A Polymer‐Based Nanopore‐Integrated Microfluidic Device for Generating Stable Bilayer Lipid Membranes

A parylene nanpore was built into multichannel microfluidic devices and a single protein channel was reconstituted in a bilayer lipid membrane (BLM) on the pore. Although BLMs on micropores used for channel recordings are usually too unstable for the exchange of solutions between the upper and lower sides, the BLMs formed in our nano­pores were stable enough to permit solution exchange. Bilayer lipid membranes (BLMs) and membrane proteins reconstituted in BLMs1-7 have received much attention for their application in areas such as drug discovery,8 single molecular detection,9-13 and the next generation of DNA sequencing.14, 15 Multichannel devices made of membrane protein arrays have recently been developed for high-throughput screening (HTS) of drugs, proteomics, and nanopore DNA sequencing.16-20 These devices are usually fabricated using microfabrication technology, which the major advantage that microfluidic channels composed in such devices can deliver the desired solution of biological materials to individual spots of the array. Applying this advantage, we established a reproducible preparation procedure for BLMs in devices through microfluidic channels.21, 22 The devices have open chambers and flow channels placed at the upper and lower sides of the BLMs; thus, the composition of the solutions at the BLMs can be changed using the microfluidic channel. However, the stability of the BLMs is still a problem for these devices. The BLMs on microsized support pores, which are used mainly in conventional devices, are susceptible to pressure fluctuations and mechanical vibrations during the solution exchange. Thus, solution exchange through microfluidic channels has not yet been performed. Although significant improvements in BLM stability have been achieved by sandwiching the bilayer with gel-phase materials23-25 and polymerizing lipid bilayers,26 these methods have not been applied to microfluidic devices. Here, we adopt a simple strategy to create stable BLMs using extremely small pores. Recently, White and co-workers achieved stable BLMs in the presence of alpha-hemolysin (αHL) with 320 nm-diameter pores.27 We believe that this is a straightforward, accessible method for the improvement of stability. Nanopores are usually fabricated in silicon materials using an ion beam or an electron beam via transmission electron microscopy (TEM).11, 12, 28 However, inherent disadvantages of using silicon materials include the low resistivity of silicon and the high dielectric losses and shunt capacitances that result from using thin insulating dielectrics; these factors lead to increased electrical noise and restricted bandwidth. Moreover, nanofabrication requires large-scale equipment, and it is time-consuming and costly. Polymer-based materials such as Teflon, polysulfone, polycarbonate (PC), and polyethylene teleftarate (PET) have high resistivity and low dielectric constants and are used for channel recordings.29 Whitesides and co-workers previously reported that the stability of BLMs on Teflon pores against an applied voltage was improved by decreasing the pore diameter (from 650 μm to 2 μm diameter).30 In addition, the current noise was improved by using smaller pores because the BLM capacitance causes overall system noise in protein channel recordings.30 Reduction of the bilayer capacitance using smaller diameter pores potentially allows for an increase in the current and temporal resolution of the recordings. Consequently, we aimed to fabricate polymer-based nano­pores integrated with microfluidic devices for stable BLMs. In this paper, we focused on poly(p-xylylene) (parylene) as an inexpensive support material for BLMs and attempted to make nanosized pores by depositing parylene onto micropores. Parylene has the following properties that make it useful as a BLM platform material: i) A GΩ seal, which is necessary for channel recordings with low noise, can be formed by BLMs on parylene pores. ii) Parylene films are polymerized during chemical vapor deposition (CVD), and, thus, the diameter of a micropore can be reduced by conformal deposition of parylene. iii) The design of parylene platforms is flexible because they are fabricated with a standard soft lithography. The parylene film can be peeled off31, 32 and integrated into the microfluidic devices using thermal bonding because the film is self-standing and easy to handle. We report here that BLMs were formed on nano­pores (PNPs), and that single αHL reconstitution and solution exchange were readily performed using a 3D microfluidic device integrated with PNPs. Additionally, the exceptional stability of the BLM on the PNP permitted long-term usage (ca. 120 h in the example presented here), allowing for an improved statistical treatment of the multichannel systems. The fabrication process for the PNP sheet is illustrated in Figure 1a. First, a 5 μm-thick parylene (paryleneC) film was polymerized in vapor and deposited on a single-crystalline silicon substrate. Aluminum was then