ABSTRACT

The accumulation of insoluble amyloids in the pancreatic islets is a pathological hallmark of type II diabetes and correlates closely with the loss of β-cell mass. The predominant component of these amyloid deposits is the islet amyloid polypeptide (IAPP). The factors contributing to the conversion of IAPP from a monomeric bioactive peptide hormone into insoluble amyloid fibrils remain partially elusive. In this study, we investigated the effect of the oxidative non-enzymatic post-translational modification induced by the reactive metabolite 4-hydroxynonenal (HNE) on IAPP aggregation and cytotoxicity. Incubation of IAPP with exogenous HNE accelerated its self-assembly into β-sheet fibrils and led to the formation of a Michael adduct on the His-18 side chain. To model this covalent modification, the imidazole N(π) position of histidine was alkylated using a close analogue of HNE, the octyl chain. IAPP lipidated at His-18 showed a hastened random coil-to-β-sheet conformational conversion into fibrillar assemblies with a distinct morphology, a low level of binding to thioflavin T, and a high surface hydrophobicity. Introducing an octyl chain on His-18 enhanced the ability of the peptide to perturb synthetic lipid vesicles, to permeabilize the plasma membrane, and to induce the death of pancreatic β-cells. Alkylated IAPP triggered the self-assembly of unmodified IAPP by prompting primary nucleation and increased its capacity to perturb the plasma membrane, indicating that only a small proportion of the modified peptide is necessary to shift the balance toward the formation of proteotoxic species. This study underlines the importance of studying IAPP post-translational modifications induced by oxidative metabolites in the context of pancreatic amyloids.

 key physiological hallmark of type 2 diabetes (T2D) consists of the accumulation of amyloid fibrils in the pancreatic islets.¹ While the causative link between amyloid formation and T2D pathogenesis remains the subject of active debates, the progressive deposition of insoluble protein aggregates is closely linked with the gradual loss of insulin- producing β-cells.² The main proteinaceous constituent of pancreatic amyloid deposits is the islet amyloid polypeptide (IAPP), or amylin IAPP is a 37-residue peptide hormone that is co-synthesized, co-packaged, and co-secreted with insulin by pancreatic β-cells and is implicated in the regulation of satiety and glucose homeostasis. This natively disordered polypeptide that transiently populates helical secondary structures is highly prone to self-assembly into amyloid fibrils.⁵ The association between pancreatic amyloid deposits and the progression of T2D has historically led to the postulate that amyloid fibrils are causing degeneration of β-cells.⁶ However, recent studies have revealed that the cytotoxicity of IAPP is dependent on its quaternary assembly state and that the toxic proteospecies are mainly the transient nonfibrillar oligomers,whereas defined fibrils are poorly toxic. IAPP rarely deposits in the islets of Langerhans of healthy individuals, although its totoxicity. Identifying these factors, or conditions, of the pancreatic microenvironment modulating IAPP self-assembly and associated toxicity is critical for better defining the role(s) of IAPP deposition in T2D etiology.Studies have revealed that T2D patients are afflicted with chronic oxidative stress, as exemplified by the increase in the levels of nitrotyrosine and oxidized low-density lipoprotein and by the decrease in the use of antioxidant defense mecha-

 MATERIALS AND METHODS

Synthesis  of  N(α)-Fmoc-[N(π)-(octyl)]-L-histidine.Modification of IAPP with an alkane analogue of HNE was performed by alkylation on the histidine imidazole N(π) groupnisms.Several conditions are known to increase oxidative stress in the pancreatic islets, including local inflammation, hyperglycemia, high levels of free fatty acid, and lip- otoxicity.²¹⁻²⁴ In turn, the loss of balance of the redox by adapting a previously reported method.³⁸ Briefly, N(α)- Fmoc-N(τ)-(trityl)-L-histidine was methylated using a DMF/ MeOH (50:50) solution containing N,N-diisopropylethylamine (DIPEA, 1.1 equiv) and 2-(6-chloro-1H-benzotriazol-homeostasis can lead to an excessive generation of damagingreactive oxygen species (ROS), which promotes non- enzymatic post-translational modifications (PTMs) of proteins, such as methionine oxidation and tyrosine nitration, and the formation of oxidized metabolites.²⁵ Interestingly, accumu-1-yl)-1,1,3,3-tetramethylaminium hexafluorophosphate (HCTU, 1.1 equiv). The solvent was removed on a rotary evaporator, and the residue was partitioned with ethyl acetate (EtOAc) and water. The organic extract was washed with aqueous 0.1 M HCl, NaHCO , and NaCl successively, dried lation of intracellular ROS and lipid peroxidation products has (MgSO ), and concentrated been observed for immortalized pancreatic β-cells treated with in vacuo. The N(π)-alkylated exogenous IAPP. In Goto-Kakizaki rats, a non-obese model of T2D, high levels of the oxidative stress markers 8- oxo-2′-deoxyguanosine and 4-hydroxynonenal (HNE) were product was obtained using in situ-generated alkyl triflate. To a stirred solution of triflic anhydride (1.1 equiv) in CH₂Cl₂ under nitrogen at −75 °C was added dropwise a solution of 1-octanol (1.1 equiv) and DIEA (1.1 equiv) over 10 min. The measured in the islets of Langerhans. HNE is a prevalent and very toxic lipid-derived aldehyde formed by the oxidation of ω-6 polyunsaturated fatty acids.²⁹ This highly reactive aldehyde is known to covalently modify a diversity of biomacromolecules, altering their physicochemical properties and biological functions.

