Role of dietary antioxidant (−)-epicatechin in the development of β-lactoglobulin fibrils

Biochimica et Biophysica Acta 1864 (2016) 766–772

Contents lists available at ScienceDirect

Biochimica et Biophysica Acta journal homepage: www.elsevier.com/locate/bbapap

Role of dietary antioxidant (−)-epicatechin in the development of β-lactoglobulin fibrils M. Carbonaro a, A. Di Venere b, A. Filabozzi c, P. Maselli d, V. Minicozzi c,⁎, S. Morante c, E. Nicolai b, A. Nucara d, E. Placidi e, F. Stellato c a

Consiglio per la Ricerca in Agricoltura e l'Analisi dell'Economia Agraria, CREA-NUT, Via Ardeatina 546, 00178 Rome, Italy Dipartimento di Medicina Sperimentale e Chirurgia, Università Degli Studi di Roma “Tor Vergata”, Via Montpellier 1, 00133 Rome, Italy Dipartimento di Fisica, Università degli studi di Roma “Tor Vergata” and INFN, Via della Ricerca Scientifica 1, 00133 Rome, Italy d Dipartimento di Fisica, Sapienza Università di Roma, Piazzale Aldo Moro 2, 00185 Rome, Italy e Istituto di Struttura della Materia, CNR, Via del Fosso del Cavaliere 100, 00133 Rome, Italy b c

a r t i c l e

i n f o

Article history: Received 24 November 2015 Received in revised form 4 March 2016 Accepted 28 March 2016 Available online 2 April 2016 Keywords: β-Lactoglobulin (−)-Epicatechin Fibrils Aggregates

a b s t r a c t Under specific physico-chemical conditions β-lactoglobulin is seen to form fibrils structurally highly similar to those that are formed by the amyloid-like proteins associated with neurodegenerative disorders, such as Alzheimer and Parkinson diseases. In the present study we provide insights on the possible role that the dietary flavonoid (−)-epicatechin plays on β-lactoglobulin fibril formation. Fibril formation is induced by keeping βlactoglobulin solutions at pH 2.0 and at a temperature of 80 °C for 24 h. Atomic Force Microscopy measurements suggest that, by adding (−)-epicatechin in the solution, fibrils density is visibly lowered. This last observation is confirmed by Fluorescence Correlation Spectroscopy experiments. With the use of Fourier Transform IR spectroscopy we monitored the relative abundances of the secondary structures components during the heating process. We observed that in the presence of (−)-epicatechin the spectral-weight exchange between different secondary structures is partially inhibited. Molecular Dynamics simulations have been able to provide an atomistic explanation of this experimental observation, showing that (−)-epicatechin interacts with β-lactoglobulin mainly via the residues that, normally in the absence of (−)-epicatechin, are involved in β-sheet formation. Unveiling this molecular mechanism is an important step in the process of identifying suitable molecules apt at finely tuning fibril formation like it is desirable to do in food industry applications. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Protein aggregation is a crucial process in several phenomena having medical, technological and nutritional implications. Proteins are generally highly susceptible to aggregation in vitro and in vivo. Indeed, accumulation of misfolded proteins, finally leading to the formation of specific aggregates called amyloid fibrils, is a key event in the development of neurological disorders such as Alzheimer, Parkinson, Hungtinton and Creutzfeldt-Jakob diseases [1]. Irrespective of the native protein structure, amyloid fibrils show common features with protein β-strands arranged perpendicular to the elongation axis. However, recent studies suggested that their toxicity may be related to certain differences in the fibril structural and biochemical properties [2,3]. Unfortunately, a precise relation between fibril polymorphism, resulting from aggregation of monomers and oligomers with different conformational properties, and toxicity is difficult to assess [4].

⁎ Corresponding author. E-mail address: [email protected] (V. Minicozzi).

http://dx.doi.org/10.1016/j.bbapap.2016.03.017 1570-9639/© 2016 Elsevier B.V. All rights reserved.

There exist a number of proteins that, though not related to any particular disease, have been recently demonstrated to be able to form, under proper controlled conditions, amyloid-like fibrils structurally similar to those associated with pathological diseases [5,6]. Among these proteins, β-lactoglobulin (BLG), the major protein of whey from bovine milk, is a particularly interesting prototypical model system to study. Indeed, BLG fibrils formation mechanism has been extensively investigated. BLG fibrils are usually produced by heating the protein for 24 h at 80 °C, pH 2.0 and low ionic strength [7,8]. It has been shown that peptides, rather than the intact protein, are the fibrils building blocks [9]. The existence of disulfide-bound fragments in BLG fibrils has also been shown [10,11]. Many structural, biochemical and nutritional properties of BLG are fairly well known [12,13]. As a member of the lipocalin family, BLG is able to bind in different sites a variety of compounds, notably retinol, fatty acids and polyphenols. Among the latter ones there are several interesting dietary compounds like catechins, quercetin, rutin and resveratrol [14]. BLG shows a remarkable binding versatility, being able to complex ligands not only in the native, but also in its unstructured form [10].

