Biochimica et Biophysica Acta 1854 (2015) 239–248

Contents lists available at ScienceDirect

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

Influence of the protein context on the polyglutamine length-dependent elongation of amyloid fibrils Céline Huynen a, Nicolas Willet b, Alexander K. Buell c, Anne-Sophie Duwez b, Christine Jerôme d, Mireille Dumoulin a,⁎ a

Laboratory of Enzymology and Protein Folding, Centre for Protein Engineering, University of Liege, Liege, Belgium Nanochemistry and Molecular Systems, Department of Chemistry, University of Liege, Liege, Belgium Department of Chemistry, University of Cambridge, Lensfield Road, Cambridge CB2 1EW, UK d Center for Education and Research on Macromolecules (CERM), Department of Chemistry, University of Liege, Liege, Belgium b c

a r t i c l e

i n f o

Article history: Received 11 September 2014 Received in revised form 20 November 2014 Accepted 1 December 2014 Available online 6 December 2014 Keywords: Quartz crystal microbalance (QCM) Atomic force microscopy (AFM) Fibril nucleation Fibril elongation Amyloid fibrils BlaP-polyQ chimeras

a b s t r a c t Polyglutamine (polyQ) diseases, including Huntington's disease, are neurodegenerative disorders associated with the abnormal expansion of a polyQ tract within nine proteins. The polyQ expansion is thought to be a major determinant in the development of neurotoxicity, triggering protein aggregation into amyloid fibrils, although non-polyQ regions play a modulating role. In this work, we investigate the relative importance of the polyQ length, its location within a host protein, and the conformational state of the latter in the amyloid fibril elongation. Model polyQ proteins made of the β-lactamase BlaP containing up to 79Q inserted at two different positions, and quartz crystal microbalance and atomic force microscopy were used for this purpose. We demonstrate that, independently of the polyQ tract location and the conformational state of the host protein, the relative elongation rate of fibrils increases linearly with the polyQ length. The slope of the linear fit is similar for both sets of chimeras (i.e., the elongation rate increases by ~1.9% for each additional glutamine), and is also similar to that previously observed for polyQ peptides. The elongation rate is, however, strongly influenced by the location of the polyQ tract within BlaP and the conformational state of BlaP. Moreover, comparison of our results with those previously reported for aggregation in solution indicates that these two parameters also modulate the ability of BlaP-polyQ chimeras to form the aggregation nucleus. Altogether our results suggest that non-polyQ regions are valuable targets in order to interfere with the process of amyloid fibril formation associated with polyQ diseases. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Nine neurodegenerative disorders, referred to as polyQ diseases, are associated with the abnormal expansion, in successive generations, of an unstable CAG trinucleotide repeat in the coding region of nine unrelated genes [1,2]. This CAG expansion is translated into an expanded polyQ tract in the corresponding proteins. The polyQ disorders include Huntington's disease, the most common one, and several spinocerebellar ataxias (SCA) [1,3]. They all share a number of features which suggest a common physiopathological mechanism [1,2,4,5]: (i) There is a threshold in the number of glutamine residues above Abbreviations: AFM, Atomic force microscopy; EDC, N-(3-dimethylaminopropyl)-N′ethylcarbodiimide hydrochloride; htt1, N-terminal fragment of huntingtin; MUA, Mercaptoundecanoic acid; NHS, N-hydroxy succinimide; PBS, Phosphate buffer saline; NHS, N-hydroxy sucinimide; PolyQ, Polyglutamine; QCM, Quartz crystal microbalance; SCA, Spinocerebellar ataxia; TEM, Transmission electron microscopy ⁎ Corresponding author at: Centre for Protein Engineering, Laboratory of Enzymology and Protein Folding, Institute of Chemistry, Bat. B6c; University of Liege, B4000 LiegeSart Tilman, Belgium. Tel.: +32 4 366 3546; fax: +32 4 366 3364. E-mail address: [email protected] (M. Dumoulin).

http://dx.doi.org/10.1016/j.bbapap.2014.12.002 1570-9639/© 2014 Elsevier B.V. All rights reserved.

which the polyQ tract, qualified as expanded, causes disease; it is generally comprised between 35 and 45 residues [7]. (ii) Above the threshold, the longer the polyQ tract, the earlier the onset and the more severe the disease [3,4,8,9]. (iii) Expanded polyQ tracts trigger the formation of amyloid fibrils that accumulate into neuronal intranuclear inclusion bodies [1,3,9,10]. Since the nine proteins share no sequence homology, except the presence of an expanded polyQ tract, it is widely accepted that this latter is a critical determinant for the development of neurotoxicity. A large body of evidence indeed suggests that it confers a toxic gain of function to the proteins, associated with protein misfolding and aggregation into amyloid fibrils; the identity of the toxic species (i.e., misfolded monomers, soluble oligomers, and/or insoluble amyloid fibrils) remains however unresolved [1,2,4,5,6,10–12]. On the other hand, a growing number of studies performed on fulllength disease-associated proteins, or fragments thereof, or model polyQ proteins have shown that the aggregation process, induced by the polyQ expansion, is modulated by the properties of the host protein [2,11,13–16]. The nature of the non-polyQ regions influences not only the Q-threshold for amyloid fibril formation [17], but also the kinetics and the pathway of aggregation, as well as the aggregate stability

240

C. Huynen et al. / Biochimica et Biophysica Acta 1854 (2015) 239–248

[13–16,18–21]. Since the early reports demonstrating the influence of non-polyQ regions [13–16,22], many studies have aimed at providing a better understanding of the molecular determinants of this complex interplay between intrinsic aggregation properties of the expanded polyQ tract and the modulating role of non-polyQ regions, in order to develop preventive or curative treatments against these devastating disorders. These studies have been carried out with several ataxins [23,24], the N-terminal fragment of huntingtin (htt1) [13,18,25], respectively associated with SCAs and Huntington's disease, and with various model polyQ proteins [19,21,26]. The results indicate that non-polyQ regions can favor the aggregation into amyloid fibrils by providing an additional aggregation prone domain [18,23,24]. In contrast, non-polyQ regions can protect from aggregation into amyloid fibrils by applying constraints to the polyQ tract preventing it from adopting an aggregation-prone conformation [13,18,19] or by increasing the saturation concentration of the protein [25]. The aggregation process of polyQ proteins has been described as a nucleation-dependent polymerization mechanism, similar to many other amyloid-forming polypeptides [2,27–31]. This mechanism is characterized by a nucleation step, during which monomers assemble into nuclei, followed by a growth phase, during which the nuclei are elongated by sequential addition of monomers into amyloid fibrils [32, 33]. Each of the individual molecular processes is likely to have a different dependence on the various molecular determinants, such as polyQ length, its position within the protein, and the properties of non-polyQ regions. It is of crucial importance to understand how these steps are affected by both the polyQ length and the properties of non-polyQ regions in order to develop therapeutic targets highly specific to the aggregation process of polyQ proteins. Up to now, very little information is available concerning the effect of non-polyQ regions on each individual step of amyloid fibril formation, especially on the growth of aggregates, the elongation step. Data from studies on polyQ peptides indicate that the elongation rate increases with the Q-repeat length [34,35], and that the conformation of monomeric polyQ peptides affects the elongation rate [35]. For example, the insertion of a β-turn template D-Pro-Gly in the center of a Q20 peptide increases the elongation rate several fold [35]. However, no data is currently available concerning (i) the polyQ length-dependent elongation rate in the context of a host protein, (ii) the influence of the exact location of the polyQ tract within a host protein and (iii) the influence of the host protein conformation on the elongation properties. The aim of this study is therefore to address these open questions using model polyQ proteins based on a 30.4 kDa host protein that is tolerant to the insertion of a variety of polypeptide sequences, the βlactamase BlaP from Bacillus licheniformis 749/C [36], and using quartz crystal microbalance (QCM) measurements in combination with atomic force microscopy (AFM). QCM has been shown to allow accurate measurements of the elongation rate of preformed fibrils (“seeds”) immobilized onto a gold-coated quartz crystal sensor upon incubation with soluble protein monomers [37,38]. We have previously designed and characterized BlaP-polyQ chimeras with up to 79Q residues inserted at position 197 and 216, within two flexible, solvent-exposed loops flanked by α-helices. The chimeras with insertion at position 197 are referred to as BlaP197 chimeras or BlaP197Qx, with x = 0, 23, 30, 55 or 79, while chimeras with insertion at position 216 are referred to as BlaP216 chimeras or BlaP216Qx, with x = 0, 30, 55 or 79 ([39], Thorn, D., C. Pain, C. Huynen, S. Preumont, C. Duez, G. Brisotto, A. Matagne, M. Dumoulin and N. Scarafone, in preparation) (Fig. 1). These two insertion sites share a common flanking sequence in between the two sites since they are only 19 amino acids apart. However, the 197 insertion site is located within a folded domain (the α-domain), while the 216 insertion site is located at the interface between the two folded domains of BlaP. These chimeras allow comparison of subtle differences in the protein environment of the polyQ tract (two different polyQ locations within the same amino acid sequence and the same overall protein structure) and therefore constitute a unique and relevant model to