 HNE spontaneously forms stable reaction mixture was stirred at −75 °C for 20 min. A solution of N(α)-Fmoc-N(τ)-(trityl)-L-histidine methyl ester (0.78 mmol, 1 equiv) in CH₂Cl₂ was added then dropwise, and the mixture was allowed to return to room temperature (RT) and stirred overnight. The mixture was poured into aqueous NaHCO and then extracted with CH Cl . Deprotection was adducts with proteins by reacting with the nucleophilic groups  of cysteine, histidine, and lysine, via the formation of a Schiff base or through Michael addition.²⁹ In diabetic patients afflicted with Alzheimer’s disease (AD), it has been observed that neuronal IAPP formed adducts with HNE,³¹ although the performed by treatment with a solution of CH₂Cl₂, trifluoro- acetic acid (TFA), and triisopropylsilane (TIS) [35:3.5:1 (v/v/ v)] for 2 h at RT. The solvent was removed in vacuo, and the mixture was purified by silica gel column chromatography bygradient elution with a CH Cl /MeOH mixture (from 99:1 to effect of this modification on IAPP aggregation has not been investigated. In addition, the oxidative metabolite HNE has been shown to modulate the self-assembly and/or cytotoxicity of amyloidogenic proteins whose aggregation is associated with (neuro)degenerative diseases. For instance, immunohisto- chemical analyses of brain samples from AD patients revealed high levels of HNE in association with amyloid plaques, while biophysical studies demonstrated that the HNE modification of the amyloid-β (Aβ) peptide promotes its aggregation, increases its affinity for lipid membranes, and contributes to its toxicity.Similarly, high levels of HNE−protein conjugates have been reported within nigral neurons of patients afflicted with Parkinson’s disease.³⁶ Incubation of α-synuclein with HNE led to multiple Michael adducts, promoting the 90:10). The identity of the compound was validated by high- resolution mass spectrometry (HRMS) and nuclear magnetic resonance (NMR): ¹H NMR (600 MHz, CDCl₃) δ 8.01 (s, 1H), 7.69 (d, J = 7.2 Hz, 2H), 7.49 (d, J = 7.2 Hz, 2H), 7.33 (t,

J = 7.2 Hz, 2H), 7.24 (t, J = 7.2 Hz, 2H), 6.90 (s, 1H), 5.67 (d,

J = 6.6 Hz, 1H), 4.51 (d, J = 6.6 Hz, 1H), 4.33 (m, 2H), 4.12

(t, J = 6.6 Hz, 1H), 3.84 (d, J = 7.2 Hz, 2H), 3.70 (s, 3H), 3.09

(ddd, J = 42.6, 10.2, 5.4 Hz, 2H), 1.67 (m, 2H), 1.20 (m, 10H), 0.79 (t, J = 6.6 Hz, 3H); ¹³C NMR (300 MHz, CDCl₃) δ 170.86, 155.82, 143.62, 143.52, 141.35, 127.85, 127.13,

124.96, 122.72, 120.08, 67.21, 53.13, 53.01, 47.05, 46.04,

46.02, 31.66, 30.46, 29.03, 28.96, 26.50, 22.57, 14.04.

Demethylation was performed under nucleophilic conditions to give the final protected amino acid. Lithium iodide (6 formation of compact oligomers with high mesencephalic cells.toxicity equiv) was added to a stirred solution of N(α)-Fmoc-N(τ)- (octyl)-L-histidine methyl ester in dry EtOAc under argon, and

the reaction mixture was heated at reflux for 20 h. The reaction was quenched by the addition of 10% aqueous HCl, and the mixture extracted with EtOAc. The final product was purified by silica gel flash chromatography by gradient elution with a CH₂Cl₂/MeOH mixture (95:5 to 80:20), and the identity of the compound was validated by HRMS.

Peptide Synthesis, Purification, and Characterization. Peptides were synthesized on a solid support based on Fmoc chemistry and the HCTU coupling strategy, as previously described.³⁹ Pseudoproline dipeptide derivatives were incorpo- rated to facilitate peptide synthesis.⁴⁰ Crude peptides were purified by preparative HPLC using a C18 column with a linear gradient of acetonitrile in H₂O/TFA [at 0.6% (v/v)]. Collected fractions were analyzed by analytical HPLC using a C18 column and then confirmed by accurate mass measure- ments by analytical LC-MS using an Agilent 1200 series HPLC instrument coupled to an electrospray time-of-flight mass spectrometer (ESI-TOF). Fractions corresponding to the desired peptide with a purity of >95% were pooled and lyophilized. Peptides were cyclized with 100% DMSO overnight and then diluted to be purified by HPLC a second time.

IAPP Monomerization and Sample Preparation. Peptides were dissolved in 100% hexafluoro-2-propanol (HFIP) at a concentration of 1 mg/mL. The HFIP solution was then filtered through a 0.22 μm hydrophilic polyvinylidene difluoride (PVDF) filter and sonicated for 30 min before being lyophilized. The resulting peptide powder was solubilized for a second time in HFIP (1 mg/mL) and sonicated for 30 min, and the solution was aliquoted and lyophilized again. Monomerized  peptides  were  stored  dry  at  −20  °C,  but  for no longer than 4 weeks.

Formation of Amyloid Fibrils. Stock solutions were prepared by dissolving the monomerized lyophilized peptide at a concentration of 50 μM, or as otherwise specified, in 20 mM Tris-HCl (pH 7.4). HNE-IAPP was generated by incubating the peptide at 50 μM with 1 mM HNE in 20 mM Tris-HCl (pH 7.4). Self-assembly occurred over a 48 h period, or as otherwise specified, at RT under quiescent conditions, i.e., without agitation. After the indicated incubation times, aliquots were taken, diluted in Tris-HCl buffer, and analyzed by fluorescence and circular dichroism (CD) spectroscopy, atomic force microscopy (AFM), and transmission electron microscopy (TEM).

Identification of the Site of HNE Modification by LC-

MS/MS. IAPP (HNE-treated and the respective control) was solubilized in 2% TFA in 20% methanol and digested using pepsin [enzyme:protein ratio of 1:40 (w/w)] in a total volume of   400   μL   overnight   (37   °C,   pH   2−2.5).   Digestion   was stopped by adding 500 μL of water, and the digest was cleaned up on a 1 cm³ (30 mg) Oasis hydrophilic−lipophilic balanced (HLB) SPE cartacetonitrile and injected (20 μL) onto an Aeris PEPTIDE XB- C18 100 mm × 2.1 mm column, with solid core 1.7 μm particles (100 Å) using a Nexera Shimadzu UHPLC system with water (A) and acetonitrile (B), both containing 0.1% modification on cysteine, histidine, and lysine residues. The search was performed for +2 to +4 charge states at a MS tolerance of 0.05 Da on precursor ions and 0.1 Da on fragments. Detected HNE-modified peptides were confirmed upon exact mass and charge state and by being absent in the control sample using PeakView software. A list including the retention time and m/z values of the verified peptides was created, and samples were re-injected for higher-quality MS/ MS data. A second IDA analysis was performed for the four most intense precursor ions (excluded after three occurrences) with an MS/MS accumulation time of 150 ms and a total cycle time of 0.9 s.