M. Carbonaro et al. / Biochimica et Biophysica Acta 1864 (2016) 766–772

All these findings have suggested that BLG can be a valuable model protein for the study of the mechanisms and the biochemical conditions that direct (inhibit or promote) amyloid fibril formation. In this context, a very interesting issue is the possible role played by small molecules (of natural and/or synthetic origin) in the modulation of fibril formation. Indeed, several dietary antioxidant and anti-inflammatory polyphenols of plant origin, such as flavonoids, have been proved to have health promoting physiological effects. Among them curcumin, the phenolic (diferuloylmethane) yellow curry pigment, has been suggested to inhibit formation of amyloid fibrils in vitro [15]. Unfortunately there are limited evidences [16] that polyphenols of plant origin might exert some protection towards development of neurodegenerative diseases. BLG has been demonstrated to be able to give rise to structured networks of amyloid fibrils. Furthermore, recent studies of the interaction of polyphenol (−)-epicatechin ((−)EC) with BLG has revealed that, at neutral pH and at stoichiometric concentration, (−)EC induces on BLG a conformational change that promotes the dissociation of the BLG dimer [17]. In the present paper the effect of (−)EC on BLG aggregation and fibril formation at pH 2.0 has been monitored while keeping the system at 80 °C, making use of a number of complementary experimental and numerical techniques. The experimental techniques which will be employed in this paper are Thioflavin-T (ThT) fluorescence, Fluorescence Correlation Spectroscopy (FCS), Fourier Transform IR (FTIR) spectroscopy, and Atomic Force Microscopy (AFM). Whenever possible, data will be compared and interpreted theoretically with numerical results coming from Molecular Dynamics (MD) simulations.

767

aluminium foil. Working solution was prepared by diluting 5 times in phosphate-NaCl buffer the ThT stock, thus reaching a final ThT concentration of 12 μM. In the assay the protein sample was added to the working solution in the amount necessary to reach a 12 μM final protein concentration so as to have [BLG]:[ThT] = 1:1. ThT fluorescence was monitored with a PC1 spectrofluorimeter (ISS Inc, Urbana-Champaign, IL USA) exciting the samples with a Xenon arc lamp at λ = 440 nm and collecting the emission spectra in the range 460–600 nm. A ISS-ALBA spectrometer equipped with a Nikon inverted microscope was employed to perform fluorescence correlation spectroscopy on the same samples. Two-photon excitation (880 nm) was provided by a Ti:Sapphire mode-locked laser (Chameleon Ultra; Coherent Inc, Santa Clara, CA, USA). The instrument alignment was performed using a dilute solution (10 nM) of Rhodamine 110. FCS data were analyzed with the help of the ISS FCS analysis software, that assumes Brownian motion for each species in the solution. The model provides the equations

Gðt Þ ¼ n X

! !−1 !−1=2 pffiffiffi n X 2 2 8D t 8D t pffiffiffi 2  f i 1 þ 2i 1 þ 2i ; ω0 z0 π πω0 z0 hC i i¼1

f i ¼ 1;

ð1Þ

ð2Þ

i¼1

2. Materials and methods

where ω0 (the beam waists) and z0 (the beam heights) are defined as the values at which the laser intensity decays by a factor 1/e 2. 〈C〉 is the solute average concentration and Di is the diffusion coefficient of the i-th species and fi its percentage fraction.

2.1. Samples preparation

2.4. FTIR measurements

Bovine BLG (L0130, ≥ 90% pure), (−)EC and ThT were purchased from Sigma Aldrich (St. Louis, MO, USA). Protein and ThT solutions were freshly prepared according to the procedure reported by Loveday and co-workers [7]. In particular, once the protein was dissolved in HCl at pH 2.0, the solution was stirred for 12 h at 4 °C and then centrifuged at 22,000g for 30 min using an Optima MAX-XP Ultracentrifuge (Beckman Coulter, USA). Finally, the solution was filtered through a 0.22 μm syringe filter. The final protein concentration was 0.25 mM, as estimated from the absorbance measured at 280 nm with a Perkin Elmer lamba 18 spectrophotometer using a value of the extinction coefficient of 17,600 M−1 cm−1 (MW = 18.000 Da). Protein fibrillation was achieved by letting BLG incubate at 80 °C in a block heater. At given time intervals aliquots of the sample were taken and subjected to measurement. Two sets of samples were prepared: one in the absence of (−)EC and a second one in the presence of (−)EC with molar ratio [BLG]:[(−)EC] = 1:1.

Infrared experiments require a special sample preparation. To this aim 1 μl drops of the sample were extracted from the incubating solution at given time intervals, for an integrated time of 120 h. Drops were dried on a polished CaF2 substrate. The samples were housed on an infrared microscope (IRscope, Bruker) equipped with a nitrogen cooled MCT detector and coupled to a IFS66V Bruker Interferometer. Data were acquired co-adding 256 interferograms with a spectral resolution of 2 cm−1, reaching a noise-to-signal ratio lower than 1% at 1600 cm −1. Further corrections due to residual atmospheric absorption and baseline mismatch were performed with the help of the OPUS code for data analysis. Since the light spot on the focal plane of the microscope is about 200 μm in diameter, we averaged several spectra collected at different positions in order to account for the possible inhomogeneities of the film's thickness. The absorption data were further analyzed with the OPUS software.