Fig. 1. X-ray crystal structure of the β-lactamase BlaP from Bacillus licheniformis 749/C. The residue numbering is based on homology to class A β-lactamases [40]. The structure was produced using PyMOL (DeLano Scientific LLC, South San Francisco, USA) with the PDB ID 4BLM [59]. BlaP is made of two domains: the α-domain (dark blue) and the α/β-domain (pink). The serine of the active site (S70) is highlighted in green. The two insertion sites, located between α-helices 8 and 9 for position 197, and between α-helices 9 and 10 for position 216, are both indicated in light blue. A SmaI restriction site has been introduced within the gene of BlaP to allow poly(CAG) sequence insertion. This introduction results in the addition of a PG dipeptide between D197 and K198 or T216 and G217 of BlaP. The polyQ tract is inserted between P and G [39].

study the principles underlying the complex interplay between the propensity of expanded polyQ tracts to aggregate into amyloid fibrils and the modulating role of non-polyQ regions. The biophysical characterization of both sets of chimeras indicates that (i) while the polyQ tract is disordered, it does not drastically perturb the structure of BlaP and (ii) the aggregation properties in solution of these polyQ proteins recapitulate those of disease-associated polyQ proteins (i.e., there is a Q-threshold for fibril formation, and above the threshold, the longer the polyQ, the faster the aggregation process). Moreover, the aggregation properties of chimeras are extremely sensitive to the polyQ environment: they depend on both the structural integrity of the host protein (i.e., BlaP) and on the position of the polyQ tract (i.e., 197 or 216) (Table 1).

Table 1 Aggregation properties of 197 and 216 chimeras of BlaP at 110 and/or 40 μM ([39], Thorn et al., in preparation). Native conditions, PBS pH 7.5, 37 °C; denaturing conditions, PBS pH 7.5, 1.85 and 3.5 M urea, 25 °C. Aggregation kinetics were monitored by measuring the remaining soluble protein concentration as a function of time. BlaP197Q55 forms amyloid fibrils under denaturing conditions while under native conditions it forms amorphous aggregates. This suggests that the native conformation of the BlaP moiety imposes constraints to the 55Q tract inserted in position 197 that block its propensity to induce amyloid fibril formation [39]. The BlaP chimera containing a 30Q tract at position 216 is not able to aggregate into amyloid fibrils, while BlaP216Q55 does form amyloid fibrils (Thorn et al. in preparation).

Aggregation properties: Fibril formation under native conditions Fibril formation under denaturing conditions Aggregation rate under native conditions

BlaP197 chimeras

BlaP216 chimeras

55Q b threshold b 79Q

30Q b threshold b 55Q

30Q b threshold b 55Q

30Q b threshold b 55Q

+

++

C. Huynen et al. / Biochimica et Biophysica Acta 1854 (2015) 239–248

Our QCM results on the kinetics of fibril elongation and their comparison with results reported for unseeded aggregation in bulk solution ([39], Thorn et al., in preparation) allow therefore a better understanding of the role of the polyQ tract and of the protein context on the nucleation and the elongation steps of amyloid fibril formation. They indicate that, similarly to what was observed for polyQ peptides [34], there is also a linear correlation between polyQ length and elongation rate within a larger protein context. Our data moreover demonstrate that the conformational state of the host protein and the location of the polyQ repeat within the latter strongly influence the propensity of the polyQ tract to elongate amyloid fibrils, as well as the formation of the aggregation nucleus. Non-polyQ regions therefore constitute an additional and specific therapeutic target in order to prevent and/or to arrest the disease development by interfering respectively with the nucleation and/or elongation of amyloid fibrils by polyQ proteins. 2. Materials and methods 2.1. Protein expression and purification BlaP197Q0, BlaP216Q0, and their respective chimeras (BlaP197Q23, BlaP197Q30, BlaP197Q55, BlaP197Q79, BlaP216Q30, BlaP216Q55, and BlaP216Q79) were expressed from the constitutive expression vector pNY in E. coli JM109 strains (Promega) [41]. BlaP197Q0 and BlaP216Q0 correspond to BlaP with a PG dipeptide inserted in position 197 or 216, respectively (Fig. 1). The proteins were expressed and purified as described previously [39,41], except that both the culture and preculture were carried out in the absence of ampicillin, previously used to select strains expressing correctly folded β-lactamase chimeras [36,39]. After the purification, the fractions containing more than 95% of the proteins of interest [as assessed by 15% (w/v) SDS-PAGE] were pooled and dialyzed extensively against PBS (phosphate buffer saline, 50 mM sodium phosphate, 150 mM NaCl) pH 7.5. The integrity of the chimeras was verified by determining their molecular mass using electrospray ionization quadrupole time-of-flight mass spectrometry at GIGA proteomics platform (Liege, Belgium). Proteins were stored at −20 °C in PBS pH 7.5 at an approximate concentration of 15 μM. After defrosting and filtering protein solutions on a 0.22 μm porosity membrane before QCM experiments, dynamic light scattering (15 acquisitions of 5 s, DynaPro Nanostar, Wyatt technology, Germany) experiments indicate that the soluble fraction contains more than 95% of monomeric molecules (data not shown). 2.2. Quantification of protein concentrations The concentration of the different protein solutions was determined by the measurement of the absorbance at 280 nm (A280) using the molar extinction coefficient experimentally determined using BlaP197Q0 (33000 M−1.cm−1 at 280 nm) [39]. 2.3. Formation of amyloid fibrils by BlaP197Q79 250 μL to 1 mL of a 110 μM BlaP197Q79 solution, filtered on a 0.22 μm porosity membrane, were incubated in PBS pH 7.5, 0.2% sodium azide, at 37 °C for one week [39]. Airtight tubes (Multiply Safecup, Sarstedt, Nümbrecht, Germany) or 1.5 mL tubes tightly closed with a cap and parafilm were used to avoid evaporation. After incubation, samples were centrifuged 50 min at 16873 g. The quantity of fibrils formed (pellet) was deduced from the determination of protein concentration in the soluble fraction (supernatant) by absorbance measurement at 280 nm. Fibrils were stored in PBS pH 7.5 at 4 °C at a concentration of 1–2 mg.mL−1 for several weeks. Various stocks of fibrils differently aged were used for the experiments and the elongation rates measured for a given polyQ-chimera did not show a dependence on the maturation stage of the seed fibrils.

