Micropatterning Fluid Lipid Bilayers on Solid Supports

See allHide authors and affiliations

Science  31 Jan 1997:
Vol. 275, Issue 5300, pp. 651-653
DOI: 10.1126/science.275.5300.651


Lithographically patterned grids of photoresist, aluminum oxide, or gold on oxidized silicon substrates were used to partition supported lipid bilayers into micrometer-scale arrays of isolated fluid membrane corrals. Fluorescently labeled lipids were observed to diffuse freely within each membrane corral but were confined by the micropatterned barriers. The concentrations of fluorescent probe molecules in individual corrals were altered by selective photobleaching to create arrays of fluid membrane patches with differing compositions. Application of an electric field parallel to the surface induced steady-state concentration gradients of charged membrane components in the corrals. In addition to producing patches of membrane with continuously varying composition, these gradients provide an intrinsically parallel means of acquiring information about molecular properties such as the diffusion coefficient in individual corrals.

The patterning of surfaces into defined regions of contrasting properties has recently received considerable attention. Successes in this field include the development of micrometer-scale patterns in surface wettability (1) and microcontact printing, which has produced patterned self-assembled monolayers with submicrometer features (2, 3). At the same time, light-directed synthesis has been used to generate spatially addressable combinatorial libraries of DNA and protein sequences on solid substrates (4). Here, we describe patterning of surfaces with properties resembling those of living cells: fluid bilayer membranes on solid supports. This work has been stimulated by the desire to create integrated devices capable of manipulating molecules in a bilayer membrane and to pattern spatially addressable libraries of chemically distinct, fluid membrane patches.

Supported membranes are self-assembling, two-dimensional (2D) fluid systems (Fig. 1). The bilayer membrane consists of two opposed leaflets of phospholipid molecules and is the basic structure central to all living cell membranes. Bilayers on solid supports were originally developed for studies of interactions between living cells, where they have proven highly useful (5, 6). Supported membranes can be formed by spontaneous fusion of lipid bilayer vesicles with an appropriate hydrophilic surface such as oxidized Si (7, 8). Interactions between membranes and surfaces involve electrostatic and hydration forces as well as attractive contributions from long-range van der Waals forces. An energetic minimum tightly traps the membrane near the surface. Supported bilayers are typically separated from the solid substrate by a thin (∼10 Å) film of water (911) and retain many of the properties of free membranes, including lateral fluidity. The fluidity is long range, with mobile components of both leaflets of the bilayer freely diffusing over the entire surface of the substrate.

Fig. 1.

Schematic diagram of a supported bilayer partitioned by a microfabricated barrier. The size of the membrane is exaggerated to illustrate its structure; actual membranes are typically 50 Å thick and are separated from the surface by a ∼10 Å layer of water. The height of the barriers ranged from 100 Å (Au) to 1 μm (photoresist) with widths between 1 and 30 μm. The functionality of a barrier depends on the surface properties of the material, not the topography.

The lateral fluidity of supported membranes is a key feature that distinguishes them from other surfaces. Although fluidity provides the membrane with a variety of unique properties, it presents an intrinsic difficulty in that membrane components are continually mixing. Micropartitioning of fluid membranes allows fabrication of surfaces that preserve their distinctive properties while gaining precise control over the diffusive mixing and flow of membrane components.

Oxidized Si and fused silica (SiO2) wafers are suitable substrates for supported membranes (12). Standard positive photoresist (13) was spun onto wafers to a thickness of 1 μm. The wafers were exposed to ultraviolet light through a photolithographic mask and developed. After a 3-min etch in Ar plasma, the photoresist-partitioned surfaces were ready for membrane deposition. For fabrication of Al2O3 and Au barriers, the photoresist was exposed with the inverse mask and used as a template for lift-off of electron beam-evaporated thin films. Supported bilayers were formed by fusion of small unilamellar vesicles (diameter ∼25 nm) consisting primarily of L-α-phosphatidylcholine (PC) molecules doped with 1 mol% of the fluorescently labeled lipid Texas Red DHPE (14). The vesicle fusion process is quite general, accommodating a variety of lipid compositions as well as the incorporation of membrane proteins (7, 15, 16). Vesicles spontaneously assembled from solution in a matter of seconds to form a continuous single bilayer on the exposed portion of the SiO2 substrate. Excess vesicles were rinsed away while the membrane was kept under water at all times. Supported membranes were stable under water and retained their uniformity and fluidity for weeks.

Membrane-coated surfaces can be viewed directly by fluorescence microscopy and appear uniform to the diffraction-limited resolution of the microscope. A supported membrane was partitioned into an array of isolated membrane patches by micropatterned lines of photoresist (Fig. 2A). The photoresist barriers appear dark against the fluorescence from the Texas Red lipid probe in the membrane (17). These 200 μm by 200 μm corrals are part of a 1-cm2 array of 2500 identical regions. The smallest corrals we observed were 5 μm by 5 μm, with 1-μm-wide barriers (18). About 2.8 million such corrals are contained in the 1-cm2 array.

Fig. 2.

(A) Epifluorescence photograph of four corrals (200 μm by 200 μm) of a fluid, supported bilayer on an oxidized Si wafer patterned with photoresist. The photoresist barriers appear as dark lines against the bright fluorescence from the Texas Red DHPE lipid probe incorporated in the membrane. (B) An array of corrals (100 μm by 100 μm) in which certain individual corrals were photobleached. Diffusive mixing of molecules within each corral has caused the circular bleached spot to spread, filling the square corral, while the photoresist barriers have prevented mixing between separate corrals (19). (C) Epifluorescence photograph of an array of corrals (50 μm by 50 μm, with barriers 2 μm wide) in which certain individual corrals were selectively photobleached with varying exposure times. Corrals were exposed in increments of 30 s, with the darkest corral receiving a 3-min exposure. We created similar patterns with varying concentrations of the fluorescent probe with a single exposure by masking variable fractional areas in different corrals.

