Introduction

Biological channel proteins, embedded in lipid bilayer, act as nanochannels regulating the transmission of various biomolecules and ions, which is vital to life process1,2. Scientists modify the nanochannels, obtained by extraction or biomimetic synthesis, with function elements (FEs) for the manipulation of ion translocation3,4,5,6,7,8,9,10. During the decades, widespread advances have been achieved on nanochannels correspondingly, including nanochannel based DNA sequencing11,12,13,14, single-molecular detections15,16,17,18, smart sensors19,20,21,22, energy transfer/storage23,24,25, and so on26,27,28,29,30. However, researchers focus all interests on the contribution from the FEs at the inner wall (IW) of nanochannels (FEIW), little attention has been paid on the contribution from the FEs at the outer surface (OS) of nanochannels (FEOS)10,31,32,33. Currently, theory simulations have forecasted the respective contribution from the FEs at OS and IW to ionic gating34,35. However, till now, experimental demonstration is rare. An essential issue has been proposed: what roles do the FEOS play in nanochannels? If we failure to settle this issue, it will be an obstacle to the nanochannel-based applications such as DNA sequencing36,37,38, molecule sensing39,40,41,42,43, and so on24,25,26. For instance, recognition a nucleotide base in sequencing typically relies on a subtle distinction of current and dwell time signatures by the blockade of inner nanochannels, which, however, cannot be actualized when the base has not access to the inner of nanochannel44. On the other hand, it thus may be envisioned that elaborate FEOS would be probably beneficial to differentiate a specific base in a DNA sequence.

In a very recent work, we made a preliminary experimental to distinguish the contribution of FEIW and FEOS to ionic gating45. The FEOS were proved to synergistically enhance the ionic gating efficiency. However, the explicit partition of regional functionalization of FEIW and FEOS has not been demonstrated. Furthermore, the independent function of the FEOS has not been achieved.

Additionally, the nanochannel, could also be nonspecifically blocked by interfering molecules, producing even false signals especially in a complex system, which would severely reduce the performance for DNA sequencing, molecule sensing and et al.46,47 We expect that the rational FEOS would block the interference molecules to reduce the false signal. Furthermore, it is expected that distinct FEOS and FEIW would invest a nanochannel with distinct but coordinate functions to accomplish the above tasks in complex environments, which only could be achieved by the functions of FEOS and FEIW rather than sole FEIW. Unfortunately, to date, all these concerns remain untapped.

Herein, we achieve the explicit partition of FEOS and FEIW based on the accurate regional-modification of OS and IW. The FEIW are served for ionic gating, and the chosen FEOS (hydrophobic or charged) are served for blocking the interference molecules into the nanochannels, decreasing the false signals for the ionic gating in complex environments (Fig. 1). Furthermore, we also define a composite factor, areas of a radar map, to evaluate the FEOS performance for blocking interference molecules.

Fig. 1
figure 1

Nanochannels with dual functions. Traditional nanochannels with single function, while ours with dual functions. a FEIW at IW are considered as the main contributions to ion gating, while FEOS at OS are nearly ignored in traditional nanochannels. b The explicit partition of FEOS and FEIW are achieved based on the accurate regional-modification of OS and IW. More importantly, FEIW are served for ionic gating, and the chosen FEOS (hydrophobic or charged), to our knowledge, are served for blocking the interference molecules into the nanochannels, decreasing the false signals for the ionic gating, even in complex environments

Full size image

Results

Explicit spatial partition of OS and IW

To create distinct OS and IW, we deposited ITO and Au alternately at the one side of the AAO membrane’s OS (Supplementary Fig. 1)48. The deposited speed is as low as 0.01 nm s−1. The linear scanning of energy dispersive spectrometer along the z-axil direction of nanochannels demonstrates that the elements as Au and Indium exist at the tips of a nanochannel (Supplementary Fig. 2). We designed four stacking orders of Au and ITO, on purpose of creating four modes of Au distributions as shown in Fig. 2a–m (none means that there is no FEs at the specific region, and all names of used samples are listed in Supplementary Table 1 and all abbreviations are listed in Supplementary Table 2).

