Lipid Bilayer Coated Al2o3nanopore Sensors: Towards a Hybrid Biological Solid-State Nanopore

Lipid Bilayer Coated Al2O3Nanopore Sensors: Towards A Hybrid Biological Solid-State Nanopore

B. M. Venkatesan1,2, J. Polans1,2, J. Comer 3,4, S. Sridhar 1,2, D. Wendell 6,7,

A. Aksimentiev3,4& R. Bashir 1,2,3,5(Corresponding author)

Email:

1 Department of Electrical and Computer Engineering, 2 Micro and Nanotechnology Laboratory, 3 Department of Physics, 4 Beckman Institute,5Department of Bioengineering,

University of Illinois at Urbana Champaign, Illinois, IL, USA, 61820

6 College of Medicine, 7 College of Engineering,

University of Cincinnati, Cincinnati, OH, USA 45221

Supplementary Information

Molecular Dynamics Simulations

Generation of the solid-state membrane

To produce an atomically detailed model of a solid-state membrane, silicon and oxygen atoms (with a 1:2 ratio) were randomly distributed within a hexagonal prism with an inscribed diameter of 11.0 nm and a thickness of 10.0 nm along the z axis. Atoms were also not placed within the region that would form the nanopore. This nanopore ran down the center of the two hexagonal faces along the z axis and had a circular cross section and a parabolic profile such that the diameter of the circle varied from 3.6 nm at the center of the membrane to 5.0 nm at z = –5.5 and +5.5 nm.

The system first underwent energy minimization and was then melted and annealed using the following regimen: 20 ps at 7000 K, 20 ps at 5000 K, 50 ps at 2000 K, 50 ps at 300 K. The simulations were performed in NAMD using a 1 fstimestep without the use of multiple timestepping. Interactions between pairs of atoms were calculated using the BKS potential energy functions (van Beest et al. 1990; Vollmayr et al. 1996). The electrostatic portion of the BKS potential was calculated using the particle-mesh Ewald method(Batcho et al. 2001) for full long-range electrostatics. A smooth 0.54–0.55 nm cutoff was used for the non-electrostatic and explicit short-range electrostatic portions of the interaction. The temperature was maintained at each phase of the annealing process by a Langevin thermostat using a damping constant 5 ps–1. An external potential was applied to maintain the shape of the membrane and pore during the annealing process (Wells et al. 2007).

The solid-state membrane used in all subsequent simulations was created using the final atom positions from the above annealing simulations. Covalent bonds (with bond constants of 6.95 nN/nm) were added between all pairs of silicon and oxygen atoms closer than 0.22 nm. The atoms of the membrane were also restrained to their final positions by the energy function U(ri)=K|ri–Ri|2, where Ri is the final position of atom i and K=13.9 nN/nm. The charges of the silicon and oxygen atoms in the solid-state membrane were set to 1.0e and –0.5e, respectively, in accord with the force field of Cruz-Chu et al.(Cruz-Chu et al. 2006), except that the charge of the oxygen atoms on the surfaces (in contact with water), were uniformly shifted to produce four solid-state membrane having different surface charge densities (–1, 0, +1, or +2 e/nm2).

Lipid Surface Separation

Lipid surface separation, as plotted in figures 3b and 3c of main text, was calculated according to the following equation: , where the z axis is normal to the solid-state membrane surface, ,is the z coordinate of the center of mass of the nitrogen atom in the choline moiety of DOPC, is the z coordinate of the center of mass of the oxygen atoms on the surface of the solid-state membrane, and 0.25 nm is an offset to account for the finite size of the choline moiety and oxygen atoms.

