Electrografting a Hybrid Bilayer Membrane via Diazonium Chemistry for Electrochemical Impedance Spectroscopy of Amyloid-β Aggregation

Herein, a novel hybrid bilayer membrane is introduced as a platform to study the aggregation of amyloid-β1–42 (Aβ1–42) peptide on surfaces. The first layer was covalently attached to a glassy carbon electrode (GCE) via diazonium electrodeposition, which provided a highly stable template for the hybrid bilayer formation. To prepare the long-chain hybrid bilayer membrane (lcHBLM)-modified electrodes, GCE surfaces were modified with 4-dodecylbenzenediazonium (DDAN) followed by the modification with dihexadecyl phosphate (DHP) as the second layer. For the preparation of short-chain hybrid bilayer membrane (scHBLM)-modified electrodes, GCE surfaces were modified with 4-ethyldiazonium (EDAN) as the first layer and bis(2-ethylhexyl) phosphate (BEHP) was utilized as the second layer. X-ray photoelectron spectroscopy (XPS) and time-of-flight secondary ion mass spectrometry (ToF-SIMS) were used to characterize the bilayer formation. Both positively charged [Ru(NH3)6]3+ and negatively charged ([Fe(CN)6]3-/4-) redox probes were used for electrochemical characterization of the modified surfaces using cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS). EIS results showed a decrease in charge transfer resistance (Rct) upon incubation of Aβ1–42 on the hybrid bilayer-modified surfaces. This framework provides a promising electrochemical platform for designing hybrid bilayers with various physicochemical properties to study the interaction of membrane-bound receptors and biomolecules on surfaces.

Amyloid-β (Aβ) is a hallmark protein implicated in Alzheimer's disease (AD). The ability of Aβ peptides to disrupt membrane integrity have been studied by numerous research groups in the past decade [61][62][63][64]. Lindberg et al. [65] reported that lipid membranes of dioleoylphosphatidylcholine catalysed the fibril formation of Aβ 1-42 through lipid-fibril interactions that reinforced secondary pathways. Single molecule microscopy has been applied to track individual Aβ peptide diffusion on lipid bilayers, with Chang et al. [66] reporting that trimers and larger oligomers were immobilized on the lipid bilayer. Kandel et al. [67] studied the cholesterol-dependent membrane pore formation of Aβ [25][26][27][28][29][30][31][32][33][34][35] peptides. Additional evidence was provided by Capone et al. [68] suggesting that Aβ peptides induced an ion channel-like ion flux in cellular membranes that was independent from the postulated ability of Aβ to modulate intrinsic ion channels or transporter proteins. Lal et al. [69] reviewed the high-resolution 3D structure of Aβ channels and their relevance to the amyloid channel paradigm. Recently, it has been established that Aβ oligomers had a profound detergent-like effect on lipid membrane bilayers as imaged by AFM and electron microscopy. Since the aggregation of Aβ peptides has been a topic of significant interest, there have been numerous studies to follow this process using electrochemical techniques [70][71][72][73][74][75][76][77][78][79][80].

Electrode Pretreatment
Glassy carbon electrodes (GCEs) (3.0 mm diameter) were purchased from CHInstruments Inc. (Austin, TX, USA). To prepare the modified electrodes, GCEs were first polished with alumina powder (1, 0.3, and 0.05 µm in sequence), followed by sonication in deionized water and subsequently ethanol (95%) for 30 min each. The electrode surface was then acid-activated by running a cyclic voltammogram in 1 M H 2 SO 4 for 15 scans between −1.2 and +1.2 V (vs. Ag/AgCl) at a scan rate of 0.1 V/s. DHP and BEHP solutions (10 mM) were prepared in deionized water.

Synthesis of 4-dodecylbenzenediazonium tetrafluoroborate (DDAN)
4-Dodecylbenzenediazonium was prepared from 4-dodecylaniline using procedures as reported in the literature for similar molecules [81]. Briefly, 4-dodecylanaline (0.5 g, 2 mmol) was mixed into a solution that contained equal volumes of glacial acetic acid and concentrated propionic acid (7 mL in total), to which 2.5 mL of HBF 4 was added. The solution was cooled to 6 • C, to which 0.2 g of NaNO 2 was slowly added. The mixture was stirred for 1 h at 6 • C and was subsequently vacuum filtered to isolate the product, which was washed with ethanol and dried. The yield was measured to be approximately 85.1%. MS ( Figure S1), FT-IR ( Figure S2), and NMR ( Figure S3) spectra of the product are shown in Figures S1-S3, respectively. The isolated product (yellow-orange) was stored in a sealed container over CaCl 2 at 4 • C until further use.