2.2. AFM measurements

2.5. MD simulations

AFM images have been acquired in air using a Veeco Multiprobe IIIa (Santa Barbara, CA) instrument. After 24 h of incubation at 80 °C, droplets containing 3 μl of the two solutions, i.e. prepared as described above in the presence and in the absence of (−)EC, were deposited on a clean silicon surface and let dry in air for about 15 min. Experiments were carried out at room temperature in tapping mode by using Si tips with a force constant of about 40 N/m and a typical curvature radius on the tip of 7 nm. AFM images were analyzed using the Gwyddion 2.34 Data Processing Software [18]. The fibrils height and the periodicity were estimated by line profiles.

MD simulations were performed employing the GROMACS package [19–22] making use of the GROMOS96 43a1 force field [23] and were carried out in the NpT-ensemble. The temperature was held fixed at 353 K using the v-rescale thermostat [24] with a coupling time of 0.1 ps. The pressure was kept constant at the reference pressure of 1 bar with a coupling time of 1 ps and an isothermal compressibility of 4.5 ⋅ 10−5 bar −1, exploiting the features of the Berendsen barostat [25]. The single point charge (SPC) model was used for water molecules. The (−)EC force field was computed with the help of the PRODRG server [26] and implemented in GROMOS96 43a1. Periodic boundary conditions were used throughout the simulation. The Particle Mesh Ewald algorithm was employed to deal with the longrange Coulomb interactions [27]. A time step of 2 fs was used. A nonbond pair list cut-off of 1.4 nm was used, and the pair list was updated every 10 steps. The simulation strategy that we adopted for

2.3. Fluorescence measurements: ThT assay A stock solution of 60 μM ThT in phosphate-NaCl buffer pH 7.0 was prepared and stored at 4 °C in a brown glass bottle covered with an

768

M. Carbonaro et al. / Biochimica et Biophysica Acta 1864 (2016) 766–772

both systems (BLG with or without (−)EC) is a quite standard procedure [28,29] and can be summarized as follows. We start by relaxing in vacuum the all-atom model of the system via a steepest descent minimization. Then water and an appropriate number of counter-ions (necessary to get system neutrality) was added. At this point the solvent was relaxed for 10 ps in the NVT-ensemble at 200 K leaving the solute untouched. Then the whole system, solute and solvent, was equilibrated for 50 ps in the NVT-ensemble at 300 K. At this point the final 120 ns long NpT MD trajectory at 353 K was started. The analysis of the numerical data was carried out using GROMACS tools. Simulations were performed at the realistic values 353 K and pH 2.0 were fibrillation is known to occur. The desired pH value was achieved by having histidine residues protonated and positively charged, and C-terminus, glutamic and aspartic acid all protonated and neutral. In Table 1 we give details of the two simulated systems. For short in the following we will refer to these systems with the nicknames reported in the first column. 3. Results and discussion 3.1. AFM results AFM measurements provide a direct “look” at the fibril structure. As shown in Fig. 1 BLG fibrils are visible both in the sample in the absence (panel a) and in the presence (panel b) of (−)EC. However, it is at the same time quite evident that (−)EC somehow lowers fibrils density. We will show in Section 3.2 that this qualitative observation is confirmed by FCS quantitative comparison of the percentage of BLG fibrils present in the solution with and without (−)EC. The average fibril periodicity and height are not affected by the (−)EC presence, being always 38 ± 3 nm and 4.7 ± 0.3 nm respectively, where errors are statistical dispersions computed over many line profiles on the surface. The values we measured are in very good agreement with those reported by Adamcik [30] for BLG fibrils composed by 2 filaments. 3.2. Fluorescence results Fibril formation evolution versus time incubation is monitored by fluorescence measurements after adding ThT to the 80 °C heated protein solution. ThT is a fluorescence probe that is known to significantly increase its fluorescence intensity while binding fibrils [7,29,31–34]. The time evolutions of the fluorescence intensity of the two samples, in the presence (full squares) and in the absence (open squares) of (−)EC, are reported all along the incubation process at 80 °C and compared (see Fig. 2). During the first 5 h of incubation the ThT fluorescence intensity does not change appreciably and is not modified by the (−)EC presence. The situation changes drastically after about 10 incubation hours, as the fluorescence intensity starts to increase exponentially, but with a rate that is significantly lower in the presence of (−)EC. Indeed an exponential fit of the kind I(t) = I0 exp (t/τ) yields τ ≈ 10.3 h for the BLG + (−)EC sample and τ ≈ 8.7 h for pure BLG.