Analysis of lipid diffusivity

Measuring the lateral diffusivity of lipids in MD simulations is complicated by collective motion of lipids (Wohlert et al. 2006; Falck et al. 2008) and different regimes of diffusion which depend on the timescale (Edholm 2008; Flenner et al. 2009). To facilitate comparison, we followed a protocol similar to Siu et al.(Siu et al. 2008). First, the trajectories of each DOPC nitrogen atom (with a sampling rate of ~0.2 ns) were extracted from all simulations of the systems discussed in this work. As in Siu et al.(Siu et al. 2008), the trajectories were generated relative to the center of mass of the given lipid’s bilayer leaflet to partially correct for collective motion. The trajectories were then cut into 11ns segments. The trajectories were categorized by whether the associated lipid was in the top or bottom leaflet. To obtain the values shown in figure 3c, the trajectories for each leaflet were placed into six bins (centered on 0, 0.3, 0.6, 0.9, 1.2, and 1.5 nm) depending on the mean separation between the entire DOPC bilayer and the solid-state surface during the given trajectory, which was calculated as described in the main text. The square displacement of the nitrogen atoms from their position at the beginning of each trajectory was calculated and averaged for all trajectories in a given bin. A linear least-squares fit of the mean-square displacement was performed considering only the last 6 ns of each trajectory. The diffusivity values, given by 1/4 of the slope of this line, are plotted in figure 3c, with each data point representing between 900 to 6000 individual 11 ns lipid trajectories. The error bars in figure 3c show the values of the diffusivity obtained by randomly partitioning the trajectories into two sets and applying the same fitting procedure to each set. Presumably the lipids above the pore have different kinetic behavior than those above the solid-state surface; however, this effect was not accounted for due to limited statistics.

Additional simulations

The initial conditions of the alternate simulations of the four systems characterized in supplementary figure 3a began with a different set of random velocities than those characterized in figure 3b. Three additional simulations were performed to investigate the effect of surface charge reversal. At t = 18.4 and 171.3 ns for the –1 e/nm2 system and t = 34.1 ns for the neutral (0 e/nm2) system, the simulations were halted and the original membrane and associated counterions were replaced by a +2 e/nm2 surface with Cl–counterions, whilst maintaining the original configurations of the lipids. As shown in supplementary figure 3b, the introduction of an instantaneous positive surface charge resulted in DOPC bilayers moving rapidly away from the surface. This was particularly evident for the two systems in which the initial lipid–surface separation was less than 0.3 nm, confirming a repulsive force between the DOPC bilayer and the positively charged surface for small lipid–surface separations.

Lipid Bilayer Movies

The movies show molecular dynamics simulations of a DOPC bilayer above a solid-state surface having various surface charge densities. They illustrate only a portion of the complete system. See the main text for the actual dimensions. The choline groups of the DOPC are illustrated by red spheres; the remainder of the DOPC is illustrated by pink tubes. The average positions of the atoms of the solid-state surface are illustrated by blue and gray spheres; the small motions of the atoms about these average positions are not shown. Water molecules and ions are also not shown. A fog effect is used to give a sense of depth. DOPC molecules may appear to disappear into the fog as the visible image of the bilayer drifts in xy plane; however, the DOPC bilayer always fills the xy plane due to the periodic boundary conditions.

Movie 1: (10544_2011_9537_MOESM3_ESM.mpg)

171 ns simulation of a DOPC bilayer above a solid-state surface having a charge of –1 e/nm2. With this value of the surface charge, the bilayer quickly approaches the surface.

Movie 2: (10544_2011_9537_MOESM2_ESM.mpg)

77 ns simulation of a DOPC bilayer above a net-neutral solid-state surface. Due to a high proportion of negatively charged oxygen atoms at the solid-state surface, the qualitative behavior of the bilayer is similar to simulations in which a negatively charged solid-state surface is used.

Movie 3: (10544_2011_9537_MOESM4_ESMmpg)

162 ns simulation of a DOPC bilayer above a solid-state surface having a surface charge density of +2 e/nm2. Unlike the in the simulations with negatively charged or neutral surfaces, the bilayer does not approach the surface, but slowly drifts away.

Supplementary Figure 1

Supplementary Figure 1 Qualitative FRAP results on Al2O3 (Scale bar is 100 μm in all images). 400 nm extruded DOPC vesicles containing 1% fluorescent lipid were incubated and ruptured on the Al2O3 surface using osmotic pressure and Ca2+ to form fluid bilayers (a) Negatively charged TR-DHPE fluorescent lipid, clear fluorescence recovery is seen in the photobleached region (b) Zwitterionic (net neutrally charged) NBD-PC fluorescent lipid, clear fluorescence recovery is seen in the photobleached region. Results confirm that bilayer formation on Al2O3 is dependent on enhanced DOPC lipid-lipid interactions and DOPC-surface interactions in the presence of high salt and Ca2+ and is independent of the fluorescent lipid molecule used.