Fluorescence Spectroscopy. During a specific incubation period, peptide solutions were mixed with 8-anilino-1- naphthalenesulfonic acid (ANS) or thioflavin T (ThT)  to yield final dye concentrations of 50 and 40 μM, respectively. Excitation wavelengths were 355 and 440 nm for ANS and ThT, respectively. Emission spectra from 385 to 585 nm (ANS) and from 450 to 550 nm (ThT) were recorded. The kinetics of amyloid self-assembly monitored by ThT fluorescence  were  carried   out  at  25   °C  in   sealed  96-well black-wall, clear-bottom, and non-binding surface microplates, as previously described.⁴² The excitation wavelength was set at 440 nm, and the emission at 485 nm was recorded every 10 min. Data obtained from triplicate wells were averaged, corrected by subtracting the corresponding control reaction, and plotted as fluorescence versus time.

Circular Dichroism Spectroscopy. During a specific incubation period, peptide solutions were incorporated into a 2 mm path length quartz cell. Far-ultraviolet (far-UV) CD spectra were recorded from 190 to 260 nm using a model J-815 CD spectropolarimeter. The wavelength step was set at 0.5 nm with an average time of 10 s/scan at each wavelength step. Each collected spectrum was background subtracted with peptide-free buffer. Raw data were converted to mean residue ellipticity (MRE): formic  acid,  at  a  flow  rate  of  300  μL/min  (40  °C).  The gradient started at 5% B, was held for 2.5 min, and was linearly increased to 30% B at 40 min, to 50% B at 26 min, and to 85% B at 26.5 min. MS and MS/MS spectra were recorded on a Sciex high-resolution quadrupole time-of-flight (Q-TOF) TripleTOF 5600 mass spectrometer equipped with a DuoSpray ion source in positive ion mode set at a source voltage of 5 kV, a source temperature of 500 °C, and GS1/GS2 gas flows of 50 psi, with a declustering potential of 80 V. The instrument performed a survey TOF-MS acquisition from m/z 140 to 1250 (accumulation time of 250 ms), followed by MS/ MS on the 15 most intense precursor ions from m/z 250 to 1250 (excluded for 20 s after two occurrences) using information-dependent acquisition (IDA) with dynamic back- ground subtraction. Each MS/MS acquisition had an accumulation time of 50 ms and a collision energy of 30 ± 10 V. The total cycle time was 1.05 s. MS/MS data were searched against a user-created IAPP. FASTA file by ProteinPilot software using Paragon algorithm⁴¹ with a detection protein threshold of unused ProtScore > 0.05 (confidence > 10%). The protein search algorithm was changed to consider a probability of 0.80 for HNE

The CD spectra were smoothed with the Savitsky−Golay algorithm using a smoothing moving window of 11 data points.Atomic Force Microscopy. Peptide solutions were diluted in 1% acetic acid to yield a final concentration of 10 μM and immediately applied to freshly cleaved mica. Micas were washed twice with deionized water and air-dried for 24 h. Images were acquired on a Veeco/Bruker multimode AFM instrument using ScanAsyst-air mode with a silicon tip (2−12 nm tip radius, 0.4 N/m force constant) on a nitride lever. Images were taken at 0.5 Hz and 1024 scans/min.

Transmission Electron Microscopy. Peptide solutions were diluted to 10 μM and applied on a glow-discharged copper carbon-coated 300 mesh grid. Samples were negatively stained with 1.5% (w/v) uranyl formate for 1 min and air-dried for 24 h. Images were recorded using a FEI Tecnai G2 Spirit Twin microscope microscope operating at 120 kV and equipped with a Gatan Ultrascan 4000 4k × 4k CCD camera system.

Critical Aggregation Concentration. Pyrene was solu- bilized in ethanol (1 mM) and diluted in 20 mM Tris-HCl (pH 7.4). Monomerized peptides were solubilized in 100% HFIP and sonicated for 5 min before being solubilized into the

Leakage of Large Unilamellar Vesicles. Large uni- lamellar vesicles (LUVs) were formed using 1,2-dioleoyl-sn- glycero-3-phosphocholine (DOPC) and 1,2-dioleoyl-sn-glyc- ero-3-phosphoglycerol (DOPG) at different molar ratios and supplemented, or not, with 20% cholesterol (Chol). Lipids were solubilized in 100% chloroform, and the solvent was evaporated with a nitrogen gas stream. The lipid film was then rehydrated in 20 mM Tris-HCl buffer (pH 7.4) containing 70 mM calcein for 30 min. The solution was frozen and thawed five times before being extruded through a 100 nm nucleopore membrane for 15 cycles. Non-encapsulated calcein was removed using a Sephadex G-25 column. The lipid concentration was determined using an inorganic phosphate detection colorimetric assay. LUV solutions were used immediately, and fresh LUVs were used for each experiment. Peptide solutions were prepared by dissolving the lyophilized powder in 100% DMSO and sonicated for 5 min before being diluted in 20 mM Tris-HCl (pH 7.4) containing 150 mM NaCl. The final peptide concentration was 25 μM with 0.5% DMSO, and LUVs were added to reach a final lipid concentration of 500 μM. The control used to determine 100% leakage (F_max) consisted of calcein and LUVs treated with 0.2% Triton X-100. Dye leakage was reported using the equation

% of leakage =F − Fbaseline

Fmax − Fbaseline

Cellular Assays. Rat INS-1 cells were seeded in black-wall, clear-bottom, 96-well plates (tissue culture-treated) at a density of 15,000 cells/well in RPMI-1640 medium supplemented with 10% FBS, 100 units/mL penicillin, 100 mg/mL streptomycin, 10 mM HEPES, 1 mM sodium

RESULTS

The Lipid-Derived Aldehyde HNE Hastens IAPP Aggregation into β-Sheet-Rich Fibrils. Previous studiesbhave revealed that the reactive lipid peroxidation products, HNE and 4-oxo-2-nonenal (ONE), modulate the self-assembly of amyloidogenic polypeptides, including the Aβ peptide,α-synuclein,³⁷ and the immunoglobulin light chain. We initially investigated how exogenous HNE affects IAPP aggregation into cross-β quaternary structures. IAPP amyloid formation in the presence or absence of HNE was evaluated over incubation time by monitoring the fluorescence of ThT, a small dye whose fluorescence increases upon binding to cross- β structures.⁴⁷ Recording the change in ThT fluorescence intensity is a method commonly used to monitor the kineticsof amyloid formation.⁴² The addition of 20 equiv of HNE accelerated the formation of ThT positive IAPP proteospecies  In fact, in the presence of HNE, the ThT signal peaked after incubation for 6 h under quiescent conditions, while in the presence of the EtOH vehicle control, the plateau of ThT fluorescence occurred after incubation for 12 h. Similar results were observed by measuring the formation of hydrophobic clusters by ANS. Interestingly, the presence of HNE in the aggregation mixture induced the formation of IAPP aggregates with greater exposure of hydrophobic clusters, as observed with the ANS fluorescence intensity measured after incubation for 24 h . The secondary conformational transition associated with amyloid formation was evaluated by CD spectroscopy. As expected, immediately after its solubilization in Tris-HCl buffer, monomerized IAPP displayed a random coil conformation, as inferred from the CD spectra characterized with a single minimum at ∼200 nm