Fig. 2
figure 2

Explicit spatial partition of OS and IW. We design four samples with varied stacking orders of Au and ITO as a [none@OS(Au) + none@IW(ITO)], e [none@OS(ITO) + none@IW(Au)], i [none@OS(Au) + none@IW(Au)] and m [none@OS(ITO) + none@IW(Au)] by atomic layer deposition (ALD). Corresponding SEM images taken from vertical and sectional view reveal that nanochannels remain open even after the deposition processes. The element distributions of the four samples are characterized by TOF-SIMS in b, f, j, and n, respectively. The depth with the 5% of the peak intensity is defined as the start or the end of Au or ITO deposition (the gray dashed line in b, f, j, and n, respectively). Thus, the corresponding 3D reconstructions (c, g, k, and o) of the four samples are plotted from the intensity distribution of TOF-SIMS data. The surface area of Au and ITO on the OS or IW are calculated (d, h, l, and p) according to the element depth measured from TOF-SIMS and the pore size measured from SEM (Supplementary Fig. 5). Scale bars, 500 nm

Full size image

Accurate characterizations of elements along nanochannels have been done by a time of flight secondary ion mass spectrometry (TOF-SIMS). The analyzing area is around 60 μm × 60 μm on the X–Y plane beginning from the outmost membranes. The deposition depths (sectional-view of SEM images in Supplementary Fig. 3) of Au and ITO along nanochannels are characterized by the intensity distribution vs. depth from TOF-SIMS. We define the boundary at 5% of the peak intensity (Fig. 2b–n) as the start or the end of Au or ITO deposition. Corresponding 3D reconstructions are shown in Fig. 2c–o. Importantly, the depth of Au or ITO on IW can reach several micrometers, in contrast with the several nanometers at OS. For example, for the 500-s Au deposition, the depth of Au can reach 1.37 μm on the IW, while the thickness of Au is just 5 nm on the OS (Supplementary Fig. 4). Furthermore, after the depositions, the tips of channels remain open as shown from not only the top view but also the sectional view (Right in Fig. 2a–m). Statistics of pore sizes have been done by counting around 9000 pores for each kind of deposited membrane. The average pore size decreases slightly from the original AAO with 25 nm diameter to 22.3 nm (Fig. 2a), 21.1 nm (Fig. 2e), 19.7 nm (Fig. 2i), 17.8 nm (Fig. 2m), respectively (Supplementary Fig. 5). And accurate area of OS and IW are listed in four samples (Fig. 2d–p). From the above, explicit spatial partition of OS and IW is realized with the control accuracy of dozens of nanometers.

Explicit regional partition of FEIW and FEOS

Since explicit spatial partition of OS and IW has been accomplished, accurate partition of FEIW and FEOS with different functional groups (–SH, –OH, et al.) can be realized next. DNA oligomers are chosen as the FEs for regional modification and further ionic gating due to the facility for labeling and design of programmable structure (Fig. 3a–n). In detail, Au is modified with the thiol-DNA through Au-thiol binding, while ITO and AAO with hydroxyl groups are modified with the amino-DNAs through silane coupling reactions (Supplementary Fig. 6). We designed a linear 27 base DNA sequence with fluoresceine isthiocyanate (FITC) and Cyaine-5 (Cy5) respectively, used in our previous works (all sequences listed in Supplementary Table 2)49. Laser scanning confocal microscopes (LSCM) were applied to characterize the distribution of FEs at the OS and IW respectively. As shown in the LSCM images (Fig. 3d–p), the fluorescent distributions along nanochannels are divided into two or three regions, which are in line with the distribution of Au, ITO, and AAO. However, compared with TOF-SIMS, the fluorescent distribution of DNA-Cy5 (Fig. 3c–o) is far greater than the depth of the deposited Au (Fig. 2a–m). The large disparity is attributed to the following reasons: (1) There are great disparities for the measuring accuracy between TOF-SIMS (about 10 nm) and LSCM (about 500 nm); (2) The measurement depth of Au from fluorescence is larger than the actual value; (3) The measurement depth of Au from TOF-SIMS is less than the actual value (see the detail discussion in the 4th part in Supplementary Information). Nevertheless, both results from TOF-SIMS and LSCM show the same trend for the depth of Au (Figs. 2d–p, 3e–q), which means that the data from the two instruments support each other. FEOS and FEIW, therefore, could be rationally coated on the desired OS and IW, respectively.