241

2.4. Transmission electron microscopy Aggregated samples, negatively stained using 2% uranyl acetate (w/v), were analyzed by transmission electron microscopy (TEM, Philips CM100), as described by Scarafone et al. [39]. 2.5. Functionalization of the quartz crystal sensors Amyloid fibrils of BlaP197Q79 were covalently attached to goldcoated quartz crystal sensors (QSX 301, 50 nm gold coating, Qsense, Västra Frölunda, Sweden) via an activated self-assembled monolayer (SAM) of mercaptoundecanoic acid (MUA) according to protocols described in the literature [37,42]. This compound contains carboxyl groups that are activated by the coupling reagents, N(3-dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride and N-hydroxy succinimide (NHS) to form amide bonds with amine groups of proteins [37,43]. First, the quartz crystals were immersed in 5 mL of 0.5 mg.mL−1 MUA in ethanol for 24 h, rinsed with ethanol and dried with nitrogen gas. Then, they were activated in 5 mL of aqueous solution of 0.3 mg.mL−1 EDC and 0.5 mg.mL−1 NHS for 20 min, rinsed with milli-Q water and dried with nitrogen gas. 100 μL of 0.2 to 1 mg.mL−1 sonicated amyloid fibrils (i.e., seeds) were loaded onto quartz crystals in a water-saturated atmosphere for 15 min. Fibrils were sonicated 5 times 5 s on ice, with a 5 s break between each sonication period (sonicator MSE 12/74 MK2, MSE Scientific instruments, Sussex, England) prior to surface attachment. This protocol served to reduce their length and increase the number of fibril ends that could be elongated. It therefore allows to increase the rate of mass addition due to fibril elongation and thus it facilitates the measurements of elongation rates [44]. The quartz crystals were then rinsed with milli-Q water, dried with nitrogen gas, and incubated with 100 μL of 1 M ethanolamine pH 8.5 in a water-saturated atmosphere for 20 min. Ethanolamine blocks all unbound-sites to avoid covalent attachment of the monomeric protein with the surface of the sensor. Finally, the quartz crystals were rinsed with milli-Q water, dried with nitrogen gas and introduced into the microbalance flow cell (Q-sense E4, Q-sense AB, Västra Frölunda, Sweden) for QCM measurements. As discussed in the results section, the surface density of attached seed fibrils determined by AFM imaging was found to be homogeneous at the surface of each individual quartz crystal, but was subject to a certain variability between different quartz crystals. 2.6. Quartz crystal microbalance measurements QCM experiments were performed using one flow cell of the Q-sense E4. QCM is a technique that has been shown previously to be a powerful strategy for monitoring protein aggregation [37,38,45,46]. The formation and growth of aggregates is measured through an increase in surface-bound mass (Δm), leading to a detectable decrease in the resonant frequency of the quartz crystal. When seeds are attached to the sensor surface, the measurement is highly selective for the elongation reaction, and the rate of changes in frequency (Δf) is proportional to the elongation rate [38]. Due to the irreversible covalent attachment of the seeds [37,42], a constant ensemble of seeds can be probed repeatedly and changes in elongation rate upon changes in conditions, such as temperature and protein sequence [44], can be very accurately resolved. A previous study further revealed that the relative change in fibril elongation rate caused by a change in solution condition is independent of the fact whether the fibril elongates on a surface or in solution [47]. When the QCM is operated in air or in vacuum, the mass sensitivity is given by the Sauerbrey equation (Eq. (1)) [48]:  pffiffiffiffiffiffiffiffiffiffi 2 Δf ¼ ‐ 2f 0 =A ρq μ q Δm

ð1Þ

where f0 is the fundamental frequency, A the electrode area, ρ the density, and μ the shear modulus of the quartz crystal. For the sensors

242

C. Huynen et al. / Biochimica et Biophysica Acta 1854 (2015) 239–248

employed in our experiments, the mass sensitivity for a thin, rigid layer is 17.7 ng.Hz−1. However, it was recently reported that for amyloid fibril growth, the mass sensitivity can be higher by up to a factor of four compared to the simple Sauerbrey case [49]; this effect is mainly due to the roughness of the fibril layer that traps water which is moved with the crystal during the vibration. For the experiments described in the present study, the mass sensitivity and therefore the absolute rate of mass addition are however not required. The quartz crystal was inserted into a microbalance flow cell and was equilibrated with 700 μL of PBS pH 7.5 which was first pumped over the quartz crystal surface with a peristaltic pump at a flow rate of 0.4 mL.min− 1. The pump was then switched off and the PBS buffer was left in contact with the quartz crystal until a stable baseline was reached. 400 μL of a solution of one of the BlaP chimeras (~ 15 μM) were brought into contact with the quartz crystal (at a flow rate of 0.4 mL.min−1), the pump was then switched off, and the protein solution was left in contact with the quartz crystal for 15 min. Between each injection of protein solutions, the crystal was rinsed with 700 μL of the respective buffer: PBS pH 7.5 (for experiments under native conditions [39]), or PBS pH 7.5 containing 1.85 M urea (for experiments under denaturing conditions [39]) to wash the protein solution out of the liquid cell. The crystal was left in contact with the buffer for ~15 min until the baseline was stable. All experiments were performed at a controlled temperature of 25 °C or 37 °C. A given chimera was pumped 2 to 3 times over the quartz crystal during the same QCM experiment (i.e., on the same sensor) and each experiment was independently carried out in triplicate (i.e., on different sensors) using different batches of production for most of the chimeras. Since, as mentioned above and discussed in the results section, the density of seeds covalently attached at the sensor surface can vary from one sensor to another, elongation rates measured using different sensors cannot be directly compared with satisfying accuracy: the elongation rate, defined as Δf/Δt, measured for one given chimera at a specific concentration will vary from one sensor to another. Thus, the elongation rate measured for one chimera (e.g., BlaP197Q79), at least in triplicate on one quartz crystal, was used as reference (i.e., 100%) to determine the relative elongation rates of the other chimeras on the same sensor. Then the relative elongation rates measured on different sensors can be compared. To facilitate the presentation of the QCM results, all results are shown for the overtone N = 7 of the resonant frequency. 2.7. Atomic force microscopy AFM was used to image the fibrils, to determine the surface density of seeds on the QCM sensor after attachment, and to measure the length of the seed fibrils. AFM has already been successfully used in the past to investigate the morphology of species formed on the pathway of amyloid fibril formation [45,50,51]. Dried quartz crystals with attached fibrils were imaged by AFM (PicoPlus 5500, Keysight Technologies) operated in intermittent contact mode in air. AFM probes (POINTPROBE® PLUS Silicon-SPM-Sensor, NanoAndMore GmbH, Germany) with resonant frequencies of ~300 kHz were used. Comparison of the topography images recorded before and after QCM experiments was used to check whether or not the changes in the QCM frequency were indeed due to the elongation of the surface-bound seeds. For chimeras associated with measured elongation rates close to the limit of detection of QCM (i.e., BlaP197Q30), we compared the topography images of sensor surfaces with attached seed fibrils recorded before and after an extended period of incubation with a 15 μM solution of the chimera. The surfaces were treated exactly in the same way as for elongation experiments monitored by QCM (i.e., incubation with MUA in ethanol, then with EDC and NHS, then addition of seed fibrils, and finally passivation with ethanolamine, as described in Section 2.5). Comparison of the average lengths of the attached fibrils before and after incubation with the chimera allows confirming the absence or presence of effective elongation. A negative control experiment was carried out