After incubation for 6 h in the absence of HNE, a partial conformational shift was observed, i.e., a decrease in the magnitude of the random coil signal and the appearance of a β- sheet signal at 220 nm. In sharp contrast, after incubation for 6 h in the presence of HNE, IAPP has completely shifted toward a β-sheet-rich secondary structure. It is worth mentioning that Effect of histidine alkylation on IAPP amyloid self-assembly. (A−C) Peptides were incubated at 25 °C in 20 mM Tris-HCl (pH 7.4). (A) ThT fluorescence, ANS fluorescence, and far-UV CD spectra of 25 μM IAPP and H18-Oc IAPP immediately after solubilization (0 h) and after incubation for 24 h. (B) Representative ThT kinetics of amyloid formation at a low peptide concentration (4 μM). (C) Representative TEM images of IAPP and H18-Oc IAPP immediately after solubilization (0 h) and after incubation for 24 h. The scale bar is 200 nm. (D) Critical aggregation concentration (CAC) of IAPP and H18-Oc determined by pyrene fluorescence.

the presence of EtOH [3% (v/v)] used to solubilize HNE accelerated IAPP amyloid formation . The accelerating effect of HNE on IAPP amyloid formation was evaluated by imaging the aggregates assembled after 6 and 24 h by AFM and TEM. As shown in panels D and E of , the presence of HNE promoted the formation of long and linear amyloid fibrils. Microscopy analyses revealed that the terized by a double bond and a carbonyl group, as for HNE, allows the localization of partial positive charge at the β- carbon, rendering it strongly electrophilic.⁵¹ Addition of HNE to Cys, His, and Lys side chains occurs mainly via Michael addition, while Schiff base formation involving the Lys primary amine group can also take place. Michael adducts represent the vast majority of HNE-mediated protein modifications.

 The presence of HNE in the aggregation mixture did not IAPP sequence contains four residues susceptible to profoundly modify the morphology of the quaternary assemblies, as both samples showed a mixture of twisted and flat ribbon fibrils, characteristic of the polydispersity of IAPP fibrils assembled in vitro.^48,49 Overall, these results indicate that the reactive aldehyde HNE accelerates the aggregation of IAPP into prototypical β-sheet fibrillar structures.

HNE Selectively Modifies Histidine 18. The high reactivity of HNE is associated with its three functional groups, the aldehyde, the C  C bond, and the hydroxyl. Each of these groups can participate alone, or in sequence, in chemical reactions with different macromolecules depending on electron delocalization. A conjugated system charac-HNE-mediated Michael addition: Lys-1, Cys-2, Cys-7, and His-18. Cys-2 and Cys-7 are involved in a disulfide bond, which decreases significantly their reactivity. To identify the main sites of covalent modification, IAPP was incubated in the presence of HNE and analyzed by LC-MS/MS after pepsin digestion. Following database searching, two peptides, both containing His-18, were confidently identified as HNE- modified (with 99% confidence).  the high- resolution extracted ion chromatograms (±0.01 Da) of each of these two peptides (¹⁶LVHSSNNFGAIL²⁷ and ¹⁵FLVHSSNN- FGAIL²⁷) in their unmodified and HNE-modified forms in the HNE-treated IAPP sample. The control. Alkylation of His-18 prevents the inhibitory effect of acidic pH on IAPP amyloid formation. (A−C) Peptides were incubated at 25 °C in 20 mM acetate buffer (pH 4.5) at 25 μM, and at 0 and 24 h, the aggregation mixture was analyzed by (A) ThT and ANS fluorescence and far-UV CD spectroscopy, and (B) TEM (scale bar of 200 nm). (C) Representative ThT kinetics of amyloid formation of IAPP and the H18-Oc derivative (25 μM) in 20 mM acetate buffer (pH 4.5).

Synthesis of IAPP Specifically Alkylated at His-18. To investigate the molecular mechanisms of the effect of HNE on

IAPP self-assembly, we prepared a substituted IAPP derivative by introducing an octyl chain on the imidazole ring of His-18 (H18-Oc IAPP). This strategy was inspired by the work of Qahwash and colleagues, who used the octanoyl group as a model to study the effects of HNE on Aβ aggregation and cytotoxicity.⁵⁴ To facilitate the synthesis of H18-alkylated IAPP, we first modified the imidazole ring in solution before introducing the Fmoc-protected alkylated His during elonga- tion on the solid support. We used a regiospecific synthetic approach to prepare the N(π)-substituted histidine  Briefly, after methyl esterification of commercially available N(α)-Fmoc-[N(τ)-Trt]-protected histidine (1), alkylated product 3 was obtained using in situ-generated alkyl triflate. Removal of side chain Trt protecting group was performed with TFA to give product 4 and demethylation was performed under nucleophilic con- ditions to give the final Fmoc-protected His (5) in a satisfactory yield of approximatively 70%. Two-dimensional NMR analysis (NOESY) of product 4 confirmed the alkylation at the N(π) position of the imidazole ring of His . An additional correlation was observed between the proton aindicative of a dipolar interaction. Purified N(α)-Fmoc-[N(π)-octyl]-His (5) was introduced into IAPP synthesis on Rink Amide resin, yielding H18-Oc IAPP.