Fig. 1. AFM images of BLG fibrils in the absence (panel a) and in the presence (panel b) of (−)EC. (−)EC is added to BLG in stoichiometric proportion ([BLG]:[(−)EC] = 1:1). Images have been acquired after 24 h of incubation at 80 °C and pH 2.0.

are super-imposable to those in the presence of (−)EC (blue pluses in the figure). After a 24 h incubation (t = 24), the G(t) behaviours in the absence (black diamonds) and in the presence (red crosses) of (−)EC are significantly different. The continuous lines are obtained by fitting FCS data with the expression of G(t) given in Eq. (1) which corresponds to the ansatz of a pure Brownian motion for the species in the solution. The fits are initially performed assuming a two-species model (i = 1, 2 in Eq. (1)), like the one described by Berland and co-workers [35], and leaving the percentage of each component, fi, and their diffusion coefficients, Di, as free parameters (see Eq. (1)). The use of this fit to analyse the autocorrelation curves is an over simplification to describe the molecules size distribution after heating. However this simplistic viewpoint is able to detect the effect of (−)EC on BLG. With this in mind we notice that at t = 0 (recall that before incubation starts, the presence of (−)EC is ineffective) one of the two components, namely the one with the lowest diffusion constant, D1, is essentially absent. After 24 h of incubation, instead, in the absence of (−)EC the component with the highest diffusion constant (D2) is almost absent, but it reappears in the presence of (−)EC (red cylinder). This result is qualitatively depicted in panel b of Fig 3, where the cylinder heights are proportional to the percentages of each of the two

3.3. FCS results Further quantitative information on the possible influence of (−)EC on BLG aggregate dimensions can be obtained from FCS experiments. Fluorescence correlation data are reported in Fig. 3 for different incubation times and for the two different samples. At t = 0, (before incubation) in the absence of (−)EC (data not shown) experimental points Table 1 Simulated systems: nicknames (left column) and compositions (right column). System name

Composition

blg blgEC

BLG monomer + 16,545 H2O + 48 Na+ + 69 Cl − BLG monomer + (−)EC + 16,486 H2O + 47 Na+ + 68 Cl −

Fig. 2. Fluorescence intensity of ThT added to BLG in the absence (open squares) and in the presence (closed squares) of (−)EC (see Materials and methods section). In the inset, the ThT fluorescence spectra at t = 24 h in the presence (red line) and in the absence (black line) of (−)EC are reported. For reference we also report the ThT fluorescence spectrum at t = 0 (blue curve) when fibrils are not yet formed: no effect due to (−)EC is visible.

M. Carbonaro et al. / Biochimica et Biophysica Acta 1864 (2016) 766–772

769

Fig. 3. Panel a: FCS measurements of the BLG and BLG+(−)EC samples at pH 2.0. Again the two samples, with and without (−)EC, at t = 0, have identical behaviours (blue pluses). At t = 24, data for BLG in the absence of (−)EC are displayed as black diamonds, while those in the presence of (−)EC as red crosses. Solid lines represent the best fit provided by the ansatz of Eq. (1). In panel b, the percentage of each species is pictorially compared.

components. Cylinder colours are in correspondence with the plots of panel a. Looking at the diffusion coefficient values obtained from the fitting procedure, it is quite interesting to notice that the largest one (i.e. the one corresponding to the smallest component) has a value D2 ≈ 250 μm2/s. This value is very similar to that of Rhodamine, a 443 Da fluorescent probe whose diffusion coefficient is measured to be 280 μm2/s [36–38]. Rhodamine molecular weight is similar to that of ThT (367 Da), thus suggesting that the only visible component at t = 0 is due to the fluorescence intensity fluctuation of free ThT. This confirms that fibrils are absent before incubation both in the presence and in the absence of (−)EC in solution. As already mentioned, at t = 24 h in the absence of (−)EC (black diamonds in panel a and black cylinders in panel b of Fig. 3), only the component with the lowest diffusion coefficient (D1 ≈ 10 μm2/s) is present. A straightforward interpretation is that there is no more free ThT (highest diffusion constant) because it is now totally bound to the heavy (lowest diffusion constant) molecular aggregates, like the long fibrils observed in the AFM experiments. Data at t = 24 in the presence of (−)EC (red crosses in panel a of Fig. 3) show a shoulder between about 10−3 and 10−2 s, which is suggestive of the presence of more than one species. We performed a two components fit by holding D1 and D2 fixed at the values obtained from the previous fits and leaving only the two corresponding fractions, f1 and f2, as free parameters. The best fit to data returns f1 ≈ 68% and f2 ≈ 32% as shown by the red cylinders in panel b of Fig. 3. A 32% of ThT molecules free in solution, in the presence of (−)EC, is the symptom of a decrease in the concentration of fibrils, as fibrils are the only aggregates to which ThT can bind.

