Electrical characteristics of amyloid beta peptides in vertical junctions

Assembled amyloid beta (Aβ) peptides have been considered pathological assemblies involved in human brain diseases, and the electron transfer or electron transport characteristics of Aβ are important for the formation of structured assemblies. Here, we report the electrical characteristics of surface-assembled Aβ peptides similar to those observed in Alzheimer’s patients. These characteristics correlate to their electron transfer characteristics. Electrical current–voltage plots of Aβ vertical junction devices show the Aβ sequence dependence of the current densities at both Aβ monomers (mono-Aβs) and Aβ oligomers (oli-Aβs), while Aβ sequence dependence is not clearly observed in the electrical characteristics of Aβ planar field effect transistors (FETs). In particular, surface oligomerization of Aβ peptides drastically decreases the activity of electron transfer, which presents a change in the electron transport pathway in the Aβ vertical junctions. Electron transport at oli-Aβ junctions is symmetric (tunneling/tunneling) due to the weak and voltage-independent coupling of the less redox-reactive oli-Aβ to the contacts, while that at mono-Aβ junctions is asymmetric (hopping/tunneling) due to redox levels of mono-Aβ voltage-dependently coupled with contact electrodes. Consequently, through vertical junctions, the sequence- and conformation-dependent electrical characteristics of Aβs can reveal their electron transfer activities. Improvements to biomolecule/vertical device interfaces have given researchers a new tool to characterize the neurotoxins associated with brain disease. Assembly of amyloid beta peptides into fibrils has been identified as contributing to the onset of Alzheimer’s disease. Sohyeon Seo and Hyoyoung Lee from Sungkyunkwan University in Suwon, South Korea, and colleagues report that conductivity measurements can help predict the chemical activity of amyloid beta peptide structures. The team’s device integrates the peptides into a vertical junction layout by immobilizing them onto monolayer-coated gold electrodes. Correlating the device’s conductivity with microscopically determined chemical structures revealed that current flow depended on the peptide’s amino acid sequences. Drastic drops in conductivity were observed when the electrode-bound monomer peptides assembled into fibrils, indicating the presence of a different mechanism of current flow. Experimental descriptions of Aβ oligomers (oli-Aβs) assembled on surfaces as compared with Aβ monomers (mono-Aβs) are crucial to understanding changes in chemical reactivity. Here, we fabricated Aβ molecular junctions between linker molecular layered electrodes using different Aβ segments and report that electron transport pathways changed from asymmetric hopping across monomeric Aβ junctions to symmetric tunneling across oligomeric Aβ junctions.


Introduction
Electron transfer and electron transport processes of proteins are important for understanding and gaining insight into their biological functions such as enzymatic reactions 1,2 . However, it is difficult to predict the mechanism of electron transport in proteins due to the size of the protein containing a number of amino acid residues. In biomolecular junctions composed of target biomolecules incorporated between two electrodes, longrange electron transport processes through biomolecules were found to be strongly influenced by molecular size and coupling effects between molecules and electrodes 3,4 . In addition, junction configurations can directly affect the transport efficiency across molecules with regard to the presence of molecular redox activity [5][6][7] . For example, junction conductance across proteins is influenced by the electron transport efficiency of interactions between redox-active residues and electrodes 7 . Thus, electron transport behaviors across protein junctions could be involved in structural changes, regardless of whether proteins contain redox-active residues. This behavior is important to consider when studying the mechanical structural changes that cause certain proteins to become toxic.
Amyloid beta (Aβ) peptide molecules are critical components of Alzheimer's disease [8][9][10] . Concentration-or time-dependent morphological changes that occur as a result of fibrillization via oligomerization of Aβ monomers (mono-Aβs) during aggregation have been intensively investigated [11][12][13][14] . Moreover, electronic conductance studies of Aβ peptide structures have revealed the importance of structural parameters such as the sequences and assembly conditions of Aβ 15,16 . However, understanding how mechanical pathway-dependent chemical activity (or toxicity) results from oligomerization of Aβ remains a significant challenge, although oligomers and fibrils are known to play an active role [17][18][19] . Based on previous reports 11,15,20 , electron transport pathways of amyloid-based proteins whose structural conformation can be determined by a mechanical transition from monomers to oligomers result in changes in chemical activity, such as the redox activity of amino acid residues. Thus, oligomerization of Aβ could induce a change in electron transport across structurally altered Aβ molecules (e.g., Aβ oligomers, oli-Aβs) on a surface. The electron transport efficiency of monoand oli-Aβs can determine the corresponding mechanical pathways and chemical activities during aggregation processes.