Fig. 3
figure 3

Explicit partition of FEIW and FEOS. a DNA labeled with FITC or Cy5 are bonded at ITO, Au and AAO surface using silane coupling, Au-S bonding and silane coupling, respectively. Schematics (b, f, j, n) described the varied modification labeled DNA at the OS and IW of four samples which are introduced in Fig. 2a–m, respectively. The intensity distributions of fluorescence along the depth of AAO membrane are listed in c, g, k, o, respectively. The red lines and green lines represent the DNA with Cy5 and the DNA with FITC, respectively, based on the Z-axil of LSCMs for the four samples (d, h, l, p). e, i, m, q Comparison between the Au depth (gold) measured by TOF-SIMS and the DNA-Cy5 (red) depth measured by LSCM

Full size image

Contribution from FEOS and FEIW to ionic gating

The contributions from FEOS and FEIW, respectively, to ionic gating were investigated, referring to the increment of gating ratio (∆gr) after modifying IW and OS with DNAs, respectively (Fig. 4b, c). The four patterns of the DNA-Cy5 distributions were designed for electrochemical characterizations in Fig. 4d–s. The fluorescent intensity referring to the amount of DNA-Cy5 was characterized by LSCM (Supplementary Fig. 7). A two-electrode system, previously reported49,50, contained two symmetric Ag/AgCl electrodes which were placed in the two symmetric chambers filling with the tris buffer (pH = 7.4, 500 mM NaCl, 10 mM MgCl2) as shown in Fig. 4a. The resistances under the three states were measured, as the following: 1) without FE (Rnone-FE), 2) with FE (RssDNA-FE) and 3) with FE and its complementary DNA strand (RdsDNA-FE). The following are the formulas:

$$\Delta {\mathrm{gr}}_{{\mathrm{ssDNA - FE}}}{\mathrm{ = }}\left( {R_{{\mathrm{ssDNA - FE}}}-R_{{\mathrm{none - FE}}}} \right)/R_{{\mathrm{none - FE}}},$$
(1)
$$\Delta {\mathrm{gr}}_{{\mathrm{dsDNA}} - {\mathrm{FE}}}{\mathrm{ = }}\left( {R_{{\mathrm{dsDNA - FE}}}-R_{{\mathrm{none - FE}}}} \right)/R_{{\mathrm{none - FE}}}.$$
(2)
Fig. 4
figure 4

Measurements of ion gating. ΔgrDNA-FE, from four kinds of samples with varied FEOS and FEIW. The I–V are applied to characterize the resistance of samples, reflecting the gating contribution of the IW and OS in a solid-state membrane after being modified with FEIW and FEOS on the IW and OS, respectively. a The cartoon of two-electrode system is used in electrochemical characterizations. Ions go through the nanochannels under the direct current (DC). There are several states of FEOS and FEIW at Au surface: b none-FE, ssDNA-FE grafting and dsDNA-FE grafting. c none-FE, cpDNA-FE grafting and sswDNA-FE grafting. Schemes illustrate the distribution of FEOS and FEIW at OS and IW, in four samples which are described in (d, i, n, s, respectively. I–V curves characterize the current of the state as: none-FE grafting, ssDNA-FE grafting and dsDNA-FE grafting in four samples which are described in e, j, o, t; none-FE grafting, cpDNA-FE grafting and sswDNA-FE grafting in four samples which are described in g, l, q, v. Varied Δgr are defined in the bottom part. Calculated Δgr are shown in f, k, p, u for DNA hybridization and h, m, r, w for sswDNA formation. Error bars represent standard deviations of the measured samples. Five experimental replicates for each data