Synthesis of 4-ethylbenzenediazonium (EDAN)
4-Ethyldiazonium was prepared in situ using a solution of 4-ethylaniline (2 mM) with an equimolar sodium nitrite in 1.25 M HCl solution according to previous reports [54]. This solution was purged with nitrogen gas for at least 10 min.

Electrode Modification
For the modification of GCE surfaces with lcHBLMs, GCEs were modified by the 4-dodecylbenzene moiety by first dissolving 4-dodecylbenzene (2 mM) and tetrabutylammonium tetrafluoroborate (10 mM) in acetonitrile. CV was run between −0.9 V and +0.6 V for 30 scans at 50 mV/s (Figure 1), resulting in the formation of the first layer. The electrodes were taken out of the solution, rinsed with deionized water, and sonicated in deionized water for 1 min. For the adsorption of the second layer, a solution of vesicles was prepared by mixing 10 mM DHP in deionized water along with equimolar NaOH to assist dissolution. The solution was then sonicated for 2 h to form a homogenous mixture. Subsequently, GCEs that were formerly modified with the first layer were incubated with 100 µL of the solution overnight. For the modification of GCE surfaces with short-chain hybrid lipid membranes (scHBLMs), immediately following the synthesis of EDAN in situ, CV was applied between 0.6 V and −1.2 V vs. Ag/AgCl for 10 cycles at 50 mV s −1 scan rate ( Figure S4) [54]. The electrodes were taken out the solution, rinsed with deionized water, and sonicated in deionized water for 1 min. Electrodes modified with 4-ethylbenzene were then incubated with a 10 mM solution of BEHP in deionized water mixed with equimolar NaOH to assist dissolution.

Electrochemical Measurements
The modified GCEs (both ethyl and dodecyl bilayers) were analyzed electrochemically using cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) in either a 2.5 mM solution of [Fe(CN) 6 ] 3−/4− with 50 Mm NaBr or a 2 mM solution of [Ru(NH 3 ) 6 ] 3+ with 50 mM NaBr (NaBr was added to the aforementioned solutions to act as a counterion). CV measurements were performed at a scan rate of 50 mV s −1 unless otherwise stated. EIS data were collected as Nyquist plots with an applied bias of −0.2 V and +0.2 V for measurements performed using [Ru(NH 3 ) 6 ] 3+ and [Fe(CN) 6 ] 3−/4− , respectively at a frequency ranging from 100 MHz to 100 kHz.

Aβ 1-42 Aggregation Studies
Aβ 1-42 was obtained from AnaSpec Inc. (Fremont, CA). Aβ 1-42 was first treated using HFIP as described before [82]. Briefly, 0.5 mL of HFIP was added to 1 g of Aβ 1-42 , which was then sonicated for 15 min. The solution was then stored at 4 • C overnight. The solution was then aliquoted, after which the HFIP was dried under N 2 . Aβ 1-42 films were then stored at −20 • C until required for the experiment. Immediately before measurement, 1 mL of 0.01 M phosphate buffer (pH 7.4) was added to dissolve the peptide. lcHBLM-modified GCEs were then incubated with 100 µL of the peptide solution for 10 min, 24 h, and 48 h. Aβ 1-42 -modified lcHBLM surfaces were gently washed with deionized water immediately before EIS measurements.