Site-Specific Alkylation of His-18 Accelerates IAPP Amyloid Formation. Having access to an IAPP derivative specifically alkylated on the His-18 imidazole ring, we could now probe the mechanistic effects of HNE conjugation on IAPP self-assembly and associated toxicity. The addition of an octyl   chain   to   the   His-18   side   chain   increases   IAPP hydrophobicity, which required presolubilization of the monomerized peptide in HFIP or DMSO. Thus, IAPP and H18-Oc IAPP were first solubilized in 100% HFIP or DMSO at 5 mM and then diluted into 20 mM Tris-HCl buffer (pH 7.4) to reach a final peptide concentration of 25 μM and 0.5% (v/v) organic solvent.  alkylation dramatically altered the kinetics of IAPP fibrillogenesis. In fact, immediately after the addition of H18-Oc IAPP to the aqueous buffer (time 0 h), ThT and ANS fluorescence positive signals were obtained . In comparison to unmodified IAPP, the proteospecies generated after incubation of the H18-Oc IAPP derivative for 24 h were characterized by low ThT binding capacity and high ANS fluorescence, suggesting that the alkylation of IAPP induces the formation of aggregates with a low capacity to bind ThT and high surface hydrophobicity. CD analyses revealed that His-18 alkylation confines IAPP to a secondary β-sheet-rich conformation immediately after solubilization. Interestingly, no prototypical lag phase was observed for H18-Oc IAPP in the ThT kinetics of amyloid formation, even at a low peptide concentration of 4 μM ). By TEM imaging, short (proto)fibrils could be observed immediately after solubiliza- tion of H18-Oc IAPP, whereas no aggregates were visible for the unmodified peptide. TEM images also revealed macroscopic differences between IAPP and H18-Oc IAPP fibrils assembled after incubation for 24 h. While IAPP formed long, thick, and polymorphic fibrils, H18-Oc fibrils were shorter and thinner.  Effect of site-specific alkylation of His-18 on IAPP cytotoxicity and lipid membrane perturbation. (A) INS-1E cells were treated with different peptide concentrations, and the cell viability was measured after treatment for 24 h. (B−E) INS-1E cells were treated with 25 μM IAPP, or H18-Oc IAPP, and the cell viability and plasma membrane perturbation were measured after different incubation times. (C) Quantification of cell viability measured by the Live/Dead assays and (D) representative confocal images of INS-1E cells treated with 25 μM IAPP.

The scale bar is 20 μm. (E) Membrane perturbation evaluated by LDH activity in the cell media. In panels A−C and E, data represent the means ± SEM of at least four individual experiments performed in triplicate. (F) Leakage of calcein from 500 μM DOPC/DOPG (90:10) and DOPC/DOPG/Chol (70:10:20) LUVs induced by 25 μM peptide.average height of 6.1 ± 2.3 nm, while H18-Oc fibrils displayed an average height of 3.2 ± 1.4 nm . Finally, the CAC was evaluated using pyrene fluorescence, which probes the formation of micellar oligomeric aggregates known to correlate closely with the amyloidogenic propensity.⁵⁵ The observed CAC for H18-Oc was ∼10 times lower than that observed for the unmodified IAPP , indicating that alkylation of the His-18 imidazole ring potentiates IAPP aggregation. It is worth mentioning that fibrils assembled from H18-Oc IAPP diverge from the fibrils formed from wild-type IAPP in the presence of HNE. In fact, self-assembly of IAPP in the presence of soluble HNE led to a dense network of ThT positive long fibrils , whereas H18-Oc fibrils were shorter with a weak ThT signal . This divergence could be related to the fact that HNE modification on the His- 18 imidazole ring occurs concurrently with IAPP self-assembly and that a small proportion of IAPP molecules is modified prior to nucleation and elongation, in agreement with the MS data .

Moreover, the differences in the phys- icochemical properties between HNE and the octanoyl chain can also lead to structural and morphological variations within the resulting fibrils. Additionally, HNE can lead to intra- and intermolecular cross-linkage, which is not the case for the alkyl chain. In addition, both fibril preparations, i.e., IAPP incubated in the presence of HNE and H18-Oc IAPP, showed seeding capacities similar to that of fibrils assembled from the unmodified IAPP.

Alkylated IAPP Self-Assembles at Acidic pH. In vitro at neutral pH, IAPP readily self-assembles into amyloid fibrils, even at concentrations in the low micromolar range. In sharp contrast, in the secretory granules of pancreatic β-cells, IAPP remains in a quasi-soluble state, although its local concentration is estimated to be roughly between 0.8 and 4 mM.⁶ It has been proposed that the pH value of 5.5 of the β-cell granules contributes to the inhibition of IAPP aggregation⁵⁶ In fact, protonation of the imidazole ring of His-18 at acidic pH has been shown to prevent IAPP aggregation, through electrostatic repulsion between adjacent monomers.

 Accordingly, we evaluated how site-specific alkylation modu- lates the propensity of IAPP to self-assemble at acidic pH by incubating the H18-Oc derivative in a pH 4.5 acetate buffer. As previously reported, incubation of unmodified IAPP at acidic pH inhibited amyloid formation . On the contrary, H18-Oc IAPP formed linear fibrils characterized by a high surface hydrophobicity and a β-sheet-rich secondary structure. Nonetheless, the ThT fluorescence intensity of H18- Oc IAPP fibrils assembled at pH 4.5 is reduced, compared to that of the ThT signal measured at neutral pH . This difference in the final ThT fluorescence intensity could be ascribed to alterations in the affinity of the dye and/or binding mode, caused by the protonation of the dimethylamino group of ThT.Thus, as previously reported with the introduction of an ethyl acetate group on His-18,⁶⁰ alkylation of His-18 prevents the inhibitory effect of acidic pH on IAPP aggregation.

This observation suggests that generation of the reactive lipid aldehyde HNE within the secretory granules could trigger IAPP amyloid formation upon its covalent conjugation at His-18.His-18 Alkylation Potentiates the Cytotoxicity ofIAPP and Its Capacity to Perturb Lipid Membranes. As shown above, site-specific alkylation of the imidazole ring of His-18 hastens IAPP amyloid formation and decreases the critical concentration to form micellar aggregates, likely by increasing peptide hydrophobicity. In turn, this could severely H18-Oc IAPP accelerates unmodified IAPP amyloid formation and potentiates its cytotoxicity. (A and B) IAPP and a mixture of IAPP and H18-Oc IAPP (99:1) were incubated at a final concentration of 25 μM in 20 mM Tris-HCl buffer (pH 7.4), and after incubation for 0, 5, and 24 h, aliquots of the aggregation mixtures were taken and analyzed by (A) ThT and ANS fluorescence and far-UV CD spectroscopy. (B) TEM images of 25 μM unmodified IAPP incubated alone or in the presence of 1% H18-Oc IAPP after 5 h. The scale bar is 200 nm. (C) Representative kinetics of amyloid formation of IAPP and an IAPP/H18-Oc IAPP mixture (99:1) at a final concentration of 25 μM monitored by ThT fluorescence. (D) Viability of INS-1E cells after peptide treatment. (E) Perturbation of the plasma membrane of INS-1E cells measured by the LDH activity in the media after different treatment times with the peptide. (D and E)