Full size image

I–V responses were performed by using the sample with the pore size of 25 ± 5 nm and the thickness of 40 μm for dsDNA hybridization (Fig. 4e–t) and for sswDNA formation (Fig. 4g–v). The ∆grssDNA-FE and ∆grdsDNA-FE in Fig. 4k–m are of 9.1 ± 2.2% and 31.6 ± 2.8%, significantly greater than that in Fig. 4f–h (0.24 ± 0.13% and 1.2 ± 0.26%, respectively). Interestingly, the ∆grssDNA-FE and ∆grdsDNA-FE in Fig. 4p–r are approximately equal to that in Fig. 4k–m, revealing the negligible contribution to the ionic gating from FEOS in Fig. 4p–r. Furthermore, the negligible contribution to the ionic gates from FEOS is further proved by using the deposition membrane with the larger pore size (Supplementary Fig. 8 and Supplementary Fig. 9).

Based on the above foundation, we endeavor to construct a high-efficiency ion gating system through generating DNA superstructure, wherein concatenated DNA strands repeatedly hybridized (Supplementary Fig. 10). These DNA superstructures are designed as FEs to amplify ∆gr by enhancing steric hindrance in the nanochannels (with the pore size of 25 ± 5 nm and the thickness of 40 μm)50. The ∆grcpDNA-FE and ∆grsswDNA-FE in Fig. 4e, g are the smallest among the four patterns of the membranes (Supplementary Fig. 11). The ∆grcpDNA-FE and ∆grsswDNA-FE in Fig. 4p–r are approximately equal to the ones in Fig. 4k–m. The approximately equal ∆grcpDNA-FE or ∆grsswDNA-FE also suggests the negligible contribution of FEOS to ionic gating, similar to the conclusions in the last paragraph. When prolong the deposition duration of Au, the ∆grcpDNA-FE or ∆grsswDNA-FE is further enhanced (Fig. 4u–w). Furthermore, the negligible contribution of ionic gates from FEOS is further proved by using the deposited membranes with a high gating ratio up to 4635% and further applied in ATP detections (Supplementary Fig. 12 and Supplementary Fig. 13). Electrochemical impedance spectroscopy (in Supplementary Fig. 14 and Supplementary Fig. 15) also demonstrate that when linear DNAs act as the FEs for ionic gating49, the impact from the FEOS are proved to be negligible, while the major contribution to ionic gating is resulted from the FEIW in our system.

The above conclusion seems to be inconsistent with our previous results, but it is not. We employed further calculations for the density of FEOS and FEIW at OS and IW (Supplementary Fig. 16), respectively, from our previous work. The amount of FEIW is about 7.2 times of the one of FEOS, while the amount of FEIW is only 2.3 times of the one of FEOS in our previous work45, which suggest that the ratio between the FEOS and FEIW is vital for the contribution to ion gating. The contribution from FEOS to ion gating in this work is negligible due to the much less ration between FEOS and FEIW than that in the previous work.