Covalent Modification of GCE Surfaces
To prepare the hybrid bilayer, DDAN and EDAN were utilized as precursors to covalently attach the alkylaryl chains onto the GCE surfaces. As described in Section 2 (also shown in Scheme S1), DDAN was synthesized, isolated, and characterized using NMR, MS, and FTIR (Figure S1-S3). EDAN was prepared in situ using 2-ethyl aniline (Scheme S2). To graft DDAN onto GCE surfaces, thirty consecutive cycles of CV were performed at a scan rate of 0.05 V s −1 between 0.6 V and −0.9 V vs. Ag/AgCl (Scheme 1). Every time, the HBLMs were prepared on a new, bare GCE to avoid any remnants from prior surface modifications interfering with the study. Figure 1 shows the first, second, and last two voltammograms for DDAN grafting. As shown in Figure 1, a broad irreversible peak can be seen at 0 V for the first scan, which represented the reduction and loss of N 2 as previously reported for other diazonium modifications [50,54,58]. Our investigations showed that after thirty cycles, the current difference between the last two voltammograms displayed less than 2% difference for almost all measurements.

Preparation of the Hybrid Bilayer Membrane (HBLM)
To prepare the second layer, DHP was chosen for the electrodes with the first layer made of DDAN. For electrodes with EDAN as the first layer, BEHP was chosen as the second layer. A solution of 10 mM DHP was prepared and sonicated for 2 h to form vesicles prior to incubation. Figure S6 shows TEM images of these vesicles. Subsequently, 100 µL of the prepared solution was incubated on the DDAN-modified surfaces at room temperature overnight. The modified GCEs were then gently rinsed with deionized water followed by a 1 min sonication of the GCE-DDAN-DHP, which is coined here as the lcHBLM), to remove the non-specifically adsorbed DHP molecules/vesicles (Scheme 2). The same procedure was performed to synthesize GCE-EDAN-BEHP using a 10 mM solution of BEHP, which is coined here as the scHBLM. Scheme 1 shows an illustration of the lcHBLM. Contact angle measurements (Table 1) performed on the bare GCE surfaces displayed approximately 73.4 • . The contact angles increased after the first layer was deposited to about 85.6 • due to the increased hydrophobicity resulting from the nonpolar alkyl chains populating the GCE surface for both lcHBLM and scHBLM. Following the construction of the second layer, contact angle measurements displayed a decrease to approximately 79.1 • for both lcHBLM and scHBLM. This was attributed to the fact that the second layer consisted of negatively charged phosphate groups, which resulted in the increased hydrophilicity of the modified surfaces. Furthermore, ellipsometry studies were conducted to measure the length of the bilayer, which showed that the measured length was 84.8 Å for the monolayer of the lcHBLM. However, the calculated length was only 17.3 Å. We hypothesize that this is attributed to the first and second layers having different defections due to some multilayers possibly having been formed during electrografting. For future studies, the formation of multilayers can be prevented by adding a radical scavenger, such as 2,2-diphenyl-1-picrylhydrazyl, in excess or by electrodepositing diazonium moieties that possess a methyl group that is meta to the position of grafting [83][84][85]. Research in our laboratory to verify this hypothesis is in progress.
Surface characterizations of the HBLMs were performed using X-ray photoelectron spectroscopy (XPS) and time-of-flight secondary ion mass spectrometry (ToF-SIMS). XPS results showed a distinctive phosphorus peak for the modification of the GCE surface with DDAN-DHP ( Figure 2). ToF-SIMS data for bare GCE, GCE-DDAN, and GCE-DDAN-DHP are shown in Figure 3A,B. In Figure 3A, the spectrum with negative polarity showed the mass of 4-dodecylbenzene, which formed the first layer of lcHBLM (DDAN) for both GCE-DDAN and GCE-DDAN-DHP, and no considerable peaks for the bare GCE. The spectra with positive polarity are shown in Figure 3B with the mass of dihexadecyl phosphate (DHP; the second layer of lcHBLM) only for GCE-DDAN-DHP and no considerable peaks for bare GCE and GCE-DDAN.

