Lipid bilayer formation on an ion image sensor and measurement of time response of potential dependency on ion concentration

Cells are the constituent units of all living organisms, and are divided from the external environment by their cell membranes. Cell membranes have functions as reaction fields for transporting substances, information, and energy in and out of cells via membrane proteins. These reactions trigger physiological activities such as metabolism, neurotransmission and cellular recognition, and their disruption directly relates to diseases. Membrane proteins account for a large part of drug discovery targets,¹⁾ therefore they are important research subjects in the fields of medicine and drug discovery. Ion channels are one of representative membrane proteins, and essential for controlling ion permeation though cell membranes. The fundamental structure of cell membranes are lipid bilayers, whose hydrophobic core effectively prevents the permeation of water-soluble substances including ions. Ion channels play roles of signal transducers responding to external stimulus, such as inherent binding of ligands, mechanical stimulation, and membrane potential. Ion channels make ions pass through the lipid bilayer membrane transiently according to the electric potential and ion concentration gradient.

Ion channels need to be embedded in a lipid bilayer membrane to maintain their proper structures and functions, as with other membrane proteins. Artificial lipid bilayer systems are used for in vitro assays of ion channels, e.g., liposomes, black membranes, supported lipid bilayers, and suspended lipid bilayers.^2,3) The most popular method to investigate ion channels with an artificial lipid bilayer membrane is ion current recording using the black membrane and a patch clamp amplifier.²⁾ The black membrane is composed of two lipid monolayers on a hydrophobic substrate that have a small aperture, where the lipid monolayers combine to form a bilayer. High resistivity on the giga ohm order is necessary to measure the ion channel current through single ion channel, which is typically the order of pA. The channel current recording with the patch clamp amplifier is a sophisticated technique that achieves real-time monitoring of single ion channel protein.⁴⁾ However, a low-noise and high-frequency amplifier is needed for each measurement point, and generally fabricating "giga ohm seal" is a matter of proficient. The number of simultaneous measurements are limited, and thus the improvement of throughput is demanded.^5–7)

Supported lipid bilayers (SLBs) are artificial lipid bilayer membranes situated at solid–liquid interfaces.^8–12) The lipid bilayer does not directly adhere to the substrate, because a thin water layer of 1−2 nm exists between the lipid bilayer and the substrate.^13–15) Lipid molecules laterally diffuse in the SLB, and the macroscopic fluidity of the membrane is maintained on the substrate. It is used to investigate the fundamental physicochemical properties and molecular distribution of lipid bilayers with atomic force microscope (AFM)^11,12,16) and fluorescence-microscope-based methods.^17–19) On the other hand, it is known that the SLB system is not suitable to the ion channel recording because the ion leakage through the water layer between the lipid bilayer makes formation of the giga ohm seal difficult.²⁾ In previous studies, channel measurements using SLB have been used in impedance spectroscopy.²⁰⁾

Semiconductor-based solid biosensors are applied to lipid bilayer membranes and cells in recent decades.^13,21–29) Integrating a large number of sensors on a chip achieves multi-points parallel measurements and spatial distribution of biological activities. Action potential, ion release, and vesicle release from cells are measured using these sensors. Applying the semiconductor-based solid sensors to the ion channel may lead to a large-scale parallel measurement. However, the channel current detection on a solid sensor is not achieved, because the SLB on a sensor surface also have the problem of the leak current as described above.²⁾

We aim to construct an artificial lipid bilayer system on the charge transfer type ion image sensor for detecting ion channel activities, as a step towards the large-scale multipoint measurement of ion channels. Figure 1 shows the schematic of the system in this study, in which we used the charge transfer type ion image sensor with an ion-selective plasticized polyvinyl chloride (PVC) membrane^27,30) and formed a SLB of phosphatidylcholine on the sensor. There are several essential points to achieve this system. First, the sensor surface needs to be covered with a SLB with sufficiently small leakage, to recognize the ions though ion channels from the leak ion flow. Second, the ion channels need to be reconstructed in the SLB, without damaging the lipid bilayer and the ion channel itself. Third, the SLB containing the ion channel need to be robust against the solution exchange for adding and removing inhibitors of the ion channel. We demonstrate the proof-of-concept of the system in Fig. 1, using gramicidin A (GA) as a model ion channel.

Zoom In Zoom Out Reset image size

2.1. Ion image sensor and K⁺ concentration measurement

2.2. Preparation of lipid vesicle suspensions and supported lipid bilayers

2.3. Fluorescence microscopy and atomic force microscopy

In the SLB system, the substrate affects the structure and properties of the lipid bilayer.¹²⁾ Figure 2(a) shows the AFM topography and cross-section profile of the PVC membrane on the Si₃N₄/Si substrate. The surface of the PVC membrane was flat, and its roughness was 0.29 nm (rms, N = 6) [Fig. 2(a)]. The roughness of the PVC membrane in this study was comparable to that of a thermally oxidized SiO₂ layer on a Si wafer.³⁷⁾ A streak pattern existed on the PVC membrane, although they were shallow enough to be hardly recognized in the cross-section profile. We attribute the streaks to a kind of self-organized patterns appearing during the solvent evaporation.^38,39) We confirmed that the streaks were not artifacts due to external noise, by observing the same sample position with different scanning rates and view sizes. The water contact angle (WCA) of the PVC membrane was 72°.