The chosen FEOS endow OS with the function as anti-interference

Since the FEOS in Fig. 4 are demonstrated to be negligible for the ionic gating, we propose to introduce FEOS to OS, which is expected to endow OS with a function (Fig. 5). In order to achieve the function of OS, the anti-interference is taken as an example, and then several different FEOS are chosen (Figs. 5, 6) to endow OS with the anti-interference function. Thus, 3 different processes were conducted to evaluate whether the FEOS, we chose, bring OS the anti-interference ability (the cpDNA-Cy5 is chosen as the interference molecules in Fig. 5). There are three different processes: Process 1 (Fig. 5a, step 0 to step 1); Process 2 (Fig. 5f, step 0 to step i); Process 3 (Fig. 5k or p, step 0 to step i). The gating ratio (∆gr1 = Rf/Rb −1) in Process 1 (Fig. 5c–e) is the reference value, and the gating ratio (∆gr2 = Ri/Rb −1) in Process 2 (Fig. 5i, j) and (∆gr3 = Ri/Rb −1) in Process 3 (Fig. 5n–t) are compared with ∆gr1 to test the sample’s anti-interference ability. The deviations from the test system can be ignored (Supplementary Fig. 17 and Supplementary Fig. 18). As expected (Fig. 5u), ∆gr2 is about 242% of the reference value (∆gr1), which suggests that the sample in Fig. 5f without anti-interference ability, thus, induces the false positive signal. In contrast (Fig. 5u), ∆gr3 for the samples with the FEOS (both perfluorooctyltriethoxy silane (POTS) in Fig. 5k and polyacrylic acid (PAA) in Fig. 5p) in Process 3 are around 106 and 111%, respectively, of the reference value (∆gr1), which suggests that the samples with the FEOS endow OS with the function, anti-interference.

Fig. 5
figure 5

The chosen FEOS endowing OS with the function, anti-interference. The FEOS is hydrophobic or negative molecular. Three different processes (Process 1 (a), 2 (f) and 3 (k/p)) were conducted to evaluate whether the FEOS, we chosen, bring OS the anti-interference ability. cpDNA-Cy5 is chosen as the interference molecule. [none@OS(ITO) + none@IW(Au)] in Fig. 5a are in the 0 state in Process 1. [none@OS(ITO) + DNA@IW(Au)] in Fig. 5f are in the 1 state in Process 2. [PTOS@OS(ITO) + DNA@IW(Au)] in Fig. 5k are in the 2 state in Process 3. [PAA@OS(ITO) + DNA@IW(Au)] in Fig. 5u are in the 2 state in Process 3. Fluorescent images (b, d, g, l, q), fluorescent profile (h, m, r), resistances increment (e, i, n, s), I–V curves (c, j, o, t) are corresponding to the steps in Process 1, 2 and 3, respectively. u Comparison of the Δgr from the three processes mentioned above, and the following is the definition for them: Δgr1 = Rf/Rb-1 (for Process 1); Δgr2 = Ri /Rb-1(for Process 2); ∆gr3 = Ri /Rb-1 (for Process 3). Scale bars, 10 μm. Error bars represent standard deviations of the measured samples. Five experimental replicates for each data

Full size image
Fig. 6
figure 6

The chosen FEOS endowing OS with the function, anti-interference. The FEOS is positive molecular. Three different processes (Process 1 (a), 2 (f) and 3 (k)) were conducted to evaluate whether the FEOS, we chosen, bring OS the anti-interference ability. AIEgen is chosen as the interference molecule. [none@OS(ITO) + none@IW (Au)] in a are in the 0 state in Process 1. [none@OS(ITO) + DNA@IW (Au)] in f are in the 1 state in Process 2. [PEI@OS(ITO) + DNA@IW (Au)] in Fig. 6k are in the 2 state in Process 3. Fluorescent images (b, d, g, l), fluorescent profile (h, m), resistances increment (e, i, n), I–V curves (c, j, o) are corresponding to the steps in Process 1, 2 and 3, respectively. p Comparison of the Δgr from the three processes mentioned above, and the definition for them: Δgr1 = Rf/Rb − 1 (for Process 1); Δgr2 = Ri/Rb − 1(for Process 2); ∆gr3 = Ri/Rb − 1(for Process 3). Scale bars, 10 μm. Error bars represent standard deviations of the measured samples. Five experimental replicates for each data