α-Helix, turn and random coil secondary structures are not resolved in the present spectra: all of them contribute to the band centered at 1660 cm −1 that we call hereafter “unordered” secondary structure contribution. On the contrary, the contribution from β-sheet secondary structure is well resolved and appears to be located at 1630 cm −1. FTIR confirms previous NMR studies, according to which BLG at pH 2.0 and low ionic strength [39] was found to display both ordered β-barrel and “unordered” structures. The low-intensity bands at 1610, 1690 are assigned to aggregated β-sheet structures while the 1720 cm −1 band is attributed to the CO group of the protein backbone [40]. In Fig. 5 the relative intensity of β-sheet (Wβ) and “unordered” structures (Wu), the only two secondary structures that have been shown to be sensitive to thermal treatment, are reported for pure BLG (panel a) and for BLG in the presence of (−)EC (panel b). Wβ gradually increases between 10 and 20 h of incubation time at the expenses of Wu.

3.4. FTIR results FTIR spectroscopy is used to monitor possible secondary structure evolution related to fibril formation. In Fig. 4 the absorption spectra of pure BLG in the amide I and II regions at four selected incubation times are shown. In order to resolve secondary structure components FTIR spectra are deconvoluted with a sum of Gaussian lineshapes. As an example of the procedure we display at the bottom of Fig. 4 the five Gaussians, centered at 1610, 1630, 1660, 1690 and 1720 cm −1, that are used to fit the t = 0 amide I band of the data set. In Supplementary Materials we give details of the deconvolution procedure.

Fig. 4. The absorption spectra of pure 0.25 mM BLG in the amide I and II regions at four selected incubation times. We show a five-Gaussian fit (dashed curve) to the amide I band of the t = 0 data (red curve).

770

M. Carbonaro et al. / Biochimica et Biophysica Acta 1864 (2016) 766–772

Fig. 5. Time evolution of the relative intensities of β-sheet (black dots) and “unordered” (brown squares) secondary structures. Panel a: BLG in the absence of (−)EC, panel b: BLG in the presence of (−)EC. Blue symbols represent the sum of the two contributions. Continuous and dashed lines are to guide the eye.

Afterwards, both relative intensities attain their asymptotic values after 20 h of heating. Data in Fig. 5 show that the overall variation of Wβ and Wu is significantly lower in the sample containing (−)EC (about ±0.13) compared to that observed in pure BLG (± 0.20). This result, together with the findings obtained from fluorescence experiments (see Fig. 2), suggests that (−)EC might inhibit fibril formation, possibly stabilising the native BLG secondary structure. It is interesting to remark that the secondary structure evolution appears to be almost exactly synchronised with fibril formation both in the absence and in the presence of (−)EC, as seen from the time dependence of the ThT fluorescence measurements.

Fig. 6. Root Mean Square Displacement of BLG molecule vs time. blg in blue and blgEC in red.

denaturation as it also happens in the case of ligands interacting with the Amyloid-β peptide [28]. We clustered together the configurations whose RMSD values differ less than a given threshold (a reasonable choice is 3 Å) with the aim of identifying leading structural features. The histogram of Fig. 7 is built by counting, for each cluster, the number of configurations that along the trajectory happen to fall into the cluster. We notice that in the blgEC case (red rectangles), only two clusters are appreciably populated, whereas in blg case (blue rectangles) four/five different clusters appear to be populated. In Fig. 8 we show the time evolution of the clusters populations in the absence (panel a) and in the presence (panel b) of (−)EC all along the trajectory. An interesting feature emerges in the case of blgEC. Actually the two most populated clusters are never populated at the same

3.5. MD simulations results To complement and help interpreting the experimental analysis illustrated in the previous sections, we carried out a numerical study of the influence of (−)EC on the BLG secondary structure by performing MD atomistic simulations of a model system where one BLG monomer is dissolved in water in the absence of (−)EC as well as in the presence of one (−)EC molecule (see Table 1 in Section 2.5). The initial BLG monomer configuration from which simulations are started is taken from the X-ray structure provided in Ref. [41] (ID:1BSY). The effect of the (−)EC presence on the BLG structural stability is monitored by measuring the evolution of the Root Mean Square Displacement (RMSD) of the BLG backbone referred to the initial structure from which the simulation was started. In Fig. 6 the behaviour of the RMSD as a function of the simulation time is displayed. We observe that, after a transient of 40 ns the system stabilizes. From 40 to 120 ns, both the RMSD mean value and its oscillation amplitude appear to be smaller in the blgEC, system where a (−)EC molecule is present, than in the blg system where no (−)EC molecule is present (Table 1). This behaviour suggests that the BLG molecule in the presence of (−)EC is more stable and deviates less from the initial MD configuration. This result is consistent with previous experimental data [42] showing that ligand binding stabilizes BLG against physical and chemical

Fig. 7. BLG structural clustering histograms. Blue refers to blg, red to blgEC.

M. Carbonaro et al. / Biochimica et Biophysica Acta 1864 (2016) 766–772

771

Fig. 8. Clusters occupation as a function of the simulation time. Panel a refers to blg, panel b to blgEC.

time. In fact, cluster 2 is only populated during the first 15 ns when it is replaced by cluster 1 that remains constantly populated till the end of the simulation. This behaviour is in contrast with the one of blg (panel a), where BLG is shown to jump quite often among structures belonging to the four/five most populated different clusters. Such a behaviour is a striking confirmation of the structural stabilisation role played by the (−)EC. The two most populated structures of BLG in the presence of (−)EC (blgEC system) are reported in Fig. 9. It is evident that after a few ns the interaction of (−)EC with BLG promotes the opening of an initially closed calyx. Once (−)EC has entered the calyx, it remains inside it all along the simulation. Calyx is a typical fold of the lipocalin protein superfamily [43]. According to the current literature [44,45] and to our experiments at pH 2.0 and 80 °C (shown in the Supplementary Materials), the calyx is expected to stay closed in the absence of other molecules interacting with BLG.