deposited on the parylene and patterned using a standard photolithographic process. Using aluminum as a mask, parylene was etched by an oxygen plasma. After the aluminum was removed, the parylene sheet with microapertures was peeled from the silicon substrate using tweezers. After parylene films (1 piece: 3 mm × 3 mm, thickness: 5 μm) with 5 μm pores had been made using the general microelectromechanical system (MEMS) above, the size of the micropores was reduced to nanosize by another parylene conformal vaporization process. a) Fabrication process of the parylene nanopore sheet: (1) 5 μm parylene film is deposited on Si wafer; (2) 5 μm pore is designed by photolithography; (3) parylene is etched by oxygen plasma; (4) Al is removed; (5) parylene film is peeled off; (6) 5 μm pore is reduced by a second conformal deposition of parylene. b) Overview and cross-sectional images of the microfluidic device. c) Illustration of the thermal bonding process of the parylene nanopore film and microfluidic substrates made of PMMA for packaging at 120 °C for 20 min. Figure 2a presents the relationship between the parylene pore diameter and the thickness of the second parylene deposition on a control substrate that was simultaneously vaporized with the pore. The pore diameter decreased with increasing parylene thickness on the control by the secondary deposition. We expected that a 2.5 μm parylene film would entirely fill the 5 μm pre-pore. However, the pre-pore stayed open until nearly 3 μm of deposition in this experiment, probably because the growth rate inside the pore was slower than that in the control.33 Finally, we succeeded in consistently obtaining pores of 400 to 800 nm diameter using a thickness of the second deposition equal to 2.9 μm. Scanning electron microscopy images are shown in Figure 2b. A single round nanopore was observed on the large film (3 mm × 3 mm). The contact between the first and second vaporized parylene layers was strong, and these layers did not exfoliate each other. This second vapor deposition method can be applied for pores not only on parylene but also on a variety of substrates, such as silicon, glass, and other polymer films. Figure 2c shows a typical current–voltage curve recorded using a PNP in 1.0 M KCl buffer with a potentiostat. The resistance of the pore opening in this membrane was 0.72 MΩ. This value roughly corresponded to the value obtained for a similar-sized glass nanopore34-36 and the value estimated37 from the SEM images. a) Relationship between the diameter of the reduced parylene pore and the thickness of the parylene membrane on the flat substrate. b) Scanning electron microscopy images of a parylene nanopore after the second deposition. A 400 nm-diameter round pore is observed in a large film (3 mm × 3 mm). c) Current–voltage curves of a parylene nanopore in 1.0 M KCl 10 mM PBS buffer at pH 7.4. The pore resistance is 0.73 MΩ. The inset shows the resistance of three independent nanopores in the microfluidic device. d) A photograph of the microfluidic device with tubes. Next, we integrated the PNP films into a microfluidic device. The PMMA microfluidic device designed had dimensions of 40 mm × 30 mm × 3 mm. As illustrated in Figure 1b–c, it consisted of four parts: 1) a top layer including upper channels (400 μm in both width and depth) for three individual recording sites with Ag/AgCl electrodes, 2) three parylene films with a nanopore, 3) a separator (supporting the parylene films and separating the top and bottom layers), and 4) a bottom layer with a lower channel and an Ag/AgCl electrode. The devices were fabricated by machining a 1.5 mm-thick plate for the top and bottom layers and a 0.2 mm-thick plate for the separator using an automated computer aided design and manufacturing (CAD/CAM) modeling machine (MM-100, Modia systems, Japan). They were bonded by heating at 120 °C for 20 min (Figure 1c). A photograph of the device is shown in Figure 2d. Teflon tubes were connected from microsyringes (Hamilton, Nevada) to each flow channel in the top and bottom layers. The resistances of the individual nanopores in the device were tested. The three individual pores showed similar resistances (Rp = 0.71 MΩ ± 0.02 MΩ, inset of Figure 2c), thus ensuring the complete sealing of the assembled device; even a slight current leaking through a pass other than the pore would spoil current channel recordings. This result indicates that the fabrication process is adequate to measure protein channel conductance at each flow channel. Here, we note that nanopores patterned on millimeter-scale parylene films fabricated by a standard process are easy to handle and integrate readily into conventional microfluidic devices. The three-step BLM formation procedure (shown in Figure 3) has been described in our previous papers.21, 22 First, buffer solution was used to fill the upper and lower flow channels. Second, 1–5 μL of 1,2-diphytanoyl-sn-glycero-3-phosphocholine (DPhPC) dissolved in n-decane (10 mg mL−1) was injected into the upper flow channel and passed through onto the surface of the PNP. At this time, the