Full size image

To explore the mechanism of the anti-interference for the OS with the FEOS, the interference molecules are labeled with fluorescence molecules, Cy5. Almost no fluorescence in Process 1 (Fig. 5b–d). Both the intensity and the depth at step i in Process 3 (Fig. 5l–r) are weaker and thinner than the ones at step i in Process 2 (Fig. 5g, h), which means that the sample FEOS (PTOS and PAA) can block the interference molecules into the nanochannels, but the sample without FEOS cannot. It is possibly that the FEOS, hydrophobic PTOS (Supplementary Fig. 19) and negative PAA (Supplementary Fig. 20), impede the hydrophilic and negative interference molecules (cpDNA-Cy5 chosen in Fig. 5) entering into the nanochannels. To further demonstrate the blocking mechanism, we performed the further destruction experiments, step d (in Process 2 and Process 3) after step i. The interference molecules were again added, and, this time, they were driven by −2 V constant voltage, entering into nanochannels. ∆R = Rd − Ri of the sample without FEOS (Supplementary Fig. 21e) at step d in Process 2 is set as the reference value, due to the sample without anti-interference ability. ∆R = Rd − Ri of the sample with FEOS (Supplementary Fig. 21e and Supplementary Fig. 21g) at step d in Process 3 are 98 and 85% of the above reference value, respectively, which further prove hydrophobic and negative FEOS blocking mechanism.

To further evaluate the conclusions in Fig. 5, we changed the interference molecules to AIEgens (Fig. 6), positive-charged small molecules, which are distinct from ccCy5-DNA in Fig. 5. Similar to the experiments in Fig. 5, here are three different processes: Process 1 (Fig. 6a, step 0 to step 1); Process 2 (Fig. 6f, step 0 to step i); Process 3 (Fig. 6k, step 0 to step i). The gating ratio (∆gr1 = Rf/Rb − 1) in Process 1 (Fig. 6c–e) is the reference value, and the gating ratio (∆gr2 = Ri/Rb −1) in Process 2 (Fig. 6i, j) and (∆gr3 = Ri/Rb − 1) in Process 3 (Fig. 6n, o) are compared with ∆gr1 to evaluate the samples’ anti-interference ability. As expected (Fig. 6p), ∆gr2 is about 202% of the reference value (∆gr1), which suggests that the sample in Fig. 6f without anti-interference ability. In contrast (Fig. 6p), ∆gr3 for the samples with the FEOS (Polyetherimide (PEI) in Fig. 6k) in Process 3 is around 143%, of the reference value (∆gr1), which suggests that the sample endow OS with the function, anti-interference. The fluorescence results demonstrate that the sample FEOS (PEI) can block the interference molecules into the nanochannels, but the sample without FEOS cannot (Fig. 6b–m). It seems that the anti-interference performance for PTOS and PAA are better than that of PEI, which is probably due to the much larger size of interference molecules in Fig. 5 (DNA) than that in Fig. 6 (AIEgens). We will further do the comparisons in the Discussion part.

To explore the mechanism of the anti-interference for the OS with positive charged PEI as the FEOS, we performed the similar experiments with the ones in Fig. 5. It is possibly, that the FEOS, positive PAA, impede the positive interference molecules (AIEgens chosen in Fig. 6) entering into the nanochannels. To further demonstrate the blocking mechanism, we performed the further destruction experiments, step d (Process 2 and Process 3 in Supplementary Fig. 22) after step i. The results of destruction experiments further prove positive FEOS blocking mechanism. Therefore, all of the chosen FEOS, indeed, endow OS with the function, anti-interference to interference molecules (not only macro-molecules, but also small molecules; not only negative, but also positive molecules).

