www.sciencemag.org/cgi/content/full/338/6109/932/DC1

Supplementary Materials for

Synthetic Lipid Membrane Channels Formed by Designed DNA

Nanostructures

Martin Langecker, Vera Arnaut, Thomas G. Martin, Jonathan List, Stephan Renner, Michael Mayer, Hendrik Dietz,* Friedrich C. Simmel*

*To whom correspondence should be addressed. E-mail: dietz@tum.de (H.D.); simmel@tum.de (F.C.S.)

Published 16 November 2012, Science 338, 932 (2012) DOI: 10.1126/science.1225624

This PDF file includes:

Materials and Methods

Texts S1 to S9

Figs. S1 to S24

Tables S1 to S3

References

2

Materials and Methods Assembly and purification of membrane channels

Molecular self-assembly with scaffolded DNA origami. Structures were designed

using caDNAno v.02 (29). DNA scaffold strands of 7249 bases length derived from the genome of bacteriophage M13 were prepared recombinantly as previously described (4). Staple oligonucleotide strands were prepared by solid-phase chemical synthesis (Eurofins MWG, Ebersberg, Germany, HPSF purification). Cholesterol modified oligonucleotides were synthesized using standard phosphoramidite chemistry, HPLC purified, and quality checked by mass spectrometry by biomers (Ulm, Germany).

Production of DNA channel structures was accomplished in one-pot reactions by mixing 20 nM scaffold strands with 100 nM of oligonucleotide staple strands in a buffer including 5 mM TRIS, 1 mM EDTA, 20 mM MgCl2, 5 mM NaCl (pH 8) (4, 9). The mixture was subjected to thermal annealing ramps that cooled from 65°C to room temperature. Shortly before purification the origami structures were incubated with 10 µM of cholesterol-modified oligonucleotides at room temperature for one hour.

Purification. 50µl of the reaction products were mixed with 450 µl of buffer containing 5 mM Tris, 1 mM EDTA, 5 mM MgCl2, 5 mM NaCl and spin filtered with a 100 kDa Amicon filter (Millipore) at 2000 rcf for 30 minutes. The filtration was repeated 5 additional times. After purification, the concentration of oligonucleotides not bound to the DNA channel is negligible as judged from gel electrophoresis.

Preparation of lipid membrane vesicles

Formation of small unilamellar vesicles (SUVs). POPC (1-palmitoyl-2-oleoyl-sn-

glycero-3-phosphocholine) lipids were dissolved in chloroform to a concentration of 5 mg/ml. A lipid film was formed by evaporating 1 ml of POPC solution in a 5 ml round bottom flask, using a rotational evaporator. The flask was kept under vacuum for at least 30 minutes to evaporate remaining chloroform. The lipid film was resuspended in 2 ml of a 150 mM aqueous KCl solution. Subsequently, SUVs were formed by sonicating the solution with a tip sonicator (Bandelin Sonoplus mini20) for ca. 5 minutes at 8 W, until the solution started to become transparent. Afterwards the SUVs were purified by centrifugation at 1500 rcf for 5 minutes. The solution was shock frozen with liquid nitrogen and kept at -80°C until usage.

Preparation of giant unilamellar vesicles (GUVs). GUVs were created by electroswelling using the following procedure: 6µl (10mg/ml) DPhPC (1,2-diphytanoyl-sn-glycero-3-phosphocholine, Avanti Polar Lipids, Alabaster, AL, USA) with 10% Cholesterol were dissolved in Chloroform. The solution was dried on two parallel platinum electrodes, d=0.5 mm, with a spacing of 2.5 mm (center to center) for 1h under vacuum. The electrodes were then inserted into a 500 µl eppendorf tube filled with 1.2 M sorbitol. The osmolarity was determined with an Osmomat 30 (Gonotec, Berlin). The osmolarity of the GUV buffer was adjusted by slightly changing the sorbitol concentration. Then a voltage of 3 Vp-p at 5 Hz was applied for 3h.

3

Imaging

Transmission electron microscopy. Particles were adsorbed on glow-discharged

formvar-supported carbon-coated Cu400 TEM grids (Science Services, Munich, Germany) and stained using a 2% aqueous uranyl formate solution containing 25 mM NaOH. Imaging was performed using a Philips CM100 transmission electron microscope operated at 100 kV. Images were acquired using an AMT 4x4 Megapixel CCD camera. Micrograph scale bars were calibrated using 2D catalase crystal lattice constants as length reference. Imaging was performed at 28500x magnification for the majority of images and 11500x magnification for overview images.

TEM image processing. Micrographs of individual particles were picked using the EMAN2 (30) boxer routine. Micrograph alignment and superposition was performed using the XMIPP mlf_2Dalign routine (31).

Fluorescence/DIC imaging of DNA channels on GUVs. Mg2+ ions strongly influence the permeability of GUV membranes. In order to study interactions of DNA channels with GUVs, fluorescence imaging experiments were performed in a Mg2+ free buffer containing 300mM KCl. Control TEM experiments confirmed the stability of DNA channels under these conditions. For staining, 4 µl of DNA channel solution were incubated with 1 µM (1µl) YOYO-1 (Invitrogen, Life Technologies GmbH, Darmstadt, Germany) for 2 hours in 600 mM KCl, 0.5xTBE. Then, 5 µl of the GUV solution were added and incubated for 1 h. Next the GUV-Origami solution was diluted in 5 steps to a final concentration of 300mM KCl and 600 mM sorbitol in order to reduce background fluorescence. (All other consumables were purchased form Sigma Aldrich, Munich, Germany). A microscopic observation chamber was assembled using vacuum grease and a cover glass, coated overnight with 5% BSA solution. For fluorescence and DIC imaging an Olympus IX81 microscope with a 60x 1.4 oil immersion objective and a Hamamatsu Orca r2 camera was used.

Electrical measurements

All electrical experiments were performed using 16 channel microelectrode cavity

arrays (MECA) (12) provided by the group of Jan Behrends, University of Freiburg. The MECA chips were interfaced using an Orbit 16 device, provided by Nanion Technologies GmbH, Munich, Germany. Lipid solutions were prepared from 1,2-diphytanoyl-sn-glycero-3-phosphocholine (DPhPC, Avanti Polar Lipids, Alabaster, AL, USA) and dissolved in octane (Sigma Aldrich, Munich, Germany) at a concentration of 10 mg/ml. As electrolyte 1M KCl solution (Merck KGaA, Darmstadt, Germany) was used, containing 10 mM Tris and 1 mM EDTA (Sigma Aldrich, Munich, Germany) at pH 8. To form lipid bilayers on the 16-well MECA chips, 100 µl of electrolyte solution were added to the measurement chamber. In the MECA setup, the ground Ag/AgCl electrode resides in the cis (upper) compartment. 1-2 µl of lipid solution were pipetted in the vicinity of the cavities followed by painting the lipid bilayer on top of the cavities using a

4

magnetic stir bar. Parallel current recordings were performed using a 16 Channel Triton+ amplifier (Tecella Inc., Costa Mesa, CA, USA) that was controlled via a custom made Labview (National Instruments Germany GmbH, Munich, Germany) measurement software. For low noise experiments on the gating behaviour of the DNA structures, individual channels were switched to a single channel patch clamp amplifier (EPC9, HEKA Elektronik GmbH, Lambrecht, Germany).

HPLC purified oligonucleotides for translocation studies (biomers, Ulm, Germany)

were dissolved in 10 mM Tris, 1mM EDTA solution at pH 8. DNA sequences: Hairpin: 5’- (dT)6 - ATCTACATATTTTTTTTTTTATGTAGAT - (dT)50 -3’ Q-T60: 5’- GGTTGGTGTGGTTGG - (dT)60 -3’ Q-T125: 5’- GGTTGGTGTGGTTGG - (dT)125 -3’

Data analysis Patch clamp data were analyzed using custom routines implemented in Matlab

(MathWorks, Ismaning, Germany). A cut-off frequency of 3 kHz was used for the built-in Bessel filter of the Tecella amplifier, but in fact the slow frequency response of the 16 channel amplifier board limited the bandwidth considerably (0.5 – 1.7 kHz) in our experiments. To determine the amplitude of measured current blockades reliably, rising edge and falling edge of all peaks were fitted with single exponentials as described in Ref. (12). All dwell time histograms feature logarithmically spaced bin widths and were fitted with a single exponential (p(t)=exp(-t/τ)) with the time constant τ being the only fit parameter. For measurements on hairpin translocation, it was necessary to introduce an amplitude scaling factor to obtain satisfying fits.

5

Supplementary Text Note S1: Quality control of DNA channel formation

The quality of folded and purified DNA channels was assessed by direct imaging using negative-staining transmission electron microscopy (TEM), before the samples were used for electrical characterization. Typical single-particle TEM micrographs are shown in Figs. S1-S2. The quality of DNA channel assembly, binding of cholesterol, and the efficacy of purification from excess DNA strands was assessed using agarose gel electrophoresis (Fig. S3).

Fig. S1 Negative-stain TEM micrographs of DNA channels. Left panel: side view transmission projections, right panel: side view projections rotated by 60° around the long axis relative to side view 1. Each micrograph is sized 110 nm x 110 nm. The first micrograph in the top row of each of the panels is an average micrograph. See also Fig. 1D.

6

Fig. S2 Negative-stain TEM micrographs of DNA channels - top view. Each micrograph is sized 60 nm x 60 nm. The first micrograph in the top row is an average micrograph. See also Fig. 1D.

Fig. S3 Electrophoretic mobility analysis. A: Laser-scanned photograph using a Typhoon FLA9500 (GE Healthcare) of a EtBr-stained 2% agarose gel slab on which the following samples where electrophoresed: M=1kb ladder, 1= scaffold DNA strand, 2: set of staple DNA stands used for the folding, 3: cholesterol-modified DNA strands, 4: products from one-pot thermal annealing of scaffold DNA and staple DNA strands: the folded DNA channel species migrates faster through the gel as the unstructured scaffold due to structural compaction, 5 = folded DNA channel after purification from short oligonucleotides, 6 = folded unpurified DNA channels after addition of cholesterol-modified DNA strands, 7: folded DNA channels that were purified from all short DNA strands after addition of cholesterol-modified DNA strands. B: The same gel as in A, but laser scanned on a different emission wavelength. One of the staple DNA strands required for folding the DNA channel was labeled with the organic dye Atto655. Note the absence of any fast-migrating staple strand species in lane 7, after purification, highlighting the efficacy of the purification method for removing short strands. C: False-colored overlay of images from A and B.

7

Note S2: Binding of the DNA channel to lipid membranes, TEM evidence for stem insertion Binding of the DNA channel to lipid bilayer membranes was promoted by hydrophobic cholesterol moieties that were attached to the DNA channels. These moieties were attached to DNA oligonucleotides which in turn were hybridized to single-stranded adaptor strands protruding at 26 positions from the face of the ‘barrel’ domain of the channel structure that faces the stem domain (cf. schematic depiction in Fig. 1A of the main paper, staple sequences in Supplementary Note S8 and design diagrams in Note S9). The cholesterol-modified DNA oligomers had the sequence 5’ CCTCGCTCTGCTAATCCTGTTA-3’.

Depending on the position of the modification (5’ or 3’) on the chemically modified DNA strand, the cholesterol moiety can either be placed in close vicinity of the channel structure or within a distance of 22 bases from the cap. Both versions were investigated in our studies and showed electrically detectable interactions with the lipid membrane. The proximal modification version was found to lead to more stable insertions into the membrane, whereas the distal version mainly caused transient current fluctuations that were not stable upon voltage reversal. We thus used the proximal modification.

We obtained transmission electron microscopy (Figs. 1E-G, Figs. S4-S9) and complementary fluorescence microscopy evidence (Fig. S10) for binding of the DNA channels to lipid bilayer membranes. DNA channels with cholesterol modification adhered readily to lipid membranes (Figs. S4-S7), while DNA channels without the modification did not show any appreciable interaction with the membranes (Figs. S5-S7). We further investigated the influence of the number of cholesterol modifications as well as their geometric arrangement on their efficacy of promoting binding of the DNA channel to membranes (Fig. S8). A statistical analysis of our samples showed that the small unilamellar vesicles had a typical projected diameter on the order of 50 nm (from the TEM image it cannot be judged whether the vesicles burst and restructured or collapsed on the TEM grid). The fraction of vesicles with at least one DNA channel increased with the number of cholesterols, with good binding starting above ≈ 10 cholesterols per channel (cf. Fig. S8).