In this work, the electrical characteristics of Aβ peptides were presented to describe the conformational status of the peptides. Surface-immobilized Aβ peptide layers (e.g., Aβ−linker−gold or Aβ−linker/graphene (G)) were used. Electron transport pathways through Aβ were investigated by using three Aβ peptides (e.g., Aβ 1-40 , Aβ 1-28 , and Aβ [12][13][14][15][16][17][18][19][20][21][22][23][24][25][26][27][28] in vertical gold/biomolecule-gold junctions ("/" denotes physical contact and "−" denotes chemical contact). Electrical characteristics of the vertical junctions showed Aβ sequence dependence, while the current flowing in the field effect transistor (FET) with an Aβ peptide/G channel was weakly dependent on the peptide sequences. In particular, the surface oligomerization of the Aβ peptides drastically changed the electron transfer characteristics by shifting the electrochemical potentials associated with the change in chemical activity. This was clearly revealed in the electrical characteristics of Aβ vertical junctions rather than Aβ/G FETs. (1) Redoxactive peptide residues of mono-Aβs strongly coupled with the linker monolayer led to asymmetric electron transport characteristics. (2) However, oligomeric Aβ junctions exhibited only symmetric electron transport characteristics due to weak redox activity. These findings indicate that the intermolecular assembly of Aβ peptides results in changes in surface chemical activities (i.e., electron transfer activities), leading to changes in the electron transport pathways.

Results and discussion
To study the electrical characteristics of Aβ peptides through a vertical junction, it is important to have welldefined molecular assembly layers on metal electrodes. However, the formation of molecular layers of proteins or polypeptides might be complicated by a lack of spatial control. Thus, a self-assembled molecular linker monolayer of 1-mercaptopropionic acid (MPA) was used to immobilize Aβ onto a gold electrode, which was activated by coupling reactions with 1-ethyl-3-(3-dimethylamino) propyl carbodiimide/N-hydroxysuccinimide (EDC/NHS) (Scheme 1). Mono-Aβ−MPA molecular layers were formed via chemical bonding of lysine (Lys 16 or 28 denoted with red circles) residues onto the activated MPA (EDC/NHS−MPA) molecular layer (denoted by a blue circle). Then, the vertical junction of top gold−MPA//Aβ −MPA−bottom gold was fabricated.

Conformation-and sequence-dependent electron transfer of Aβ peptide-immobilized surfaces Characteristics of in-solution oligomerization
Changes in the molecular conformation and chemical activity (i.e., electron transfer activity) due to the oligomerization of Aβ molecules were verified by circular dichroism (CD) spectroscopy and differential pulse voltammetry (DPV). In Fig. 1a, the CD spectrum of mono-Aβ 1-40 solution showed a characteristic absorption minimum of unstructured monomers at approximately 200 nm 19,21 . In-solution oligomerization under our conditions (1 h incubation) resulted in a shift in the absorption Scheme 1 a The formation of an Aβ 1-40 molecular layer via a 1-mercaptopropionic acid (MPA) linker monolayer. Terminal carboxylic acid (-COOH) groups of an MPA SAM were activated by EDC/NHS and turned to NHS-ester groups (denoted by a blue circle). The amino group of a lysine residue (Lys 16-NH 2 denoted by a red circle) was coupled with the NHS-ester group to form an amide bond (denoted by a purple circle). b The molecular junctions comprised (top)gold−MPA//Aβ 1-4 −MPA−(bottom)gold. The geometric structure of Aβ 1-40 was obtained from the Protein Data Bank (PDB; 1ba4). Oli-Ab 1-40 was formed by on-surface oligomerization.
were decreased, and the peaks shifted to positive potentials. Furthermore, the process of in-solution Aβ oligomerization (or fibrillization via ordered assembly) was verified by scanning electron microscopy (SEM) (Fig.  S1a-d). Apparently, oligomerization of the mono-Aβ 1-40 solution by 1 h incubation could not change the morphology of an oli-Aβ 1-40 -deposited surface compared with that of a mono-Aβ 1-40 -deposited surface. After 4 and 24 h incubations, short protofibrils and long mature fibrils were observed on the oli-Aβ 1-40 -deposited surfaces, respectively.