Supplementary Figure 2

Supplementary Figure 2 (a) AFM scans of SiO2, ALD deposited Al2O3 and sputtered TiO2 with rmsroughnesses of 152 pm, 241 pm and 941 pm respectively. Lipid diffusion coefficients of 1.75 μm2/s, 2.7a μm2/s and 2.4 μm2/s were extracted on these surfaces respectively suggesting that on the sub-nanometer scale, the chemical properties of the substrate are more important than surface topography (b) XPS spectra from SiO2, Al2O3 and TiO2 coated substrates confirming the correct composition of each substrate prior to vesicle incubation and rupture (c) Water contact angle measurements on SiO2, Al2O3, TiO2 and PDMS before and after surface treatment with 1 min O2 plasma at 100 W. Al2O3 substrates remained hydrophilic (contact angle of < 5°) even 3 hours after treatment.

Supplementary Figure 3

Supplementary Figure 3 (a) The average separation between the DOPC lipid bilayer and the solid-state surface as a function of time for the four differently charged surfaces. The simulations characterized here were identical to those used to produce figure 3b of the main text, except that the atoms began with a different set of random velocities. (b) The average separation between the DOPC lipid bilayer and the solid-state surface as a function of time in three simulations in which the areal charge density of the membrane was changed to +2 e/nm2 from an initial value of either –1 e/nm2 or 0 e/nm2 during the simulation. The original simulations from which the initial conditions were derived are also shown. The results seem to suggest electrostatic repulsion between the bilayer and the positively charged +2 e/nm2 surface at close distances. (c) The component of the electric dipole moment of the DOPC headgroup perpendicular to the plane of the bilayer as a function of the distance between the lipids and the solid state surface. The data for two different surfaces are shown. The dipole points away from the bilayer when the bilayer is far from either surface. However, the dipole reverses direction when the bilayer is placed near the positive surface.

Supplementary Figure 4

Supplementary Figure 4 (a) Experimental setup used for typical electrical measurements integrating perfusion setup with nanopore chip. After vesicle incubation, DI perfusion followed by 1M KCl, 10 mMTris, 5mM CaCl2 pH 8.0 through this setup induced bilayer formation on Al2O3nanopores (b) Comparison of the electrical properties of a BLM formed in a Teflon aperture, solid-state membrane with no pore and a bilayer grafted Al2O3nanopore. IV characteristics were fitted to extract a resistance that is comparable in all three cases, confirming that a GΩ seal is achieved using a bilayer grafted Al2O3nanopore.

References

P. F. Batcho, D. A. Case and T. Schlick, The Journal of Chemical Physics, 115(9): 4003-4018 (2001)

E. R. Cruz-Chu, A. Aksimentiev and K. Schulten, The Journal of Physical Chemistry B, 110(43): 21497-21508 (2006)

O. Edholm, Computational Modeling of Membrane Bilayers, 60: 91-110 (2008)

E. Falck, T. Rog, M. Karttunen and I. Vattulainen, Journal of the American Chemical Society, 130(1): 44-+ (2008)

E. Flenner, J. Das, M. C. Rheinstadter and I. Kosztin, Physical Review E, 79(1): - (2009)

S. W. I. Siu, R. Vacha, P. Jungwirth and R. A. Bockmann, The Journal of Chemical Physics, 128(12): 125103-12 (2008)

B. W. H. van Beest, G. J. Kramer and R. A. van Santen, Physical Review Letters, 64(16): 1955 (1990)

K. Vollmayr, W. Kob and K. Binder, Physical Review B, 54(22): 15808 (1996)

D. B. Wells, V. Abramkina and A. Aksimentiev, The Journal of Chemical Physics, 127(12): 125101-10 (2007)

J. Wohlert and O. Edholm, Journal of Chemical Physics, 125(20): - (2006)