4. Conclusions Protein aggregation and fibril formation have recently received an increased attention due to their relevance in health, nutritional and technological issues. Various antioxidants have been reported to inhibit fibrils formation or to destabilize preformed ones. Trying to clarify this mechanism, we focused on BLG as a valuable model protein and on the polyphenol (−)EC as a potential fibril inhibitor. In this work we have provided experimental and computational evidence that (−)EC is able to slow down BLG aggregation process. AFM analysis clearly indicates that (−)EC has the effect of reducing BLG fibril formation. These qualitative findings are confirmed by FCS measurements which show that in the presence of (−)EC a reduced percentage of BLG fibrils is present.

Moreover, both ThT fluorescence and FTIR data reveal that aggregates formation starts concurrently with the onset of secondary structural changes of the molecule. This is consistent with FTIR data which show that fibril formation is accompanied by an enhancement of the β-sheet secondary structure fraction at the expenses of the “unordered” (turn, random coil and α-helix) structure. Furthermore, FTIR data indicate that the native secondary structure, where a large amount (40%) of segments of BLG molecule at pH 2.0 and 80 °C are poorly structured, is not completely lost upon fibril formation, thus suggesting a most likely partial structural rearrangement in BLG. MD simulations unambiguously give evidence that (−)EC stabilizes the BLG secondary structure. This result indirectly confirms FTIR and ThT fluorescence results which suggest that, in the presence of (−)EC, the BLG structural changes are both slowed down and reduced in size. The atomic structural description provided by MD simulations allows us to determine the mechanism by which (−)EC stabilizes the BLG structure. Following the time evolution of the (solvated) blgEC system, one can conclude that (−)EC promotes the opening of the protein calyx where it enters and remains stably bound. Summarizing we can say that the various experimental and numerical results we have collected build up a consistent picture where (−)EC has the effect of both slowing down and reducing BLG fibrils formation. Conflict-of-interest statement All the authors who took part in this study declare that they have nothing to disclose regarding competing interests or funding from industry with respect to this manuscript. Research involving human participants and/or animals This article does not contain any studies with human participants or animals performed by any of the authors.

Fig. 9. The BLG protein structures of the two most populated clusters (see Fig. 7) of the blgEC system. Panel a shows a representative of the structures most visited along the first 15 ns of the simulation. Panel b shows a representative of the structures most visited along the final 105 ns. In both panels only the amino acid residues at the calyx entrance are explicitly displayed as van der Waals spheres. The rest of the protein is only sketched. The (−)EC molecule is drawn in the ball and stick representation.