Characteristics of on-surface oligomerization
To promote the formation of oli-Aβ on a substrate, the on-surface oligomerization of mono-Aβ was conducted. Aβ 1-40 molecules oligomerized on a quartz substrate were assembled into oli-Aβ 1-40 and physically adsorbed on the substrate (Fig. 2a) for the identification of structural changes. Unlike the oligomerization of Aβ that occurred in solution, the on-surface oligomerization resulted in a drastic CD spectral change where a pair of positive and negative peaks appeared at approximately 190 and 210 nm, characteristic of a beta-sheet structure 19 , indicating the ordered assembly of oli-Aβ 1-40 (Fig. 2b). This indicates the influence of the surface on the oligomerization of Aβ 24 ; this surface oligomerization is distinguishable from in-solution oligomerization (Fig. 1). Furthermore, for the formation of oli-Aβ molecular layers on an MPA self-assembled monolayer (SAM), the onsurface oligomerization of mono-Aβ − MPA molecular layers was conducted in 0.01 M phosphate buffer (PB) solution containing 0.15 M NaCl at 37°C for 1 h (Scheme 1a and Fig. 2c). Mono-Aβ-tethered gold or indium tin oxide (ITO) electrodes were used for electrical and electrochemical measurements, respectively. As shown in Scheme 1a, lysine residues with a charged long chain would be relatively favorable for the coupling reaction with NHS ester compared with the primary amines at the C-or N-termini, which have short chains that can be statically hindered by nearby peptide chains. Unfortunately, CD of the oligomerized amyloid peptide MPA-ITO could not be observed, probably due to insufficient amounts of amyloid peptides on the MPA SAM. In particular, an ITO electrode was suitable for not only the formation of an MPA SAM but also to maintain a stable potential window in PB electrolyte solution. In Fig. 2d, mono-Aβ-tethered ITO electrodes (Aβ 1-4 −MPA−ITO) showed a change in electron transfer activities to shift oxidation peaks to approximately 0.6 and 1.05 V Ag|AgCl , as shown in Fig. 1. This electrochemical activity changed significantly after the oligomerization on the surface was allowed to proceed for 1 h, whereas physically adsorbed oli-Aβs that were oligomerized in solution showed a gradual decrease in electrochemical activity as incubation time increased (Fig. 1b). Only one peak at approximately 0.8 V Ag|AgCl was observed for the oli-Aβ 1-4 −MPA−ITO. One of the possible reasons is that electrochemically active residues on the Aβ molecules tethered on the electrodes via chemical binding are probably more hindered by bulky side residues than physically adsorbed Aβ molecules. Thus, electron transfer from the electrode could be blocked, even at an early stage of oligomerization (e.g., 1 h incubation) 23 . The other possible reason could be a change in electron transfer activity due to the strong interaction with the electrode leading to a change in chemical reaction potentials.
Oligomeric changes in Aβ on the surface induced a change in the surface hydrophilicity (Fig. S2), which can change the electrochemical activity of the surface in the electrolyte. Both MPA and EDS/NHS−MPA SAMs are hydrophilic and can interact positively with hydrophilic residues of mono-Aβs. After modification of the SAMs with Aβ molecules, the surfaces retained their hydrophilicity (Fig. S2a). However, oligomerization of mono-Aβ SAM led to a decrease in surface hydrophilicity (Fig. S2b). This surface change could result in the electrochemical hindrance of oli-Aβ SAMs, leading to the disappearance of electrochemical reactivity. Clean HOPG, on the other hand, exhibited hydrophobicity (Fig. S2c), inducing repulsive interactions with hydrophilic residues of Aβ molecules, which resulted in the hydrophilic surfaces of the physically adsorbed oli-Aβ molecules (Fig. S2d). Thus, the electrochemical activity of Aβ molecules in oli-Aβs/ HOPG was retained for a few hours of incubation, unlike that of the Aβ molecules in oli-Aβs−MPA−ITO.