The total peptide concentration for IAPP and IAPP with H18-Oc IAPP is 25 μM. Data represent means ± SEM of at least four individual experiments performed in triplicate. impact the capacity of the peptide to perturb the plasma membrane, which is one of the proposed upstream mechanisms associated with IAPP-induced pancreatic cell death,⁹ among engagement of the RAGE receptor,⁶¹ activation of the pro-apoptotic FAS receptor,⁶² and defects in activity after these treatment periods . By fluorescence microscopy, we observed that the decrease in metabolically active cells in green (calcein-AM) and the loss of plasma membrane integrity in red (ethidium homodimer-1) occur after treatment for 3 h with the alkylated IAPP autophagy.⁶³ Accordingly, we compared the cytotoxicity of the H18-Oc IAPP derivative to that of its unmodified counterpart using β-cell line INS-1E. A concentration-depend- ent decrease in cell viability was observed for both peptides upon treatment for 24 h.

Strikingly, alkylated IAPP was significantly more toxic than unmodified IAPP for all concentrations evaluated. Next, the cell viability after different treatment periods with monomerized peptides was evaluated by means of measurement of the metabolic activity and the Live/Dead assay. Interestingly, 25 μM H18-Oc IAPP decreased the viability of INS-1E cells after incubation for 3 and 6 h, while the unmodified IAPP had no effect on metabolic. In contrast, for the unmodified IAPP, these events were observed only after treatment for 24 h. Next, perturbation of the plasma membrane induced by the peptide was evaluated by measuring the leakage of cytoplasmic LDH. The LDH activity measured in the cell media after treatment for 6 h indicated that both peptides avidly disrupt the plasma membrane . Again, H18-Oc IAPP induced a significantly higher LDH activity at short incubation times, i.e., 2 and 3 h . This observation suggests that alkylation of IAPP potentiates its capacity to perturb the plasma membrane, either by favoring the formation of membrane-active proteospecies or by increasing the capacity of the monomeric/oligomeric peptides to interact with the lipid bilayer. The capacity of the peptides to disrupt lipid membranes was further explored using synthetic anionic LUVs composed of DOPC and DOPG (9:1), a molar ratio of anionic lipids that closely mimics the composition of the pancreatic β-cell plasma membrane, which is estimated to be ∼2−13 mol % of total phospholipids.⁶⁴ In the presence of unmodified IAPP, the increase in fluorescence, which is associated with a decrease in the level of calcein self- quenching caused by its leakage from LUVs, occurred after incubation for 4 h .

In sharp contrast, H18-Oc IAPP induced an immediate increase in calcein fluorescence,with a quasi-absence of a lag phase, and greater perturbation after incubation for 14 h. The same trend was observed when the anionic DOPC/DOPG LUVs were supplemented with 20% cholesterol, which was used to better mimic the lipid composition of the eukaryotic plasma membrane and because cholesterol is known to modulate interaction of IAPP with lipid membranes. As it is known that the percentage of anionic lipids and the presence of cholesterol modulate the capacity of IAPP to induce membrane

ThT kinetics, with the T₅₀ shifting from 6.5 ± 0.1 to 3.2 ± 0.3 h with the addition of 1% H18-Oc IAPP . In addition, the final ThT fluorescence intensity after incubation for 16 h was decreased in the presence of 1% alkylated IAPP, suggestive of a change in the quaternary structure of the heterogeneous fibrils. These biophysical data indicate that alkylation of His-18 on a subpopulation of the IAPP peptide promotes the aggregation of unmodified IAPP through cross- assembly, which could have implications in the context of HNE conjugation of IAPP in T2D. Accordingly, we evaluated how H18-Oc IAPP modulates the cytotoxicity of its unmodified counterpart using INS-1E cells. Interestingly, a25 μM monomerized IAPP/H18-Oc IAPP mixture (99:1) induced a more pronounced decrease in cell viability after incubation for 24 h in comparison to that with 25 μM IAPP  It is worth mentioning that a 0.25 μM H18-Oc IAPP alone, which corresponds to the 1% molar ratio and is below the CAC for this peptide, had no effect on the viability of INS-1E cells. Evaluation of cytoplasmic LDH release upon treatment with the peptide mixtures revealed a majorpermeabilization,⁶⁷ we also used different lipid compositions of LUVs. Strikingly, for all lipid compositions evaluated, ranging from 2% to 30% of anionic lipids, H18-Oc IAPP showed an enhanced capacity to permeabilize LUVs compared to that of its unmodified counterpart. As observed by ANS fluorescence, the H18-Oc fibrils showed a high level of exposure of hydrophobic clusters , and thus, we evaluated the ability of preassembled H18-Oc fibrils to induce the perturbation of anionic LUVs. In contrast to fibrils assembled from the unmodified peptide, H18-Oc fibrils disrupted synthetic anionic LUVs, with or without supple- mental cholesterol . Nonetheless, the capacity of H18-Oc fibrils to disrupt synthetic lipid vesicles did not correlate with cytotoxicity toward pancreatic β-cells . This divergence could be ascribed to the complexity and heterogeneity of the cell plasma membrane, including the presence of proteoglycans known to exacerbate IAPP-mediated cytotoxicity.¹⁴ Taken together, these observations indicate that the site-specific alkylation of IAPP on His-18 significantly potentiates its ability to disrupt lipid membranes and to induce cell death.