The TEM data obtained with the DNA channel version featuring 26 cholesterols further clearly revealed that the cholesterol modifications at the stem-face of the barrel promote binding of the DNA channel to lipid membranes in the desired orientation in which the stem faces the membrane (Figs. S4-S9). There are a number of additional implications of the TEM data that also support the notion of membrane-insertion of stem domain of the DNA channel. a) The most frequently observed images are of the type in Figure 1E (see also Figs. S4-S6), where one or two DNA channels adhere to small vesicles that are comparable in size to the channel itself. In these images, the transmission signal of the vesicles is very high, suggesting that the interior of the vesicle does not contain the heavy ions of the staining

8

agent (uranyl formate in this case) which are responsible for the contrast. If the stem domain penetrates into a stain-excluding vesicle, it will thus be invisible (case 1). If the stem domain were above or below the vesicle (case 2), it should add to the observed transmission contrast which is not observed. The data in Figs. 1F and S9 (see also b) and new experiments (see d) clearly show that when the stem domain is accessible to the staining agent (case 2), it can be detected. Another possible explanation of the data in figure 1E is physical rupture of stem domain from the channel (case 3). The rupture, however, must then be specifically induced by attachment of the channel to a membrane since objects in the same micrographs that are not attached to a vesicle all have intact stem domains. The data in Figs. 1F and S9 (see b) and new experiments (see d) show that membranes do not induce rupture of the stem domain. Thus the data in figure 1E favors case 1, i.e. insertion of the stem into the membrane. b) Some images such as those in Figs. 1F and S9, where multiple DNA channels have adhered to larger medium-size vesicles show the stem as apparently pinching through the lipid membrane. Note that the transmission contrast of the larger vesicles is significantly lower than that of the smaller ones in Fig. 1E. This suggests that the vesicle actually does contain the heavy ions of the stain. The fact that the stem domain is visible for some of the DNA channels argues against the notion that attachment to membranes promotes rupture of the stem domain (case 3 in a). The data in figure 1F can be explained either by an actual insertion into a vesicle that does contain stain (case 1) or by a situation where the stem is above or below the membrane but not actually inserted (case 2). Although it cannot be decided which case applies here, what the data shows is that the stem domain can be detected either in a stain-filled vesicle or when on-top or below of membranes. The data in Figs. 1F and S9 thus clearly supports our interpretation in (a) of figure 1E. c) In TEM images of DNA channels adhered on top of larger membrane patches (such as those in Fig. 1G, shown in detail in Fig. S7), the stem domain is not visible for channels adhered on the membranes, but it still is visible for channels that are not on top of a membrane. Again, one explanation for this finding is that for channels on membranes the stem penetrates the lipid bilayer but is protected from the staining agent by the surrounding membrane patch (case 1), and is thus invisible to the observer. An alternative scenario that could explain the data would require physical rupture of the stem from the channel, but this must then have happened exclusively to channels adhered to a membrane (case 3). The data discussed in (b) and new experiments in (d) show that case 3 can be ruled out. d) We prepared new variants of the DNA channel with cholesterol modifications attached asymmetrically (rather than concentrically) with respect to the stem on the channel and acquired new TEM data (see Fig. S8). We argued that the asymmetric configuration might help creating a “false positive” control in which the channel attaches to the edge of membranes, but the stem domain may not necessarily need to penetrate the membrane in order to satisfy all possible membrane/cholesterol ‘bonds’. The resulting images show binding of these modified channels to the membrane. Importantly, images of smaller vesicles with bright transmission contrast (suggesting that the interior of these vesicles may be free from the staining agent) could be collected. In contrast to (a), however, for

9

many of these images the stem domain could now clearly be discerned. Without any assumption on whether the stem actually penetrates the membrane or not, this finding shows that the stem domain can be discerned and does not rupture in the vicinity of these stain-free, high contrast vesicles. This was a so far missing piece of evidence that now helps to rule out case 2) in the discussion of (a). Taken together, the only scenario that appears consistent with all the TEM data is the one in which the stem domain of the TEM channel can actually insert into the membrane. This note includes the following figures:

Fig. S4: Negative-stain TEM single-particle micrographs of cholesterol-modified DNA channels attached to POPC small unilamellar vesicles (SUV).

Figs. S5-S6: Negative-stain TEM field of view micrographs of DNA channels with and without cholesterol modifications added to POPC SUVs. Fig. S7: Negative-stain TEM field of view micrographs of modified and not modified DNA channels added to POPC SUVs. Fig. S8. Influence of number and arrangement of cholesterol anchors on the channel incorporation into membranes. Fig. S9. Additional TEM micrographs of DNA channels on medium-sized vesicles as in Fig 1F. Fig. S10. Fluorescence microscopy images of DNA channels on DPhPC GUVs.

10

Fig. S4 Negative-stain TEM micrographs of cholesterol-modified DNA channels attached to POPC small unilamellar vesicles (SUV). Each micrograph is sized 60 nm x 60 nm. The first micrograph in the top row contains the respective average micrograph also depicted in Fig. 1D.

11

Fig. S5 TEM micrographs of DNA channels with and without cholesterol modifications added to POPC membranes. A: Cholesterol modified DNA channels mixed with SUVs and imaged by TEM. The lipids form white circular patches. Almost all DNA channels are attached to SUVs (collapsed on the TEM grid) or larger supported lipid membrane patches. B: Same image as A, but with red circles highlighting DNA channels bound to membrane patches. C: Unmodified DNA channels mixed with SUVs show no binding – DNA channels and lipid patches typically are distributed randomly with no apparent relation to each other. Scale bars = 250 nm.

!

!

A

C

B

12

Fig. S6 Negative-stain TEM micrographs of modified and not modified DNA channels added to POPC membranes. A: Cholesterol modified channels with SUVs, B: Unmodified channels with SUVs. As in Fig. S4 at a higher magnification (28500 x). Scale bar is 100 nm.

! !

A B

13

Fig. S7 TEM micrographs of DNA channels on lipid films (top) and isolated DNA channels (bottom). The first image in each series is an average micrograph from the respective set of micrographs. In the top panel, the stem domain is hardly visible, suggesting that it does not contribute to transmission contrast because it is hidden from the staining agent by a surrounding membrane

14

Fig. S8. Influence of cholesterol anchors on the channel incorporation into membranes. A: Statistical analysis of size distribution of membrane patches and influence of the number of cholesterols per DNA channel on binding. B: Combined histograms showing the SUV size distribution (apparent diameter upon adsorption to the TEM grid) and the binding efficiency. Red circles in the schemes indicate cholesterol modifications at the bottom of the DNA channel ‘barrel’ domain, gray circles are modifications protruding from the sides of the six helix stem. C: Additional TEM micrographs of DNA channels will 26 cholesterol modifications on medium-sized vesicles. Scale bars = 25 nm. D: DNA channels with 11 cholesterol modifications oriented asymmetrically around the stem domain adhered to medium-sized lipid vesicles. Note that the stem domain can be discerned (red arrows).

15

Fig. S9. Additional TEM micrographs of DNA channels on medium-sized vesicles as in Fig. 1F. Scale bars = 25 nm.

1

Fig. S10 Fluorescence microscopy DNA channels to DPhPC GUVs. a. Differential interference contrast images of GUVs in 300 nM KCl solution b. Epifluorescence microscope images of YOYO labeled origami pores bound to the GUV membranes. The presence of labeled DNA channels in solution leads to a strong fluorescence of the GUV membranes, suggesting that the DNA channels accumulate at the membrane surface due to binding interactions. A control experiment where only YOYO dye was used showed no observable fluorescence of the vesicle membrane. TEM images confirm that the intercalating dye YOYO does not alter the origami structure at the labeling density used here.

2

Note S3: Membrane insertion of DNA channels General considerations. The energy cost associated with inserting a DNA channel into the hydrophobic core

of the membrane without rearrangement of the lipids would be extremely high (already the insertion of a single monovalent ion would cost several eV (or several 100 kJ/mol) according to the Born model). We therefore assume that the lipids dynamically re-arrange around the DNA channel with the hydrophilic heads facing the stem of the structure. The resulting lipid structure would be similar to that of “toroidal pores” that have been studied in the context of pore-forming peptides such as melittin, solvents, mechanical stress or electroporation (32, 33).

Thermodynamics of insertion. Energetically, the membrane insertion process

involves a variety of contributions: the formation of a toroidal hole of radius R in the bilayer membrane is typically estimated to cost a free energy of E = 2 π γL R - π γS R2, where γL is the specific line tension of the lipid membrane and γS is its specific surface tension. A defect exceeding a critical size Rc = γL/γS will actually continue to grow, whereas smaller defects tend to reseal. We here estimate the energy cost for formation of a small hole conservatively by keeping only the linear term which dominates for small R, i.e. by E ≈ 2 π γL R.

Experimental and theoretical values for γL lie in the range of 5 – 40 pN (34). To accommodate a six helix bundle of diameter 6 nm would therefore cost less than ≈ 100 – 800 pN nm ≈ 25 – 200 kBT. This is compensated by the free energy gained from binding of the cholesterols to the lipid membrane. Insertion of a single cholesterol from solution into a lipid bilayer results in an energy gain of order 20 – 25 kBT (35). As the cholesterols are concentrically arranged around the stem (cf. Fig. S8), the maximum number of cholesterol insertions is achieved, when the stem points perpendicularly through the membrane. In addition, in the presence of divalent ions (the DNA channels are stabilized by Mg2+ ions), there is a favorable interaction between the zwitterionic headgroups of the PC lipids used in this study (36). Again, this energy is maximized when the DNA channel is inserted perpendicularly into the membrane. This shows that the cost for pore creation can be compensated by the favorable binding energies.

Mechanism of insertion. In order to be able to insert into a lipid membrane,

however, a potentially high activation barrier first has to be surmounted. The height of this barrier will depend on the lipid type (or lipid composition for mixtures), the curvature of the membrane, the presence of phase boundaries or defects within the lipid bilayer (32,33). Membrane defects can be created thermally, by chemical or electrical gradients, or by mechanical tension (33). Membrane binding molecules are in fact thought to preferentially insert at such defects and phase boundaries (37,38).

In our work two different types of lipid bilayer membranes were used: POPC lipids were used for the fabrication of small unilamellar vesicles (SUVs), while DPhPC lipids were used for the electrophysiological measurements. This choice was made as POPC provided better SUV samples, while DPhPC provided more stable suspended bilayer membranes, which are better suited for long-term electrophysiological measurements, and which sustained higher voltages. POPC and DPhPC are quite different in their

3

physical properties: POPC is a phospholipid with an unsaturared fatty acid chain and is in the liquid disordered state in our experiments, DPhPC is a fully saturated lipid and is known to form very stable lipid membranes (in the liquid ordered state in our experiments).

In fact, we seem to observe spontaneous insertion into the lipid bilayer (as discussed in SOM text S2) only for POPC lipids, whereas for DPhPC bilayers the additional application of voltage pulses is required (see below). It seems likely that for the “more fluid” POPC bilayers, thermal fluctuations are sufficient to promote insertion of the stem into the membrane, whereas the energy barrier for spontaneous insertion into DPhPC layers is too high. Furthermore, as is also well known from pore forming proteins, insertion into SUVs is actually easier than insertion into large vesicles or membranes (39), which may be attributed to a larger separation between the lipid headgroups due to the curvature of the SUVs.

As for the exact molecular mechanism of membrane insertion of our DNA channels,

unfortunately so far no molecular dynamics studies are available that exactly correspond to this case. That penetration of large, charged molecules through a bilayer can occur in principle, however, has been shown in several cases previously. For instance, hydrophilic (charged) HIV-1 TAT peptides were shown to spontaneously translocate through zwitterionic phospholipid bilayers (40), involving large membrane fluctuations and lipid rearrangements. Similar processes have been observed in MD simulations for translocation of the peptide penetratin (41) through the lipid bilayer.

For the DNA channels, we surmise that they initially adsorb to the lipid bilayer through their cholesterol modifications. It is conceivable that then the formation of defects and a similarly complex interplay of lipid rearrangements as for the charged peptides facilitate spontaneous insertion into the lipid bilayer, at least in the case of the more fluid, fluctuating POPC bilayers. For insertion, presumably the lipid headgroups are always oriented towards the DNA structure – here the positive charge of the zwitterionic PC headgroup may directly interact with the negative phosphate, whereas electrostatic interactions between the negative lipid charge and the phosphate backbone may be mediated by Mg2+ ions bound to the origami structure (36).

Pulse protocol for insertion into DPhPC bilayers. For electrophysiological

experiments, insertion of DNA channels into planar suspended bilayers composed of DPhPC lipids was required, which is considerably less favorable than insertion into SUVs made from POPC. As mentioned above, a bias potential can assist the incorporation by local destabilization of the membrane, and we developed a dedicated pulse protocol for this procedure. Similar pulse protocols have been previously used to facilitate insertion, e.g., of α-hemolysin (14), MspA (14), or OmpF (13) channels into lipid bilayers.