Surface morphology and binding energy of Aβ peptide-MPA monolayers
In molecular electronic devices, the surface order (or roughness) of molecular layers influences the junction properties 25 . To explore changes in surface morphology by mono-Aβ modification on an MPA SAM and further on-surface oligomerization, monoor oli-Aβ 1-40 molecular SAMs were examined by scanning tunneling microscopy (STM) (Fig. 3a-c). STM images were used to evaluate the surface morphology in nanoscale molecular layers on a single-crystal electrode. The etch pits formed during thiol adsorption on Au (111) are shown. Surface roughness increased in both monoand oli-Aβ molecular layers, indicating that the surface ordering decreased with the formation of Aβ layers compared with that of the MPA layer. Features from defects in the monoor oli-Aβ molecular layers were distinguishable from those in the MPA SAM, indicating the formation of new defects in the Aβ layers. The surface morphology of the Aβ molecular layers was slightly changed by oligomerization. These observations indicated that the oli-Aβ molecular layers maintained the ordering of mono-Aβ molecular layers after oligomerization, which also revealed the advantageous surface conditions of Aβ for the fabrication of molecular layer-based molecular junctions. The microscale surface morphology of different Aβ peptide layers was evaluated by AFM on an e-beam-deposited gold electrode (Fig. S3a-c). The surface roughness was increased by an increase in the molecular length of Aβ peptides. Ellipsometry measurements showed an increase in the film thickness as the molecular length of Aβ peptides increased (Fig. S3d).
Furthermore, the Aβ peptides with the MPA molecular SAM exhibited changes in their binding energies with gold electrodes; these binding energies were verified by ultraviolet photoelectron spectroscopy (UPS) (Fig. 3d,e). Surface modification with the activated-MPA (i.e., (EDS/ NHS)−MPA) SAM led to an increase in binding energies with the electrodes compared to the binding energies of the MPA SAM according to the analysis of the secondary electron cutoff (SEC) region (Fig. 3d). This phenomenon was also observed after modification with mono-Aβs. No specific correlation between sequences of Aβ peptides and shifts in work function was found in the SEC region (the left plot in Fig. 3e). However, the energy offset between the Fermi level of the gold electrode and the nearest molecular orbital (e.g., highest occupied molecular orbital, HOMO) became larger in the HOMO region as the sequence length increased (the right plot in Fig. 3e). Thus, electron injection barriers from the electrode to the mono-Aβ layers decreased from mono-Aβ 1-40 to mono-Aβ 1-28 and mono-Aβ 12-28 , which indicates length-dependent electron transport behaviors in those junctions 26 . Moreover, as oligomerization of the Aβ peptides occurred on the surface, the binding energy contributed by chemical contacts became weaker in the SEC region, and the energy offset between the Fermi level of the electrode and the nearest molecular orbital became larger in the HOMO region. Thus, electron transport across the mono-Aβ junction can be favored to a greater extent than that across the oli-Aβ junction.
Conformation-and sequence-dependent electrical characteristics of Aβ peptides in soil-state devices Electrical characteristics of Aβ vertical junctions in a crossbar device For the fabrication of a cross-bar device, protein stability could be an issue due to nonphysiological environments during the fabrication of molecular junctions, especially contact formation of a top electrode. Soft contacts such as AFM tips 3,27,28 , soft materials 6,29 , or gold nanowires 7 have therefore been used to study the electron transport mechanisms of proteins. In this work, the vertical junction devices of Aβ were configured into top gold−MPA//Aβ−MPA−bottom gold, where top gold−MPA electrodes were assembled by a soft contact on bottom monoor oli-Aβ−MPA−gold electrodes (Fig. S4). In our previous work 30 , the electrical characteristics and stability of the vertical molecular junction assembled by soft contact were verified with length-dependent tunneling transport properties for a series of alkanethiol molecules.