Zoom In Zoom Out Reset image size

Figure 2(b) shows the AFM topography and the cross-section profile of DOPC-SLB on the PVC membrane. The surface of DOPC-SLB was uniform, and we did not find a defect or hole on it. The roughness of the DOPC-SLB was 0.31 nm (N = 6), which corresponded to that of PC-SLBs on the thermally oxidized SiO₂/Si substrate in the previous studies.^19,37) Hydrophilicity of the substrate surface is one of dominant factors affecting the SLB formation by the vesicle fusion method.¹²⁾ The PVC membrane (WCA = 72°) was less hydrophilic than a chemically cleaned SiO₂/Si surface, but still it was hydrophilic if we assume the boundary of hydrophilic and hydrophobic surfaces at WCA = 90°. Previous studies showed that a lipid bilayer forms on a thermally treated SiO₂/Si substrate with WCA = 67°,³⁷⁾ while a lipid monolayer forms on hydrophobic alkyl-modified substrates with WCA >100°.^17,40) The results in Figs. 2(a) and 2(b) showed that microscopically uniform and flat DOPC-SLB was formed on the PVC membrane.

We investigated the macroscopic uniformity and morphology of the DOPC-SLB with optical microscopy. Figure 3 shows fluorescence images and the FRAP process of DOPC-SLB on the ion image sensor coated with the PVC membrane. Their brightness is adjusted to show the fluorescence intensity at the ion-sensing region. Outside of the sensing region had higher intensity and thus overexposed because of the reflection by the metal wiring in the circuit. FRAP process indicates the fluidity and continuity of a lipid bilayer based on the recovery of the fluorescence intensity.^35,36) Dye-labeled lipids at a part of a sample are bleached with strong excitation light, and then the fluorescence intensity recovers if the lipids in the bleached dark region and the surrounding bright region exchange through the lateral lipid diffusion. The DOPC-SLB on the sensing region showed uniform fluorescence intensity [Fig. 3(a)], and the intensity recovered in the FRAP measurement [Figs. 3(b) and 3(c)]. Fluid and continuous SLB was formed on the PVC membrane of the ion image sensor.

Zoom In Zoom Out Reset image size

Figure 4 shows the dependence of the output potential of the ion image sensor on the K⁺ concentration. The average potential of all active pixels at [K⁺] = 1, 10 and 100 mM before and after the formation of SLB are plotted. The potential gradients before and after the formation of the SLB were 63 mV/decade and 61 mV/decade, respectively. These are close to Nernst response as in the previous studies.^31,41) On the other hand, the output potential at each [K⁺] decreased after the formation of the SLB. The hydrophobic core of a lipid bilayer is an insulator with a high resistance value.^2,5,42) In the ion image sensor in this study, bias voltage is applied between the reference electrode and the sensing area.^30,32,33) The potential drop occurred after the sensing region was covered with an insulating SLB as shown in Fig. 3. Although the output potential decreased, the potential gradient with respect to [K⁺] did not change. In SLB systems, the lipid bilayer does not contact directly on a solid substrate, but ∼1 nm thick water layer exists between the lipid bilayer and substrate.^13–15) Thus, the ion concentration in this water layer was measured after the SLB was formed on the PVC membrane.

Zoom In Zoom Out Reset image size

We measured the output potential of the ion image sensor after exchanging the buffer solution of [K⁺] = 1 mM with that of [K⁺] = 100 mM. The temporal change of a representative pixel is shown in Fig. 5. We show the change in potential from that at [K⁺] = 1 mM, because the absolute value of the potential shifts depending on the condition of the PVC membrane surface, e.g. with and without a SLB, as shown in Fig. 4. The potential value corresponding to [K⁺] = 1 mM quickly changed to that of [K⁺] = 100 mM after the solution exchange at 0 s on the PVC membrane without SLB [Fig. 5(a)]. Inflection during the potential increase appeared around Time = 2.3 s, because of the fluctuation of the water surface when the solution exchange was started. This artificial response appeared all the potential time course in this study. After the formation of a SLB, the potential change after the solution exchange slowed down [Fig. 5(b)]. It is because the sensor surface was covered by a SLB and the ions were shielded. Meanwhile, K⁺ ions from the edges of the lipid bilayer membrane or pinholes on a nanoscale laterally diffuse through the ∼1 nm aqueous layer between the substrate and SLB,⁴³⁾ which we described above. The potential finally reached the equilibrium value at the [K⁺] concentration, as shown in Fig. 4.