After the addition of origami channels to supported lipid membranes on the MECA chip, the voltage pulse protocol shown in Fig. S11A is applied, supported by repeated mixing. After some time, typically a clear step in the current signal is observed along with an increase of 1/f noise (Fig. S11B). Typically this sharp increase in transmembrane current occurred during the initial part of the incorporation pulse (230 mV for 20 ms),

4

suggesting a voltage-enhanced incorporation probability as known for other biological channels.

Similar current steps could be observed at 100 mV bias voltage, but less frequently. In some cases, two or more subsequent steps indicated the incorporation of multiple channels into the same membrane (see Fig. S11A inset). The flicker noise power varies greatly from channel to channel, from negligible up to values on the order of 10-6.

Fig. S11 a Top: Voltage pulse used for incorporation of origami channels into the lipid membrane. Bottom: Corresponding current response before (black dots) and during incorporation (red line). Inset: Consecutive incorporation of two origami channels at 100 mV bias voltage. b Noise power spectra before (black line) and after the incorporation of a single origami pore (red line). Current traces were recorded at 100 mV at a filter frequency of 3 kHz.

Control experiments. Using the same insertion pulse protocol, we performed

experiments with DNA channels lacking a stem. These were prepared and purified in the same way as the complete DNA channels, but with the staples responsible for folding of the protruding six-helix bundle left out (Fig. S12). For these “mutants”, we did not observe any incorporation within several hours of observation time. We also performed experiments with cholesterol-labeled anchor strands alone. These also did not lead to the formation of channels in the lipid membranes. At very high concentrations (> 10 µM), cholesterol-labeled oligonucleotides destabilized the lipid bilayer, leading to membrane rupture.

5

Fig. S12 A channel mutant without stem (schematic depiction and TEM class average of the structure) constructed for control experiments. This structure did not penetrate through lipid bilayer membranes using the pulse protocol shown in Fig. S11.

Multiple Insertions. In experiments with elevated membrane voltages (e.g., constant V=200 mV), occasionally large conductances/currents were observed that corresponded to the presence of several channels in parallel (Fig. S13). We assume that in such cases, cholesterol-mediated aggregates of DNA channels initially adsorbed to the membrane inserted into the membrane in concert. In these cases, we observed stepwise closing of single channels over time, potentially caused by clogging of the channels. Partial recovery of the channel conductance is observed after setting the bias temporarily to 0 mV (Fig. S13A). As shown in Fig. S13, current histograms display the quantized conductance of the DNA channels.

6

Fig. S13 a Time course of the ionic current across the lipid bilayer with multiple origami channels incorporated at 200 mV bias voltage. The DNA channel structures used here had a 7 nt modification at site M2 (Notes S6, S8, S9). Vertical lines mark limits of continuous recording. The recording was suspended a this positions with no applied bias for a few ms. Right: Corresponding all-point current histogram, showing a mean peak-to-peak distance of 172±15 pA This corresponds to a mean conductance change of 0.86±0.08 nS. b Current response of origami channels at 200 mV with a modification present at site M1, accompanied by the corresponding all-point current histogram. The mean peak-to-peak distance corresponds to 194±14 pA (0.97±7 nS).

7

Note S4: Conductance of the DNA channels A simple estimate for the conductance of the DNA channel is obtained by

approximating it by a cylindrical resistor of equal dimensions (length L = 42 nm, diameter d = 2 nm) with additional access resistances at the channel entrances. The conductance of such a cylinder is given by

! = !!!!

4! + !"

resulting in a conductance of 0.78 nS for a electrolytic conductivity of κ=10.86 S/m

(for 1M KCl at T=25°C (42)). This result fits our experimental findings well. As observed previously (24), we also expected a considerable lateral conductivity, allowing ionic current also to flow from the sides of the origami cap or stem. Only in the membrane region the current is expected to be strictly confined to the central channel itself. Experimentally we find a conductance slightly higher than calculated above, supporting a leaky channel structure. It is also conceivable, that the central channel is actually slightly compressed and has a smaller diameter than the assumed 2 nm, but this effect might be overcompensated by leak conductivity.

As this issue cannot be answered quantitatively at this point without more detailed structural studies - potentially also involving molecular dynamics simulations of the DNA channel -, we performed simplified electric field calculations to get an impression of the influence of leaks on its electrical properties (see Note S5).

The conductance measured is close to that of α-hemolysin, which initially inspired the cap-stem structure for our DNA channel. For 1M KCl, e.g., a conductance of G=0.93 nS is measured for α-hemolysin (I=112 ± 3 pA for V=120 mV at T=21°C (43)). One has to note, however, that the physical properties of the DNA channel and α-hemolysin are quite different. α -hemolysin is shorter (10 nm including cap), narrower (1.4 nm at its central constriction), and less charged – in addition, the cap structure of the DNA channel appears to be leaky to the flow of small ions (see Note S5). Thus, presumably the combination of a variety of counter-acting effects coincidentally makes the DNA channel conductance similar to that of α-hemolysin.

8

Measurement setup. For the electrophysiological experiments, we utilized parallel chip setup (Fig. S14) recently developed by Baaken et al. (12) that allowed us to work with small sample volumes and increase the throughput of our measurements. The setup has 16 measurement sites connected to 16 amplifiers with relatively low temporal resolution. Depending on the type of experiment, measurements on a single spot were performed with a dedicated low noise patch clamp amplifier (cf. Materials and methods).

Fig. S14 A: Micro - Electrode Cavity Array (MECA) Chip. The chip contains 16 cavities with individually addressable electrodes. B: Schematic depiction of a cavity, fabricated onto a glass substrate by lithographic structuring of an 8 µm thick layer of SU-8 photoresist. On the bottom of the cavity an Ag/AgCl electrode is created by evaporation of gold followed by electrochemical deposition of a silver layer and subsequent chlorination. After forming a lipid bilayer across the cavities, membrane channels can be incorporated and current recordings can be performed.

9

Note S5: Mapping the potential inside the pore

A qualitative study of how the origami structure’s intrinsic electrical conductivity impacts potential distribution and channel conductance was performed by means of a finite element simulation using Comsol Multiphysics to solve Poisson’s equation. To this end the DNA channel was approximated by two cylinders with diameters 6.6 nm and 20 nm, with a cylindrical channel of diameter 2.2 nm inside (Fig. S15A). The lipid bilayer was modelled as a 4 nm thick layer with zero conductivity. The symmetry axis of the channel was defined as the z axis of the coordinate system. z=0 corresponds to the boundary of the lipid bilayer facing the stem side of the structure. The channel interior and adjacent solution compartments were simulated with KCl solution bulk conductivity (11 S/m). To screen the influence of origami leakiness, the conductivity of the origami structure was varied in the simulation from 10-4 S/m (insulating) to 10 S/m. The mesh density ranged from 18 elements/nm3 (channel interior & surrounding) to 10-2 elements/nm3 (solution compartments). Fig. S15 shows the results of the simulations. As expected, the calculated channel conductance coincides with that of a simple geometrical model (Note S4) for the limit of zero origami conductivity. For origami conductivity values of 10-1 S/m and more (corresponding to 1% of the bulk solution conductivity) a considerable increase in channel conductance is observed. In the bulk solution limit, the channel conductance corresponds to a channel with a length equal to the lipid bilayer thickness and a diameter equal to the stem’s outer diameter. The electrical potential along the channel’s central axis is shown in Fig. S15D. Even at comparably low origami conductivity, a considerable inhomogeneity of the potential drop is apparent. Notably, the region of steepest slope is not at the center of the channel structure, but at the bilayer position, i.e. 26 nm from the cis side channel entrance. The local variation of electric field strength for nonzero leak conductivity can be expected to strongly influence translocation experiments, where charged molecules are pulled electrophoretically through the pore (cf. Discussion in Note S7).

10

Fig. S15 FEM Simulations of the DNA pore. A: Simulated conductance across a DNA channel incorporated into a lipid bilayer for varying intrinsic origami conductivity at 100 mV bias voltage and 11 S/m bulk solution conductivity. B: Color map of the electrical potential distribution along a y-z plane at x=0 calculated for 0.1 S/m origami conductivity. The dashed line marks the structure boundaries. C: Electric Field along the symmetry axis (z-axis) of the channel for varying origami conductivity. D: Corresponding potential drops along the channel. e Schematic illustration of the penetration depth of the structures used in translocation experiments before unzipping (cf. Figure 3 of the main paper).

11

Note S6: Channel gating

As shown in Fig. 2 of the paper, individual DNA channels displayed distinct gating behavior in which the channel conductance switched between several (sub)-conductance levels. While for high bias voltages the channels underwent full closure as seen, e.g., in Fig. S13, at lower voltages we typically observed current reduction amplitudes ΔI corresponding to fractions of a single channel conductance. We also designed channel ‘mutants’, in which one of the central helices forming the stem of the DNA channel was modified with a single-stranded heptanucleotide not incorporated into the main origami structure. This modification potentially could fluctuate, block the channel, or act as a ‘handle’ for the electric field and lead to partial unzipping of a DNA staple strand. We studied two channel mutants – for modification M1 the heptanucleotide pointed into the central channel, for modification M2 the staple extension was oriented away from the central stem (pointing into one of the “other” holes of the cap, cf. Notes S8, S9). We found that both modifications increased the gating frequency and behavior. Moreover, we never observed a current signal without gating for the mutants, whereas many of the unmodified DNA channels showed no gating at all (such as in current trace i) in Fig. 2D).

Fig. S16A shows a scatter plot of conductance change vs. channel closure time for an unmodified DNA channel structure. A single population is observed with a mean conductance reduction of ΔG/ΔG0 of 45% and an exponentially distributed dwell time with a mean value of 10.5 ms.

For modification M2 (Fig. S16B), a second population occurs. As expected for this mutant, the additional staple pointing away from the central stem does not lead to a stronger blockade amplitude – however, it causes gating fluctuations at short timescales.

As seen in Fig. S16C, for modification M1 (heptanucleotide pointing into the central channel) a clear second population occurs with a higher blockade amplitude of 70% and a considerably shorter time constant of 0.3 ms. We also constructed a mutant M1’ with a modification at the same position as M1, but with a heptanucleotide of different sequence (A7) (Fig. S16D). This modification causes a similar gating pattern as M1.

We also studied the voltage dependence of gating fluctuations of the M2 mutant (Fig. S17). Apparently, the closure time increases slightly with increasing voltage, indicating the coupling of the closure events to electrostatic forces. Notably, this clearly differentiates gating events from translocation events through the channel, which become faster with increasing voltage (Fig. S19)!

12

 

 

A B

 

 

1 10 100

0.4

0.6

0.8

0

5

10  

G/G0

0 1 2

Frequency  (%)

Frequency  (%)Closure  Time  (ms)1 10 100 1000

0.4

0.6

0.8

0

5

10

0 1 2 3

Frequency  (%)

Frequency  (%)Closure  Time  (ms)

G/G0

 

 

1 10 100

0.4

0.6

0.8

0

5

10

0 1 2

Frequency  (%)

Frequency  (%)Closure  Time  (ms)

G/G0

C D

G/G0

Frequency  (%)

Frequency  (%)Closure  Time  (ms)1 10 100

0.4

0.6

0.8

0

5

10

 

 0 1 2

 

Fig. S16 Scatter plots of the relative conductance change versus dwell time for the closure states of A an unmodified origami channel, B a channel with modification (sequence TTTCCGG) on site M2, C with modification (sequence TTTCCGG) on site M1, D with a modification on site M1 but with different sequence (A7) at 100 mV bias voltage. Each data point corresponds to a single closing event. In total 697 events (red), 2532 events (blue), 623 events (grey), and 931 events (black) were recorded. On the right and on the top the corresponding histograms are shown. The lines correspond to single exponential (no modification or modification M2) or double exponential (with modification M1), (top) and Gaussian fits (right).

13

 

 

 

   

151050

0

5

10

0

5

1 10 100 1000Closure  Time  (ms)

75  mV

100  mV

125  mV

Frequency  (%)

Fig. S17 Closure time for mutant M2 as a function of applied membrane voltage. The mean closure time increases with increasing voltage.

14

Note S7: DNA translocation experiments

Current blockade by DNA molecules with different tail lengths. In our DNA translocation experiments, DNA molecules (hairpin and quadruplex structures) with polythymidine tails of different lengths (60 and 125 nt) were investigated for both forward (cis-trans) and backward (trans-cis) translocation directions (cf. Fig. S15E). The fact that longer tails lead to a larger transient current reduction is another clear indication of translocation through the central channel of our DNA structures. Moreover, the current blockade amplitudes measured in these experiments can be used for a qualitative discussion of the electric field distribution within the DNA channel.

In a simple picture, the presence of a DNA strand of length LDNA reduces the effective diameter of the channel from A to A-ADNA, where ADNA is the diameter of single-stranded DNA. Hence, the total conductance of the channel is given by the series of the conductances of blocked (Gb) and unblocked (Gu) part, i.e.,

!! = !  ! − !!"#!!"#

!! = !  !