Three different peptides (Aβ 1-40 , Aβ 1-28 , and Aβ 12-28 ) formed mono-Aβ and oli-Aβ molecular junctions. Plots of current density against voltage (J/V) clearly showed that mono-Aβ molecular junctions had asymmetric characteristics (Fig. 4a, b), while oli-Aβ junctions had symmetric characteristics (Fig. 4d, e). The statistical J curves as a function of applied voltage (with 95% confidence intervals of the mean) were obtained with 50 devices for each SAM. These are molecular junctions of non-shorted devices, which allowed >80% yield. For both monoand oli-Aβ junctions, molecular lengthdependent J/V plots were obtained, indicating that the junction distance of a smaller Aβ peptide is shorter than that of a larger Aβ peptide. Thus, the J values of Aβ 12-28 junctions were larger than those of Aβ 1-28 and Aβ 1-40 junctions, as expected. For mono-Aβ junctions, the J/V rectifications indicated asymmetric electronic coupling via either chemical or physical contacts of Aβ peptides to the electrode, although compositionally symmetric junctions were fabricated for (top) Au−MPA//Aβs −MPA−Au (bottom).
In particular, the asymmetric J/V plots were found to be dependent on the bias voltage value. At a low voltage range (±0.5 V), all mono-Aβ junctions had symmetric sigmoidal curves (Fig. S5), while in the voltage range higher than ±1.0 V, asymmetric sigmoidal J/V curves were observed. This bias voltage-dependent asymmetric electron transport indicates current rectification behavior involving the asymmetric electronic structure of mono-Aβ molecules induced by a high bias voltage 31,32 . Furthermore, the J values decreased in the oligomeric Aβ junctions of (top) Au−MPA//oli-Aβ−MPA−Au (bottom). Thus, oligomerization of Aβ peptides in molecular layers can equalize the electronic and/or contact properties in their molecular junctions.
The effects of the asymmetric contacts were examined by using control devices (Fig. S6). Junctions composed of asymmetric contacts, such as (top) gold−MPA//MPA−gold  (bottom) and (top) gold//MPA−gold (bottom) junctions without Aβ molecules, were fabricated. These asymmetric contacts exhibited symmetric behavior in the J/V curves, similar to oli-Aβ molecular junctions, indicating that asymmetric contact junctions do not strongly influence the asymmetric electron transport behavior of mono-Aβ molecules. In general, both (exponentially length-dependent) tunneling and (inversely proportional lengthdependent) hopping processes can contribute to electron transport pathways across molecular junctions 29,33,34 . Based on the molecular size-dependent J/V plots of mono-Aβ molecules (Fig. 4b, e), the electron transport mechanism was distance dependent, although junction distances formed by three Aβ molecules could not be correctly determined. In addition, bias voltage-dependent transport characteristics indicated a redox centerinvolved mechanism 26,35 . The energy level of the redox potential (E HOMO ) to vacuum can be estimated by using the formal half-wave potential (E 1/2 ) 36 ; E HOMO = E abs,NHE -E 1/2,NHE , where E abs,NHE is the absolute potential energy of the normal hydrogen electrode, NHE (E abs,NHE = −4.44 eV and E Ag|AgCl = 0.197 E NHE ). Thus, the energy levels of the redox potentials of mono-Aβ 1-40 were estimated to be at approximately −5.25 eV and −5.69 eV, whereas that of oli-Aβ 1-40 was measured to be only approximately −5.49 eV (Fig. 2d). In the case of mono-Aβ 1-40 , the HOMO energy levels can be arranged between the Fermi levels of gold at 1.0 V, which were depicted at the same location in the junctions in Fig. 4c due to the undefined locations of redox centers. However, one of the HOMO energy levels of oli-Aβ 1-40 can be arranged between the Fermi levels of gold at 1.0 V, which is depicted with a dotted square including the inactive redox center in Fig. 4g.
