Mechanism of amyloid β−protein dimerization determined using single−molecule AFM force spectroscopy

Aβ42 and Aβ40 are the two primary alloforms of human amyloid β−protein (Aβ). The two additional C−terminal residues of Aβ42 result in elevated neurotoxicity compared with Aβ40, but the molecular mechanism underlying this effect remains unclear. Here, we used single−molecule force microscopy to characterize interpeptide interactions for Aβ42 and Aβ40 and corresponding mutants. We discovered a dramatic difference in the interaction patterns of Aβ42 and Aβ40 monomers within dimers. Although the sequence difference between the two peptides is at the C−termini, the N−terminal segment plays a key role in the peptide interaction in the dimers. This is an unexpected finding as N−terminal was considered as disordered segment with no effect on the Aβ peptide aggregation. These novel properties of Aβ proteins suggests that the stabilization of N−terminal interactions is a switch in redirecting of amyloids form the neurotoxic aggregation pathway, opening a novel avenue for the disease preventions and treatments.


Materials
The sequences of all five peptides were (substitutions were underlined): A40 (CDAEFRHDSGYEVHHQKLVFFAEDVGSNKGAIIGLMVGGVV); A42 (CDAEFRHDSGYEVHHQKLVFFAEDVGSNKGAIIGLMVGGVVIA); [VPV] A40 (CDAEFRHDSGYEVHHQKLVFFAEDVGSNKGAIIVLMPGVVV); [VPV] A42 (CDAEFRHDSGYEVHHQKLVFFAEDVGSNKGAIIVLMPGVVVIA); [pP] A42 (CDAEFRHDSGYEVHHQKLVFFAEDVGSNKGAIIGLMPPGVVIA). Each peptide included an additional residue (Cys) at its Nterminus to allow covalent attachment to the AFM tip or substrate. All peptides were synthesized using 9fluorenylmethoxycarbonyl (FMOC) chemistry and purified by reverse phase high performance liquid chromatography (RPHPLC). The identity and purity (usually > 97%) of the peptides were confirmed by amino acid analysis followed by mass spectrometry and RPHPLC. Stock solutions of cysteinylAβ peptides were prepared as previously described [1][2][3] . Briefly, for each Aβ peptide, the lyophilized form was dissolved in trifluoroacetic acid (2 mg/ml) by ultrasonication for 5 min to destroy preformed aggregates and then dried immediately using a vacuum centrifuge (Vacufuge, Eppendorf). The white powder of Aβ peptide was dissolved at 2 mg/ml in dimethyl sulfoxide (DMSO) as a stock solution and then diluted in DMSO before being used. The concentration of each Aβ peptide's stock solution was determined by spectrophotometry (Nanodrop ® ND1000).
The molar extension coefficients were 1280 cm -1 ·m -1 and 120 cm -1 ·m -1 for tyrosine and cysteine, respectively. Aliquots of each peptide were stored at 20°C.

Functionalization of AFM tips
The AFM tips were modified as described previously 1,4,5 . Briefly, silicon nitride (Si 3 N 4 ) AFM tips (MSNL10, Bruker Nano, Santa Barbara, CA) were immersed in 100% ethanol solution for 30 min, thoroughly rinsed with water, dried with argon, and then exposed to UV light (CL1000 Ultraviolet Crosslinker, UVP, Upland, CA) for 30 min. The AFM tips were immersed in an aqueous solution of 167 M MAS for 3 h and then rinsed with DI water. To monomerize A molecules by breaking any intermolecular disulfide bonds, a 20 nM A peptide solution in phosphate buffer was processed with 20 M TCEP hydrochloride for 15 min before functionalization. The MASmodified AFM tips were submerged into the above mentioned peptide solution for 1 h to covalently attach the peptides. After rinsing with pH 7.4 sodium phosphate buffer, the A peptidetethered AFM tips were treated with 10 mM mercaptoethanol solution for 10 min to quench the unreacted maleimide moieties. Finally, the A peptidecoated AFM tips were washed with phosphate buffer and stored in the same buffer solution. Typically, the storage time was less than 24 h at room temperature.

Modification of mica surfaces
Procedures for mica modification and peptide immobilization were similar to those described 1,4,5 . Briefly, mica sheets (AshevilleSchoonmaker Mica Co., Newport News, VA) were cut into1.5 cm × 1.5 cm squares and glued to glass slides with epoxy. The freshly cleaved mica surfaces were treated with APS for 30 min followed by conjugation with 167 M MALPEGSVA in DMSO. After activation for 3 h, the mica squares were rinsed sequentially with DMSO and water to remove unbound MALPEGSVA, and then dried with argon flow.
The activated maleimide groups were then ligated with 20 nM cysteinylA peptides for 1 h to make the Afunctionalized mica substrates. The remaining immobilization steps were the same as those described above for the AFM tips.

