Total chemical synthesis and biophysical properties of a designed soluble 24 kDa amyloid analogue† †Electronic supplementary information (ESI) available. See DOI: 10.1039/c8sc01790e

A soluble amyloid analogue was designed and prepared by total chemical synthesis using native chemical ligation.


LC-MS: Peptide masses were measured on a LC/MS instrument equipped with a Thermo
Scientific Accela UHPLC (Hypers II GOLD column, 50×2.1 mm, 1.9 μm) integrated with a Thermo Scientific LCQ Fleet ion-trap. Deconvolution of data was performed in MagTran 1.03 (Amgen, Thousand Oaks, CA). More precise MS measurements for larger protein products have been performed using a Thermo Scientific Q Exactive hybrid quadrupole-Orbitrap mass-spectrometer. Direct infusion into electrospray ionization source was performed from solutions of proteins in H 2 O/CH 3 CN (1:1, v/v) containing formic acid (0.1%, v/v).
NMR spectroscopy: 1D 1 H spectra shown in Figure S7 were recorded on a Bruker Avance III 600 MHz spectrometer. The spectra and data depicted in Figure 4c were set to 30 ppm and 116.5 ppm, respectively and the quadrature detection was obtained using a STATES-TPPI phase program. The 15 N relaxation rates were measured from HSQC type experiments as described. (1) For 15 N T 1 relaxation, intensities were extracted from a set of 9 spectra recorded with relaxation delays ranging between 20 ms and 1 s. For 15 N T 2 relaxation, intensities were extracted from a set of 10 spectra recorded with relaxation delays ranging from 0 to 130 ms, with 15 N 180° pulses applied every 0.9 ms at a field strength of 4.2 kHz. The 2D spectra were processed using nmrPipe program (2) and the intensities were measured using CcpNmr. (3) The relaxation rates were computed from non-linear fits of the time dependent intensities by either a single or double exponential decay function using inhouse Python scripts. The values of the T 1 and T 2 relaxation times provide an estimate of the global correlation time (τ c ) of TofT-3 using the approximation provided by Farrow et al: (1) = √ − , where T 1 = spin-lattice relaxation time constant, T 2 = spin-spin relaxation time constant and ω N = 15 N Larmor frequency.
This gives a value between 9.8 and 13.2 ns, in agreement with the expected correlation time of a protein of ~200 amino-acids at 35 °C (11.5 ns as estimated from the model of Daragan). (4) DOSY spectra were recorded for TofT-3 as a function of temperature to get an estimate of the hydrodynamic radius using TRIS as an internal reference. (5)  first, backbones were fixed and models were minimized, thus allowing side-chain and central linker reorganization; and a second minimization afforded reorganization of peptide backbone and side chains simultaneously. C-N linkers 1 and 2 were then introduced and the models were minimized using the same protocol as before. Finally, N-methylations were inserted by replacing the hydrogen atom of the amide bonds by a methyl group and models were minimized.

Chemical synthesis
Preparation of the hydrazide (2-CT-NHNH 2 ) resin: (7) The 2-CT-OH resin (3 g, 1.6 mmol/g active groups, mesh size 200-400) was placed in a three-necked round bottom flask. It was flushed several times with argon. Dry dichloromethane (28 mL) was added. The mixture was gently stirred allowing the resin to swell. Thionyl chloride (450 µL, 6.2 mmol, 1.3 equiv) was slowly added at 0°C. The suspension was stirred under argon for 2 h allowing it to slowly warm up to room temperature. The solvent was removed and the resin was washed with DCM and DMF. The 2-CT-Cl resin was then swollen for 20 min in DMF (18 mL). A mixture of hydrazine monohydrate (700 µL, 14.4 mmol, 3 equiv) and DIEA (2 mL, 11.5 mmol, 2.4 equiv) in DMF (4 mL) was added to the resin at 0°C. The suspension was stirred at r.t. for 90 min. The reaction was quenched by the addition of methanol (300 µL).
The resin was washed with DMF, water, DMF, methanol and diethyl ether and dried under reduced pressure. Final mass of the resin: 3 g.