by incubating a sensor without immobilized seed fibrils with BlaP197Q30 for 24 h; this sensor was treated in the same way as all the others (incubation with MUA in ethanol, then with EDC and NHS and finally passivation with ethanolamine) except that the incubation with a fibril-suspension was omitted. AFM images were analyzed using the software Gwyddion, version 2.36. When indicated, the length of the seeds was calculated and averaged from 40 to 50 independent seeds or fibrils present at the surface of the quartz crystal, using this software. 3. Results 3.1. Optimization of the surface concentration of fibrils immobilized onto the gold surface of quartz crystals Initial experiments were performed in order to optimize the density of seed fibrils that are immobilized onto the gold surface of the quartz crystals to allow both a significant elongation rate to be measured by QCM, and to be visualized by AFM. For this purpose, amyloid fibrils made of BlaP197Q79 were first formed under native conditions (PBS pH 7.5, 37 °C) and purified as described previously [39], and the presence of fibrillar aggregates was demonstrated by TEM (not shown). Fibrils were sonicated before being covalently attached to the gold surface of the quartz crystals. 100 μL of seeds at three different concentrations of total protein, 0.2, 0.5 and 1 mg.mL−1, were brought into contact with the chemically activated surface of the quartz crystal. AFM images (Fig. S1) show that seeds immobilized onto the sensors at each of the three concentrations used have an average length of ~ 60 nm (56 nm ± 23 nm, 58 nm ± 30 nm, and 58 nm ± 19 nm, respectively on Fig S1 A, B, and C). At a concentration of 1 mg.mL−1, the seeds are very densely packed on the sensor surface (Fig. S1 A). This concentration was therefore not used for elongation experiments, since it may lead to a rapid change in the number of growing surfacebound seeds during elongation experiments, due to surface crowding effects. In addition, such a high density renders precise determination of the average length increase by AFM difficult. On the other hand, small and isolated fibrils are observed when immobilized at 0.2 or 0.5 mg.mL−1 (Fig. S1 B and C). Moreover, at these two concentrations, the density of seeds on the gold surface was found high enough to record a significant and reproducible decrease in the resonant frequency upon addition of the monomers of BlaP197Q79 (Fig. 2A). For all experiments described in the next sections, the immobilization was therefore carried out with 100 μL of seeds at 0.2 mg.mL−1. Following such an immobilization, the density of seed fibrils was found by AFM imaging to be homogeneous at the surface of each individual quartz crystal, but was subject to a certain variability between different quartz crystals. The surface density varies approximatively from 38 ± 6 to 207 ± 27 seed fibrils per μm2 (the standard deviation is given from calculation at four different locations on the sensor surface). The largest variability factor of seed fibril surface density between two different sensors is therefore of 5.5. This variability can be due to several factors such as the use of different fibril batches as seeds whose concentration may vary slightly, the previous treatment of the gold surface (whether a new sensor or a cleaned used sensor is employed), the efficiency of the sonication treatment, errors in the volume of fibrils deposited on the surface, and/or variations of the incubation time of the sensor with the fibrils. 3.2. Elongation of preformed fibrils of BlaP197Q79 by the BlaP197 chimeras under native conditions The ability of the different BlaP197 chimeras to elongate immobilized fibrils made of BlaP197Q79 was first investigated under native conditions (PBS pH 7.5, 37 °C) [39]. The addition of a ~15 μM solution of BlaP197Q79 and BlaP197Q55 on immobilized preformed fibrils made by BlaP197Q79 leads to a significant linear decrease in the resonant frequency (Fig. 2 A). This linear

C. Huynen et al. / Biochimica et Biophysica Acta 1854 (2015) 239–248

243

Fig. 2. A. Representative graph of QCM experiments carried out under native conditions (PBS pH 7.5, 37 °C) with solutions of BlaP197Qx, with x = 79 (Q79, 15.3 μM), 55 (Q55, 14.5 μM), 30 (Q30, 15.6 μM), 23 (Q23, 16.8 μM), and 0 (Q0, 14.9 μM). The black curve corresponds to the overtone 7 of the resonant frequency; the red lines indicate the linear regressions associated with the addition of monomers (f(t) = y0 + at; a = Δf.Δt−1, proportional to the elongation rate); the colored zones highlight the periods of time during which the sensor is incubated with the solutions of chimeras; the white zones correspond to the periods of washing of the QCM flow cell and sensor surface and equilibration with PBS pH 7.5. B–E, AFM topography images of quartz crystals: B, before (T0) and after (Tf) a QCM experiment performed sequentially with BlaP197Q79 and BlaP197Q55, C, D and, E, before and after 24 h of incubation of BlaP197Q30 (C and D) and BlaP197Q0 (E), at 15 μM (PBS pH 7.5, 37 °C), at the surface of a quartz sensor on which BlaP197Q79 fibrils were immobilized (C and E) or not (D). The white double arrows represent the fibril length.

decrease is reproducibly observed for the different injections of these chimeras during one single experiment and for independent repeats of these experiments (data not shown). This is interpreted as a mass increase at the quartz crystal surface associated with fibril elongation. AFM images confirm that the decrease of the resonant frequency, observed during QCM experiments, is associated with the elongation of the seeds. Indeed, the length of the seeds was found to have increased from 56 nm ± 23 nm to 193 nm ± 30 nm on average (Fig. 2 B). Injection of a solution of BlaP197Q30 induces a much lower decrease of the resonant frequency as a function of time (Fig. 2A): the elongation rate only represents 9 ± 6% of the elongation rate measured with a solution of BlaP197Q79 (Table S1). This chimera therefore seems to be able to elongate preformed fibrils of BlaP197Q79 only at a very low rate. This elongation rate being within the limit of detection of the QCM experiment, the ability of BlaP197Q30 to elongate BlaP197Q79 fibrils was further checked by AFM. A quartz crystal surface, on which BlaP197Q79 fibrils were immobilized, was imaged before and after 24 h of incubation with 100 μL of a 15 μM solution of BlaP197Q30 (Fig. 2C). The length of the fibrils is significantly increased after 24 h of incubation with BlaP197Q30 (from 59 nm ± 23 nm to 178 nm ± 57 nm), confirming that BlaP197Q30 elongates preformed fibrils. On

the other hand, no fibrils are observed on a quartz crystal surface without attached seed fibrils incubated for 24 h with BlaP197Q30 (Fig. 2 D). Moreover, the surface density of seed fibrils of the sensor with BlaP197Q79 fibrils immobilized is the same before and after the 24 h incubation with BlaP197Q30, i.e., 22 ± 6 and 20 ± 7 seed fibrils per μm2, respectively (the standard deviation is given from calculation at four different locations on the sensor surface). Altogether, these observations indicate that the elongated fibrils observed on Fig. 2 C are due to elongation of the attached seed fibrils and not due to spontaneous formation in solution or on the surface. On the other hand, the injection of BlaP197Q0 and BlaP197Q23 monomers is not associated with a significant decrease in the resonant frequency as a function of time (Fig. 2 A): the rate of the resonant frequency change (Δf.Δt−1) associated with the injection of BlaP197Q0 and BlaP197Q23 is similar to that observed for the buffer alone, and represents respectively 0.4 ± 6.2% and −0.4 ± 11.3% of the one measured for BlaP197Q79. These two chimeras cannot therefore elongate preformed fibrils made of BlaP197Q79. The absence of elongation of BlaP197Q79 fibrils upon 24 h of incubation with a 15 μM solution of BlaP197Q0 is confirmed by AFM imaging (Fig 2 E). These results indicate that there is a threshold for the elongation of fibrils under native

244

C. Huynen et al. / Biochimica et Biophysica Acta 1854 (2015) 239–248

conditions for BlaP197 chimeras, which is comprised between 23 and 30Q. Note that the elongation process described above is not reversible since no shift of the resonant frequency is observed upon both washing the protein solution off the sensor and equilibrating the sensor with PBS. Moreover, the elongation rate is not significantly affected whether BlaP197Q79 elongates seeds after itself, after BlaP197Q30 or after BlaP197Q55 (Fig. 2A). This observation is also true for BlaP197Q55 and BlaP197Q30. Furthermore, when comparing the polyQ length dependence of the relative elongation rate, a linear correlation is observed (Table S1 and Fig. 3): the relative elongation rate decreases by ~1.9% per each deleted glutamine residue, relative to the reference elongation rate of BlaP197Q79 under native conditions. 3.3. Effect of the unfolding of the BlaP moiety of the BlaP197 chimeras on the elongation of preformed fibrils Scarafone et al. have previously demonstrated that the native conformation of BlaP imposes constraints on the 55Q tract inserted in position 197 that block its ability to induce the formation, in solution, of amyloid fibrils. Once the BlaP moiety is unfolded, the 55Q tract can however trigger the aggregation into amyloid fibrils of BlaP197Q55 [39]. To better understand the role of the constraints imposed by the native conformation of BlaP on the elongation step, elongation of BlaP197Q79 fibrils by the 197 chimeras of BlaP was monitored by QCM under denaturing conditions (~ 15 μM, 1.85 M urea, PBS pH 7.5, 25 °C). At this urea concentration, it was previously determined that ca. 50% of molecules are unfolded for BlaP197Q30, BlaP197Q55 and BlaP197Q79, while ca. 18% of BlaP197Q23 molecules are unfolded, and all molecules of BlaP197Q0 are native; the latter being used as negative control [39]. On each sensor used for this study, the elongation rate of a 15 μM solution of BlaP197Q79 was measured under native conditions (PBS pH 7.5, 0 M urea, 25 °C), and serves as a positive control. 3.3.1. Effect of the conformation of BlaP on the Q-threshold for fibril elongation The injection of a solution of BlaP197Q55 and BlaP197Q79 onto the quartz surface is reproducibly associated with a decrease in frequency