! − !!"#

!!"! =!!!!!! + !!

= !  !(! − !!"#)

! − !!"! ! + !!"#!!"#

Compared with an unoccupied channel the relative current reduction (conductance

reduction) is then given by: !"! =

!"! = 1−

! − !!"# !! − !!"# ! + !!"#!!"#

=!!"#!!"#

! − !!"# ! + !!"#!!"#

A 125 thymidine tail fully stretched to its contour length has a length of LDNA ≈ 50

nm, i.e., it spans the whole length of the DNA channel. Assuming an effective diameter of 1 nm for ssDNA, we would expect a current/conductance reduction of ΔG/G of ADNA/(A-ADNA) ≈ 33% (L=LDNA in the formula above), which is considerably higher than what we observe (around 10%). This may be attributed in part to the lateral leak conductance of the channel structure discussed in Note S5. A 60 nt long tail corresponds to a length of LDNA ≈ 24 nm (assuming a base-to-base distance of 0.4 nm), from which we expect a ΔG/G of ≈ 15%. While again our experimentally observed reduction is smaller (~ 5%), this value is completely consistent with the value for the 125 tail if one takes the leak conductance into account.

As discussed in SOM note S5, a partially conductive DNA channel structure would result in a more complicated potential landscape/electric field distribution than the simple pore picture used for the estimations above. In Fig. S15E, the lengths of the DNA molecules together with the electric fields for different values of the leak conductivity are depicted. For finite leak conductivity, we expect a much stronger influence on the conductance from DNA molecules with long tails – traversing the high field region defined by the lipid bilayer membrane – than for the short tails. In fact, it is expected that

15

short-tailed DNA molecules may feel a considerably smaller electric force depending on the exact field distribution. Furthermore, we expect short-tailed molecules to be sensitive to the translocation direction. As seen in the scatter plot Fig. 3c for DNA hairpins with a 50 nt tail, backward translocation results in faster translocations with a higher current amplitude, indicating that the short tail penetrates deeper into the high field region for the backward direction.

Length of the DNA tails in the channel. For translocation experiments we expect the

DNA tails to be fully stretched in the channel, and therefore their lengths are simply given by the number of bases times the base-to-base-distance for ssDNA (0.4 nm – 0.5 nm). As the channel width (2 nm) is close to this length as well as to the persistence length of ssDNA, the usual theories for polymer conformation inside a narrow channel are not applicable - these refer either to the limiting case where the channel diameter is much larger than the persistence length (resulting in a scaling treatment) or much smaller than the persistence length (resulting in a mechanical deflection model) (44). A better molecular understanding of the conformation of DNA strands inside a DNA channel will require molecular modeling and simulation (45).

In any case, for 100 mV applied bias over a length of L=42 nm, we expect electrostatic forces to be well in the higher pN regime (also supported by the fact that we observe force-induced melting of DNA duplexes in the translocation experiments). In free solution experiments, a single-stranded random coil structure would be completely stretched out at these forces, and we assume the same to be the case in the context of our translocation studies.

Voltage dependence of hairpin translocation. In order to demonstrate that the

observed translocation signals are not “gating” signals, we also studied their voltage dependence. As shown in Fig. S18, hairpin translocation can be measured at V=100 mV with typical dwell times in the range 1-10 ms. At this voltage, the event rate as well as the SNR is very low and we were therefore not able to collect enough data for performing a statistical analysis. In addition to the data presented in the main text (Fig. 3C) which was collected at V=200 mV, we also performed translocation studies at V=250 mV (Fig. S19). The dwell times decrease from 100 mV over 200 mV to 250 mV, which is a clear sign of voltage-assisted unfolding/translocation of the hairpin structures. By contrast, the “gating times” increase with increasing voltage as can be seen in Fig. S17.

Measurements at V=250 mV are already performed close to the breakdown voltage of the lipid bilayer membrane, and thus also demonstrate that the stability of the DNA channels in an electrical field is not a limiting factor.

Quadruplex translocations in backward direction. In Fig. S20A, a scatter plot

generated from backward translocation events (from trans to cis) of Q-T125 DNA at negative bias potential is shown. The scatter plot histograms yield a blockade amplitude that is 27% lower (11.2 pA) and a characteristic dwell time that is 2.3 times shorter (3.5 ms) than for the forward direction. This observation indicates a directionality of the translocation process itself, as the integrated electrophoretic force along the DNA channel should be independent of the direction of translocation (as the T125 tail is expected to

16

traverse the whole central channel). A possible reason for this effect may be steric or binding interactions at the entrance of the asymmetrically designed origami structure.

Long-lived events for Q-T125 translocation at long measurement times. For

prolonged measurement times for the Q-T125 translocation experiment in forward direction (cf. Fig. 3C), a sudden increase in long-lived blockade events was observed. A scatter plot including all Q-T125 measurement data recorded for these events is shown in Figure S20B. In this plot, a second population is clearly visible. It features a higher blockade amplitude (20.5 pA) than the main population and a considerably longer characteristic dwell time (46 ms). A comparable population was not observed for the Q-T60 measurement or for the Q-T125 backward translocation measurement. One possible explanation for this time dependent effect is the agglomeration of DNA molecules at the channel entrance over the course of a measurement run. In fact, after addition of quadruplex DNA the event rate continuously increased, indicating that the DNA concentration at the pore entrance rises over time. This may be a result of insufficient mixing upon addition (we mixed very gently to avoid destruction of the membrane) or interaction of DNA with single-stranded scaffold loops protruding from the DNA channel structure.

Life-time experiments with alpha hemolysin pores. In Fig. 3F of the main paper it can be noted that the two G quadruplex structures

with different tails lengths have approximately the same dwell time, even though the electrical forces applied to them should be quite different. This is in accord, however, with studies using α-hemolysin, which found a relatively weak voltage dependence of the quadruplex lifetime. We included data on hairpin stability and quadruplex stability as a function of voltage (Fig. S21) measured in nanopore force spectroscopy experiments as previously described (46). This peculiar behavior can be attributed to the different energy landscapes with quite different positions of the transition state for unfolding for hairpin and quadruplex structures.

Fig. S18 Examples for hairpin translocations at 100 mV with dwell times in the range 1-10 ms.

 

20  ms

5  pA

2.9  ms 2.8  ms 1.3  ms 2.7  ms

1.1  ms 4.6  ms 3.0  ms 2.4  ms 10.2  ms

17

Fig. S19 Comparison of hairpin translocations at 200 mV and 250 mV. The black lines correspond to exponential fits with τ=1.5 ms and τ=0.9 ms.

Fig. S20 Additional data for quadruplex translocations. a Scatter Plot of the current amplitude vs. dwell time for the backward translocation of Q-T125 through the artificial ion channel at negative bias voltage (-200mV), accompanied by the corresponding histograms. 266 events were detected. The solid lines correspond to single exponential (top) and Gaussian (right) fits to the histograms. b Long-lived population: Scatter Plot of the current amplitude vs. dwell time for forward translocation of Q-T125 (see Fig. 3), including data from the last part of the measurement. Each data point corresponds to a single translocation event. The lines correspond to single exponential (top) and Gaussian (right) fits to the histograms.

1 100

5

10

Rel

ativ

e F

req

uen

cy (

%)

Dwell Time (ms)

250 mV200 mV

1 10 1000

10

20

0

5

10

   Frequency  (%

)

 

 

Amplitude  (pA)

Dwell  Time  (ms)0 5 10

   

 

Frequency  (%)1 10 1000

10

20

0

5

10

   Frequency  (%

)  

Amplitude  (pA)

Dwell  Time  (ms)0 5 10

   

 

Frequency  (%)

A B

18

Fig. S21 Dwell time measurements as a function of transmembrane voltage performed with α hemolysin channels. Data for hairpins are shown in blue, data for G quadruplexes in red. A larger stability as well as very shallow voltage dependence for G quadruplex structures is apparent.

0 50 100 150 20010

10

10

100

102

Voltage  (mV)

(s)  

 

19

Note S8: DNA sequences

Table S1. Staple strand sequences for DNA-based membrane channel. Core staples are indicated by gray bars, adaptors to cholesterol anchors by red bars.