Electrochemistry of lcHBLM-and scHBLM-Modified Surfaces
Ruthenium hexamine ([Ru(NH 3 ) 6 ] 3+ and ferri/ferrocyanide ([Fe(CN) 6 ] 3−/4− , having overall positive and negative charges, respectively, revealed the electrochemical characteristics of lcHBLM-and scHBLM-modified surfaces. Figure 4A Figure 4B. The blue graph (insertion) shows the small R ct of [Ru(NH 3 ) 6 ] 3+ on a bare GCE. The red graph shows a relatively high R ct when the electrode was modified with the first layer (GCE-DDAN). Similar to CV, EIS for [Ru(NH 3 ) 6 ] 3+ in the presence of a second layer (GCE-DDAN-DHP), a decrease in R ct was observed compared to the monolayer (green graph). These behaviors were reproducible for all measurements performed on different days with new solutions and different GCE surfaces. A similar behavior was observed for scHBLM (which consists of shorter molecules for both first and second layers) as shown in Figure S5.   Table S1. added to the semicircle diameter in EIS (green), which was attributed to the repulsion of the negatively charged groups in the second layer. For scHBLM, a similar behavior was observed, which is shown in Figure S2C,D. To simulate all EIS graphs, a modified Randles equivalent circuit ( Figure 4B,D) was utilized with all values of the equivalent circuit elements as summarized in Table S1.
Furthermore, when the C dl of the lcHBLM in the context of [Ru(NH 3 ) 6 ] 3+ (Table S1) was examined, an increase in capacitance was observed (from 48.2 to 740 nF) between the GCE-DDAN and GCE-DDAN-DHP, which was hypothesized to be due to the increase in charge presented by the negatively charged phosphate groups of the DHP and the positively charged [Ru(NH 3 ) 6 ] 3+ , resulting in increased charge separation. However, a slight decrease in C dl was observed for [Fe(CN) 6 ] 3−/4− (from 60.5 to 57.2 nF) indicating that the overall charge separation remained relatively unchanged. Furthermore, when the Z w was examined, an increase in diffusion was observed for [Ru(NH 3 ) 6 ] 3+ (from 5.81 to 38.6 µMho·s −1/2 ), which was hypothesized to be due to the attraction of the positively charged ions to the negatively charged DHP. This was further verified by the slight decrease in Z w for the [Fe(CN) 6 ] 3−/4− (from 5.18 to 2.86 µMho·s −1/2 ), which was thought to be due to the slightly more hindered diffusion of the probe as a result of the repulsion between the negatively charged [Fe(CN) 6 ] 3−/4− and DHP layers.
To investigate the mechanical stability of the bilayer, we stirred both EDAN-modified GCEs and EDAN-BEHP-bilayer-modified GCEs at 500 rpm for 5 min. EIS was performed using [Ru(NH 3 ) 6 ] 3+ as the positively-charged redox probe before and after high-speed rotation ( Figure 5). Values of the simulated equivalent circuit elements are shown in Table S2.  Table S2. A comparison of each EIS result showed that both monolayer and bilayer were stable after a high-speed rotation. Work towards electrochemical kinetic investigations using HBLM-modified rotating disc electrodes are in progress in our laboratory.