amphiphilic lipid molecules became spontaneously ordered at the interface between the n-decane and the buffer solution. Third, the buffer solution was introduced sequentially into the upper channel. The bilayer formed autonomously via self-assembly of the lipid molecules when the lipid layer became thin. The resistance of the nanopore region was monitored using a patch-clamp amplifier. The DPhPC/n-decane system is commonly used to study planar BLMs and their physical properties are well known. In this study, the pore resistance was increased from 0.72 MΩ to 3–30 GΩ by BLM formation. It is known that the resistance of BLMs is more than 1 GΩ.29 a) Procedure for BLM preparation in the microfluidic device shown in an illustration (top, not to scale) and a photograph (bottom). b) A current–time trace corresponding to (i) the baseline current associated with a BLM spanning a parylene pore, (ii) the insertion of a single αHL channel (50 pA current flow), and (iii) binding of individual heptakis(6-O-sulfo)-β-cyclodextrin (s7βCD) molecules to the αHL channel after solution exchange. Data were recorded at a DC bias of 50 mV in a 1.0 M KCl in 10 mM phosphate buffered saline (PBS) solution with pH 7.4 at 23 °C ± 1 °C. c) Current histograms before (between αHL formation and solution exchange) and after the solution exchange. Solid lines are Gaussian fits yielding peak values. The clogged channel state with the s7βCD binding appears after the solution exchange. d) Apparent membrane capacitance versus time trace of BLM on PNP without flowing. The BLM was stable for 119.6 h in this microfluidic device. Generally, BLM formation on micropores can be observed directly as the formation of a black membrane by optical microscopy or can be monitored by the membrane capacitance. However, the black membrane state of the BLMs in the nanopore cannot be observed by a microscope. Furthermore, the precise capacitance of the BLM on PNP could not be monitored in this device. The apparent capacitance consists of the membrane and a parasitic capacitance of the device was used. To confirm BLM formation, αHL solution (0.3 μM in the buffer) was introduced from the lower channel while monitoring the current. An abrupt increase in the current appeared after several minutes under a 50 mV bias as shown in Figure 3b. The average conductance was 0.9 nS ± 0.1 nS. The single αHL channel conductance in 1.0 M KCl has been reported to be ≈1 nS;38 the present data thus suggest that a BLM formed at the PNP. When αHL was not reconstituted, the apparent capacitance of BLM was around 4 pF. In contrast, the αHL channel formation was observed when the apparent capacitance was around 10 pF in this device. As a result, we can estimate whether or not a BLM forms by moni­toring the apparent capacitance in this device. In our previous studies, the BLM formed in the micropore was unstable in a device with upper and lower flow channels. Thus, we typically used a device with a single flow channel and an open chamber at the BLM.19, 19-22 The major cause of the poor stability is that the hydrodynamic pressures on both sides of the pore are different; as a result of the imbalance, the unstable BLM easily ruptures. Here, we tried to exchange the solution at the BLM on the PNP through the microfluidic channels. After single αHL reconstitution, we exchanged the solutions in both the upper and lower channels. The operation was performed by hand rather than with a syringe pump. Surprisingly, BLMs with a protein channel stayed intact during this operation. The αHL solution was removed carefully from the lower channel and replaced by a buffer without any analytes. Removal of the protein solution suppresses additional αHL reconstitution and prevents complicated data processing due to multiprotein channel conductance. In addition, BLMs in the presence of multiproteins break easily. Therefore, maintaining the single-channel state is vital for channel recordings. The protein solution removal revealed that BLMs in PNPs had sufficient stability to withstand solution exchange. This result can be explained by the small area of the pore: the area of a 400 nm pore is more than 104 times smaller than the area of a 50 μm pore that was commonly used in previous studies. To demonstrate the advantages of our device and its solution exchange performance, we used s7βCD (50 μM) as the analyte for protein channel recordings. The upper channel was filled with a buffer without analytes and then replaced with an analyte solution in the single protein channel state. s7βCD is a cyclic molecule composed of seven glucose units, each possessing a sulfonate group located at the sixth position.39 s7βCD has previously been shown to enter αHL from the β-barrel (trans) side and to noncovalently bind to the lumen. One hundred micromolar s7βCD in buffer was injected into the upper channel. Figure 3b shows the protein channel insertion and blocking currents after the solution exchange, suggesting single binding events of s7βCD to αHL. Reversible binding of s7βCD results in a temporary 92% reduction in the conductance of the