Dynamic force spectroscopy
The dynamic force spectroscopy force measurements were conducted with a Molecular Force Probe 3D AFM system (MFP3D, Asylum Research, Santa Barbara, CA). AFM probes with nominal spring constants of 0.03 N/m were used throughout the experiments. The apparent spring constants were determined by the thermal noise analysis method with the Igor Pro 6.04 software (provided by the manufacturer). A low trigger force (100 pN) was exerted on the AFM probes. The approach velocity was constantly kept at 400 nm/s. To get a wide cover of loading rates, the retraction velocities of each DFS experiment were allowed to vary from 100 nm/s up to 4000 nm/s. When the retraction velocities reached 1000 nm/s or above, the dwell time was set for 0.3 s. At each retraction velocity, force curves were obtained by probing over an area of 7×7 µm, creating force maps sized 60×40 points. The yield of rupture events varied between 2%and4%. The nanomolar concentration of A peptides was used at the peptide immobilization step resulting in a low surface density. The low surface density enabled us to avoid multiple interactions complicating the single molecule measurement. According to the paper of Merkel et al 6 , 90% detection of specific interactions is achieved if 1 out of 710 attempts is a specific event.
Generally, at least 100 force curves with typical rupture events were analyzed at each velocity to ensure reliable statistical analysis. Control experiments were conducted by eliminating the A functionalization of either the tips or the mica substrates. Force curves with "rupturelike" events rarely occurred in control experiments. These force curves showed a "rupture" with a very short contour length (˂ 25 nm) and a small separation distance, an observation that could be attributed to nonspecific interactions between tips and linkers/substrates.

Estimation of the contour length of PEG 3400
The contour length of PEG was estimated with the following equation 7 : Where L c (F) is the contour length, N s is the average number of monomers, L planar is the length of monomers with planar conformation, L helical is the length of monomers with helical conformation, ΔG(F) is the free energy difference at zero appling force. The N s is 77 ± 10 for 3400 Da PEG. The L planar and L helical are 3.58 Å and 2.8 Å, respectively. The ΔG(F) is fixed at 3 k B T. The contour length of PEG was thus estimated to be 22.0 ± 0.9 nm.

Contour length analysis
As described previously 8 , the data points in the loading rate range of 50007000 pN/s were used for the contour length analysis. The narrow range of loading rates was chosen to exclude the potential influence of different loading rates on the contour lengths. Throughout the contour length analysis, the arrangement of Adimers was assumed to be symmetrical. This premise is reasonable based on work by Mitternacht et al. 9 in which near symmetrical A42 dimers were confirmed by Monte Carlo simulation, and by the demonstration of the existence of symmetrical and inregister alignments of adjacent Amolecules within A fibrils [10][11][12]  Therefore, the total length of all linkers was 24.5 ± 0.9 nm. In addition, the distance between Ntermini of the two peptides is ~1.0 nm. The contribution of the disordered segments of the peptides ranges from 0 nm to ~33 nm. The calculated combined contour length was divided equally between the Amonomers and was converted into the number of residues.

Dynamic force spectroscopy data analysis
Three rules were applied for identification of specific interactions 1,13 : (1) according to the thermal noise of the experimental setup, the rupture force larger than 20 pN were considered; (2) according to the control experiments, the contour length (the length at maximum physical extension of the interaction system determined after the wormlike chain (WLC) analysis) should be greater than 25 nm; and (3) the rupture events at distances of the tipsample separation (the projection of the distance between the AFM tip and the mica substrate on the vertical axis) smaller than 20 nm correspond to nonspecific interactions between the tip and these force curves were not analyzed.
The wormlike chain (WLC) model was used for fitting the forcedistance curves 1,2,5 : where F(x) is the force at the distance of x, k B is the Boltzman constant, T is the absolute temperature, and L p and L c are the persistence length and the contour length, respectively. The persistence length of PEG was fixed at 0.35 nm 14 . The contour lengths were obtained with the Igor Pro 6.04 software package from the WLC fit of forcedistance curves.
The apparent loading rates were calculated by using the following equation 5 : (2) where F p = k B T/L p , k c is the spring constant (pN/nm), v is the tip retraction velocity, F is the rupture force, and r is the apparent loading rate (pN/s).
The most probable rupture force for each grouped data set was approximated with the probability density function 15 : where p(F) is the probability density of rupture force, k off is the offrate constant at zero external force, and x  is the distance between the transition state and the bound state.
After obtaining the most probable rupture forces at a series of apparent loading rates, the dynamic force spectrum could be constructed and the corresponding data could be fitted with the BellEvans model 6 : Two important parameters, k off and x   were thus extracted. The height of the energy barrier, ΔG # , can be determined by the following equation 16,17 : where h is the Planck constant. All data errors were shown by ±SEM.  x  = 0.207 ± 0.031 nm k off = 8.7 ± 0.5 s -1 0.11 ± 0.01 x  is the potential barrier location and k off is the dissociation rate without applied force. The lifetime () was calculated by  = 1/k off . Exp denotes experiment. Errors are ±SEM.