SUPPLEMENTARY INFORMATION

Transcription

1 SUPPLEMENTARY INFORMATION Supplementary Information (Methods and Supplementary Figures 1 12) Methods DPN patterning A commercial dip-pen nanolithography (DPN) instrument equipped with an environmental chamber (NScriptor, NanoInk) was used with a one-dimensional tip array with 26 tips (Type F, A26). Optical alignment procedures were used to align the gratings on prefabricated waveguides. Ink preparation (including the use of fluorescentlylabelled, biotinylated and metal chelating lipids) was carried out as described elsewhere 1. The tips were dipped in the inkwells for up to 30 minutes at 23 C and 70% relative humidity. The gratings were written on poly(methyl methacrylate) (PMMA) sheets (HESA Glas HT, Notz Plastics) and treated with isopropanol and ultrapure (Satorius) water (5 minutes sonication both). For total internal reflection fluorescence (TIRF) imaging, PMMA(107 kd, PSS Mainz) was spin-coated from a 15% toluene solution onto glass coverslips to a thickness of 90 nm. The writing process took place at 45% relative humidity and at tip velocities of ranging from 0.1 to 10 µm/s. Substrates before use and the samples after production were stored in a nitrogen atmosphere or vacuum, which minimised hydration-induced spreading. Grating characterisation Atomic force microscopy measurements were done with a Dimension 3100 (Veeco) in a clean room. Microscopic diffraction images were taken by an inverted TE 2000 fluorescence microscope (Nikon) with a 6 objective and a colour camera (Nikon Digital Sight). The gratings were illuminated in transmission through the PMMA substrate by a 150-W halogen cold light source (Schott KL 1500 LCD) at an angle of ca. 70 to the surface normal perpendicular to the grids. Colour spectra were measured with nature nanotechnology 1

2 supplementary information a DC 480 microscope (Leica) with a 10 objective connected to a spectrometer (NanoCalc 2000) and a halogen light source (DH-2000 FHS, Mikropack) through four optical fibres in one wire. Fibres 1 and 2 led to the halogen lamp, which illuminated two 40-µm-diameter spots on the surface for orientation purposes; fibres 3 and 4 led to the spectrometers. A 3-mm-wide slot in the optical light path between the objective and the eye piece functioned as a monochromator. TIRF micrographs were taken by the inverted TE 2000 fluorescence microscope (Nikon) with a 100 objective, the W-TIRF illuminator (Nikon), T-PFS perfect focus unit (Nikon), and a CCD camera (Hamamatsu Orca-ER). Waveguide coupling A single-mode optical strip waveguide of 8-µm width was obtained by exposure of PMMA to DUV radiation through a quartz-chromium mask 2. Although the guiding is relatively weak because the low refractive index contrast (ca ) between that of the exposed surface and the substrate (refractive index of 1.48) 3, this method offers a costefficient way to fabricate optical waveguides for visible wavelengths. The substrate was 1 mm thick. We used a mask aligner (EVG 620) operating at a wavelength of λ DUV = nm applying an exposure dose of 2 J/cm 2. A bake at 70 C for 4 hours after the DUV exposure expelled volatile degradation products and durable radicals from the waveguide. Protein detection Protein-sensing experiments were carried out under liquid conditions on an inverted microscope, and time-lapse diffraction images were taken as described above. A home-made rectangular polydimethylsiloxane (PDMS) barrier was placed around the grating and filled with 200 µl of a buffer solution (PBS) containing 0.5% BSA and allowed to incubate for 10 minutes before addition of the streptavidin solution. Unless 2 nature nanotechnology