Long-range fluidity within the corrals could be observed by fluorescence recovery after photobleaching. A circular spot of light was used to sequentially photobleach the fluorescent probe molecules in five individual corrals (Fig. 2B). Long-range fluidity of the membrane is indicated by the spreading of the bleached region to fill each square corral. If the membrane were not fluid, a circular bleached (dark) region would remain (19). Lines of photoresist are effective barriers to lateral diffusion, as demonstrated by their ability to confine the photobleached region of membrane within the square corral. Bleach patterns such as those shown in Fig. 2, B and C, were stable for many days, whereas spots photobleached into membranes without any such barriers diffused away completely in about 30 min (20). The barriers shown in Fig. 2, A and B, are 10 μm wide; those in Fig. 2C are 2 μm wide, and the narrowest lines tested to date, 1 μm, were equally functional as diffusion barriers (18). These results demonstrate that, within a given corral, diffusive mixing occurs as in free fluid membranes and that there is no intermixing between different corrals.

A noteworthy point is that the barriers used to confine the fluid membrane are not mechanical. Supported bilayers are highly conformal and will follow the contour of corrugated surfaces without disruption. We have deposited membranes on SiO2 substrates with topographies identical to those of the photoresist-patterned surfaces and observed no difference in lateral diffusion properties compared with flat, unpatterned SiO2 surfaces. Thus, it is the difference in the surface electrostatic and chemical properties of the substrate and barrier material that give rise to the partitioning of the membrane. We have also found that Au and Al2O3 are effective as barrier materials. In principle, any substance that does not support fluid membranes can form an effective barrier to lateral diffusion (21). However, photoresist is the easiest material to pattern with a photomask, and nonconductive media are most convenient for electrophoretic experiments (see below).

Micropartitioned membranes are suitable substrates for further lithographic patterning. A light-directed chemical transformation (photobleaching of the fluorescent probe) can be used to selectively alter the composition of the membrane in individual corrals. Corrals were photobleached sequentially with varying exposure times to generate an array of patches with differing concentrations of the fluorescent probe molecule (Fig. 2C). Because of the fluidity of the membrane, such an array can also be created with a single exposure if the area of a bleach spot is varied within individual corrals. Lateral diffusion will make the composition within each corral homogeneous in a matter of minutes. We have found that creating variable composition arrays by either of these techniques works equally well. However, the single-exposure method has the advantage of being compatible with the use of a photomask to pattern large numbers of corrals simultaneously.

Another way of creating patterns in the supported bilayer that takes advantage of both fluidity and confinement is electrophoretic redistribution of charged membrane components. When an electric field is applied parallel to the membrane, the charged molecules drift in the plane of this 2D fluid (22, 23). The PC molecules that constitute the bulk of the membrane are neutral, and thus it is expected that they will be unaffected by the field. Steady-state, electric field-induced concentration profiles (24) of the negatively charged fluorescent probe can be clearly seen to build up against the barriers in confined regions of the membrane (Fig 3, A and B). Patterns of photoresist barriers are also fluorescent and appear as an intermediate shade of red in these photographs. Images of the same region with the electric field in opposite polarities are shown, demonstrating the reversibility of these profiles. At a field strength of 15 V/cm, it took ∼25 min for a steady-state concentration gradient to form in the 200-μm-wide corrals; a complete reversal in the direction of the profile took the same amount of time. The profiles could be switched numerous times without any apparent effect on the membrane or the barriers. When the field was turned off, it took ∼30 min for gradients in the 200-μm-wide corrals to relax back to uniformity by diffusive mixing.

Fig. 3.

Epifluorescence photographs of electric field (E)-induced concentration gradients of the negatively charged fluorescent probe lipid (Texas Red DHPE) in membranes confined by microfabricated corrals. Each of the upper two corrals is 200 μm wide. Reversal of the field causes the direction of the profile to reverse. These are steady-state gradients at a field strength of 15 V/cm. The time required to switch from one direction (A) to the other (B) is ∼25 min.

The electrophoretic response can also be used to acquire physical information about the membrane in individual corrals. The profile of the steady-state gradient in dilute systems can be quantitatively described by competition between lateral diffusion and field-induced drift (24, 25). Image analysis thus provides an intrinsically parallel means of monitoring these molecular properties in numerous corrals. More complex profiles have been observed in highly concentrated domains of membrane-tethered proteins; such systems may be useful in the study of intermolecular interactions (15).

The techniques described here provide new opportunities for the study of physical and biological properties of membranes. In particular, secondary lithographic modification of the membrane could be generalized by combination with light-directed synthesis (4). Spatially addressable molecular libraries (peptide sequences, for example) could be synthesized and displayed on the surface of confined patches of fluid membrane. In some ways, this approach is analogous to phage display, except that here the peptide sequence is defined by its location in the array, thus providing immediate identification. Such libraries would be especially useful for cell screening because of the native-like surface provided by the supported membrane. For example, purified major histocompatibility complex protein incorporated into a supported membrane can effectively replace the antigen-presenting cell in the presentation of a preprocessed antigen to a helper T cell (5, 6). Other studies have underscored the similarities between supported membranes and cell surfaces by demonstrating a substantial effect from the lateral mobility of adhesion proteins in the supported bilayer on the specific binding of cells (26, 27). Additional possibilities include the construction of electrically or photochemically addressable barriers and further development of photolithographic techniques to directly change the physical, chemical, or biological properties of the fluid membrane patches.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
View Abstract

Stay Connected to Science

Navigate This Article