Discussion

Since there are three different molecules, PTOS, PAA, and PEI, worked as FEOS, respectively, how to select FEOS and how to evaluate their performance for anti-interference are two key questions. For the first one, the rules for selecting FEOS are: (1) FEOS would not react with the interference agents; (2) FEOS have the opposite chemical or physical properties with the interference agents, for example, (hydrophobic to hydrophilic; negative to negative and positive to positive). For the second one, we need to check the data in Figs. 5, 6 and Supplementary Fig. 21, 22 again, and the value of RDestruction for sample with PTOS as FEOS is greater than the one for sample with PAA, which means blocking effect from PAA is superior to the one from PTOS. Both the intensity and the depth of sample at step 4 in Process 3 (Fig. 5q, r) are weaker and thinner than the ones of sample at step 4 in Process 3 (Fig. 5l, m), which also means that the blocking effect from PAA is superior to the one from PTOS. However, it is difficult to evaluate the effect among PAA, PTOS, and PEI due to different interference agents. We, therefore, designed a Radar map containing five parameters: the resistance variation from step 2 to step i, gating rate, fluorescent intensity, fluorescent FWHM and anti-interference ability after 2 V driving (calculated by area of triangle in Supplementary Fig. 23). The bigger area, the polygon is, and the greater anti-interference the OS perform (Fig. 7). Apparently, blocking effect from PAA is the best, and PTOS is superior to the one from PEI. It is possibly due to the fact that it is hard to block small molecules (AIEgens) compared with the macromolecules (cpDNA-Cy5).

Fig. 7
figure 7

The radar map reflecting the anti-interference performance of FEOS. The radar map integrates the five parameters reflecting the variation by the interference of cpDNA-Cy5 or AIEgens after infiltration and after 2 V driving. The five parameters are considered as the resistance increment from step 1 (or step 2) to step i (AAΔR), gating ratio (AAgr), fluorescent intensity (AAFI), fluorescent FWHM (AAFWHM). The calculated formulas are listed as AA∆R = (∆R(process 2) − ∆R(process 3))/∆R(process 2), AAFI = (FI(process 2) − FI(process 3))/FI(process 2), AAFWHM = (FWHM(process 2) − FWHM(process 3))/FWHM(process 2), AA∆gr = (∆gr(process 2) − ∆gr(process 3))/∆gr(process 2) (where the Δgr were calculated by taking RB as 100%). The fifth parameter is the area of triangle in Supplementary Fig. 23 to value the integrated anti-interference ability after 2 V driving. The bigger area of the polygon, the greater anti-interference of OS

Full size image

Coincidentally, in a nuclear pore complex, the similar divisions of function exist naturally. Analogous to the OS, the densely packed Phe-Gly Nup meshwork at the entrance of nuclear pore complex physically exclude spurious macromolecules into nuclear pore complex (analogous to FEOS), avoiding the interference on the gating properties of nuclear pore complex (Supplementary Fig. 24)51. Overall, this work could be to blaze a trial not only for the role of FEOS at OS in nanochannels, but even for the biomimetic system in the porous membrane.

Methods

Preparation of metallic deposited AAO membrane

Firstly, the AAO membranes were immersed in the 1 M HCl solution under ultrasonic treatment for 2 min. Then, as-prepared AAO membranes were washed with distilled water and drying with nitrogen gas. Deposition has been performed by using AE Nexdap PVD platform (Angstrom Engineering Inc.)45. Two kinds of deposited target as Au and ITO were used respectively. The circular targets were approximately parallel to the AAO membranes, ensuring the deposition direction perpendicular to the membranes. The layer by layer depositions were taken by the successive deposition without replacing the target or releasing vacuum. The successive deposition ensured no secondary pollution at the first deposited layer. The extreme low deposited speed was applied as 0.01 nm s−1. The deposition duration was 100 s, 500 s, and 1000 s, respectively. The deposited speed was calibrated by the deposited thickness on the surface of the flat silicon wafer at nanometer level. The platform will calibrate the parameters automatically.

DNA modification and hybridization

Modification of thiol-modified DNA. The graft of thiol-modified DNA strings was achieved by immersing the Au deposited AAO membranes into a 5′-thiol modified DNA string (1 μM) of 10 nM tris solution (pH = 7.4, 500 mM NaCl, 1 mM MgCl2) for 10 h50. The membranes were rinsed with distilled water and dried with nitrogen gas. The as-prepared membranes were then applied for further EC or LSCM characterizations. After the tests, the membranes were rinsed with distilled water and dried with nitrogen gas. For hybridization, the membranes grafting with single sequence string were then immersing into the DNA target (1 μM) for 10 h. The rinse and dry were done.