3 supplementary information otherwise noted, the streptavidin solution was added by replacement of half (100 µl) of the solution from the fluid cell with a protein solution of twice the target concentration. Optical noise from sources such as reflection from the fluid cell and air-water interface, solvent evaporation, thermal drift, and general background light was cancelled by division of the signal from the target binding gratings by the signal from pure 1,2- dioleoyl-sn-glycero-3-phosphocholine (DOPC) control grating monitored in parallel on the same surface unless otherwise noted. In order to ensure reproducible immersion of the lipid gratings in solution without induction of structural changes, we added the solution within a glove box in a nitrogen atmosphere, where the lyotrophic lipid multilayers are 'frozen' by dehydration. Further stability upon immersion was ensured by addition of the 0.5% BSA blocking agent to the solution. The data were analysed with ImageJ (v 1.38x) and Origin 6.1 by measurement of the intensity of light measured from a grating and subtraction of the background intensity from it. nature nanotechnology 3

4 supplementary information Supplementary Figure 1 Optical spectra of lipid multilayer gratings. a, Colour diffraction image of a grating array made by different tips in a parallel array. b, Higher magnification of the section highlighted in a and the corresponding optical spectra of light diffracted from the gratings of different pitch. Supplementary Figure 2 Fused grating. a, atomic force microscopy (AFM) topographical image of a high grating where the grating lines have fused, producing lower diffraction intensity. b, Cross-section of a. 4 nature nanotechnology

5 supplementary information Supplementary Figure 3 Height variation along a single grating line. The variation in height (or roughness) along a single grating line increases with the thickness of the grating lines and is about 10% of the grating height. Theory for gratings with a variety profiles generally predicts an exponential increase in diffraction efficiency 4 (and therefore intensity here) for gratings with aspect ratios (or height:period ratios) below 1:10. We have, however, observed a linear dependence. We attribute this difference from theory to the increase, with grating height, of variation in the height along a single grating line, or number of defects, and the resulting only linear increase in efficiency with grating height. PMMA, poly(methyl methacrylate). nature nanotechnology 5

6 supplementary information Supplementary Figure 4 Humidity-induced spreading. a, Time-lapse fluorescence images of 10 μm 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) gratings (doped with 1 mol% Rhodamine labelled lipid) upon exposure to humidity. After 15 minutes of exposure to humidity, the grating structure has completely spread producing an ultrathin film of homogeneous fluorescence intensity. The spreading is irreversible, and the spread film does not spread farther upon longer exposure to humidity (observed for ~1 hour). b, Decrease in the intensity of light diffracted from gratings upon exposure to relative humidity above 40% as measured by a fibre spectrometer, as described in Supplementary Figure 1. 6 nature nanotechnology

7 supplementary information Supplementary Figure 5 Spreading and immersion in solution. a, AFM topographical image of a film spread from the edge of a lipid multilayer grating. A 3.6 (± 0.2) nm step is reproducibly observed and is consistent with the thickness of a single lipid bilayer 5,6. Bar = 1 μm. b, Fluorescence micrograph of a rhodamine-labelled and biotinylated grating after immersion in PBS buffer containing 0.5% BSA. The arrow indicates a defect line in the grating that was introduced upon immersion. Bar = 5 μm. c, The same image in b is shown with contrast adjusted so that the spread layer can be seen. d and e, Hypothetical structures of the spread lipid layer on hydrophobic surfaces in air and in aqueous solution that are consistent with our observations and the literature. The tendency for fluid lipid multilayers to spread under solution, as well as to be disrupted upon crossing the air-water interface, poses a challenge for immersion. We found that the DOPC-based gratings could be immersed and remained stable under water for at least several days when patterned on hydrophobic PMMA surfaces and when the gratings were immersed in solution containing 0.5% BSA at humidity well below 40% (e.g., in a nitrogen atmosphere). The reason for immersion at low humidity is to freeze the lipids effectively into place so that they are not disrupted by the air-water interface. Further, nature nanotechnology 7