Zoom In Zoom Out Reset image size

Figure 5(c) shows the temporal potential change after the buffer exchange, after GA was added to the SLB. We added GA corresponding to 10 GA molecules, thus 5 GA dimmers, on each ion sensing pixel, assuming all GA molecules were reconstructed in SLB. A steep potential response recovered after the addition of GA [Fig. 5(c)], in comparison with that of SLB without GA [Fig. 5(b)]. GA was reconstructed in SLB, and opened paths for K⁺ ions. We added Ca²⁺, which is a blocker of GA,^44–46) to the GA-reconstructed SLB, and measured the temporal potential change [Fig. 5(d)]. The potential response after the solution exchange delayed. The result showed that the channel activity of GA was inhibited by Ca²⁺.

We evaluate the temporal response after the [K⁺] change from 1 mM to 100 mM with t_const, at which the output potential reached 95% of the equilibrium value at [K⁺] = 100 mM. We summarize t_const obtained from the SLBs on the ion image sensor in this study in Table I. The PVC-coated ion sensor without SLB [Fig. 5(a)] showed t_const = 3.1 s. After the PVC membrane was covered with SLB [Figs. 3 and 5(b)], the potential response slowed down resulting in t_const = 31.2 s. Adding GA to SLB recovered the time response (t_const = 17.9 s). The K⁺ permeation through GA was blocked with Ca²⁺ in the solution (t_const = 21.9 s). The SLB formation, GA reconstruction, and blocking GA were clearly distinguished from t_const values. We demonstrated the reconstruction of model ion channel, and measurement of the inhibitor effect.

Generally, a lipid bilayer membrane with high resistivity on the order of GΩ is needed to record the single ion current of pA order. It is known that the SLB system has a disadvantage in ion channel measurement because of the ion leakage through the ∼1 nm aqueous gap between the lipid bilayer and the substrate, and defects in the lipid bilayer.²⁾ The leak currents through these paths exceeds the ion current signal through an ion channel. In this study, however, we measured the ion concentration in the aqueous gap, rather than the ion current. The [K⁺] change from 1 mM to 100 mM through the aqueous gap takes ∼30 s after SLB covered the sensor surface. This retard is sufficient to distinguish the formation of a SLB, and inhibition of GA with Ca²⁺. Reconstructed GAs in SLB opens paths for K⁺ diffusion, and recover the rapid [K⁺] change. As shown in Fig. 3, the overall sensor surface including the sensing area and the surrounding region was covered with a continuous lipid bilayer. The diffusion coefficient of K⁺ in bulk water at 25 °C is 1957 μm² s⁻¹,⁴⁷⁾ but the cation diffusion through the aqueous gap is restricted.⁴³⁾ The ion permeation through the aqueous gap in this study was suppressed lower than the current through approximately five GA dimers. Thus, we distinguished GA activity based on the time course of the output potential. We attribute this efficient ion shielding to the flat PVC membrane on the scale of subnanometer [Fig. 2(a)], and the defect-free and full-coverage SLB on in [Figs. 2(b) and 3].

The patch clamp technique is sophisticated enough to monitor the activity of ion channels on the single molecule level, but improvement in throughput is still demanded.^5,6) Using the ion image sensor, we can obtain the output potential of each pixel independently. The ion image sensor in this study, 128 × 128 pixels simultaneously measure the time course of the potential. We note that the methodology in this study is dedicated to the detection of the ion channel activity, in compensation for a precise real-time and quantitative single channel recording. The number of reconstructed GA, thickness of the aqueous nanogap, existence of pinhole in the SLB may vary among 16 K pixels of the ion image sensor. The time course of the potential would also vary, and some pixel may get fault, nevertheless thousands of independent and parallel measurement points are available. It potentially becomes a complementary method for the patch clamp method, for example rapid pre-screening prior to single channel recording.

We formed an artificial lipid bilayer system containing an ion channel peptide on the ion image sensor based on a K⁺-selective PVC membrane. Defect-free DOPC-SLB covered the ion sensor, and effectively prevented K⁺ diffusing to the ion sensor surface. Rapid ion permeation through the SLB after the addition of GA showed the reconstruction and channel formation of GA in the SLB. Recovery in the ion shielding due to Ca²⁺ addition indicated that the SLB stably kept its structure even after the addition of GA and repeated solution exchange. We demonstrated the detection of the ion channel activity and the blocking experiment with an inhibitor, using the charge transfer type ion image sensor. The proof-of-concept results in this study are valuable for developing a high throughput pre-screening method for the ion channel that is complementary to the patch-clamp-based techniques.

This work was supported by JSPS KAKENHI Grant No. JP15H03768, JST-CREST Grant No. JPMJCR14F3, JST-A-STEP, Casio Science Promotion Foundation, and Electronics-Inspired Interdisciplinary Research Institute (EIIRIS) Project from TUT.