Fig. 4. A. Representative graphs of QCM experiments carried out under denaturing conditions (PBS pH 7.5, 1.85 M urea, 25 °C) with BlaP197Qx, with x = 79 (Q79), 55 (Q55), 30 (Q30), 23 (Q23), and 0 (Q0), at 15 μM. The black curve corresponds to the overtone 7 of the resonant frequency; the red lines indicate the linear regressions associated with the addition of monomers (f(t) = y0 + at; a = Δf.Δt−1, proportional to the elongation rate); the colored zones highlight the periods of time during which the sensor is incubated with the solutions of chimeras; the white zones correspond to the periods of washing of the QCM flow cell and sensor surface and equilibration with PBS pH 7.5, 1.85 M urea. B, AFM topography images of BlaP197Q79 seeds immobilized at the quartz crystal surface before (T0) and after (Tf) a QCM experiment performed with BlaP197Q79, BlaP197Q55, and BlaP197Q30 in the presence of 1.85 M urea. White arrows indicate seeds at T0 and fibrils at Tf.

in the presence of 1.85 M urea (Fig. 4A). This observation indicates that these two chimeras significantly elongate preformed fibrils made of BlaP197Q79 under denaturing conditions. AFM images of the sensor surface confirm that the QCM signal is associated with fibril elongation (Fig. 4 B). Furthermore, the injection of a solution of BlaP197Q30 in the presence of 1.85 M urea induces a decrease in the resonant frequency (Fig. 4 A). Its associated slope (Δf.Δt−1) represents 24 ± 6% of the slope measured upon injection of BlaP197Q79 in the presence of 1.85 M urea (Table S1). On the other hand, in the presence of 1.85 M urea, the injection of BlaP containing 0 and 23Q is not associated with a significant decrease in the resonant frequency with time (Fig. 4A): these two chimeras cannot elongate preformed fibrils under these conditions. The elongation threshold under denaturing conditions for BlaP197 chimeras is therefore comprised between 23 and 30Q, which is in the same range as the one obtained under native conditions. The denaturation of the BlaP moiety therefore does not significantly affect the elongation Q-threshold-value.

Fig. 3. Relative elongation rates as a function of the polyQ length for BlaP -polyQ chimeras containing the polyQ tract in position 197 under native conditions (ο, – –), and denaturing conditions (▲, ∙∙∙), or in position 216 under native conditions (•, —), or for polyQ peptides (×, – —, data calculated from Slepko et al., [35]).

3.3.2. Effect of the conformation of BlaP on the elongation rate Comparing the relative elongation rate of the different BlaP197 chimeras under denaturing conditions with their respective elongation rate under native conditions (Fig. 5) indicates that: (i) increasing the urea concentration from 0 to 1.85 M significantly decreases the fibril elongation rate by BlaP197Q79 at 25 °C; (ii) in contrast, the rate of fibril elongation by BlaP197Q55 and BlaP197Q30 in the presence of 1.85 M urea is significantly higher than that of their respective monomers

C. Huynen et al. / Biochimica et Biophysica Acta 1854 (2015) 239–248

245

Fig. 5. Relative elongation rates monitored by QCM of seeds made of BlaP197Q79 by BlaP197Qx, with x = 79 (197Q79), 55 (197Q55) and 30 (197Q30), at 15 μM, under native and denaturing conditions (PBS pH 7.5, 25 °C, 0 and 1.85 M urea respectively). For a given chimera (BlaP197Q79, BlaP197Q55 or BlaP197Q30), the slopes (Δf.Δt−1) measured under native conditions, corresponding to the elongation rates, are averaged and normalized to 100%; this value is used as the reference. The relative elongation rate of the same chimera in the presence of 1.85 M urea is given as a percentage of the ratio between the averaged slopes in the presence of 1.85 M and the averaged slopes of the reference.

determined in the absence of urea at 25 °C, i.e., it represents 140 ± 10% and 600 ± 300% respectively of the elongation rate in the absence of urea (Fig. 5, Table S1). The unfolding of the BlaP moiety therefore decreases the free energy barriers associated with fibril elongation by the 30Q and 55Q tract variants. Most interestingly, by comparing the elongation rate measured in the presence of 1.85 M urea for the different BlaP197 chimeras to that measured for BlaP197Q79 (Table S1 and Fig. 3), we observe that the relative elongation rate is still linearly correlated to the polyQ length, from 30 to 79Q. However, the slope of the linear regression is lower than that determined under native conditions: the elongation rate increases by ~1.5% per Q residue (Fig. 3). Altogether these results suggest a dual role of urea in the elongation of preformed fibrils by BlaP chimeras. First, urea is likely to reduce the strength of the intermolecular interactions between the monomer and the fibril end. This effect results in a decrease of the elongation rate in the presence of 1.85 M urea. Second, the urea-induced unfolding of the BlaP moiety reduces the constraints (conformational and/or steric) applied to the polyQ tract. This should consequently increase the elongation rate by monomers of BlaP chimeras. The net effect of urea on the elongation rate results from the sum of these two opposing effects and varies depending on the polyQ length: the longer the polyQ within BlaP, the more pronounced the effect of weakened intermolecular interactions and the less pronounced the effect of reduced constraints. 3.4. Influence of the location of the polyQ insertion inside BlaP on the elongation of preformed fibrils To investigate the influence of the location of the insertion of the polyQ tract inside BlaP (216 versus 197) on the elongation process, elongation experiments of BlaP197Q79 fibrils were performed under native conditions with chimeras of BlaP containing 0, 30, 55 and 79Q in position 216. On each sensor used for this study, the elongation rate of a 15 μM solution of BlaP197Q79 was measured under native conditions (PBS pH 7.5, 0 M urea, 37 °C), and serves as a positive control and reference. 3.4.1. Effect of the polyQ insertion location on the Q-threshold for fibril elongation The addition of a solution of BlaP216Q55 and BlaP216Q79 to BlaP197Q79 fibrils immobilized onto the quartz crystal is associated

Fig. 6. A, C, Representative graph of QCM experiments carried out under native conditions (PBS pH 7.5, 37 °C) with BlaP197Q79 (Q79), and BlaP216Qx, with x = 79 (Q79), 55 (Q55), 30 (Q30) and 0 (Q0). The black curve corresponds to the overtone 7 of the resonant frequency; the red lines indicate the linear regressions associated with the addition of monomers (f(t) = y0 + at; a = Δf.Δt−1, the elongation rate); the colored zones highlight the periods of time during which the sensor is incubated with the solutions of chimeras; the white zones correspond to the periods of washing of the QCM flow cell and sensor surface and equilibration with PBS pH 7.5. B, AFM topography images of BlaP197Q79 seeds immobilized at the quartz crystal before (T0) and after (Tf) a QCM experiment performed with BlaP197Q79, BlaP216Q79 and BlaP216Q55. White arrows indicate seeds at T0 and fibrils at Tf.