Start End Sequence Length Color52[125] 25[139] GTCAGTTCCTGCAATAGGAATTGATAATCGGCTGTCTTTCCT 4230[104] 12[91] AGCTATTTTTGAGAACTAGCACTAAAACAATTGTGTCAATCA 420[104] 32[84] AGTACCACCACTACGAATACATGTCAATCCAAAAACAGGAAGATTGTAT 4926[97] 6[91] TTTCATTGTCGCTGCAGCTTG 2126[146] 8[133] ATTTACGCTGTTTACAGACGACGACAATCATTTTCAAGGTGA 420[153] 53[153] GAGGCTGGCCTTTAGGCCGGAATAGAAGAACAAGCCAGCAAA 420[160] 0[168] GTATTAAGCCGTTCCAGTAAGCGTCATATCTGAAA 350[195] 3[202] GCCTATTGGTCATAGCCCCCTCGCCTCCAGCCGCC 350[202] 3[195] GCCCCCTAGTGTACTAAATCCTCATTAACCACCAG 351[77] 36[77] ATAGGTGACCGAACTGACCAAACAAGAACCGGATACTCCAGC 421[91] 42[91] GTACTCAAGTTTCGTTTTAAGAAACGAACAACGCGGTGAGAC 421[119] 36[119] TCAGAACAGACTTAGCCGGAATGCTCATTCAGTGATATTGAG 422[118] 39[118] AACCCATACCAGTCAGGACAACAGAATAAGGCTTGCATACGA 423[91] 43[104] GTAGCATGTTTACCAAAGGAAGGCAAAATTGCCCCAGCAGGC 423[126] 52[126] ACGATCTTTGCTAAGATTAGTGGGAGGTAGTTGAAATATCTG 423[168] 0[161] TCAGAGCACCGCCACCCTCAGACTGTAGCATGAAA 353[203] 5[216] GCCAGCAAAACAAATGGTAATAAGTTTTTATAAACGTTTGCC 424[90] 46[91] AAGGAACGACTTCAAATATCGCAAATGCCGTCCAAGGGCGCT 424[139] 43[139] TGGGATTAAAGTTTAGAATAAATTAACTAATATACTAACGGA 424[153] 22[147] ATGAATTCAGCTACTTAGCGA 214[195] 0[182] CTCAGAGCCGCCACCATTTTCTCGGAAC 284[216] 0[203] GCCACCACCGGAACTATTAGCAGTTAAT 285[67] 20[67] AAAAAGGCTCGGAATTGCGAATAAAGATTAAGAGGAAGAAAACGAGAA 485[105] 18[105] ATCAGCTTGCTTACTCAACAGTCATAGTAGCAACACTATCAT 425[140] 32[140] CAAGTTTAGACTCCAAGGGCGCACAATCAATTCTTCATAATT 425[154] 2[154] GCGTCAGAGAGTAAACGTTCCACCACCCCTCTGAATTTACCA 426[125] 27[125] GATAGCACCACCGACTTGAGCAGAAACCAATCAGTGCCAGCT 428[132] 28[119] ATTATCACCGTCGCATTTTAGGCAGAGGAAACAACATGTTAA 4252[104] 27[104] CGCCAGCAGGCGGTGCTCAACCCATTAGGCTATATTACAGGC 4252[90] 53[83] ACAGGAAAAACGCTGCAGAAG 219[84] 29[83] TAAAATAAAAGACAGCATCGGCAAATCA 289[133] 16[133] AAGACAATCAAGAGCTCAGAAAGGGATAAGCAAGAGAATTGA 4210[111] 25[118] GGCAAAAGAAGGCAGGATCGTCAGGGAGCTGTTTAATACATTTCGCAAA 4910[125] 30[112] ATGGTAATTGTATCATATGCGGCCAACATGTAATAATGCCGG 4232[125] 32[126] GTTTAGATAATCAGCGTCTGGCCTTCCTAAATTTAAAAGCCT 4212[90] 4[91] TAAGGGATATCACCAAGTGCCAGCCTTTAATTGTAGAATAGA 4212[132] 23[132] CAAACGTCGCCACCAAGGATTAGTAGCGAACCATCCCAATAG 4212[139] 40[126] ATGTTAGCGAGGAACCGAACATGAGTGAGAAACAGTACATAA 4232[139] 37[139] ACTAGAAATGGTTTAATTTCAAAATGCTTAGGTTGTGAGAAG 4213[91] 41[90] CCCAAATACCAGAATCCGCTCCCTAATGAGTGAGCAATCGGC 4214[104] 34[105] GAGAAACCAACGTAATCGGCCCGCATCGTAACCGTTTCATCA 4233[84] 38[91] CAATAGGAACGCCAATGTGAGAGATGGGTCAGGAATCCCAGTCTCTAGA 4915[98] 21[118] AACTGGCTCATTATGTACCGTACAGCCCTTTCAGCATTCGAGCTTCATT 4915[140] 35[153] AACAATGCAGATAGACGCAATTCCGGCTGATGCAAATCCAAT 4216[90] 49[90] CTAACGGAACGCCAAGACGACTGGATAGTTTAAACGTGTTTT 4216[111] 45[111] AATCTACGCATAGTAAGAGCGTTTGATGGTGGTTCGAAGGGA 4234[146] 30[147] TTTAGTTGAAATACCGACCGTCCGGAATACCAGTAAGAATCG 4218[104] 11[111] AACCCTCTCCACAGAACACTGGGAGGTTCAGACGGTCGAAATCCGCGAC 4918[146] 11[146] TACAGAGTGTCGTCCATTTTCCCGCCACACGCAGTGGTGGCA 4218[153] 50[140] CAGCCTTAAGAAACGCGTCTTACTCGTAATTAGAGCCGTCAA 4234[153] 7[153] AATATATACATATAATTTTGTACATTCAGAAATTAAGCCAGC 4219[105] 47[118] GAATCGTCATAAAAGGGCGAAAGGAGCGGCTTTGACGAGCAA 4220[83] 9[83] AGTTCAGCCCGAAAAACTAAACAAAAGGGTCGAGAAAACGGG 4220[118] 48[105] TAATATTCATTGAACCACCGAGTAAAGAGAACGTATAACGTG 4235[112] 10[126] GCATCTGCCAGTACTATATGTTCTTCTGAATACATAATTCAT 4221[119] 9[132] AAATCAAACAACTTCGTAATCAGGATTAAATTACCAGCGCCA 4222[104] 29[104] CCGGAAGTTAGAGCTCTTAAAAGGCTTGCACCCTCTTCAACC 4236[118] 13[132] GGGACGAACGACGGCCAGTTTAAGACGCGGTTATAAGGAAAC 4224[90] 50[91] AAATATGATAAAACAGAGGTGCATTGCAACTATCGAATTAAC 4224[118] 52[105] CATATAATGCTGTACAGTATTAACACCGGGGAACAATATTAC 4224[132] 50[133] TTCATCGCAGTGCCACGCTGATCAATCAAGGAATTTAGATAA 4224[139] 19[139] TTTTATTCAAGCAAGACTTGCTGCTATTCCTAATTATCCCAA 4237[63] 12[63] GCGATTACCGGCACAATCTTGCTTTGAA 2825[84] 22[91] TAGATTTAGTTTGAATGTTTTGGATGGCCAAACTC 356[118] 22[105] GCTCGAGGTGAATTTTAATTGCTGAATTCCAAGCGAACCAGA 425[168] 3[181] CGCGTTTTCATCGGCCTCAGACGCCACC 2826[153] 25[160] CATCCTACCAAGAACGGGTAT 2138[139] 2[126] ATAGCGATAGCTTAATATATGAAGTTACTAATAAGGCAAGCC 4227[126] 5[139] AATGCAGAACGCGCAGCATGTCATTTGGCCAATGAACAGAAT 4228[104] 0[91] GCCTCAGAGCATAATATGATAAGCAGCGCGTAATGGGCGGAT 4228[118] 1[118] AATTAAGCTGATAATTTTGCGCCAACCTGCGGGGTGCCACCC 4238[153] 3[153] GAATCCTTGCTTCTAAGTAAGAAATAGCCCACCCTTTTCCAG 4229[84] 29[76] CCATCAAAGCTAAATCGGTTGTACCAAAAGACAGT 3529[105] 7[118] GTTCTAGCAATAAAAAGGCAAAGAATTACAATAACTTAAAGG 4229[140] 11[139] GAGCCAGCCATATTTAACAACTTATACAAATAGAAACATAAA 4229[147] 27[160] TAATAAGTTCTGTCTCAACAATAGATAA 2851[147] 25[146] GTTATCTCAAATATCAAACCCGAGCCAGAAGCCGTTATCATT 4240[153] 17[153] TACCTTTGAAAACAGAGATAACTGAACA 2831[112] 34[112] ACCCCGGCGAAAGACTGCTCCATGTTAAACCTGTAGCCAGCT 4239[119] 14[105] GCCGGCAGAGCCAAGCTTGCATGCCTGCCACACAACCCTGAC 4251[105] 51[104] TGGTAATATCCACATAAGAGTCTGTCCATCACGCAGCCTTGC 4251[140] 8[140] GAGGAAGACCTCCCATCAGATAACGTCAGAATTAGTTCATTA 4249[91] 19[104] TATAATCCGTTAGACGCGTACTATGGTTGGCGCTATACTGCG 4234[104] 30[105] ACATTAATCAAAAATAATTCGAAAAGCCCATATGTAGAGGGT 4241[112] 41[111] TTGCGTATTGGGGAATAAGCATAAAGTGTAAAGCCAGGCGGT 4249[133] 19[146] CAAACAAATTATCAGGAATTATCATCATGTTTGGATCCAAAT 4241[126] 20[126] AGATGATGTTAAGCCGGGAGACATAAAAATTATTTTGCCAGT 4246[132] 20[119] TCTGAATAAGGAGCTTTTGCGGAACAAAAGTATTATACAAAA 4236[76] 13[90] CAGCTTTAGTTGGGTAACGCCAGGGTTTGATCGCATTCATTA 4246[139] 16[140] TTATACTTCAAAATTATTTGCTCAGATGGAACACCCCCACAA 4245[112] 16[112] AGAAAGTTAGAACCGTACCTTCATTAGACCGTTGGGAAGAAA 4244[90] 16[98] TCCCTTACCGGCGAACGTGGCGAGAAAGCGAAATCTTACGAGGTTAATA 4937[105] 0[105] TTGTAAACGACAGTACAAAGCCGAGGCGTAGTACCTTTGCTC 4237[140] 1[153] AGTCAATAGTGAATTTTAACCAATAACGCCTTATTCCTCAGA 4247[119] 2[119] CCACCAGAATGGAACAGCCATACAGGGATAGCGTACAATAGG 4249[126] 23[125] GACTTTATACATTTGAGGATTAATCAACTTTGAAGACCGCGC 4241[133] 45[132] GAAACAAAATTACCTGAGCAAGAAACAAAGTAACATACCATA 4248[104] 5[104] CTTTCCTAGTGAGGTCCCCCTCGTTTTAGGAGTGATCGGTTT 4248[153] 4[140] AAAGTTTGAGTAACTTCGACATCCAGAGTTGCACCTTCTGTA 4242[104] 37[104] TTCACCACGGGGAGTGGGGTGACAATTCAGGTCGACACGACG 4245[70] 16[77] CTTGACGGGGAAAGTAAATCAGATACATAACAACA 3542[111] 42[112] TTTTCTTGAAAATCCTGTTACATCGGGAAACGCCAGGGTGGT 4242[153] 47[153] GAATTATGATTGCTTTTAACGACGTAAACCTGATTATTCCTG 4243[140] 38[140] TTCGCCTTCATTTCACATCAATTTAATGATAACCTTGAAAAC 4226[186] 7[165] TAACAGGATTAGCAGAGCGAGGAAGAAAAATAATATCCAAAATCACCAGT 5024[186] 23[165] TAACAGGATTAGCAGAGCGAGGTACCGCACTCATCGAGGCTTATCCGGTA 5042[186] 42[154] TAACAGGATTAGCAGAGCGAGGAAAATCGCGCAGAGGC 3828[186] 29[165] TAACAGGATTAGCAGAGCGAGGACAAAAGGTAAAGTAAAGAATATAAAGT 5036[186] 37[165] TAACAGGATTAGCAGAGCGAGGTCTGAGAGACTACCTTTTATCAAAATCA 5044[186] 43[165] TAACAGGATTAGCAGAGCGAGGTGCGTAGATTTTCAGGTTGAATACCAAG 5022[186] 21[165] TAACAGGATTAGCAGAGCGAGGAAGAACGCGAGGCGTTAATTTTATCCTG 5014[186] 39[165] TAACAGGATTAGCAGAGCGAGGAAGCCCTTTTTAAGAAGTAAATCGTCGC 5016[186] 41[165] TAACAGGATTAGCAGAGCGAGGGAGCGCTAATATCAGAAAATTAATTACA 5020[186] 19[165] TAACAGGATTAGCAGAGCGAGGTTACCAACGCTAACGAGATTTTTTGTTT 5046[186] 45[165] TAACAGGATTAGCAGAGCGAGGAATTCATCAATATAATACAGAAATAAAG 5018[186] 17[165] TAACAGGATTAGCAGAGCGAGGTCAAAAATGAAAATAGAAGTCAGAGGGT 508[230] 1[195] TAACAGGATTAGCAGAGCGAGGAGAACCAAGCCAGATTTGATGATACAGG 5010[186] 11[165] TAACAGGATTAGCAGAGCGAGGACCACGGAATAAGTTTAAAGAAACGCAA 5048[186] 47[165] TAACAGGATTAGCAGAGCGAGGAACGTTATTAATTTTAATTATCAGATGA 5040[187] 40[154] TAACAGGATTAGCAGAGCGAGG?ACAATTTCATTTGAAT 3938[186] 38[154] TAACAGGATTAGCAGAGCGAGGAATTAATTTTCCCTTA 3834[186] 35[165] TAACAGGATTAGCAGAGCGAGGCGAGAAAACTTTTTCACGCAAGACAAAG 5050[186] 49[165] TAACAGGATTAGCAGAGCGAGGCACTAACAACTAATAGTTAAATCCTTTG 5012[186] 13[165] TAACAGGATTAGCAGAGCGAGGTGGCATGATTAAGACTGAATACCCAAAA 5032[186] 33[165] TAACAGGATTAGCAGAGCGAGGTAAATAAGAATAAACAGTGATAAATAAG 508[186] 9[165] TAACAGGATTAGCAGAGCGAGGAAGGTAAATATTGACGACCGATTGAGGG 506[230] 2[168] TAACAGGATTAGCAGAGCGAGGCTATTATCATGGCTATGGAAAGCGCAGT 506[186] 6[154] TAACAGGATTAGCAGAGCGAGGCCATTACCATTAGCAA 3830[186] 31[165] TAACAGGATTAGCAGAGCGAGGCAGTAGGGCTTAATTGTAAAGCCAACGC 5052[186] 51[165] TAACAGGATTAGCAGAGCGAGGTCACCTTGCTGAACCTAAAATATCTTTA 50

20

Table S2. Staple strand sequences for DNA-based membrane channel with modification M1 (inside of the central stem). Core staples: grey bars, cholesterol adaptors: red, modification M1: blue.