The energy levels of the oxidation of Aβ residues are likely located between those of the gold electrodes under bias voltages (e.g., 1.0 V), as shown in Fig. 4c, although this does not take into account coupling effects contributed by chemical contacts between the molecular orbitals and the gold electrode. Thus, electron transport can occur in the forward direction (V = 1.0 V) via the energy levels of redox-active residues in mono-Aβ junctions, suggesting hopping conductance. At the reverse direction (V = − 1.0 V), the energy levels of redox centers can hardly allow for transporting electrons. In contrast, oli-Aβs did not exhibit voltage-dependent transport characteristics, indicating that redox centers are unlikely to be involved in the transport mechanism (Fig. 4f), which was consistent with the electrochemical activity results (Fig. 2d). Nonetheless, the molecular size-dependent J/V plots and strong coupling of monoand oli-Aβ to the gold electrode via MPA revealed electron tunneling traversing the energy barriers. Thus, at V = ± 1.0 V, oli-Aβs can follow tunneling electron transport, while mono-Aβs can follow tunneling electron transport at V = − 1.0 V and the hopping electron transport at V = 1.0 V.

Electrical characteristics of Aβ planar devices: Aβ−PBASE/ graphene FETs
Additionally, the electrical characteristics of a monoor oli-Aβ assembly in a planar device were explored. The Aβ-immobilized graphene channel was fabricated by using a self-assembled molecular linker monolayer of 1pyrenebutanoic acid succinimidyl ester (PBASE) (Fig. 5a). Transformation of mono-Aβs to oli-Aβs was conducted on mono-Aβ-immobilized surfaces under the same conditions for the formation of vertical oli-Aβ junction devices. In the Raman fingerprint of graphene (Fig. 5b), a sharp and symmetric 2D band at approximately 2682.9 cm −1 was clearly observed and shifted to approximately 2695.5 cm −1 after modification with PBASE, indicating hole doping on pristine graphene. A D band at approximately 1348.0 cm −1 revealing structural defects on graphene appeared for both pristine graphene and PBASE-modified graphene. The G band of pristine graphene shifted from approximately 1595.0 cm −1 to approximately 1598.6 cm −1 , indicating that the PBASE molecules induce hole doping on pristine graphene. A Dʹ band appeared at approximately 1630.9 cm −1 , indicating another weak disordered feature after modification with PBASE, which is superimposed with one of the peaks 37 of the pyrene structure at 1242.2 cm −1 , 1393.7 cm −1 , and 1630.9 cm −1 . The peak positions of the Raman bands were negligibly changed by further modification with mono-Aβs and oli-Aβs. The specific influence of mono-Aβ sequences was not observed in the peak positions of the Raman spectra (Fig. S7). In addition, the peak ratio of the intensity of the G bands to 2D bands (I G / I 2D ) increased with PBASE modification, indicating hole doping 37 . The I G /I 2D ratios increased with mono-Aβ −PBASE modification and then decreased again with oli-Aβ−PBASE modification.
The hole-doping effect of PBASE and Aβ peptides on graphene was observed in the electrical characteristics of Aβ − PBASE/G-FETs (Figs. 5c and S7c, d). The pyrene linker PBASE on graphene induced a shift of the charge neutrality point to large positive values, indicating the highly concentrated doping of hole carriers. Then, the immobilization of mono-Aβs on PBASE/G increased the doping concentration of hole carriers produced by PBASE/G, and the charge neutrality point was positively shifted further. Furthermore, oligomerization of surface-tethered mono-Aβ resulted in a negative shift of the charge neutrality point produced by mono-Aβ−PBASE/G. These phenomena were noticeably observed at all Aβ−PBASE/G-FETs and were presumably associated with an increase in the surface hydrophobicity (or a decrease in the hydrophilic redox activity) due to the buried redox-active residues. This result was in agreement with the change in the water contact angle after the oligomerization of Aβ. The longer Aβ peptide led to a larger shift in the charge neutrality point by both monoand oli-modifications. Unlike oli-Aβs in vertical junctions, nonetheless, the electrical characteristics of oli-Aβ−PBASE/G-FETs according to peptide sequences were not distinguishable from those of mono-Aβ−PBASE/G-FETs. The bottom line shows that both Aβ vertical junctions and Aβ/G FETs (especially Aβ 1-40 ) can provide new insights into understanding the electrical characteristics of oligomeric peptides discriminated from those of monomeric peptides, correlating with their electron transfer characteristics.