Alkylated IAPP Accelerates the Self-Assembly of Unmodified IAPP and Increases Its Cytotoxicity. We next investigated if the co-incubation of a small molar ratio of H18-Oc IAPP with its unmodified counterpart could modulate its kinetics of self-assembly and toxicity. IAPP was incubated with the monomerized H18-Oc derivative at a 99:1 molar ratio (IAPP:H18-Oc IAPP; final concentration of 25 μM), and amyloid formation was followed by ThT and ANS fluorescence and CD spectroscopy. As observed by ThT and ANSfluorescence, the presence of 1% H18-Oc in the aggregation mixture accelerated unmodified IAPP amyloid formation, as observed by the plateau of fluorescence occurring after incubation for 5 h . Similarly, the presence of a small amount of His-18-alkylated IAPP accelerated the random coil-to-β-sheet conformational shift of unmodified IAPP . TEM images of the peptide solutions after incubation for 5 h showed the presence of a network of well- defined and long fibrils for IAPP incubated with 1% H18-Oc IAPP, while a mixture of short fibrils and oligomers was observed for the unmodified peptide incubated alone . An accelerating effect of the presence of H18-Oc IAPP on the kinetics of IAPP amyloid formation was also noticed by aprovoked by unmodified IAPP . In particular, the IAPP/H18-Oc IAPP mixture (99:1) induced significant plasma membrane perturbation after incubation for 2 and 3 h, while the effect of the homogeneous unmodified peptide solution was observed after 24 h. Again, 0.25 μM H18-Oc IAPP alone had not effect on the permeability of the INS-1E membrane , indicating that the effect is associated with cross- assembly. Overall, these results suggest that the modification of His-18 by HNE could trigger amyloid nucleation of nearby unmodified IAPP peptides and potentiate perturbation of lipid membranes.

DISCUSSION

Studies have reported that the oxidative stress in the pancreatic islets of T2D patients correlates with loss of insulin-secreting β-cells and amyloid deposition, suggesting that ROS and reactive metabolites could play an important role in disease etiology.¹ Interestingly, a close association between the reactive metabolite HNE, which is the product of the peroxidation of ω-6 polyunsaturated fatty acids, and protein misfolding has been observed in amyloid-related diseases. Nevertheless, although the peptide hormone IAPP has been reported to form adducts with HNE in vivo,³¹ the mechanistic contributions of HNE to IAPP aggregation has never been investigated. In the study presented here, we initially showed that the incubation of IAPP with exogenous HNE accelerates the formation of cross-β fibrils and leads to formation of HNE−IAPP covalent adducts. Previous studies have revealed that the effect of HNE on protein aggregation is modulated by the extent of conjugation, by the residues that are modified, and/or by the structural and physicochemical properties of the protein. For instance, modification by HNE promoted the formation of stable oligomers of α-synuclein,⁶⁸ while HNE conjugation of the immunoglobulin light chain at multiple His residues accelerated the conformational conversion associatedwith the formation of fibrillar aggregates.⁴⁶ Herein, we observed that the incubation of IAPP with HNE leads to the formation of a Michael adduct on His-18, whereas no Schiff base adducts were detected. HNE exhibits a high reactivity toward thiols and amines, reacting with nucleophilic residues in the following order: Cys > His > Lys.⁶⁹ Preferential reactivity toward Cys is ascribed to the selectivity principle, i.e.,that soft electrophiles react preferentially with soft nucleo- philes, and is dependent on the thiol−thiolate equilibrium.⁶⁹ IAPP encompasses two cysteine residues forming a disulfide bond within the N-terminal domain, making these thiols considerably less nucleophilic. Reaction of HNE with Lys can lead to Schiff base formation, although the kinetics of reaction is slow and the Schiff base products are reversible, making Michael adducts predominant,⁷⁰ as observed in the study presented here.

The multireactivity of HNE makes this aldehyde particularly efficient for the formation of intermolecular polypeptide cross- links through Michael addition at C-3 followed by Schiff base formation. Thus, cross-linkage could contribute, with the identified adduct, to the effect of HNE on IAPP aggregation. Nonetheless, to investigate the effects of His-18 alkylation, we decreased the complexity of the system by site-specifically introducing an octyl chain on the imidazole ring. The N(α)- Fmoc-[N(π)-octyl]-His derivative was prepared and intro- duced into IAPP elongation on a solid support. Biophysical analyses revealed that His-18 N-π-alkylation accelerates amyloid formation and decreases the IAPP critical aggregation concentration. The lag phase was abolished for H18-Oc IAPP, which forms short β-sheet-rich protofibrils immediately after its solubilization. IAPP comprises a highly hydrophobic sequence, with only two charged residues at pH 7.4, and this hydrophobicity is known to drive initial self-recognition.¹⁰ The hydrophobicity conferred by the octyl chain likely contributes to intermolecular hydrophobic collapse. This hypothesis is in agreement with a previous study showing that the substitution of His-18 with Gln, a residue with similar volume and hydrophobicity, moderately alters the kinetics of amyloid formation, while the H18L substitution strongly hastens IAPP self-assembly.⁷¹ In addition, it has been proposed using molecular dynamics that the accelerating effect of the site- specific modification of His-18 by diethyl pyrocarbonate (DEPC) on IAPP self-assembly could be associated with the modulation of the α-helix secondary structure through destabilization of intramolecular hydrogen bonds.⁶⁰ In fact, the helical intermediate model proposes that IAPP self- assembly is thermodynamically linked to α-helix formation, in a manner similar to coiled coil formation.

The transient helical structure of IAPP spans from Ala-5 to His-18 and extends to Asn-22 in minor conformers. Accordingly, the His-18 side chain projects on the same α-helix face as Asn-14 and Asn-21 (i, i + 4, and i + 7), likely participating in a hydrogen bond network.⁷⁴ Whether these helical intermediates are on-pathway to amyloid formation,³⁹ His-18 alkylation decreases the stability of the transient α-helix monomers and/ or helical oligomers, ultimately affecting self-recognition. His-18 alkylation also affects imidazole protonation, which isknown to influence the kinetics of IAPP self-assembly. For instance, at a pH below the imidazolium pK_a, IAPP amyloid formation is slowed, and this has been ascribed to His-18 protonation. Herein, we observed that H18-Oc IAPP, in contrast to its unmodified counterpart, assembles promptly into fibrils at pH 4.5, which was used to mimic the acidic environment of the pancreatic secretory granules. This observation suggests that IAPP aggregation could be triggered in the secretory granules by the formation of the HNE covalent adduct.

His-18 alkylation not only hastened the IAPP conforma- tional transition associated with self-assembly but also led to fibrils with different morphology and physicochemical properties. Whereas H18-Oc IAPP formed linear fibrils, these assemblies were thinner and displayed a low level of ThT binding compared to that of their unmodified counterparts. Moreover, H18-Oc fibrils showed high surface-exposed hydrophobicity, which was associated with their capacity to perturb anionic LUVs. The His imidazole ring enables five types of interaction (cation−π, π−π stacking, hydrogen−π, hydrogen bond, and coordination complex) and two protonation forms, each having their contributions to protein stability, aggregation, and (supra)molecular structure.⁷⁶ According to the models of IAPP fibrils inferred from solid-state NMR (hairpin conformation) and, more recently, from cryo-EM (S-shaped conformation), the His-18 side chain projects outward from the protofilament, likely participating in packing into mature fibrils.