protein.39 The histograms of the current amplitudes before and after the solution exchange are shown in Figure 3c. A peak corresponding to the open protein channel was observed before the solution exchange. After the s7βCD solution was introduced, open and clogged channel currents were observed. Each binding event corresponds to a 0.82 nS decrease in the conductance, representing 91% ± 3% of the open channel conductance, in good agreement with the value reported in the literature.39 In addition, the dwell time of s7βCD in the protein pore was 0.65 s ± 0.14 s (n = 25) at 23 °C ± 1 °C. This value is also consistent with the reported value.39, 40 We terminated the current recordings after the second αHL formation, which usually occurred several minutes after the first channel formation. The BLMs in PNPs showed reproducible tolerance to solution exchange (n = 4), indicating the high stability of this system. Finally, we examined the long-term stability of the BLM on the PNP. The stability was investigated by monitoring the apparent capacitance of the BLM on a PNP in the microfluidic device without flowing. Figure 3d shows the capacitance–time trace of the most stable BLM demonstrating that the capacitance was ca. 10 pF for 119.6 h (ca. 5 days). This capacitance indicates that an intact BLM was maintained for around 60 to 120 h (n = 3). The BLM was ruptured spontaneously after 119.6 h. The lifetime of the BLM in this device is measured in days, whereas it is at most a few hours for conventional BLMs formed on micropore. We believe that this robust property of the BLM will encourage the multichannel screening technology using microfluidic devices. In conclusion, we have succeeded in fabricating nano­pores in parylene films using a simple conformal deposition method. A BLM was formed in the PNPs using microfluidic channels and single a αHL was inserted in the BLM. In addition, we demonstrated solution exchange through the detection of single s7βCD molecules. Solutions at two different sides of a membrane protein channel can be controlled independently using our device, which is impossible in experiments using living cells. Surprisingly, the BLM on a PNP was stable for ca. 120 h in this device. This BLM device can be applied to biological or medical analysis targeting membrane proteins, HTS for drug discovery, and DNA sequencing. Also, the device is inexpensive and can be used to perform reproducible analysis for single molecule channel recordings. Additionally, PNPs have great potential for other applications such as membrane filters, a Coulter counter, and automated high-throughput patch clamping of living cells because the size and design of the parylene pores can be controlled. Materials: Reagents were obtained as follows: parylene monomer (Parylene Japan, Japan); a single-crystalline silicon substrate (Specialty Coating Systems, Indiana, USA); poly(methyl methacrylate) substrate (Mitsubishi rayon, Japan); KCl, K2HPO4, KH2PO4, and EDTA (Wako, Japan); 1,2-diphytanoyl-sn-glycero-3-phosphocholine (Avanti Polar Lipids, Alabama, USA); wild-type alpha-hemolysin monomer (Sigma-Aldrich, MO, USA); n-decane (Sigma-Aldrich, MO, USA); and heptakis(6-Osulfo)-β-cyclodextrin (TRC, Canada). Determination of the pore size of a PNP: Scanning electron microscopy was carried out using FE-SEM SU8000 (Hitachi, Japan). The size of the pore was analyzed by ImageJ (open source software from the National Institutes of Health). Current versus voltage measurements between −0.1 and +0.1 V at 0.05 V s−1 were performed using an ALS1206a (BAS, Japan) as a potentiostat in 1.0 M KCl, 10 mM PBS, 1 mM EDTA, pH = 7.4 at 23 °C ± 1 °C. Channel recordings: The channel currents were monitored using a patch-clamp amplifier, model CEZ-2400 (Nihonkoden, Japan). The signal was detected through a 0.2 kHz low-pass analog filter at a sampling frequency of 1 kHz in 1.0 M KCl, 10 mM PBS, 1 mM EDTA, pH = 7.4 at 23 °C ± 1 °C. The BLM resistance and capacitance were measured by applying an AC signal (V = 5 mV peak to peak, f = 100 Hz) and they were calculated from fitted results. The electric current was recorded using a digital data acquisition system Digidata 1440A (Molecular Devices, CA, of current amplitudes and was performed using 10 Devices, CA, and data are as value ± standard of BLMs in was used in this is to αHL To a thin we performed the following When the resistance did not increase 1 the second and of the three-step BLM formation were After the GΩ seal, we the of a thin lipid layer by application of a V bias the pore. A formed BLM on a nanopore has a voltage of In the event that was the was to be clogged with the lipid solution. In this the solution was removed by buffer from the upper to the lower and then the second and were until the lipid layer became thin. After a thin lipid layer we injected αHL solution. The apparent capacitance after this increased from around 4 to 10 pF. αHL channels can be formed in this BLM. measurements were performed at 23 °C ± 1 °C in a This was by the and for on in The and for their in the

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.