with a significant decrease in the resonant frequency as a function of time (Fig. 6 A). These proteins therefore readily elongate BlaP197Q79 fibrils, and the elongation is confirmed by AFM imaging: fibrils have been elongated from 65 nm ± 22 nm to 267 nm ± 65 nm on average (Fig. 6 B). A smaller decrease in the resonant frequency is also observed after injection of BlaP216Q30 monomers: its elongation rate represents 20 ± 6% of the one measured for the positive control, BlaP197Q79 (Fig. 6C, Table S1). Consequently, like BlaP197Q30, BlaP216Q30 is able to slowly elongate preformed fibrils of BlaP197Q79. On the contrary, the injection of BlaP216Q0 does not lead to a significant decrease of the QCM resonant frequency (Fig. 6C). This chimera is therefore not able to elongate the seeds made of BlaP197Q79. All together these results show that the threshold for fibril elongation is below 30Q for both BlaP197 and BlaP216 chimeras under native conditions. 3.4.2. Effect of the location of the polyQ tract on the elongation rate We then compared the relative elongation rates of BlaP197Q79 seeds determined for the different BlaP197 and BlaP216 chimeras (Fig. 7 and Table S1). Interestingly, the relative elongation rate measured for BlaP216Q30, BlaP216Q55, and BlaP216Q79 is significantly higher than that measured for chimeras having respectively 30, 55 and 79Q in position 197. Indeed, for BlaP chimeras containing 30, 55

246

C. Huynen et al. / Biochimica et Biophysica Acta 1854 (2015) 239–248

Fig. 7. Relative elongation rates monitored by QCM of seeds made of BlaP197Q79 by BlaP chimeras containing 0, 23, 30, 55 and 79Q in position 197 or 216 (197Q0, 197Q23, 197Q30, 197Q55, 197Q79, 216Q0, 216Q30, 216Q55 and 216Q79, respectively). The slopes (Δf.Δt−1) of the reference (BlaP197Q79, PBS pH 7.5, 37 °C), corresponding to the elongation rates, are averaged and normalized to 100%. The relative elongation rates of the different chimeras are given as percentages of the ratio between their averaged slopes and the averaged slopes of the reference.

and 79Q in position 216, it is 2.3 to 3.5 times higher than the rate recorded for their counterpart in position 197. Moreover, the elongation rate of seeds by BlaP216Q79 is similar whether it elongates seeds after the elongation by BlaP197Q79 or after the elongation by BlaP216 chimeras. This is also true for the elongation by BlaP216Q55. Finally, we observed that the relative elongation rate recorded for the different BlaP216 chimeras is again linearly correlated to the number of glutamines (Table S1 and Fig. 3). Most interestingly, the slope of the linear regression is virtually identical to that observed for BlaP197 chimeras under native conditions, i.e., ~1.9% per Q residue. 4. Discussion In this work, we have investigated the ability of a series of BlaPpolyQ chimeras to elongate seed fibrils made of BlaP197Q79, i.e., the β-lactamase BlaP containing 79 glutamine residues in position 197. We have specifically investigated the importance of the polyQ length, its location within the host protein, and the conformational state of the host protein for the elongation of amyloid fibrils. The first conclusion is that for both BlaP197 and BlaP216 chimeras, there is a threshold in the polyQ repeat length to elongate preformed fibrils. This threshold is situated below 30Q independently of the location of the polyQ tract within BlaP and independently, at least for BlaP197 chimeras, of the conformation of the BlaP moiety. A Q-threshold for fibril elongation was already observed for polyQ peptides [35,52]. Nonetheless this threshold, comprised between 10 and 15Q, is significantly lower than the threshold determined here for BlaP-polyQ chimeras. This finding clearly indicates that non-polyQ regions affect the ability of the polyQ repeat to elongate preformed fibrils. Importantly, the Qthreshold for fibril elongation by BlaP-polyQ chimeras is significantly lower than the threshold observed for their spontaneous aggregation into amyloid fibrils ([39], Thorn et al., in preparation, Table 1). Indeed, a 30Q tract inserted within BlaP, either in position 197 or 216, is not able to induce aggregation and amyloid fibril formation in solution, independently of the conformation of the BlaP moiety. Moreover, BlaP197Q55 forms amyloid fibrils in solution only if the BlaP moiety is unfolded, it otherwise forms amorphous aggregates [39]. These

observations indicate that the native conformation of the BlaP moiety applies constraints to the 30Q tract inserted at both locations and to the 55Q tract inserted at position 197 preventing them from forming ordered amyloid fibrils. However, in the present work, we show that these polyQ tracts can elongate preformed fibrils. This finding suggests that the nucleation step requires a larger conformational flexibility of the polyQ tract than the elongation step and that it is likely that seed fibrils assist the polyQ tracts within BlaP, via templating glutamine-glutamine interactions, to overcome the energy barrier associated with the conversion into the conformation that the monomer needs to adopt in the fibril [52,53]. Finally, BlaP197Q79 elongates fibrils faster under native conditions than in the presence of 1.85 M urea, while it spontaneously forms amyloid fibrils faster in the presence of 1.85 M urea [39]. These observations therefore suggest that the constraints applied by the native BlaP moiety also influence the ability of the 79Q tract to nucleate amyloid fibril formation. The ability of a polyQ tract to nucleate fibril formation results therefore from a delicate balance between the propensity of the polyQ tract to form intermolecular interactions and the constraints applied by the host protein to the polyQ tract. As previously shown for the full process of fibril formation [39], the longer the polyQ tract, the less pronounced the effects of the constraints. Our results also demonstrate the importance of the host protein moiety for the efficiency of polyQ tracts to elongate amyloid fibrils. First, the relative elongation rate measured for BlaP197Q30 and BlaP197Q55 is increased once the BlaP moiety is unfolded in the presence of 1.85 M urea in comparison to the rates measured under native conditions by a factor 6 ± 3 and 1.4 ± 0.1 respectively (Fig. 5, Table S1). The native structure of BlaP is therefore likely to impose constraints to the 30 and 55Q tracts that reduce their efficiency to elongate amyloid fibrils. Second, the fibril elongation rate is significantly faster for BlaP chimeras containing a given expanded polyQ tract inserted in position 216 than in position 197. This observation is in agreement with the fact that BlaP216Q79 aggregate, in solution, more rapidly into amyloid fibrils than BlaP197Q79 (Thorn et al., in preparation). This difference in the aggregation propensity of polyQ tracts inserted at position 197 and 216 under native conditions could be attributed to the differences in the protein environment of the loop within which they reside. Indeed, while the 197 insertion site is located within a folded domain (i.e., the α-domain), the 216 insertion site is located at the interface between the α-domain and the α/β-domain. The insertion at the latter location could perturb the whole interface between the two folded domains of BlaP that may lead to a significant decrease in the constraints applied to the polyQ tract (Thorn et al., in preparation), and may therefore result in a higher elongation propensity of these BlaP-polyQ chimeras. The difference in aggregation propensity of the two sets of chimeras could also depend on the inherent properties of amino acid sequences flanking the polyQ tract at position 197 and 216. The flanking sequences of the first exon of huntingtin (htt1) have been shown to significantly contribute to the aggregation propensity of htt1. The 17 amino acids N-terminal (N17) to the polyQ tract enhance the overall rate of aggregation of htt1 [18] while the C-terminal proline rich region of htt1 modulates the type and the toxicity of aggregates formed [13]. Crick et al., recently demonstrated that the N- and C-terminal flanking sequences of the polyQ tract in htt1 interfere differently with the aggregation properties of the polyQ tract. The 17 amino acids N-terminal to the polyQ tract decrease the saturation concentration of soluble htt1 and accelerate fibril formation by destabilizing non-fibrillar intermediates. On the other hand, the 38 amino acids C-terminal (C38) to the polyQ tract increase the htt1 solubility and thus decrease the driving force for insoluble aggregation [25]. The authors proposed that, together, the N17 and C38 promote the formation of ordered amyloid fibrils, at the expense of the formation of heterogeneous amorphous aggregates. Therefore, in a similar manner, the differences in aggregation propensity of BlaP197 and BlaP216 chimeras could arise from differences in solubilizing properties of 197 and 216 flanking sequences.