Procedures for the on-resin functionalization of the ornithine residue (0.1 mmol scale):
Boc protection of the last coupled residue (This step is only required for the synthesis of peptides B and H (see Tables S1 and S2) Alloc deprotection of the linker: Same procedure as before (repeated twice). Chloranil test was positive.
Bromoacetylation of the linker: The resin was swollen in DMF. A solution of bromoacetic acid (167 mg, 1.2 mmol, 12 equiv) and N,N'-diisopropylcarbodiimide (86 µL, 1.1 mmol, 11.2 equiv) in 2 mL of DMF was prepared (2 min of pre-activation) and added to the resin. It was let to react at r.t. for 60 min. Chloranil test was negative.
It was let to react at -10°C for 2 h under argon. The reaction mixture was diluted and directly injected into preparative HPLC and the product was purified using a Phenomenex Kinetex XB-C18 (250 x 21.2 mm, 100 Å, 5 µm) column with a gradient of water with TFA (0.1%, v/v) and acetonitrile with TFA (0.08%, v/v). Pure fractions of product were combined and lyophilized to give a white solid. This procedure was also employed for the synthesis of NMT-A-2 (see Table S2).

Continued on p. 13
peptide D: sequence:

NMT-B-2
Same as in the synthesis of TofT-2 (see Table S2)

Peptides used in the synthesis of covalent trimers:
peptide C: same as in the synthesis of TofT-1 (see Table S1 The isolated yield: 54 mg (17% based on the resin loading).
peptide G: Same as in the synthesis of TofT-2 (see Table S2) peptide H: Same as in the synthesis of TofT-2 (see Table S2) peptide I: Same as in the synthesis of TofT-2 (see Table S2) peptide J: Same as in the synthesis of TofT-2 (see Table S2 Table S1).           Fig. S11. T 1 relaxation times of TofT-3 measured at 35 ºC. The T 1 relaxation curves display clear bi-exponential decay behavior with a contribution of a very short T 1 (in the range of 20 to 100 ms) and a contribution of a long T 1 (in the range of 1 s). The occurrence of a fast component for the T 1 relaxation suggests the presence of a supplementary relaxation mechanism, in addition to the dipolar and chemical shift anisotropy mechanisms that usually dominate the relaxation of an isolated 15 N. A possible explanation is the occurrence of a slow exchange between soluble TofT-3 and minute amounts of high molecular weight aggregates with the rate slower than the T 2 (100 ms).
Fig. S12. The T 2 relaxation curves displayed expected mono-exponential decays for all five resonances, including the low intensity GLY10. Most amide nitrogens display fast T 2 relaxation times between 40 and 60 ms, in agreement with homogeneous dynamic behavior of the construct. Only the peak [2] is characterized by a longer T 2 , suggesting a faster correlation time for this site. c) The diffusion coefficients were obtained after the Laplace transform of the DOSY spectra using the PALMA server (https://arxiv.org/abs/1608.07055). d) As a function of temperature, the diffusion coefficient displays the expected linear dependence when corrected to the variation of viscosity and corresponds to molecular weight of ~20 kDa. (5) (8) This also indicates that the structures of the central linker (in CT it is different from "covalent trimer" [20-41]β2m, see Figure S1) are not detrimental in the amyloid self-assembly mechanism.
Fig. S16. Transmission electron microscopy (TEM) analysis of the aggregates obtained upon completion of the experiment described in Figure 5 in the main manuscript. Panel 1 depicts the long fibers observed for the "covalent trimer" [20-41]β2m (reference construct). Panels 2-7 show morphologies of different aggregates obtained in the presence of Nmethylated peptide inhibitors (their molecular structures are shown to the right of the corresponding TEM images). In the presence of covalent dimers and trimers, the fibrillar morphology was not observed indicating their higher potency to modify the amyloid selfassembly.