Conclusion
The electrical characteristics of surface-tethered monoand oli-Aβ peptides were investigated. The structural characteristics of the Aβ and the corresponding redox properties were significantly changed by oligomerization on surfaces. The electron transfer of Aβ hardly proceeded after surface oligomerization; this observation was strongly related to a change in the electron transport pathway through Aβ. Different conformations of Aβ 1-40 , Aβ 1-28 , and Aβ 12-28 were used to explore the electrical characteristics of surface-tethered Aβ peptides in two types of devices (e.g., vertical junction devices and graphene FET devices). In the vertical junction device, the electron transport pathway across the top gold−MPA// mono-Aβ−MPA−bottom gold junctions was asymmetric with a strong dependence on the bias voltage. The redoxactive peptide residues of the mono-Aβs were strongly coupled with the bottom electrode through an MPA monolayer linker, which induced asymmetric coupling of the mono-Aβ with gold electrodes at a high voltage (e.g., 1.0 V). This coupling led to electron transport via the energy levels of redox centers through a hopping process. In contrast, the oli-Aβ junctions exhibited symmetric characteristics due to reduced redox activity. The less redox-reactive molecular oli-Aβ was symmetrically coupled with the electrode contacts, where the energy levels of redox centers were not positioned between the Fermi levels of gold electrodes and the electron transport could follow a tunneling process. However, the current flowing through the FETs with an Aβ−PBASE/G channel was weakly dependent on the peptide sequences and oligomerization. Consequently, investigation of sequencedependent electron transfer and transport characteristics of Aβ peptides in vertical junctions can predict the chemical activity of ordered aggregates and distinguish the electron transport mechanism of oli-Aβ peptides from that of mono-Aβ peptides.

Materials
Amyloid beta (Aβ) 1-40 human (Sigma-Aldrich, USA), [Glu 11 ] Aβ 1-28 human (Aldrich, USA), and Aβ 12-28 human (Sigma-Aldrich, USA) were purchased and used directly without further purification. Lyophilized Aβ peptides were stored at −20°C. For the experiments, peptides were dissolved in 10 mM PB (pH 7.4, purchased from Sigma-Aldrich, USA) containing 1% ammonium hydroxide (NH 4 OH) solution (80 μL of 1% NH 4 OH solution was used for 1 mg of Aβ peptides) to obtain a final concentration of 25~50 μM Aβ. After vortexing for a few seconds, these solutions were used immediately. These samples were placed into a 50 mL glass vial that was sealed to prevent water evaporation and incubated at 4°C for 6-12 h. Then, the Aβ−MPA SAMs were washed with PB and DI water and dried with inert gas. To fabricate the oli-Aβ−MPA SAMs, a 50 μL droplet of PB solution containing 0.15 M sodium chloride (NaCl, Sigma-Aldrich, USA) was dropped on the Aβ−MPA SAM substrates, and the substrates were placed into a 50 mL glass vial. The vial was sealed to prevent water evaporation and then incubated at 37°C in an incubator for a predetermined period of time (e.g., 1−4 h). The wellwashed and dried mono-Aβ−MPA or oli-Aβ−MPA samples were stored in a vacuum desiccator in the dark.

Fabrication of Aβ vertical junction devices
A cross-bar device for Aβ vertical junctions was fabricated as reported in previous works 30,35 . Thin-film polydimethylsiloxane (PDMS)-coated polyethylene terephthalate (PET) substrates were used to pattern gold electrodes. PDMS solution was spin-coated on cleaned PET substrates and dried in a vacuum oven at 80°C overnight. Gold electrodes were patterned on PDMS/PET substrates by electron-beam (e-beam) deposition of gold using a shadow mask at a slow evaporation rate. Au electrode-patterned PDMS substrates were treated with oxygen plasma to become hydrophilic just before chemical modification.
The reliability test of the devices was performed by using different alkanethiol SAMs after device fabrication via top//bottom contact assembly (top electrode//molecular SAM-bottom electrode) as described in our previous report 30 . For the device fabrication of Aβ vertical junction devices, a top electrode-patterned substrate modified with an MPA SAM was gently placed onto a bottom electrodepatterned substrate modified with an Aβ−MPA SAM, and the top and bottom electrode arrays were aligned to form vertical junctions (top gold−MPA//Aβ−MPA−bottom gold) (Fig. S3).

