induced by the increasing concentration of DMSO in D2O (Figure 1A). The band remains virtually intact up to 25 wt % concentration of DMSO. However, at even higher concentration of DMSO, nonmonotonic spectral changes take place, which eventually render the peak considerably broadened and blue-shifted compared to the native form (Figure 1 B). The position of the amide I band of insulin in pure DMSO at 1663 cm-1 corresponds to free amide carbonyl groups, which indicates the complete unfolding of the protein.11 Random coil sustained through the interactions with DMSO may be the starting conformation for either folding or aggregation of insulin.15,16 It was thus of interest to see how disabling the denaturing effect of DMSO through elution with a weak and inert toward proteins proton-donor would affect the conformation of insulin. Figure 2 presents an FT-IR spectrum of the precipitate formed within seconds upon diluting 5 wt % solution of bovine insulin in DMSO with a 30-fold excess of chloroform. The position of the amide I band centered around 1631 cm-1 implies that β-sheet is the main secondary structural component of the precipitate.16,17 Through a peak-fitting procedure, the total β-sheet content has been estimated at ca. 55% (Supporting Information), below the level expected for typical insulin amyloid fibrils, wherein the content of extended conformation approaches 85% (Figure 2).

In contrast to insulin amyloid, the blue-shifted position of the amide I band suggests relatively weak hydrogen bonding within the β-pleated conformation.7,17 The inset second derivative spectrum reveals the presence of both parallel (at 1630 cm-1) and antiparallel (1682 cm-1 - [ν˜ (ο,π)], and 1612 cm-1 [ν˜ (π,ο)]) β-sheet components. The latter observation suggests that intermolecular stacking follows the de noVo refolding and precipitation of insulin in the chloroform phase. However, the most important difference between the β-pleated conformer and insulin amyloid concerns the apparent lack of any fibrous morphology. AFM measurements have consistently failed to produce any evidence of fibrils, which was in accordance with negligible thioflavin T fluorescence, a selective probe of amyloid fibrils as shown in Figure 3. At the high chloroform/DMSO ratio used to dilute samples for AFM, only round large shapes are observed, which are likely to reflect artifacts of the surfacetension-driven movement of small protein particles upon solvent evaporation (Figure 3A). On the other hand, a typical SEM image of dry insulin precipitate in Figure 3 B reveals a pattern of cracks lacking any finer topological features that could be associated with the periodic order of fibrils, as was observed in a control AFM experiment on insulin fibrils obtained in aqueous environment8 (Figure 3C). These observations were paralleled by the negligible intensity of fluorescence of thioflavin T-stained β-pleated conformer contrasting with insulin amyloid fibrils, as shown in Figure 3D. The fact that the elution of DMSO with chloroform and following reestablishing of intramolecular hydrogen bonding in the nonaqueous environment of chloroform does not lead to the native R-helical state of insulin, but to a β-sheet structure is interesting in light of the pronounced propensity of insulin amino acid sequence to the former (Supporting Information).18,19 Unlike water, the hydrophobic milieu of chloroform exercises no thermodynamic pressure on a burial of nonpolar amino acid side chains. Should such

8746 J. Phys. Chem. B, Vol. 112, No. 29, 2008

Fulara et al.

Figure 3. AFM (A) and SEM (B) images of the DMSO/chloroforminduced insulin conformer. In a control experiment, amyloid fibrils obtained from insulin aqueous samples incubated at pH 1.9 and 60 °C 8 were assayed for AFM imaging (C). Fluorescence emission spectra of Thioflavin T bound to insulin fibrils (dotted line), native insulin (dashed line), and the DMSO/ chloroform-induced conformer (solid line) are shown (D). Figure 5. Cartoon depicts the cycle of solvent-induced conformational changes leading from the native protein (A) to DMSO-induced random coil (B). Elution of DMSO with chloroform and the formation of hydrogen-bonded complexes between the two solvents is accompanied by the precipitation of the protein and formation of the metastable β-pleated aggregates (C), which spontaneously refold to the native form in an aqueous environment.

Figure 4. FT-IR (A) and far-UV CD (B) spectra of native bovine insulin (solid line) and the DMSO/chloroform-induced insulin conformer (dotted line) dissolved in D2O (A), or H2O (B). In either case pD (pH) of the samples were adjusted at 1.9.

residues be exposed to water, this would cause considerable frustration of water molecules solvating the protein,1 in the end decreasing stability of the β-pleated conformer. This hypothesis has been verified in following experiments wherein the chloroform-precipitated insulin was resuspended in water in order to assess stability of the β-sheet structure in an aqueous environment over time. In a sharp contrast to β-sheet-rich insulin fibrils, the amorphous precipitate dissolved in acidified D2O within seconds. Moreover, as FT-IR and far-UV CD spectra shown in Figure 4 prove, the solvent-induced extended structure refolded to the nativelike conformation with the amide I band shifting to 1653 cm-1, as well as the appearance of double 208/ 223 nm CD band, both of which are hallmarks of helical conformation. It should be stressed that, due to insolubility and strong light-scattering properties, the DMSO/chloroform-induced conformer of insulin is inaccessible to CD spectroscopy.