Start End Sequence Length Color47[119] 2[56] CCACCAGAATGGAACAGCCATACAGGGATAGCGTACAATAGGAATTTCCGG 512[116] 39[118] CCCATACCAGTCAGGACAACAGAATAAGGCTTGCATACGA 400[104] 32[84] AGTACCACCACTACGAATACATGTCAATCCAAAAACAGGAAGATTGTAT 4926[153] 25[160] CATCCTACCAAGAACGGGTAT 2114[104] 34[105] GAGAAACCAACGTAATCGGCCCGCATCGTAACCGTTTCATCA 4252[125] 25[139] GTCAGTTCCTGCAATAGGAATTGATAATCGGCTGTCTTTCCT 4251[147] 25[146] GTTATCTCAAATATCAAACCCGAGCCAGAAGCCGTTATCATT 4252[104] 27[104] CGCCAGCAGGCGGTGCTCAACCCATTAGGCTATATTACAGGC 4251[105] 51[104] TGGTAATATCCACATAAGAGTCTGTCCATCACGCAGCCTTGC 4252[90] 53[83] ACAGGAAAAACGCTGCAGAAG 2151[140] 8[140] GAGGAAGACCTCCCATCAGATAACGTCAGAATTAGTTCATTA 4249[126] 23[125] GACTTTATACATTTGAGGATTAATCAACTTTGAAGACCGCGC 4249[91] 19[104] TATAATCCGTTAGACGCGTACTATGGTTGGCGCTATACTGCG 4249[133] 19[146] CAAACAAATTATCAGGAATTATCATCATGTTTGGATCCAAAT 4248[153] 4[140] AAAGTTTGAGTAACTTCGACATCCAGAGTTGCACCTTCTGTA 4248[104] 5[104] CTTTCCTAGTGAGGTCCCCCTCGTTTTAGGAGTGATCGGTTT 4226[146] 8[133] ATTTACGCTGTTTACAGACGACGACAATCATTTTCAAGGTGA 4224[90] 50[91] AAATATGATAAAACAGAGGTGCATTGCAACTATCGAATTAAC 4246[139] 16[140] TTATACTTCAAAATTATTTGCTCAGATGGAACACCCCCACAA 4224[118] 52[105] CATATAATGCTGTACAGTATTAACACCGGGGAACAATATTAC 4245[112] 16[112] AGAAAGTTAGAACCGTACCTTCATTAGACCGTTGGGAAGAAA 4245[70] 16[77] CTTGACGGGGAAAGTAAATCAGATACATAACAACA 3546[132] 20[119] TCTGAATAAGGAGCTTTTGCGGAACAAAAGTATTATACAAAA 4244[90] 16[98] TCCCTTACCGGCGAACGTGGCGAGAAAGCGAAATCTTACGAGGTTAATA 4943[140] 38[140] TTCGCCTTCATTTCACATCAATTTAATGATAACCTTGAAAAC 420[153] 53[153] GAGGCTGGCCTTTAGGCCGGAATAGAAGAACAAGCCAGCAAA 426[118] 22[105] GCTCGAGGTGAATTTTAATTGCTGAATTCCAAGCGAACCAGA 426[125] 27[125] GATAGCACCACCGACTTGAGCAGAAACCAATCAGTGCCAGCT 425[140] 32[140] CAAGTTTAGACTCCAAGGGCGCACAATCAATTCTTCATAATT 425[105] 18[105] ATCAGCTTGCTTACTCAACAGTCATAGTAGCAACACTATCAT 424[216] 0[203] GCCACCACCGGAACTATTAGCAGTTAAT 285[168] 3[181] CGCGTTTTCATCGGCCTCAGACGCCACC 285[154] 2[154] GCGTCAGAGAGTAAACGTTCCACCACCCCTCTGAATTTACCA 428[132] 28[119] ATTATCACCGTCGCATTTTAGGCAGAGGAAACAACATGTTAA 423[203] 5[216] GCCAGCAAAACAAATGGTAATAAGTTTTTATAAACGTTTGCC 424[139] 43[139] TGGGATTAAAGTTTAGAATAAATTAACTAATATACTAACGGA 425[67] 20[67] AAAAAGGCTCGGAATTGCGAATAAAGATTAAGAGGAAGAAAACGAGAA 4812[139] 40[126] ATGTTAGCGAGGAACCGAACATGAGTGAGAAACAGTACATAA 423[126] 52[126] ACGATCTTTGCTAAGATTAGTGGGAGGTAGTTGAAATATCTG 423[168] 0[161] TCAGAGCACCGCCACCCTCAGACTGTAGCATGAAA 354[195] 0[182] CTCAGAGCCGCCACCATTTTCTCGGAAC 284[153] 22[147] ATGAATTCAGCTACTTAGCGA 2115[98] 21[118] AACTGGCTCATTATGTACCGTACAGCCCTTTCAGCATTCGAGCTTCATT 499[84] 29[83] TAAAATAAAAGACAGCATCGGCAAATCA 2816[90] 49[90] CTAACGGAACGCCAAGACGACTGGATAGTTTAAACGTGTTTT 429[133] 16[133] AAGACAATCAAGAGCTCAGAAAGGGATAAGCAAGAGAATTGA 421[119] 36[119] TCAGAACAGACTTAGCCGGAATGCTCATTCAGTGATATTGAG 423[91] 43[104] GTAGCATGTTTACCAAAGGAAGGCAAAATTGCCCCAGCAGGC 4210[111] 25[118] GGCAAAAGAAGGCAGGATCGTCAGGGAGCTGTTTAATACATTTCGCAAA 4918[153] 50[140] CAGCCTTAAGAAACGCGTCTTACTCGTAATTAGAGCCGTCAA 4210[125] 30[112] ATGGTAATTGTATCATATGCGGCCAACATGTAATAATGCCGG 421[91] 42[91] GTACTCAAGTTTCGTTTTAAGAAACGAACAACGCGGTGAGAC 424[90] 46[91] AAGGAACGACTTCAAATATCGCAAATGCCGTCCAAGGGCGCT 4220[118] 48[105] TAATATTCATTGAACCACCGAGTAAAGAGAACGTATAACGTG 420[195] 3[202] GCCTATTGGTCATAGCCCCCTCGCCTCCAGCCGCC 351[77] 36[77] ATAGGTGACCGAACTGACCAAACAAGAACCGGATACTCCAGC 4222[104] 29[104] CCGGAAGTTAGAGCTCTTAAAAGGCTTGCACCCTCTTCAACC 4212[90] 4[91] TAAGGGATATCACCAAGTGCCAGCCTTTAATTGTAGAATAGA 420[160] 0[168] GTATTAAGCCGTTCCAGTAAGCGTCATATCTGAAA 3512[132] 23[132] CAAACGTCGCCACCAAGGATTAGTAGCGAACCATCCCAATAG 4224[132] 50[133] TTCATCGCAGTGCCACGCTGATCAATCAAGGAATTTAGATAA 4224[139] 19[139] TTTTATTCAAGCAAGACTTGCTGCTATTCCTAATTATCCCAA 4213[91] 41[90] CCCAAATACCAGAATCCGCTCCCTAATGAGTGAGCAATCGGC 4225[84] 22[91] TAGATTTAGTTTGAATGTTTTGGATGGCCAAACTC 3526[97] 6[91] TTTCATTGTCGCTGCAGCTTG 2118[146] 11[146] TACAGAGTGTCGTCCATTTTCCCGCCACACGCAGTGGTGGCA 4234[153] 7[153] AATATATACATATAATTTTGTACATTCAGAAATTAAGCCAGC 4221[119] 9[132] AAATCAAACAACTTCGTAATCAGGATTAAATTACCAGCGCCA 4227[126] 5[139] AATGCAGAACGCGCAGCATGTCATTTGGCCAATGAACAGAAT 4228[104] 0[91] GCCTCAGAGCATAATATGATAAGCAGCGCGTAATGGGCGGAT 4228[118] 1[118] AATTAAGCTGATAATTTTGCGCCAACCTGCGGGGTGCCACCC 4215[140] 35[153] AACAATGCAGATAGACGCAATTCCGGCTGATGCAAATCCAAT 4229[84] 29[76] CCATCAAAGCTAAATCGGTTGTACCAAAAGACAGT 3529[105] 7[118] GTTCTAGCAATAAAAAGGCAAAGAATTACAATAACTTAAAGG 4229[140] 11[139] GAGCCAGCCATATTTAACAACTTATACAAATAGAAACATAAA 4229[147] 27[160] TAATAAGTTCTGTCTCAACAATAGATAA 2830[104] 12[91] AGCTATTTTTGAGAACTAGCACTAAAACAATTGTGTCAATCA 4216[111] 45[111] AATCTACGCATAGTAAGAGCGTTTGATGGTGGTTCGAAGGGA 4231[112] 34[112] ACCCCGGCGAAAGACTGCTCCATGTTAAACCTGTAGCCAGCT 4232[125] 32[126] GTTTAGATAATCAGCGTCTGGCCTTCCTAAATTTAAAAGCCT 4232[139] 37[139] ACTAGAAATGGTTTAATTTCAAAATGCTTAGGTTGTGAGAAG 4218[104] 11[111] AACCCTCTCCACAGAACACTGGGAGGTTCAGACGGTCGAAATCCGCGAC 4933[84] 38[91] CAATAGGAACGCCAATGTGAGAGATGGGTCAGGAATCCCAGTCTCTAGA 4934[104] 30[105] ACATTAATCAAAAATAATTCGAAAAGCCCATATGTAGAGGGT 4234[146] 30[147] TTTAGTTGAAATACCGACCGTCCGGAATACCAGTAAGAATCG 4242[153] 47[153] GAATTATGATTGCTTTTAACGACGTAAACCTGATTATTCCTG 4238[153] 3[153] GAATCCTTGCTTCTAAGTAAGAAATAGCCCACCCTTTTCCAG 4235[112] 10[126] GCATCTGCCAGTACTATATGTTCTTCTGAATACATAATTCAT 4236[76] 13[90] CAGCTTTAGTTGGGTAACGCCAGGGTTTGATCGCATTCATTA 4236[118] 13[132] GGGACGAACGACGGCCAGTTTAAGACGCGGTTATAAGGAAAC 4219[105] 47[118] GAATCGTCATAAAAGGGCGAAAGGAGCGGCTTTGACGAGCAA 4237[63] 12[63] GCGATTACCGGCACAATCTTGCTTTGAA 2837[105] 0[105] TTGTAAACGACAGTACAAAGCCGAGGCGTAGTACCTTTGCTC 4237[140] 1[153] AGTCAATAGTGAATTTTAACCAATAACGCCTTATTCCTCAGA 4238[139] 2[126] ATAGCGATAGCTTAATATATGAAGTTACTAATAAGGCAAGCC 4241[112] 41[111] TTGCGTATTGGGGAATAAGCATAAAGTGTAAAGCCAGGCGGT 4220[83] 9[83] AGTTCAGCCCGAAAAACTAAACAAAAGGGTCGAGAAAACGGG 4239[119] 14[105] GCCGGCAGAGCCAAGCTTGCATGCCTGCCACACAACCCTGAC 4240[153] 17[153] TACCTTTGAAAACAGAGATAACTGAACA 280[202] 3[195] GCCCCCTAGTGTACTAAATCCTCATTAACCACCAG 3541[133] 45[132] GAAACAAAATTACCTGAGCAAGAAACAAAGTAACATACCATA 4241[126] 20[126] AGATGATGTTAAGCCGGGAGACATAAAAATTATTTTGCCAGT 4242[104] 37[104] TTCACCACGGGGAGTGGGGTGACAATTCAGGTCGACACGACG 4242[111] 42[112] TTTTCTTGAAAATCCTGTTACATCGGGAAACGCCAGGGTGGT 4218[186] 17[165] TAACAGGATTAGCAGAGCGAGGTCAAAAATGAAAATAGAAGTCAGAGGGT 5038[186] 38[154] TAACAGGATTAGCAGAGCGAGGAATTAATTTTCCCTTA 3836[186] 37[165] TAACAGGATTAGCAGAGCGAGGTCTGAGAGACTACCTTTTATCAAAATCA 5024[186] 23[165] TAACAGGATTAGCAGAGCGAGGTACCGCACTCATCGAGGCTTATCCGGTA 5028[186] 29[165] TAACAGGATTAGCAGAGCGAGGACAAAAGGTAAAGTAAAGAATATAAAGT 506[186] 6[154] TAACAGGATTAGCAGAGCGAGGCCATTACCATTAGCAA 3814[186] 39[165] TAACAGGATTAGCAGAGCGAGGAAGCCCTTTTTAAGAAGTAAATCGTCGC 5040[187] 40[154] TAACAGGATTAGCAGAGCGAGG?ACAATTTCATTTGAAT 396[230] 2[168] TAACAGGATTAGCAGAGCGAGGCTATTATCATGGCTATGGAAAGCGCAGT 5032[186] 33[165] TAACAGGATTAGCAGAGCGAGGTAAATAAGAATAAACAGTGATAAATAAG 508[186] 9[165] TAACAGGATTAGCAGAGCGAGGAAGGTAAATATTGACGACCGATTGAGGG 5030[186] 31[165] TAACAGGATTAGCAGAGCGAGGCAGTAGGGCTTAATTGTAAAGCCAACGC 5046[186] 45[165] TAACAGGATTAGCAGAGCGAGGAATTCATCAATATAATACAGAAATAAAG 508[230] 1[195] TAACAGGATTAGCAGAGCGAGGAGAACCAAGCCAGATTTGATGATACAGG 5048[186] 47[165] TAACAGGATTAGCAGAGCGAGGAACGTTATTAATTTTAATTATCAGATGA 5012[186] 13[165] TAACAGGATTAGCAGAGCGAGGTGGCATGATTAAGACTGAATACCCAAAA 5010[186] 11[165] TAACAGGATTAGCAGAGCGAGGACCACGGAATAAGTTTAAAGAAACGCAA 5026[186] 7[165] TAACAGGATTAGCAGAGCGAGGAAGAAAAATAATATCCAAAATCACCAGT 5016[186] 41[165] TAACAGGATTAGCAGAGCGAGGGAGCGCTAATATCAGAAAATTAATTACA 5052[186] 51[165] TAACAGGATTAGCAGAGCGAGGTCACCTTGCTGAACCTAAAATATCTTTA 5034[186] 35[165] TAACAGGATTAGCAGAGCGAGGCGAGAAAACTTTTTCACGCAAGACAAAG 5022[186] 21[165] TAACAGGATTAGCAGAGCGAGGAAGAACGCGAGGCGTTAATTTTATCCTG 5042[186] 42[154] TAACAGGATTAGCAGAGCGAGGAAAATCGCGCAGAGGC 3844[186] 43[165] TAACAGGATTAGCAGAGCGAGGTGCGTAGATTTTCAGGTTGAATACCAAG 5020[186] 19[165] TAACAGGATTAGCAGAGCGAGGTTACCAACGCTAACGAGATTTTTTGTTT 5050[186] 49[165] TAACAGGATTAGCAGAGCGAGGCACTAACAACTAATAGTTAAATCCTTTG 50

21

Table S3. Staple strand sequences for DNA-based membrane channel with modification M2 (on the outside of the central stem. Core staples: grey bars, cholesterol adaptors: red, modification M2: blue.