Interaction with Aβ 1-42
Aβ 1-42 , which is a well-described hallmark protein of AD [71][72][73][74], was used as a model system to explore the applications of lcHBLM-modified surfaces (the lcHBLM was used as opposed to the scHBLM due to the size of the protein being too large to be embedded in the scHBLM). To study the interaction of Aβ 1-42 with the lcHBLM, we prepared a fresh solution of 10 µM of Aβ 1-42 in PBS (pH 7.4) and incubated an aliquot of the peptide solution (20 mL) for 10 min, 24 h, and 48 h on the lcHBLM-modified surfaces. PBS was used in this proof-of-concept study so that any electrochemical changes detected were solely attributed to the interaction of Aβ 1-42 with the lcHBLM. In the presence of biological fluids, non-specific adsorption of interfering particles could have provided misleading results in these preliminary experiments. However, further investigations are planned to study the behaviour of Aβ 1-42 within the lcHBLM in the presence of biological fluids such as cerebrospinal fluid. After incubation, the electrodes were washed thoroughly with PBS and then deionized water before EIS measurements ( Figure 6) were taken using [Ru(NH 3 ) 6 ] 3+ and [Fe(CN) 6 ] 3−/4− as redox probes. As shown in Figure 6A,B, the R ct of Aβ 1-42 -modified surfaces decreased significantly after 10 min of incubation time (purple). This decrease in R ct was attributed to the aggregation of Aβ 1-42 on the bilayer creating disruption on the ordered membrane surface that facilitated the diffusion of redox probes to the GCE surface. The disruption of neuronal cell membranes with the oligomers and fibrils of Aβ 1-42 to form pores was previously described to cause neurotoxicity [75][76][77][78][79][80]. A significant decrease in R ct was observed with a similar trend for both redox probes, despite the fact that the charges of probes were opposite. This strengthened our hypothesis that some nano/micro pores were formed in the lcHBLM layer, which might have facilitated the charge transfer between the GCE surface and redox probes, as has been reported previously [86,87]. Additionally, when other circuit elements were examined in the context of the Aβ  (Table S3). This was hypothesized to be due to an increase in charge separation between the undisturbed portions of the lcHBLM and [Ru(NH 3 ) 6 ] 3+ probe as Aβ 1-42 further aggregated and settled within the biomimetic membrane. However, further experiments need to be performed to verify this hypothesis. Meanwhile, the Z w for both probes showed an increase from 236 to 1330 µMho·s −1/2 for [Fe(CN) 6 ] 3−/4− and from 560 to 1930 µMho·s −1/2 for [Ru(NH 3 ) 6 ] 3+ . Similar studies have been performed using octadecanethiol monolayers on Au electrodes as described by Valincius et al. [88], who observed an instantaneous increase in capacitance upon the injection of small Aβ 1-42 oligomers (1-4 nm in diameter) followed by a gradual return to near original C dl within an hour, which they attributed to the disruption of the octadecanethiol SAM. However, their studies were performed on a smaller time scale (1 h). Valincius et al. [89] also reported that Aβ 1-42 oligomers caused damage to the tethered BLMs composed of varying phospholipids, which they hypothesized to be due to the Aβ 1-42 aggregating into pore-like structures.
Further studies aiming to observe the formation of Aβ-based pores on lcHBLMmodified surfaces are in progress in our laboratory using atomic force microscopy (AFM). Another noticeable trend in EIS results was that a significant decrease in R ct was observed after a short time (10 min) of incubation with Aβ 1-42 . The difference between the R ct values obtained at 24 and 48 h of incubation was not statistically significant (Table S3). In terms of differences in R ct observed with [Ru(NH 3 ) 6 ] 3+ as the redox probe, there was a significant decrease of 72.1% in R ct between 0 and 10 min, while between 10 min and 24 h, there was a further 17.1% decrease. However, in between 24 h and 48 h incubation periods, there was a negligible decrease of 0.8% in R ct . In the case of the [Fe(CN) 6 ] 3−/4− as the redox probe, there was a 26.2% decrease between 0 and 10 min. A significant decrease of 51.8% in R ct was observed between 10 min and 24 h, but only a 3.8% decrease in R ct was detected between 24 h and 48 h incubation periods. These observations implied that the disruption of the lcHBLM layer took place at the early stages of Aβ 1-42 aggregation. Electrochemical investigations to discover small molecules that would affect the interaction of Aβ 1-42 with the lcHBLM-modified surfaces is in progress in our laboratory.  Table S3.

Conclusions
Our preliminary results displayed the synthesis and electrografting of a novel hybrid bilayer membrane on the surface of GCEs. The lcHBLM-modified GCEs were utilized to study the Aβ 1-42 aggregation process upon interaction with the surface-anchored mem-brane. This platform can be customized according to the purpose of the study by choosing appropriate molecules that are used for diazonium salts for the covalently attached first layer as well as the phospholipid (or other bipolar molecules) as the second layer. A combination of different molecules can also be used for the two layers to adjust the affinity, thickness, and other physical properties of the HBLMs. We envisage that similar EIS studies can be performed using lcHBLM-modified GCES in connection with membrane-bound biomolecules to understand their interactions with small molecules in drug screening assays. Furthermore, future studies aim to use these novel HBLMs to quantify biomolecules in the context of biological fluids by embedding antibodies, aptamers, as well as other biorecognition elements within the HBLM.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/mi13040574/s1, Scheme S1: Schematic diagram for synthesis of DDAN; Scheme S2: Schematic diagram for in situ synthesis of EDAN; Figure S1: Mass spectrometric characterization of DDAN; Figure Table S1: The values of simulated equivalent circuit elements of Nyquist plots shown in Figure 4B,D; Table S2: The values of simulated equivalent circuit elements of Nyquist plots shown in Figure 5 with the redox probe [Ru(NH 3 ) 6 ] 3+ ; Table S3: The values of simulated equivalent circuit elements of Nyquist plots shown in Figure 6.