772

M. Carbonaro et al. / Biochimica et Biophysica Acta 1864 (2016) 766–772

Transparency document The Transparency document associated with this article can be found, in online version. Acknowledgements We wish to thank Giancarlo Rossi for fruitful discussions and for reading the manuscript. This work was supported by SUMA-INFN (Italy) and by the University of Rome “Tor Vergata” (Uncovering Excellence 2014). Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.bbapap.2016.03.017. References [1] C. Dobson, Protein folding and misfolding, Nature 426 (2003) 884–890. [2] S.W. Chen, S. Drakulic, E. Deas, M. Ouberai, F.A. Aprile, R. Arranz, S. Ness, C. Roodveldt, T. Guilliams, E.J. De-Genst, D. Klenerman, N.W. Wood, T.P.J. Knowles, C. Alfonso, G. Rivas, A.Y. Abramov, J.M. Valpuesta, C.M. Dobson, N. Cremades, Structural characterization of toxic oligomers that are kinetically trapped during α-synuclein fibril formation, Proc. Natl. Acad. Sci. U. S. A. 112 (2015) 1994–2003. [3] D. Huang, M.I. Zimmerman, P.K. Martin, A.J. Nix, T.L. Rosenberry, A.K. Paravastu, Antiparallel β-sheet structure within the C-terminal region of 42-residue Alzheimer's amyloid-β peptides when they form 150-kDa oligomers, J. Mol. Biol. 427 (2015) 2319–2328. [4] M. Stefani, Structural features and cytotoxicity of amyloid oligomers: implications in Alzheimer's disease and other diseases with amyloid deposits, Prog. Neurobiol. 99 (2012) 226. [5] V. Vetri, M. D'Amico, V. Foderá, M. Leone, A. Ponzoni, G. Sberveglieri, V. Militello, Bovine serum albumine protofibril-like aggregates formation: solo but not simple mechanism, Arch. Biochem. Biophys. 508 (2011) 13–24. [6] R. Carrotta, V. Vetri, F. Librizzi, V. Martorana, V. Militello, M. Leone, Amyloid fibrils formation of Concanavalin A at basic pH, J. Phys. Chem. B 115 (2011) 2691. [7] S. Loveday, X.L. Wang, M. Rao, S.G. Anema, H. Singh, β-Lactoglobulin nanofibrils: effect of temperature on fibril formation kinetics, fibril morphology and the rheological properties of fibril dispersions, Food Hydrocoll. 27 (2012) 242–249. [8] M. Mazaheri, A.A. Moosavi-Movahedi, A.A. Saboury, F. Khodagholi, F. Shaerzadeh, N. Sheibani, Curcumin protects β-lactoglobulin fibril formation and fibril-induced neurotoxicity in PC12cells, PLoS One 10 (2015) 1–17. [9] C. Akkermans, P. Venema, A.J. van der Goot, H. Gruppen, E.J. Bakx, R.M. Boom, E. van der Linden, Peptides are building blocks of heat-induced fibrillar protein aggregates of β-lactoglobulin formed at pH 2, Biomacromolecules 9 (2008) 1474–1479. [10] P. Rasmussen, A. Barbiroli, F. Bonomi, F. Faoro, P. Ferranti, M. Iriti, G. Picariello, S. Iametti, Formation of structured polymers upon controlled denaturation of β-lactoglobulin with different chaotropes, Biopolymers 86 (2007) 57–72. [11] A.C. Dave, S.M. Loveday, S.G. Anema, T.S. Loo, G.E. Norris, G.B. Jameson, H. Singh, β-Lactoglobulin self-assembly: structural changes in early stages and disulfide bonding in fibrils, J. Agric. Food Chem. 61 (2013) 7817–7828. [12] M.Z. Papiz, L. Sawyer, E.E. Eliopoulos, A.C.T. North, J.B.C. Findlay, R. Sivapradasaro, T.A. Jones, M.E. Newcorner, P.J. Kraulis, The structure of β-lactoglobulin and its similarity to plasma retinol-binding protein, Nature 324 (1986) 383–385. [13] M. Carbonaro, F. Bonomi, S. Iametti, M. Cappelloni, E. Carnovale, Aggregation of proteins in whey from raw and heat-processed milk: formation of soluble macroaggregates and nutritional consequences, Lebensm. Wiss. Technol. 31 (1998) 522–529. [14] M. Stojadinovic, J. Radosavljevic, J. Ognjenovic, J. Vesic, I. Prodic, D. Stanic-Vucinic, T. Cirkovic Velickovic, Binding affinity between dietary polyphenols and βlactoglobulin negatively correlates with the protein susceptibility to digestion and total antioxidant activity of complexes formed, Food Chem. 136 (2013) 1263–1271. [15] F. Yang, G.P. Lim, A.N. Begum, O.J. Ubeda, M.R. Simmons, S.S. Ambegaokar, P.P. Chen, R. Kayed, C.G. Glabe, S.A. Frautschy, G.M. Cole, Curcumin inhibits formation of amyloid beta oligomers and fibrils, binds plaques, and reduces amyloid in vivo, J. Biol. Chem. 280 (2005) 5892–5901. [16] S.A. Bhat, B. Bano, Conformational behaviour and aggregation of chickpea cystatin in trifluoroethanol: Effect of epicatechin and tannic acid, Arch. Biochem. Biophys. 562 (2014) 51–61.