Start End Sequence Length Color47[119] 2[56] CCACCAGAATGGAACAGCCATACAGGGATAGCGTACAATTTTCCGG 462[116] 39[118] AGGAACCCATACCAGTCAGGACAACAGAATAAGGCTTGCATACGA 450[104] 32[84] AGTACCACCACTACGAATACATGTCAATCCAAAAACAGGAAGATTGTAT 4926[153] 25[160] CATCCTACCAAGAACGGGTAT 2114[104] 34[105] GAGAAACCAACGTAATCGGCCCGCATCGTAACCGTTTCATCA 4252[125] 25[139] GTCAGTTCCTGCAATAGGAATTGATAATCGGCTGTCTTTCCT 4251[147] 25[146] GTTATCTCAAATATCAAACCCGAGCCAGAAGCCGTTATCATT 4252[104] 27[104] CGCCAGCAGGCGGTGCTCAACCCATTAGGCTATATTACAGGC 4251[105] 51[104] TGGTAATATCCACATAAGAGTCTGTCCATCACGCAGCCTTGC 4252[90] 53[83] ACAGGAAAAACGCTGCAGAAG 2151[140] 8[140] GAGGAAGACCTCCCATCAGATAACGTCAGAATTAGTTCATTA 4249[126] 23[125] GACTTTATACATTTGAGGATTAATCAACTTTGAAGACCGCGC 4249[91] 19[104] TATAATCCGTTAGACGCGTACTATGGTTGGCGCTATACTGCG 4249[133] 19[146] CAAACAAATTATCAGGAATTATCATCATGTTTGGATCCAAAT 4248[153] 4[140] AAAGTTTGAGTAACTTCGACATCCAGAGTTGCACCTTCTGTA 4248[104] 5[104] CTTTCCTAGTGAGGTCCCCCTCGTTTTAGGAGTGATCGGTTT 4226[146] 8[133] ATTTACGCTGTTTACAGACGACGACAATCATTTTCAAGGTGA 4224[90] 50[91] AAATATGATAAAACAGAGGTGCATTGCAACTATCGAATTAAC 4246[139] 16[140] TTATACTTCAAAATTATTTGCTCAGATGGAACACCCCCACAA 4224[118] 52[105] CATATAATGCTGTACAGTATTAACACCGGGGAACAATATTAC 4245[112] 16[112] AGAAAGTTAGAACCGTACCTTCATTAGACCGTTGGGAAGAAA 4245[70] 16[77] CTTGACGGGGAAAGTAAATCAGATACATAACAACA 3546[132] 20[119] TCTGAATAAGGAGCTTTTGCGGAACAAAAGTATTATACAAAA 4244[90] 16[98] TCCCTTACCGGCGAACGTGGCGAGAAAGCGAAATCTTACGAGGTTAATA 4943[140] 38[140] TTCGCCTTCATTTCACATCAATTTAATGATAACCTTGAAAAC 420[153] 53[153] GAGGCTGGCCTTTAGGCCGGAATAGAAGAACAAGCCAGCAAA 426[118] 22[105] GCTCGAGGTGAATTTTAATTGCTGAATTCCAAGCGAACCAGA 426[125] 27[125] GATAGCACCACCGACTTGAGCAGAAACCAATCAGTGCCAGCT 425[140] 32[140] CAAGTTTAGACTCCAAGGGCGCACAATCAATTCTTCATAATT 425[105] 18[105] ATCAGCTTGCTTACTCAACAGTCATAGTAGCAACACTATCAT 424[216] 0[203] GCCACCACCGGAACTATTAGCAGTTAAT 285[168] 3[181] CGCGTTTTCATCGGCCTCAGACGCCACC 285[154] 2[154] GCGTCAGAGAGTAAACGTTCCACCACCCCTCTGAATTTACCA 428[132] 28[119] ATTATCACCGTCGCATTTTAGGCAGAGGAAACAACATGTTAA 423[203] 5[216] GCCAGCAAAACAAATGGTAATAAGTTTTTATAAACGTTTGCC 424[139] 43[139] TGGGATTAAAGTTTAGAATAAATTAACTAATATACTAACGGA 425[67] 20[67] AAAAAGGCTCGGAATTGCGAATAAAGATTAAGAGGAAGAAAACGAGAA 4812[139] 40[126] ATGTTAGCGAGGAACCGAACATGAGTGAGAAACAGTACATAA 423[126] 52[126] ACGATCTTTGCTAAGATTAGTGGGAGGTAGTTGAAATATCTG 423[168] 0[161] TCAGAGCACCGCCACCCTCAGACTGTAGCATGAAA 354[195] 0[182] CTCAGAGCCGCCACCATTTTCTCGGAAC 284[153] 22[147] ATGAATTCAGCTACTTAGCGA 2115[98] 21[118] AACTGGCTCATTATGTACCGTACAGCCCTTTCAGCATTCGAGCTTCATT 499[84] 29[83] TAAAATAAAAGACAGCATCGGCAAATCA 2816[90] 49[90] CTAACGGAACGCCAAGACGACTGGATAGTTTAAACGTGTTTT 429[133] 16[133] AAGACAATCAAGAGCTCAGAAAGGGATAAGCAAGAGAATTGA 421[119] 36[119] TCAGAACAGACTTAGCCGGAATGCTCATTCAGTGATATTGAG 423[91] 43[104] GTAGCATGTTTACCAAAGGAAGGCAAAATTGCCCCAGCAGGC 4210[111] 25[118] GGCAAAAGAAGGCAGGATCGTCAGGGAGCTGTTTAATACATTTCGCAAA 4918[153] 50[140] CAGCCTTAAGAAACGCGTCTTACTCGTAATTAGAGCCGTCAA 4210[125] 30[112] ATGGTAATTGTATCATATGCGGCCAACATGTAATAATGCCGG 421[91] 42[91] GTACTCAAGTTTCGTTTTAAGAAACGAACAACGCGGTGAGAC 424[90] 46[91] AAGGAACGACTTCAAATATCGCAAATGCCGTCCAAGGGCGCT 4220[118] 48[105] TAATATTCATTGAACCACCGAGTAAAGAGAACGTATAACGTG 420[195] 3[202] GCCTATTGGTCATAGCCCCCTCGCCTCCAGCCGCC 351[77] 36[77] ATAGGTGACCGAACTGACCAAACAAGAACCGGATACTCCAGC 4222[104] 29[104] CCGGAAGTTAGAGCTCTTAAAAGGCTTGCACCCTCTTCAACC 4212[90] 4[91] TAAGGGATATCACCAAGTGCCAGCCTTTAATTGTAGAATAGA 420[160] 0[168] GTATTAAGCCGTTCCAGTAAGCGTCATATCTGAAA 3512[132] 23[132] CAAACGTCGCCACCAAGGATTAGTAGCGAACCATCCCAATAG 4224[132] 50[133] TTCATCGCAGTGCCACGCTGATCAATCAAGGAATTTAGATAA 4224[139] 19[139] TTTTATTCAAGCAAGACTTGCTGCTATTCCTAATTATCCCAA 4213[91] 41[90] CCCAAATACCAGAATCCGCTCCCTAATGAGTGAGCAATCGGC 4225[84] 22[91] TAGATTTAGTTTGAATGTTTTGGATGGCCAAACTC 3526[97] 6[91] TTTCATTGTCGCTGCAGCTTG 2118[146] 11[146] TACAGAGTGTCGTCCATTTTCCCGCCACACGCAGTGGTGGCA 4234[153] 7[153] AATATATACATATAATTTTGTACATTCAGAAATTAAGCCAGC 4221[119] 9[132] AAATCAAACAACTTCGTAATCAGGATTAAATTACCAGCGCCA 4227[126] 5[139] AATGCAGAACGCGCAGCATGTCATTTGGCCAATGAACAGAAT 4228[104] 0[91] GCCTCAGAGCATAATATGATAAGCAGCGCGTAATGGGCGGAT 4228[118] 1[118] AATTAAGCTGATAATTTTGCGCCAACCTGCGGGGTGCCACCC 4215[140] 35[153] AACAATGCAGATAGACGCAATTCCGGCTGATGCAAATCCAAT 4229[84] 29[76] CCATCAAAGCTAAATCGGTTGTACCAAAAGACAGT 3529[105] 7[118] GTTCTAGCAATAAAAAGGCAAAGAATTACAATAACTTAAAGG 4229[140] 11[139] GAGCCAGCCATATTTAACAACTTATACAAATAGAAACATAAA 4229[147] 27[160] TAATAAGTTCTGTCTCAACAATAGATAA 2830[104] 12[91] AGCTATTTTTGAGAACTAGCACTAAAACAATTGTGTCAATCA 4216[111] 45[111] AATCTACGCATAGTAAGAGCGTTTGATGGTGGTTCGAAGGGA 4231[112] 34[112] ACCCCGGCGAAAGACTGCTCCATGTTAAACCTGTAGCCAGCT 4232[125] 32[126] GTTTAGATAATCAGCGTCTGGCCTTCCTAAATTTAAAAGCCT 4232[139] 37[139] ACTAGAAATGGTTTAATTTCAAAATGCTTAGGTTGTGAGAAG 4218[104] 11[111] AACCCTCTCCACAGAACACTGGGAGGTTCAGACGGTCGAAATCCGCGAC 4933[84] 38[91] CAATAGGAACGCCAATGTGAGAGATGGGTCAGGAATCCCAGTCTCTAGA 4934[104] 30[105] ACATTAATCAAAAATAATTCGAAAAGCCCATATGTAGAGGGT 4234[146] 30[147] TTTAGTTGAAATACCGACCGTCCGGAATACCAGTAAGAATCG 4242[153] 47[153] GAATTATGATTGCTTTTAACGACGTAAACCTGATTATTCCTG 4238[153] 3[153] GAATCCTTGCTTCTAAGTAAGAAATAGCCCACCCTTTTCCAG 4235[112] 10[126] GCATCTGCCAGTACTATATGTTCTTCTGAATACATAATTCAT 4236[76] 13[90] CAGCTTTAGTTGGGTAACGCCAGGGTTTGATCGCATTCATTA 4236[118] 13[132] GGGACGAACGACGGCCAGTTTAAGACGCGGTTATAAGGAAAC 4219[105] 47[118] GAATCGTCATAAAAGGGCGAAAGGAGCGGCTTTGACGAGCAA 4237[63] 12[63] GCGATTACCGGCACAATCTTGCTTTGAA 2837[105] 0[105] TTGTAAACGACAGTACAAAGCCGAGGCGTAGTACCTTTGCTC 4237[140] 1[153] AGTCAATAGTGAATTTTAACCAATAACGCCTTATTCCTCAGA 4238[139] 2[126] ATAGCGATAGCTTAATATATGAAGTTACTAATAAGGCAAGCC 4241[112] 41[111] TTGCGTATTGGGGAATAAGCATAAAGTGTAAAGCCAGGCGGT 4220[83] 9[83] AGTTCAGCCCGAAAAACTAAACAAAAGGGTCGAGAAAACGGG 4239[119] 14[105] GCCGGCAGAGCCAAGCTTGCATGCCTGCCACACAACCCTGAC 4240[153] 17[153] TACCTTTGAAAACAGAGATAACTGAACA 280[202] 3[195] GCCCCCTAGTGTACTAAATCCTCATTAACCACCAG 3541[133] 45[132] GAAACAAAATTACCTGAGCAAGAAACAAAGTAACATACCATA 4241[126] 20[126] AGATGATGTTAAGCCGGGAGACATAAAAATTATTTTGCCAGT 4242[104] 37[104] TTCACCACGGGGAGTGGGGTGACAATTCAGGTCGACACGACG 4242[111] 42[112] TTTTCTTGAAAATCCTGTTACATCGGGAAACGCCAGGGTGGT 4218[186] 17[165] TAACAGGATTAGCAGAGCGAGGTCAAAAATGAAAATAGAAGTCAGAGGGT 5038[186] 38[154] TAACAGGATTAGCAGAGCGAGGAATTAATTTTCCCTTA 3836[186] 37[165] TAACAGGATTAGCAGAGCGAGGTCTGAGAGACTACCTTTTATCAAAATCA 5024[186] 23[165] TAACAGGATTAGCAGAGCGAGGTACCGCACTCATCGAGGCTTATCCGGTA 5028[186] 29[165] TAACAGGATTAGCAGAGCGAGGACAAAAGGTAAAGTAAAGAATATAAAGT 506[186] 6[154] TAACAGGATTAGCAGAGCGAGGCCATTACCATTAGCAA 3814[186] 39[165] TAACAGGATTAGCAGAGCGAGGAAGCCCTTTTTAAGAAGTAAATCGTCGC 5040[187] 40[154] TAACAGGATTAGCAGAGCGAGG?ACAATTTCATTTGAAT 396[230] 2[168] TAACAGGATTAGCAGAGCGAGGCTATTATCATGGCTATGGAAAGCGCAGT 5032[186] 33[165] TAACAGGATTAGCAGAGCGAGGTAAATAAGAATAAACAGTGATAAATAAG 508[186] 9[165] TAACAGGATTAGCAGAGCGAGGAAGGTAAATATTGACGACCGATTGAGGG 5030[186] 31[165] TAACAGGATTAGCAGAGCGAGGCAGTAGGGCTTAATTGTAAAGCCAACGC 5046[186] 45[165] TAACAGGATTAGCAGAGCGAGGAATTCATCAATATAATACAGAAATAAAG 508[230] 1[195] TAACAGGATTAGCAGAGCGAGGAGAACCAAGCCAGATTTGATGATACAGG 5048[186] 47[165] TAACAGGATTAGCAGAGCGAGGAACGTTATTAATTTTAATTATCAGATGA 5012[186] 13[165] TAACAGGATTAGCAGAGCGAGGTGGCATGATTAAGACTGAATACCCAAAA 5010[186] 11[165] TAACAGGATTAGCAGAGCGAGGACCACGGAATAAGTTTAAAGAAACGCAA 5026[186] 7[165] TAACAGGATTAGCAGAGCGAGGAAGAAAAATAATATCCAAAATCACCAGT 5016[186] 41[165] TAACAGGATTAGCAGAGCGAGGGAGCGCTAATATCAGAAAATTAATTACA 5052[186] 51[165] TAACAGGATTAGCAGAGCGAGGTCACCTTGCTGAACCTAAAATATCTTTA 5034[186] 35[165] TAACAGGATTAGCAGAGCGAGGCGAGAAAACTTTTTCACGCAAGACAAAG 5022[186] 21[165] TAACAGGATTAGCAGAGCGAGGAAGAACGCGAGGCGTTAATTTTATCCTG 5042[186] 42[154] TAACAGGATTAGCAGAGCGAGGAAAATCGCGCAGAGGC 3844[186] 43[165] TAACAGGATTAGCAGAGCGAGGTGCGTAGATTTTCAGGTTGAATACCAAG 5020[186] 19[165] TAACAGGATTAGCAGAGCGAGGTTACCAACGCTAACGAGATTTTTTGTTT 5050[186] 49[165] TAACAGGATTAGCAGAGCGAGGCACTAACAACTAATAGTTAAATCCTTTG 50