8 supplementary information the 0.5% BSA blocking agent can be expected to block the background surface and slow the spreading. As the difference between spreading behaviours in humid air and under water on hydrophobic surfaces involves different interfacial energies, spreading can be expected to proceed by different mechanisms, i.e., monolayer and bilayer spreading (Supplementary Fig. 5), the kinetics of which has been quantitatively described as a balance between the spreading force and the resistive drag 7 9. In contrast to an observable change in contact angle typical for spreading of bulk sessile droplets, lipid multilayers tend to spread as molecularly thin and homogeneous layers 7. The hydrophobic chain is well known, e.g. from Langmuir-Blodgett films of amphiphilic molecules, to tend to orient toward the air at the air-water interface as well after transfer onto hydrophilic surfaces 10. Furthermore, evidence has been published that monolayers spread on hydrophobic surfaces under water 8. We therefore attribute the readier spread of lipids in humid air than under water under the conditions of our experiments to the difference between these preferred spreading mechanisms and the friction within these molecularly thin precursor films 8,11 13, as well as to surface pasivation by the BSA blocking agent. 8 nature nanotechnology

9 supplementary information Supplementary Figure 6 Dewetting of biotinylated grating lines upon streptavidin protein binding. a and b, A higher magnification of the same grating shown in Supplementary Figure 5 (as well as supplementary video 3) before and after exposure to streptavidin protein. In this experiment, 20 μl of 500 nm streptavidin protein was added to ~180 μl of buffer. Bars = 5 μm. c, The intensity profile measured along a single grating line indicated by the arrows in a and b is shown plotted against time as a waterfall plot. At t = 20 seconds, certain points along the line can be seen to increase strongly in intensity while the rest of the line decreases in intensity. As these droplets have a higher fluorescence intensity than the initial grating line, we can conclude that the lipids move within the initially continuous grating line to form round droplets rather than simply dissolving into solution. This result indicates that the lipid multilayer film is dewetting from the substrate along the direction of the grating line. nature nanotechnology 9

10 supplementary information Supplementary Figure 7 Conceptual model and simulation of droplet height as a function of the oil-water surface energy. a, Drawing of a cross-section of an adherent oil droplet in water and definition of the variables used for this model. If a circular droplet cross-section is assumed, then the height H can be defined as a function of the contact angle and contact width r by: H 2 r r cot sin 2 The equilibrium contact angle is then defined by the Young equation: cos ( )/ sw so ow where sw, so, ow are the interfacial energies of the solid-water, solid-oil, and oil-water interfaces, respectively. If r = 300 nm (as half of a 600 nm grating period) and sw = 1 and so = 2 are assumed to be reasonable values, as the literature indicates, the droplet height can be simulated as a function of ow as shown in b. The sharp increase in droplet height as ow increases above 1 is consistent with the observed dewetting effect upon 10 nature nanotechnology

11 supplementary information binding of the streptavidin protein to the oil-water interface. This very simplified model is useful for a qualitative understanding of the trends we have observed in terms of fundamental adhesion science. A more quantitative model should take into account relevant nanoscale effects such as line tension, DLVO forces, interfacial curvature, and non-equilibrium processes. Supplementary Figure 8 Intercalation of streptavidin protein into lipid multilayers. To test further the hypothesis that proteins are able to intercalate themselves within the lipid multilayer gratings, we fabricated fused squares (10 10 μm in size) of varying thickness from 5 nm to 110 nm by writing with a pitch of 100 nm. Heights were measured by AFM before immersion. After incubation with fluorescently labelled 50 nm streptavidin for 1 hour, the sample was rinsed with PBS buffer, and the fluorescence intensities of the different squares were measured by fluorescence microscopy. The intensity steadily increased with grating height, up to as much as 700 times the intensity of the fluorescence from protein bound to the spread layer for gratings of 110 nm high. nature nanotechnology 11