The observation of a spontaneous transition of β-sheet-toR-helical structure in a globular protein is particularly interesting in light of the well-known thermodynamic stability of non-native β-pleated fibrillar aggregates surpassing that of the native state.15,20–22 This has led to voicing the hypothesis, that aggregated conformations rather than the native state may correspond to the true global energy minimum of a protein.23 While our study (summarized in Figure 5) underscores the importance of avoiding thermodynamic frustration at the water-protein contacts in maintaining the thermodynamic stability of protein aggregates, it also provides a simple twosolvent system enabling studying refolding pathway of denatured globular proteins de noVo in nonaqueous environment. By using the two solvent system, we have managed to overcome the fundamental difficulty facing attempts to study refolding of proteins in nonaqueous environment. The profoundly denaturating effect of dimethyl sulfoxide ensures that the staring conformation of insulin is truly unfolded, therefore less likely to fall into kinetic traps of residual native structure during the following refolding stage. Chloroform, on the other way, is capable of eluting DMSO, while providing a relatively inert (e.g., in terms of hydrogen-bonding properties) nonaqueous environment for the protein. Dramatic perturbations of interactions with surrounding water affect proteins implicated in neurodegenerative diseases, many of which are transmembrane (Aβ-peptide), or membrane-anchored (PrPC - cellular isoform of prion precursor protein).22 Untangling physicochemical relationships between perturbed solvation and protein misfolding is urgently needed, as it may be the key to understanding

Insulin in a Nonaqueous Environment molecular basis of neurotoxicity of aggregated proteins and devising successful therapeutic strategies. Abbreviations AFM, atomic force microscopy; CD, circular dichroism; DMSO, dimethyl sulfoxide; FT-IR, Fourier transform infrared; HB, hydrogen bond; HHBW, half height band width; SEM, scanning electron microscopy. Acknowledgment. Contributions of A.F. and S.W. were equal and were supported with studentship grants from Polish Children’s Fund. Financial support from the Polish Ministry of Education and Science, Grant 77/E-72/SPB/COST/P-04/DWM38/ 2005-2007 is gratefully acknowledged by W.D. We thank Mr. Adam Presz for his kind help with SEM measurements. Supporting Information Available: Peak-fitting spectral data; secondary structure prediction for bovine insulin and proinsulin chains. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Dzwolak, W.; Grudzielanek, S.; Smirnovas, V.; Ravindra, R.; Nicolini, C.; Jansen, R.; Loksztejn, A.; Porowski, S.; Winter, R. Biochemistry 2005, 44, 8948. (2) Anfinsen, C. B. Science 1973, 181, 223. (3) Dzwolak, W.; Ravindra, R.; Nicolini, C.; Jansen, R.; Winter, R. J. Am. Chem. Soc. 2004, 126, 3762. (4) Otzen, D. E.; Sehgal, P.; Nesgaard, L. W. Biochemistry 2007, 46, 4348.

J. Phys. Chem. B, Vol. 112, No. 29, 2008 8747 (5) Smirnovas, V.; Winter, R.; Funck, T.; Dzwolak, W. J. Phys. Chem. B 2005, 109, 19043. (6) Dzwolak, W.; Ravindra, R; Winter, R. Phys. Chem. Chem. Phys. 2004, 6, 1938. (7) Dzwolak, W.; Smirnovas, V.; Jansen, R.; Winter, R. Protein Sci. 2004, 13, 1927. (8) Dzwolak, W.; Jansen, R.; Smirnovas, V.; Loksztejn, A.; Porowski, S.; Winter, R. Phys. Chem. Chem. Phys. 2005, 7, 1349. (9) Smirnovas, V.; Winter, R.; Funck, T.; Dzwolak, W. ChemPhysChem. 2006, 7, 1046. (10) Grudzielanek, S.; Jansen, R.; Winter, R. J. Mol. Biol. 2005, 351, 879. (11) Jackson, M.; Mantsch, H. H. Biochim. Biophys. Acta 1991, 1078, 231. (12) Hirota-Nakaoka, N.; Hasegawa, K.; Naiki, H.; Goto, Y. J. Biochem. 2003, 134, 159. (13) Goates, J. R.; Ott, J. B.; Reeder, J.; Lamb, J. D. J. Chem. Soc., Faraday Trans. 1 1972, 68, 2171. (14) Daniel, D. C.; McHale, J. L. J. Phys. Chem. A 1997, 101, 3070. (15) Millican, R. L.; Brems, D. N. Biochemistry 1994, 33, 1116. (16) Dzwolak, W.; Loksztejn, A.; Smirnovas, V. Biochemistry 2006, 45, 8143. (17) Goormaghtigh, E.; Cabiaux, V.; Ruysschaert, J. M. Subcellular Biochemistry; Hilderson, H. J., Ralston, G. B., Eds.; Plenum Press: New York, 1994; Vol. 23, pp 405-450. (18) Jones, D. T. J. Mol. Biol. 1999, 292, 195. (19) McGuffin, L. J.; Bryson, K.; Jones, D. T. Bioinformatics 2000, 16, 404. (20) Dobson, C. M. Nature 2003, 426, 884. (21) Weissmann, C.; Enari, M.; Klohn, P. C.; Rossi, D.; Flechsig, E. Acta Neurobiol. Exp. 2002, 62, 153. (22) Uversky, V. N.; Fink, A. L. Biochim. Biophys. Acta 2004, 1698, 131. (23) Gazit, E. Angew. Chem., Int. Ed. 2002, 41, 257.

JP8029727