C. Huynen et al. / Biochimica et Biophysica Acta 1854 (2015) 239–248

Moreover, we observed that (i) a given chimera containing a polyQ tract at position 216 elongates as efficiently seeds that have been previously elongated by BlaP197Q79 or by any other BlaP216 chimeras (i.e., with a different polyQ tract) and (ii) a given BlaP197 chimera elongates as efficiently seeds that have been elongated by any other BlaP197 chimeras (i.e., with a different polyQ tract). This suggests that the polyQ flanking sequences are not part of the core of the fibrils. The core is likely to be formed by polyQ β-strands of roughly the same size independently of the total length of the polyQ expansion within the chimera; longer polyQ tracts are likely to organize in more β-strands. This suggestion, together with the observation that the Q-threshold is smaller for fibril elongation than for the spontaneous aggregation into amyloid fibrils, further supports the recruitment mechanism that explains part of cytotoxicity associated with polyQ proteins [52,54,55]; amyloid fibrils made of a specific polyQ protein are able to recruit the same protein containing a non-pathological polyQ tract, or to recruit a different polyQ protein. Recruitment of these proteins would reduce their total cellular activity or alter their properties and thus contribute to the pathology [55]. Finally, we observed that the relative elongation rate linearly increases with the number of glutamine residues, from 30 to 79Q, for both BlaP197 and BlaP216 chimeras, independently of the conformation of the BlaP moiety (i.e., native or unfolded) (Fig. 3). Under conditions where the BlaP moiety is native, the slope of the linear fit, i.e., ~ 1.9% per Q residue, is independent of the position of the polyQ tract within BlaP, 197 or 216. A correlation between the polyQ length and the elongation rate was previously observed for polyQ peptides by Walters et al. with 20 to 24Q peptides, and by Slepko et al. with 10 to 40Q peptides [34,35]. Moreover, Slepko et al. already demonstrated the linear correlation between the polyQ length and the elongation rate with 10 to 40Q peptides; the slope of the linear fit, i.e., ~ 2.1% per Q residue, is again very similar to that measured with BlaP-polyQ chimeras (Fig. 3) [34]. Moreover, this observation indicates that determination of the elongation rate of a polyQ peptide/protein under a particular set of conditions will allow predicting the elongation rate of the same peptide/protein with a polyQ tract of different length. Thus, we show for the first time, to our knowledge, the linear correlation, in a protein context, between the rate of elongation of fibrils and the polyQ sequence length. In contrast to the results we described for the elongation of fibrils, spontaneous fibril formation in solution by polyQ proteins and peptides does not linearly depend on the polyQ length [20,56,57,59]: large changes in aggregation propensity per additional glutamine are observed at low repeat lengths, while only small changes per added residue occur for long tracts. This nonlinear dependence of the nucleation step on the polyQ length has been related to the structure and size of the aggregation nucleus that depends on the number of glutamines in the repeat [58]. 5. Conclusion Our results demonstrate that the linear correlation between the polyQ length and the fibril elongation rate is independent of the environment of the polyQ: it is observed both when the polyQ is free or inserted within a host protein, and independently of its location within the protein, and the conformational state of its host protein. However, we also demonstrate that non-polyQ regions drastically influence the elongation properties of polyQ sequences: (i) the Q-threshold for fibril elongation is significantly increased when the polyQ tract is embedded into a protein, (ii) the native conformational state of the host protein decreases the ability of short polyQ sequences to elongate fibrils, and (iii) the elongation rates are highly sensitive to the location of the polyQ tract within the host protein. The comparison of our results with spontaneous aggregation in solution also suggests that non-polyQ regions interfere with the formation of the aggregation nucleus by applying conformational constraints to relatively short polyQ tracts. Our results therefore suggest that non-polyQ regions constitute an additional potential therapeutic target to design drugs more specific than those

247

targeting polyQ sequences, to interfere with the nucleation and elongation of amyloid fibrils by polyQ proteins (e.g., ligands that are specific to non-polyQ regions of polyQ disease-associated proteins that further increase constraints applied to the polyQ tract) to prevent and/or slow down respectively the disease development. Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.bbapap.2014.12.002.

Acknowledgements The authors thank Céline Falentin for her assistance with the QCM instrument, Fabrice Boulienne and Anne-Marie Matton for their assistance with the purification of proteins, and David Thorn for critical reading. This work was supported by grants from Fonds de la Recherche Fondamentale et Collective (CJ and 2.4581.12 F to MD), Fonds de la Recherche Scientifique (FRS-FNRS, 1.C039.09 and MIS-F.4505.11 to MD), Fonds Spéciaux from the University of Liege (11/108 to MD) and the Belgian program of Interuniversity Attraction Poles administered by the Federal Office for Scientific Technical and Cultural Affairs (P7/4444 and 7/05). CH is a Research Fellow of FRS-FNRS and MD is a Research Associate of FRS-FNRS. AKB thanks Magdalene College, Cambridge, and the Leverhulme Trust for support. The funding sources had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

References [1] H.Y. Zoghbi, H.T. Orr, Glutamine repeats and neurodegeneration, Annu. Rev. Neurosci. 23 (2000) 217–247. [2] S.L. Hands, A. Wyttenbach, Neurotoxic protein oligomerisation associated with polyglutamine diseases, Acta Neuropathol. 120 (4) (2010) 419–437. [3] H.T. Orr, H.Y. Zoghbi, Trinucleotide repeat disorders, Annu. Rev. Neurosci. 30 (2007) 575–621. [4] P.O. Bauer, N. Nukina, The pathogenic mechanisms of polyglutamine diseases and current therapeutic strategies, J. Neurochem. 110 (6) (2009) 1737–1765. [5] J.M. Ordway, et al., Ectopically expressed CAG repeats cause intranuclear inclusions and a progressive late onset neurological phenotype in the mouse, Cell 91 (6) (1997) 753–763. [6] Y.W. Chen, K. Stott, M.F. Perutz, Crystal structure of a dimeric chymotrypsin inhibitor 2 mutant containing an inserted glutamine repeat, Proc. Natl. Acad. Sci. U. S. A. 96 (4) (1999) 1257–1261. [7] C.A. Ross, Polyglutamine pathogenesis: emergence of unifying mechanisms for Huntington's disease and related disorders, Neuron 35 (5) (2002) 819–822. [8] J.B. Penney Jr., et al., CAG repeat number governs the development rate of pathology in Huntington's disease, Ann. Neurol. 41 (5) (1997) 689–692. [9] L. Schols, et al., Autosomal dominant cerebellar ataxias: clinical features, genetics, and pathogenesis, Lancet Neurol. 3 (5) (2004) 291–304. [10] S.W. Davies, et al., Formation of neuronal intranuclear inclusions underlies the neurological dysfunction in mice transgenic for the HD mutation, Cell 90 (3) (1997) 537–548. [11] R. Wetzel (Ed.)Misfolding and aggregation in Huntington disease and other expanded Polyglutamine repeat diseases, K.J. D.C., M. Ramirez-Alvarado (Eds.),Protein Misfolding Diseases: Current and Emerging Principles and Therapies, 2010, pp. 305–324. [12] C.A. Ross, Intranuclear neuronal inclusions: a common pathogenic mechanism for glutamine-repeat neurodegenerative diseases? Neuron 19 (6) (1997) 1147–1150. [13] A. Bhattacharyya, et al., Oligoproline effects on polyglutamine conformation and aggregation, J. Mol. Biol. 355 (3) (2006) 524–535. [14] K. Nozaki, et al., Amino acid sequences flanking polyglutamine stretches influence their potential for aggregate formation, Neuroreport 12 (15) (2001) 3357–3364. [15] L. Masino, et al., Solution structure of polyglutamine tracts in GST-polyglutamine fusion proteins, FEBS Lett. 513 (2–3) (2002) 267–272. [16] D. Bulone, et al., The interplay between PolyQ and protein context delays aggregation by forming a reservoir of protofibrils, PLoS ONE 1 (2006) e111. [17] M.F. Perutz, et al., Glutamine repeats as polar zippers: their possible role in inherited neurodegenerative diseases, Proc. Natl. Acad. Sci. U. S. A. 91 (12) (1994) 5355–5358. [18] A.K. Thakur, et al., Polyglutamine disruption of the huntingtin exon 1 N terminus triggers a complex aggregation mechanism, Nat. Struct. Mol. Biol. 16 (4) (2009) 380–389. [19] A.L. Robertson, et al., The rate of polyQ-mediated aggregation is dramatically affected by the number and location of surrounding domains, J. Mol. Biol. 413 (4) (2011) 879–887. [20] A.L. Robertson, et al., The structural impact of a polyglutamine tract is locationdependent, Biophys. J. 95 (12) (2008) 5922–5930. [21] A.L. Robertson, S.P. Bottomley, Towards the treatment of polyglutamine diseases: the modulatory role of protein context, Curr. Med. Chem. 17 (27) (2010) 3058–3068.