 Studies have revealed that His-18 interacts with Thr-36, or Tyr-37, from the opposite protofilament.⁷ Alternatively, molecular dynamics simula- tion based on the solid-state NMR model of Tycko has proposed that His-18 forms an interlayer hydrogen bond with Ser-19 of the adjacent monomer within the same protofila- ment, contributing to elongation.⁸¹ The key implication of His- 18 in fibril formation has been further highlighted by the fact that the sequence of rodent IAPP (rIAPP), which contains Arg at position 18 and does not fibrillize, could assemble into amyloids upon the R18H substitution.⁸² These previous studies suggest that His-18 alkylation likely modulates intermolecular interactions implicated in protofilament packing and/or elongation, explaining the morphological differences between unmodified and H18-Oc IAPP fibrils.

Several cellular mechanisms associated with the toxicity of

IAPP prefibrillar proteospecies have been proposed, including oxidative stress, mitochondrial dysfunction, apoptosis, endo- plasmic reticulum stress, and plasma membrane perturba- tion.It is known that the interaction of IAPP with the outer leaflet of the plasma membrane triggers several downstream biochemical events leading to cell death. We observed that the H18-Oc IAPP is more toxic than the unmodified peptide and requires a shorter incubation time to induce INS-1E cell death. By measuring the release of cytoplasmic LDH into the media, we showed that plasma membrane permeabilization occurs significantly earlier for H18-Oc IAPP, which correlates with its high potency to induce leakage of anionic LUVs. These results could be ascribed to the hastened kinetics of self-assembly of H18-Oc IAPP, to the formation of distinct quaternary species, and/or to potent interactions with lipid membranes. Proposed mechanisms of plasma membrane perturbation by IAPP include the formation of transmembrane pores by oligomeric intermediates and membrane disassembly induced by the mechanical pressure of fibril growth at the membrane surface.

The hydrophobicity conferred by His-18 alkyla- tion could cause the formation of less compact oligomers with a high level of exposure of hydrophobic surfaces, as observed with the H18-Oc fibrils. In turn, this could potentiate their interaction with lipid membranes, as previously reported for oligomers assembled from the N-terminal domain of the HypF protein.⁸⁸ Alternatively, alkylation of IAPP could prompt fibril growth at the membrane surface, inducing disassembly of the INS-1E plasma membrane and lipid vesicles. Our results correlate with a previous study showing that HNE adduction on the Aβ peptide increases its cytotoxicity and potency to disrupt membranes.⁵⁴ In sharp contrast, modification of His-18 by DEPC led to an IAPP derivative with a low capacity to disrupt the POPG lipid bilayer and reduced cytotoxicity, whichis likely associated with the pro-aggregating effect of DEPC conjugation.⁶⁰ Co-assembly experiments revealed that a 1% molar ratio of H18-Oc IAPP accelerates the self-assembly of unmodified IAPP and potentiates its capacity to perturb lipid membranes. For instance, permeabilization of the plasma membrane was observed after incubation for 2 h in the presence of an IAPP/H18-Oc IAPP mixture (99:1), in contrast to 24 h for a homogeneous unmodified IAPP solution. This observation suggests that alkylation of His-18 favors peptide− lipid interactions and/or insertion within the lipid bilayers, which in turn serve as a nucleation site for pore formation and/or membrane disassembly by fibril growth. This suggests that covalent modification of a small proportion of IAPP by HNE triggers the aggregation of unmodified peptides, ultimately affecting the viability of pancreatic cells.

Overall, this study revealed that HNE promotes  4-Hydroxynonenal IAPP amyloid self-assembly and that the covalent alkylation of His- 18 represents a potential mechanism contributing to pancreatic β-cell degeneration. The data presented here indicate that only a small proportion of alkylated peptide is necessary to make unmodified IAPP highly prone to aggregation and to shift the balance toward cytotoxic proteospecies. Nonetheless, although the selected octyl analogue has a chain length similar to that of HNE and mimics its hydrophobicity, it does not recapitulate HNE polarity and/or potential cross-linking, requiring further mechanistic studies. Despite these limitations, our results highlight the importance of studying spontaneous non- enzymatic post-translational modifications induced by oxida- tive metabolites in the context of islet amyloids. This study, which constitutes a first step toward understanding the effect of HNE in IAPP aggregation and tissue deposition, strongly encourages additional investigation to verify if IAPP is covalently modified by HNE in the pancreatic islets of diabetic patients.

■           ASSOCIATED CONTENT

the effect of ethanol on IAPP amyloid formation, characterization of N(α)-Fmoc-[N(π)- (octyl)]-L-histidine methyl ester, analysis of peptides, characterization of amyloid fibrils, TEM images of IAPP and H18-Oc IAPP monomers at pH 4.5, and lipid Phuong Trang Nguyen − Department of Chemistry, Université du Québec àMontréal, Montreal H3C 3P8, Canada; Quebec Network for Research on Protein Function,

Engineering and Applications, PROTEO,

Mélanie Côté-Cyr − Department of Chemistry, Université du Québec àMontréal, Montreal H3C 3P8, Canada; Quebec Network for Research on Protein Function, Engineering and

Applications, PROTEO,

Nadjib Kihal − Department of Chemistry, Université du Québec àMontréal, Montreal H3C 3P8, Canada; Quebec Network for Research on Protein Function, Engineering and

Applications, PROTEO,

Noé Quittot − Department of Chemistry, Université du Québec àMontréal, Montreal H3C 3P8, Canada; Quebec Network for Research on Protein Function, Engineering and

Applications, PROTEO,

Makan Golizeh − Department of Mathematical and Physical Sciences, Concordia University of Edmonton, Edmonton, AB T5B 4E4, Canada

Lekha Sleno − Department of Chemistry, Université du Québec àMontréal, Montreal H3C 3P8, Canada;

Funding

This work was supported by grants from the Natural Sciences and Engineering Research Council of Canada (NSERC, RGPIN-2018-06209) and the Canadian Research Chair program (CRC, 950-231440) to S.B. M.B., P.T.N., and

M.C.-C. acknowledge fellowships from NSERC.

Notes

The authors declare no competing financial interest.

■           ACKNOWLEDGMENTS

The authors acknowledge Dr. Guillaume Charron for his technical assistance.