[17] A. Nucara, P. Maselli, V. Giliberti, M. Carbonaro, Epicatechin-induced conformation changes in β-lactoglobulin B monitored by FT-IR spectroscopy, SpringerPlus 2 (2013) 661. [18] D. Neas, P. Klapetek, Gwyddion: an open-source software for SPM data analysis, Cent. Eur. J. Phys. 10 (2012) 181–188. [19] H.J.C. Berendsen, D. van der Spoel, R. van Drunen, GROMACS: a message-passing parallel molecular dynamics implementation, Comput. Phys. Commun. 91 (1995) 43–56. [20] E. Lindahl, B. Hess, D. van der Spoel, Gromacs 3.0: a package for molecular simulation and trajectory analysis, J. Mol. Model. 7 (2001) 306–317. [21] D. van der Spoel, E. Lindahl, B. Hess, G. Groenhor, A.E. Mark, H.J.C. Berendsen, Gromacs: fast, flexible, and free, J. Comput. Chem. 26 (2005) 1701–1718. [22] B. Hess, C. Kutzner, D. van der Spoel, E. Lindahl, Gromacs 4: algorithms for highly efficient, load-balanced, and scalable molecular simulation, J. Chem. Theory Comput. 4 (2008) 435–447. [23] W.F. van Gunsteren, S.R. Billeter, A.A. Eking, P.H. Hiinenberger, P. Kriiger, A.E. Mark, W.R.P. Scott, I.G. Tironi, Biomolecular Simulation, The GROMOS96 Manual and User Guide, Hochschulverlag AG, Groningen, 1996. [24] G. Bussi, D. Donadio, M. Parrinello, Canonical sampling through velocity rescaling, J. Chem. Phys. 126 (2007) 014101. [25] H.J.C. Berendsen, J.P.M. Postma, A. DiNola, J.R. Haak, Molecular dynamics with coupling to an external bath, J. Chem. Phys. 81 (1984) 3684–3690. [26] A.W. Schüttelkopf, D.M.F. van Aalten, PRODRG: a tool for high-throughput crystallography of protein-ligand complexes, Acta Crystallogr. D60 (2004) 1355–1363. [27] T. Darden, D. York, L. Pedersen, Particle mesh Ewald: an N·log(N) method for Ewald sums in large systems, J. Chem. Phys. 98 (1993) 10089–10092. [28] V. Minicozzi, R. Chiaraluce, V. Consalvi, C. Giordano, C. Narcisi, P. Punzi, G.C. Rossi, S. Morante, Computational and experimental studies on β-sheet breakers targeting Amyloid-β1-40 fibrils, J. Biol. Chem. 289 (2014) 11242–11252. [29] M.G. Di Carlo, V. Minicozzi, V. Foderà, V. Militello, V. Vetri, S. Morante, M. Leone, Thioflavin-T templates amyloid-β(1–40) conformation and aggregation pathway, Biophys. Chem. 206 (2015) 1–11. [30] J. Adamcik, J.M. Jung, J. Flakowski, P. De Los Rios, G. Dietler, R. Mezzenga, Understanding amyloid aggregation by statistical analysis of atomic force microscopy images, Nat. Nanotechnol. 5 (2010) 423–428. [31] M. Levine, K. Sörgjerd, P. Hammarström, Detection and characterization of aggregates, prefibrillar amyloidogenic oligomers, and protofibrils using fluorescence spectroscopy, Biophys. J. 88 (2005) 4200–4212. [32] H. Lindgren, Stopped-flow kinetics reveal multiple phases of Thioflavin T binding to Alzheimer beta (1-40) amyloid fibrils, Arch. Biochem. Biophys. 342 (1997) 306–316. [33] M. Biancalana, K. Makabe, A. Koide, S. Koide, Molecular mechanism of Thioflavin-T binding to the surface of beta-rich peptide self-assemblies, J. Mol. Biol. 385 (4) (2009) 1052–1063. [34] M. Biancalana, S. Koide, Molecular mechanism of Thioflavin-T binding to amyloid fibrils, Biochim. Biophys. Acta 1804 (2010) 1405–1412. [35] K.M. Berland, P.T. So, E. Gratton, Two-photon fluorescence correlation spectroscopy: method and application to the intracellular environment, Biophys. J. 68 (2) (1995) 694–701. [36] D. Wang, N.V. Visser, M. Veenhuis, I.J. van der Klei, Physical interactions of the peroxisomal targeting signal 1 receptor pex5p, studied by fluorescence correlation spectroscopy, J. Biol. Chem. 278 (44) (2003) 43340–43345. [37] S.J. Gould, C.S. Collins, Opinion: peroxisomal protein import: is it really that complex? Nat. Rev. Mol. Cell Biol. 3 (2002) 382–389. [38] T. Krouglova, J. Vercammen, Y. Engelborghs, Correct diffusion coefficients of proteins in fluorescence correlation spectroscopy. application to tubulin oligomers induced by Mg2+ and paclitaxel, Biophys. J. 87 (2004) 2635–2646. [39] F. Fogolari, L. Ragona, L. Zettab, S. Romagnoli, K.G. De Kruifc, H. Molinari, Monomeric bovine β-lactoglobulin adopts a β-barrel fold at pH 2, FEBS Lett. 436 (1998) 149–154. [40] A. Barth, Infrared spectroscopy of proteins, Biochim. Biophys. Acta Bioenerg. 1767 (2007) 1073–1101. [41] B.Y. Qin, M.C. Bewley, L.K. Creamer, H.M. Baker, E.N. Baker, G.B. Jameson, Structural basis of the Tanford transition of Bovine β-Lactoglobulin, Biochemistry 37 (1998) 14014–14023. [42] A. Barbiroli, F. Bonomi, P. Ferranti, D. Fessas, A. Nasi, P. Rasmussen, S. Iametti, Bound fatty acids modulate the sensitivity of bovine β-lactoglobulin to chemical and physical denaturation, J. Agric. Food Chem. 59 (2011) 5729–5737. [43] D.R. Flower, A.C.T. North, T.K. Attwood, Structure and sequence relationships in the lipocalin and related proteins, Protein Sci. 2 (1993) 753–761. [44] M. Collini, D'Alfonso, G. Baldini, New insight on β-lactoglobulin binding sites by 1anilinonaphthalene-8-sulfonate fluorescence decay, Protein Sci. 9 (2000) 1968–1974. [45] M. Collini, L. D'Alfonso, H. Molinari, L. Ragona, M. Catalano, G. Baldini, Competitive binding of fatty acids and the fluorescent probe 1-8-anilinonaphthalene sulfonate to bovine β-lactoglobulin, Protein Sci. 12 (2003) 1596–1603.