248

C. Huynen et al. / Biochimica et Biophysica Acta 1854 (2015) 239–248

[22] M. Tanaka, et al., Intra- and intermolecular beta-pleated sheet formation in glutamine-repeat inserted myoglobin as a model for polyglutamine diseases, J. Biol. Chem. 276 (48) (2001) 45470–45475. [23] R.P. Menon, et al., Mapping the self-association domains of ataxin-1: identification of novel non overlapping motifs, Peer J. 2 (2014) e323. [24] A.M. Ellisdon, B. Thomas, S.P. Bottomley, The two-stage pathway of ataxin-3 fibrillogenesis involves a polyglutamine-independent step, J. Biol. Chem. 281 (25) (2006) 16888–16896. [25] S.L. Crick, et al., Unmasking the roles of N- and C-terminal flanking sequences from exon 1 of huntingtin as modulators of polyglutamine aggregation, Proc. Natl. Acad. Sci. U. S. A. 110 (50) (2013) 20075–20080. [26] M.D. Tobelmann, R.M. Murphy, Location trumps length: polyglutamine-mediated changes in folding and aggregation of a host protein, Biophys. J. 100 (11) (2011) 2773–2782. [27] S. Chen, F.A. Ferrone, R. Wetzel, Huntington's disease age-of-onset linked to polyglutamine aggregation nucleation, Proc. Natl. Acad. Sci. U. S. A. 99 (18) (2002) 11884–11889. [28] A.L. Robertson, S.P. Bottomley, Molecular pathways to polyglutamine aggregation, Adv. Exp. Med. Biol. 769 (2012) 115–124. [29] K. Sugaya, S. Matsubara, Nucleation of protein aggregation kinetics as a basis for genotype-phenotype correlations in polyglutamine diseases, Mol. Neurodegener. 4 (2009) 29. [30] E. Scherzinger, et al., Self-assembly of polyglutamine-containing huntingtin fragments into amyloid-like fibrils: implications for Huntington's disease pathology, Proc. Natl. Acad. Sci. U. S. A. 96 (8) (1999) 4604–4609. [31] S. Barton, et al., The length dependence of the polyQ-mediated protein aggregation, J. Biol. Chem. 282 (35) (2007) 25487–25492. [32] J.D. Harper, P.T. Lansbury Jr., Models of amyloid seeding in Alzheimer's disease and scrapie: mechanistic truths and physiological consequences of the time-dependent solubility of amyloid proteins, Annu. Rev. Biochem. 66 (1997) 385–407. [33] M.F. Perutz, Glutamine repeats and inherited neurodegenerative diseases: molecular aspects, Curr. Opin. Struct. Biol. 6 (6) (1996) 848–858. [34] N. Slepko, et al., Normal-repeat-length polyglutamine peptides accelerate aggregation nucleation and cytotoxicity of expanded polyglutamine proteins, Proc. Natl. Acad. Sci. U. S. A. 103 (39) (2006) 14367–14372. [35] R.H. Walters, et al., Elongation kinetics of polyglutamine peptide fibrils: a quartz crystal microbalance with dissipation study, J. Mol. Biol. 421 (2–3) (2012) 329–347. [36] C. Huynen, et al., Class a-lactamases as versatile scaffolds to create hybrid enzymes: applications from basic research to medicine, Biomed. Res. Int. 2013 (2013) 827621. [37] A.K. Buell, C.M. Dobson, M.E. Welland, Measuring the kinetics of amyloid fibril elongation using quartz crystal microbalances, Methods Mol. Biol. 849 (2012) 101–119. [38] T.P. Knowles, et al., Kinetics and thermodynamics of amyloid formation from direct measurements of fluctuations in fibril mass, Proc. Natl. Acad. Sci. U. S. A. 104 (24) (2007) 10016–10021. [39] N. Scarafone, et al., Amyloid-like fibril formation by polyQ proteins: a critical balance between the polyQ length and the constraints imposed by the host protein, PLoS ONE 7 (3) (2012) e31253.

[40] R.P. Ambler, et al., A standard numbering scheme for the class A beta-lactamases, Biochem. J. 276 (Pt 1) (1991) 269–270. [41] M. Vandevenne, et al., The Bacillus licheniformis BlaP beta-lactamase as a model protein scaffold to study the insertion of protein fragments, Protein Sci. 16 (10) (2007) 2260–2271. [42] A.K. Buell, et al., Surface attachment of protein fibrils via covalent modification strategies, J. Phys. Chem. B 114 (34) (2010) 10925–10938. [43] W.P. Hu, et al., Kinetic analysis of beta-amyloid peptide aggregation induced by metal ions based on surface plasmon resonance biosensing, J. Neurosci. Methods 154 (1–2) (2006) 190–197. [44] A.K. Buell, et al., Population of nonnative states of lysozyme variants drives amyloid fibril formation, J. Am. Chem. Soc. 133 (20) (2011) 7737–7743. [45] M.B. Hovgaard, et al., Quartz crystal microbalance studies of multilayer glucagon fibrillation at the solid-liquid interface, Biophys. J. 93 (6) (2007) 2162–2169. [46] J.A. Kotarek, K.C. Johnson, M.A. Moss, Quartz crystal microbalance analysis of growth kinetics for aggregation intermediates of the amyloid-beta protein, Anal. Biochem. 378 (1) (2008) 15–24. [47] A.K. Buell, et al., Electrostatic effects in filamentous protein aggregation, Biophys. J. 104 (5) (2013) 1116–1126. [48] G. Sauerbrey, Verwendung von Schwingquarzen zur Wägung dünner Schichten und zur Mikrowägung, Z. Phys. 155 (1959) 206–222. [49] T.C.T. Michaels, et al., Quantitative analysis of diffusive reactions at the solid-liquid interface in finite systems, J. Phys. Chem. Lett. 5 (2014) 695–699. [50] K.K. Sweers, et al., Atomic force microscopy under controlled conditions reveals structure of C-terminal region of alpha-synuclein in amyloid fibrils, ACS Nano 6 (7) (2012) 5952–5960. [51] R. Khurana, et al., A general model for amyloid fibril assembly based on morphological studies using atomic force microscopy, Biophys. J. 85 (2) (2003) 1135–1144. [52] S. Chen, et al., Polyglutamine aggregation behavior in vitro supports a recruitment mechanism of cytotoxicity, J. Mol. Biol. 311 (1) (2001) 173–182. [53] K. Kar, et al., d-polyglutamine amyloid recruits l-polyglutamine monomers and kills cells, J. Mol. Biol. 426 (4) (2014) 816–829. [54] E. Preisinger, et al., Evidence for a recruitment and sequestration mechanism in Huntington's disease, Philos. Trans. R. Soc. Lond. B Biol. Sci. 354 (1386) (1999) 1029–1034. [55] M.K. Perez, et al., Recruitment and the role of nuclear localization in polyglutaminemediated aggregation, J. Cell Biol. 143 (6) (1998) 1457–1470. [56] K. Kar, et al., Critical nucleus size for disease-related polyglutamine aggregation is repeat-length dependent, Nat. Struct. Mol. Biol. 18 (3) (2011) 328–336. [57] B.S. Heck, F. Doll, K. Hauser, Length-dependent conformational transitions of polyglutamine repeats as molecular origin of fibril initiation, Biophys. Chem. 185 (2014) 47–57. [58] E. Landrum, R. Wetzel, Biophysical underpinnings of the repeat length dependence of polyglutamine amyloid formation, J. Biol. Chem. 289 (15) (2014) 10254–10260. [59] J.R. Knox, P.C. Moews, Beta-lactamase of Bacillus licheniformis 749/C Refinement at 2A resolution and analysis of hydration, J Mol Biol 202 (2) (1991) 435–455.