Modification of amino-modified DNA. The membranes were washed with distilled water and dried in argon gas. After that, the membranes were immersed into a 5% acetone solution of APTES for 10 h. The membranes were thoroughly washed in acetone and baked at 120 °C for 2 h, and left overnight in 25% aqueous solution of glutaraldehyde (25 wt% aqueous solution). After thoroughly washing with distilled water and drying with nitrogen gas, the membranes were modified with 5′-aminated DNA (the capture probe, 1 μM) in 1 ml of 10 mM Tris solution (pH = 7.4, 500 mM NaCl, 1 mM MgCl2) for 10 h50. Furthermore, the modification of two kinds DNA probes was achieved by immersing the salinized membranes into the equimolar amino-modified and thiol-modified DNA (1 μM) for 10 h. The membranes were rinsed with distilled water and dried with nitrogen gas.

DNA hybridization and the formation of DNA supersandwich structure. For hybridization, the membrane grafting with single sequence string was immersed into the DNA target (1 μM) for 10 h. The DNA supersandwich structure was formed by immersing the cp-DNA modified DNA into the mixture of equimolar P1 and P2 (1 μM) for 10 h. All the membranes after DNA hybridization were rinsed with distilled water and dried with nitrogen gas after the DNA hybridization.

Superficial functionalization (FEOS@OS)

The perfluorooctyltriethoxysilane (POTS) coating was achieved by adding the PTOS solution on the side by metallic deposition for 30 min and rinsing throughout by alcohol. The PAA and PEI coating were achieved by the spin-coating of PAA solution (1 mM) and PEI solution (1 mM) at 3000 rpm. The spin-coating carried out for twice, each time adding 200 μl and lasting for 15 s. After the spin-coating, the membranes were washed throughout by water.

Time of flight secondary ion mass spectrometry (TOF-SIMS)

Secondary ion mass spectra of deposited AAO surfaces were characterized by TOF-SIMS V (IONTOF, GmbH). A Bi liquid metal primary ion source was applied with an angle of 45° relative to the sample surface with a pulsed Bi3++ primary ion beam of 30 keV and shave off fresh 60 µm × 60 µm areas for each analysis. The TOF analyzer was installed at an angle of 90° to the sample surface. Negative secondary ion spectra were collected. Mass calibration was carried out using standard procedures (mass resolving power > 5000).

Laser scanning confocal microscope (LSCM)

The fluorescent-dyed AAO membrane were clamped by cover glass and microscope slide filling with water (around 1 cm2 AAO with 100 μl water). The LSCM characterizations were performed by using LSCM system (Olympus FV1200) with a 40× objective. For FITC-modified DNA, excitation: 495 nm and emission collected: 505–535 nm. For Cy5-modified DNA, excitation: 565 nm and emission collected: 655–685 nm. For Silole-SNBr (AIE), excitation: 466 nm and emission collected: 565–595 nm. To characterize the fluorescent distribution along Z-axil, Z-axil scanning was applied by 500 nm for each step. On purpose of comparison of fluorescent intensity, the settings of channel parameter were constant, including HV = 600, Gain = 1.75 and Offset = 1. All membranes were avoided light before LSCM tests.

Electrochemical measurements

The electrochemical characterizations of the AAO and its derivative products were performed by using two-electrode system in a 10 mM tris solution (pH = 7.4, 500 mM NaCl, 1 mM MgCl2) as electrolytes. The symmetric Ag/AgCl electrodes were used as a working and a counter electrode. Bio-logic VMP3 station (Bio-logic Science Instrument Pvt Ltd) was used to measure the electrochemical responses. The I–V tests were performed in the potential from −0.2 to 0.2 V under a direct-current mode. The EIS measurements were performed under an alternating-current mode. The tests were performed in a frequency range from 1 MHz to 10 mHz at the open circuit potential with an AC perturbation of 5 mV. The electric driving was performed under constant voltage as −2 or +2 V for 10 min. Before EC measurement, each dry membrane was immersed in 10 nM tris buffer overnight to reduce the test error coming from inadequate impregnation.