12 supplementary information This strong dependence of protein binding on the thickness of fused multilayers further indicates the intercalation of protein within the multilayers. Supplementary Figure 9 Reversible intercalation of his-tagged GFP in lipid multilayer gratings. a, Chemical structure of the his-tag binding NTA-lipid 1,2-dioleoyl-sn-glycero- 3-{[N(5-amino-1-carboxypentyl)iminodiacetic acid]succinyl} (nickel salt). This lipid was mixed with DOPC at 20 mol%, and patterned simultaneously with pure DOPC control. a, Diffraction signal of the his-tag binding lipid gratings shown relative to the 12 nature nanotechnology

13 supplementary information intensity of the control gratings. Time points indicated are (1) injection of His-GFP protein to a final concentration of 12.5 μm, (2) washing with PBS buffer, (3) addition of imidizole to unbind the His-GFP, (4) washing with PBS buffer, (5) second addition of His-GFP, (6) washing with PBS buffer, and (7) second addition of imidazole. Although, upon the second addition of his-gfp, the signal did not return to the original strength of the first addition, the data indicate the semi-reversibility of the his-gfp binding. c and d, Excerpts from the time-lapse diffraction series used to generate the plot in a. c is before the first addition of His-GFP, and d is after. Arrows indicate the column of NTA-lipid gratings; the other column is control gratings composed of pure DOPC. The 5th row in the his-fused grating does not diffract but shows intensity in the green and blue channel because of incoherent scattering and fluorescence from the GFP protein. Because of the GFP fluorescence, we monitored the red channel. e and f are fluorescence micrographs of the GFP protein. e is after protein binding and PBS wash, and f is after protein unbinding by imidazole treatment. The presence of the his-gfp is apparent in both cases, especially in the thicker fused grating, further indicating that it has been encapsulated within the lipid multilayers. All images are of the same area, and the scale bar = 20 μm. As intercalation is observed in the micromolar range (or mg/l, considering molecular weight of GFP), it is suitable for multiplexed analyte detection in this concentration range, e.g., for quality control in protein purification or multiplexed detection of blood components in this range such as cholesterol, amino acids, lipids, and sugars. nature nanotechnology 13

14 supplementary information Supplementary Figure 10 Detection of streptavidin protein at 500 pm. Bar = 20 μm. After 15 minutes of incubation at 500 pm, we were unable to distinguish the signal for 5 nm from that of 0.5 nm significantly but were able to do so after 90 minutes of incubation. The figure shows the result of dividing an image of light diffracted from two gratings (left arrow, DOPC control; right arrow, biotinylated grating) taken after 90 minutes of incubation by an image taken after only 5 minutes of incubation. The decrease in diffraction intensity can be clearly seen in the biotinylated grating. The slight decrease in intensity of the control grating can be explained by slow spreading of the grating under water. Detection of analytes in serum at concentrations of nm is useful for example for determining the presence of hormones 14,15. Rapid prototyping of more complex photonic structures and improvements in the lipid chemistry are promising approaches to lowering the limit of detection. 14 nature nanotechnology

15 supplementary information Supplementary Figure 11 Protein detection in a complex mixture. The diffraction intensities of a DOPC control grating and a biotinylated grating were monitored before and after addition of 10% fetal calf serum at time = 0 (left arrow) and streptavidin protein (added to yield a final concentration of 50 nm) at time = 10 minutes (right arrow). The raw data are shown without normalisation relative to the control so that the two signals can be compared. No significant difference between the control and the biotinylated grating can be observed after addition of the serum, yet a strong difference can be observed after addition of the streptavidin. Lipids are highly resistant to non-specific protein binding 16 and the data indicate the possibility of using lipid multilayer gratings for specific detection of protein in complex mixtures such as serum. The diffraction micrographs for these data are shown in Supplementary video 3. nature nanotechnology 15