22

Note S9: cadDNAno design maps of the DNA channels

Fig. S22: caDNAno design for unmodified channel. Cholesterol adaptor sequences are shown in red

23

Fig. S23: caDNAno design for channel modification M1. Cholesterol adaptor sequences are shown in red. Channel modification with the sequence TTTCCGG is shown in dark blue.

24

Fig. S24: caDNAno design for channel modification M2. Cholesterol adaptor sequences are shown in red. Channel modification with the sequence TTTCCGG is shown in dark blue.

References

1. B. Hille, Ion Channels of Excitable Membranes (Sinauer, Sunderland, MA, ed. 3, 2001).

2. L. Song et al., Structure of staphylococcal α-hemolysin, a heptameric transmembrane pore.

Science 274, 1859 (1996). doi:10.1126/science.274.5294.1859 Medline

3. P. W. K. Rothemund, Folding DNA to create nanoscale shapes and patterns. Nature 440,

297 (2006). doi:10.1038/nature04586 Medline

4. S. M. Douglas et al., Self-assembly of DNA into nanoscale three-dimensional shapes.

Nature 459, 414 (2009). doi:10.1038/nature08016 Medline

5. E. S. Andersen et al., Self-assembly of a nanoscale DNA box with a controllable lid.

Nature 459, 73 (2009). doi:10.1038/nature07971 Medline

6. H. Dietz, S. M. Douglas, W. M. Shih, Folding DNA into twisted and curved nanoscale

shapes. Science 325, 725 (2009). doi:10.1126/science.1174251 Medline

7. N. C. Seeman, Nanomaterials based on DNA. Annu. Rev. Biochem. 79, 65 (2010).

doi:10.1146/annurev-biochem-060308-102244 Medline

8. D. Han et al., DNA origami with complex curvatures in three-dimensional space. Science

332, 342 (2011). doi:10.1126/science.1202998 Medline

9. C. E. Castro et al., A primer to scaffolded DNA origami. Nat. Methods 8, 221 (2011).

doi:10.1038/nmeth.1570 Medline

10. See supplementary materials on Science Online.

11. B. Sakmann, E. Neher, Single Channel Recording (Plenum, New York, 1995).

12. G. Baaken, M. Sondermann, C. Schlemmer, J. Rühe, J. C. Behrends, Planar

microelectrode-cavity array for high-resolution and parallel electrical recording of

membrane ionic currents. Lab Chip 8, 938 (2008). doi:10.1039/b800431e Medline

13. C. Danelon, E. M. Nestorovich, M. Winterhalter, M. Ceccarelli, S. M. Bezrukov,

Interaction of zwitterionic penicillins with the OmpF channel facilitates their

translocation. Biophys. J. 90, 1617 (2006). doi:10.1529/biophysj.105.075192 Medline

14. S. Renner, A. Bessonov, F. C. Simmel, Voltage-controlled insertion of single α-

hemolysin and Mycobacterium smegmatis nanopores into lipid bilayer membranes.

Appl. Phys. Lett. 98, 083701 (2011). doi:10.1063/1.3558902

15. J. J. Kasianowicz, E. Brandin, D. Branton, D. W. Deamer, Characterization of individual

polynucleotide molecules using a membrane channel. Proc. Natl. Acad. Sci. U.S.A.

93, 13770 (1996). doi:10.1073/pnas.93.24.13770 Medline

16. H. Bayley, P. S. Cremer, Stochastic sensors inspired by biology. Nature 413, 226 (2001).

doi:10.1038/35093038 Medline

17. S. Howorka, Z. Siwy, Nanopore analytics: Sensing of single molecules. Chem. Soc. Rev.

38, 2360 (2009). doi:10.1039/b813796j Medline

18. C. Dekker, Solid-state nanopores. Nat. Nanotechnol. 2, 209 (2007).

doi:10.1038/nnano.2007.27 Medline

19. A. R. Hall et al., Hybrid pore formation by directed insertion of α-haemolysin into solid-

state nanopores. Nat. Nanotechnol. 5, 874 (2010). doi:10.1038/nnano.2010.237

Medline

20. D. Branton et al., The potential and challenges of nanopore sequencing. Nat. Biotechnol.

26, 1146 (2008). doi:10.1038/nbt.1495 Medline

21. S. Majd et al., Applications of biological pores in nanomedicine, sensing, and

nanoelectronics. Curr. Opin. Biotechnol. 21, 439 (2010).

doi:10.1016/j.copbio.2010.05.002 Medline

22. E. A. Manrao et al., Reading DNA at single-nucleotide resolution with a mutant MspA

nanopore and phi29 DNA polymerase. Nat. Biotechnol. 30, 349 (2012).

doi:10.1038/nbt.2171 Medline

23. N. A. W. Bell et al., DNA origami nanopores. Nano Lett. 12, 512 (2012).

doi:10.1021/nl204098n Medline

24. R. Wei, T. G. Martin, U. Rant, H. Dietz, DNA origami gatekeepers for solid-state

nanopores. Angew. Chem. Int. Ed. 51, 4864 (2012). doi:10.1002/anie.201200688

25. S. Burge, G. N. Parkinson, P. Hazel, A. K. Todd, S. Neidle, Quadruplex DNA: Sequence,

topology and structure. Nucleic Acids Res. 34, 5402 (2006). doi:10.1093/nar/gkl655

26. W. Vercoutere et al., Rapid discrimination among individual DNA hairpin molecules at

single-nucleotide resolution using an ion channel. Nat. Biotechnol. 19, 248 (2001).

doi:10.1038/85696 Medline

27. A. F. Sauer-Budge, J. A. Nyamwanda, D. K. Lubensky, D. Branton, Unzipping kinetics

of double-stranded DNA in a nanopore. Phys. Rev. Lett. 90, 238101 (2003).

doi:10.1103/PhysRevLett.90.238101 Medline

28. J. Mathé, H. Visram, V. Viasnoff, Y. Rabin, A. Meller, Nanopore unzipping of individual

DNA hairpin molecules. Biophys. J. 87, 3205 (2004).

doi:10.1529/biophysj.104.047274 Medline

29. S. M. Douglas et al., Rapid prototyping of 3D DNA-origami shapes with caDNAno.

Nucleic Acids Res. 37, 5001 (2009). doi:10.1093/nar/gkp436 Medline

30. G. Tang et al., EMAN2: An extensible image processing suite for electron microscopy. J.

Struct. Biol. 157, 38 (2007). doi:10.1016/j.jsb.2006.05.009 Medline

31. S. H. W. Scheres et al., Modeling experimental image formation for likelihood-based

classification of electron microscopy data. Structure 15, 1167 (2007).

doi:10.1016/j.str.2007.09.003 Medline

32. T. J. McIntosh, S. A. Simon, Roles of bilayer material properties in function and

distribution of membrane proteins. Annu. Rev. Biophys. Biomol. Struct. 35, 177

(2006). doi:10.1146/annurev.biophys.35.040405.102022 Medline

33. S. J. Marrink, A. H. de Vries, D. P. Tieleman, Lipids on the move: Simulations of

membrane pores, domains, stalks and curves. Biochim. Biophys. Acta 1788, 149

(2009). doi:10.1016/j.bbamem.2008.10.006

34. J. Wohlert, W. K. den Otter, O. Edholm, W. J. Briels, Free energy of a trans-membrane

pore calculated from atomistic molecular dynamics simulations. J. Chem. Phys. 124,

154905 (2006). doi:10.1063/1.2171965 Medline

35. A. Kessel, N. Ben-Tal, S. May, Interactions of cholesterol with lipid bilayers: The

preferred configuration and fluctuations. Biophys. J. 81, 643 (2001).

doi:10.1016/S0006-3495(01)75729-3 Medline

36. S. Gromelski, G. Brezesinski, DNA condensation and interaction with zwitterionic

phospholipids mediated by divalent cations. Langmuir 22, 6293 (2006).

doi:10.1021/la0531796 Medline

37. D. Papahadjopoulos, K. Jacobson, S. Nir, I. Isac, Phase transitions in phospholipid

vesicles: Fluorescence polarization and permeability measurements concerning the

effect of temperature and cholesterol. Biochim. Biophys. Acta 311, 330 (1973).

38. A. Blicher, K. Wodzinska, M. Fidorra, M. Winterhalter, T. Heimburg, The temperature

dependence of lipid membrane permeability, its quantized nature, and the influence of

anesthetics. Biophys. J. 96, 4581 (2009). doi:10.1016/j.bpj.2009.01.062 Medline

39. M. K. Jain, D. Zakim, The spontaneous incorporation of proteins into preformed bilayers.

Biochim. Biophys. Acta 906, 33 (1987). doi:10.1016/0304-4157(87)90004-9 Medline

40. H. D. Herce, A. E. Garcia, Molecular dynamics simulations suggest a mechanism for

translocation of the HIV-1 TAT peptide across lipid membranes. Proc. Natl. Acad.

Sci. U.S.A. 104, 20805 (2007). doi:10.1073/pnas.0706574105 Medline

41. S. Yesylevskyy, S.-J. Marrink, A. E. Mark, Alternative mechanisms for the interaction of

the cell-penetrating peptides penetratin and the TAT peptide with lipid bilayers.

Biophys. J. 97, 40 (2009). doi:10.1016/j.bpj.2009.03.059 Medline

42. K. Pratt, W. Koch, Y. Wu, P. Berezansky, Molality-based primary standards of

electrolytic conductivity (IUPAC technical report). Pure Appl. Chem. 73, 1783

(2001). doi:10.1351/pac200173111783

43. A. Meller, D. Branton, Single molecule measurements of DNA transport through a

nanopore. Electrophoresis 23, 2583 (2002).

44. J. R. C. van der Maarel, Introduction to Biopolymer Physics (World Scientific, Singapore,

2007).

45. A. Aksimentiev, K. Schulten, Imaging α-hemolysin with molecular dynamics: Ionic

conductance, osmotic permeability, and the electrostatic potential map. Biophys. J.

88, 3745 (2005). doi:10.1529/biophysj.104.058727

46. S. Schink et al., Quantitative analysis of the nanopore translocation dynamics of simple

structured polynucleotides. Biophys. J. 102, 85 (2012).

doi:10.1016/j.bpj.2011.11.4011 Medline