Fabrication of Aβ peptide−PBASE/graphene FETs
Graphene (SAINT, Sungkyunkwan University) was grown by chemical vapor deposition (CVD) on Cu foil and transferred onto SiO 2 substrates with patterned gold electrodes by a polymethyl methacrylate (PMMA)-mediated method 38 . The transferred graphene was characterized by Raman spectroscopy with an excitation energy of 532 nm (Ntegra Spectra DUO Max, NT-MDT II). A graphene FET device was immersed in dimethylformamide (DMF) solution containing 10 mM 1-pyrenebutanoic acid succinimidyl ester (PBASE, Aldrich) for the formation of π-π stacked PBASE/graphene (PBASE/G) films at 4°C for over 7 h. The washed and dried PBASE/G FETs were used for the formation of mono-Aβ peptide−PBASE/G and oli-Aβ peptide−PBASE/G as described above.

Characterization of molecular assembly and devices
Monomers and oligomers of Aβ peptides were characterized by CD spectrometry (J-1500 CD spectrometer, Jasco, USA). Oligomerization of amyloid beta solution in 0.01 M PB was conducted by adding NaCl to a final concentration of 0.15 M and incubation at 37°C for various times. For CD measurements of Aβ films, cleaned quartz substrates were used. Mono-Aβ film on a quartz substrate was prepared by 6 h incubation at 4°C after dropping a 50 μL droplet of PB solution containing mono-Aβ in a sealed vial. An oli-Aβ film on a quartz substrate was prepared by 1 h incubation at 37°C after dropping a 50 μL droplet of PB solution containing mono-Aβ and 0.15 M NaCl in a sealed vial. Then, Aβ samples were washed with PB and DI water and dried with inert gas.
Gold/SiO 2 substrates were used for the morphology and thickness analysis of Aβ films. Morphology and thickness measurements of three monomeric Aβ films were conducted by AFM (SPM 5200, Agilent, USA) and ellipsometry (MG-1000, Nano-View, Korea). Morphological changes as a result of Aβ oligomerization were verified by SEM (JSM-7100F, JEOL, USA). HOPG was used for the electrochemical study of physically adsorbed Aβ films. The first few layers of HOPG were peeled off using tape and then used immediately. For physically adsorbed films of monoor oli-Aβs to a certain substrate, a 50 μL droplet of each solution was dropped on cleaned substrates, which were then placed into a 50 mL glass vial that was then sealed to prevent water evaporation and incubated at 4°C for 6-12 h. Then, Aβ samples were washed with PB and DI water and dried with inert gas.
Changes in surface binding energy as a result of chemical modification were verified by UPS (AXIS Supra, Kratos, UK). Water contact angle measurements (Phoenix-300, SEO, Korea) were performed for surface characterization of hydrophilicity and hydrophobicity. An Au (111) disc (MaTeck, Germany) was used for STM (SPM 5200, Agilent, USA) studies of Aβ-modified surfaces.
The Au(111) disc was chemically cleaned using piranha solution (a mixture of H 2 SO 4 and H 2 O 2 at a volume ratio of 3:1), thoroughly washed with DI water, and dried with inert gas. Then, the Au(111) disc was annealed by a hydrogen flame, cooled to room temperature, and immediately further chemically modified. The Aβ-modified Au(111) sample as a working electrode was placed into an electrochemical STM cell composed of a Pt-wire counter electrode and a reference electrode. All electrochemical STM experiments were performed in a degassed aqueous solution containing 0.1 M HClO 4 under inert gas.
Electrochemical studies of Aβ molecules were performed by DPV (CHI 620, CH Instruments, USA) using a 5 s pulse period, 0.05 V amplitude, and 0.2 s pulse width. HOPG and ITO/glass electrodes were used for physically adsorbed Aβ layers and chemically adsorbed Aβ layers, respectively. A three-electrode electrochemical cell was used; Pt wire was used as the counter electrode, and an Ag|AgCl (saturated KCl) electrode was used as the reference electrode. The electrolyte was 0.1 M PB solution degassed with an inert gas before use. The geometric area of the working electrodes was 0.2 cm 2 .