16 supplementary information Supplementary Figure 12 Dependence of the grating response on grating height. Gratings of different heights were fabricated on the same surface and exposed to streptavidin protein. The grating height was determined from the relative diffraction intensity; the intensity/height data in Fig. 2c were used as a calibration. In this experiment, 20 μl of 500 nm streptavidin protein was added to the ~180 μl of buffer in the fluid cell. The data are plotted on an exponential time scale and are broken into three phases. Phase I indicates the diffraction intensity before addition of the protein. During phase II (<30 s after protein addition) a quick dewetting response resulted in lowered intensity for the high gratings, whereas a slower intercalation process resulted in the increase in intensity over about 5 minutes (phase III) for all gratings. Dewetting understandably occurs on shorter time scales than intercalation because the protein need only bind to the surface of the lipid multilayer to change the interfacial energy rather than having to fill the multilayer volume to cause a measurable intercalation effect. The occurrence of dewetting only in the higher gratings is consistent with the observation that higher gratings tend to exhibit a larger variance in their height, as shown in Supplementary Fig. 3. The intensity of light diffracted from control gratings remained 16 nature nanotechnology

17 supplementary information constant. This experiment indicates the importance of precisely controlling lipid grating height. Supplementary References 1. Sekula, S. et al. Multiplexed lipid dip-pen nanolithography on subcellular scales for the templating of functional proteins and cell culture. Small 4, (2008). 2. Henzi, P., Rabus, D. G., Bade, K., Wallrabe, U. & Mohr, J. Low-cost single mode waveguide fabrication allowing passive fiber coupling using LIGA and UV flood exposure. Proc. SPIE 5454 (2004). 3. Punke, M. et al. Coupling of organic semiconductor amplified spontaneous emission into polymeric single-mode waveguides patterned by deep-uv irradiation. IEEE Photonic Tech. L 19, (2007). 4. Moharam, M. G. & Gaylord, T. K. Diffraction analysis of dielectric surface-relief gratings. J. Opt. Soc. Am. 72, (1982). 5. Schuy, S. & Janshoff, A. Thermal expansion of microstructured DMPC bilayers quantified by temperature-controlled atomic force microscopy. Chemphyschem 7, (2006). 6. Tristram-Nagle, S., Petrache, H. I. & Nagle, J. F. Structure and interactions of fully hydrated dioleoylphosphatidylcholine bilayers. Biophys. J. 75, (1998). 7. Nissen, J., Gritsch, S., Wiegand, G. & Radler, J. O. Wetting of phospholipid membranes on hydrophilic surfaces concepts towards self-healing membranes. Eur. Phys. J. B 10, (1999). 8. Sanii, B. & Parikh, A. N. Surface-energy dependent spreading of lipid monolayers and bilayers. Soft Matter 3, (2007).9. Radler, J., Strey, H. & Sackmann, E. Phenomenology and kinetics of lipid bilayer spreading on hydrophilic surfaces. Langmuir 11, (1995). 10. Chen, X. D. et al. Langmuir-Blodgett patterning: a bottom-up way to build mesostructures over large areas. Accounts Chem. Res. 40, (2007). 11. Leger, L. & Joanny, J. F. Liquid spreading. Rep. Prog. Phys. 55, (1992). 12. De Gennes, P. G. Wetting: statics and dynamics. Rev. Mod. Phys. 57, (1985). 13. Nissen, J., Jacobs, K. & Radler, J. O. Interface dynamics of lipid membrane spreading on solid surfaces. Phys. Rev. Lett. 86, (2001). 14. Rothwell, P. M., Udwadia, Z. F. & Lawler, P. G. Thyrotropin concentration predicts outcome in critical illness. Anaesthesia 48, (1993). 15. Isurugi, K., Kanazawa, M., Yanaihara, T. & Kambegawa, A. Responses of serum levels of testicular-steroid hormones to HCG stimulation in patients with prostatic-cancer and benign prostatic hypertrophy. Prostate (1981). nature nanotechnology 17

18 supplementary information 16. Ross, E. E. et al. Planar supported lipid bilayer polymers formed by vesicle fusion. 2. Adsorption of bovine serum albumin. Langmuir 19, (2003). 18 nature nanotechnology