Proteins on Supported Lipid Bilayers Diffusing around Proteins Fixed on Acrylate Anchors

Mobility of proteins and lipids plays a major role in physiological processes. Platforms which were developed to study protein interaction between immobilized and mobile proteins suffer from shortcomings such as fluorescence quenching or complicated fabrication methods. Here we report a versatile platform comprising immobilized histidine-tagged proteins and biotinylated proteins in a mobile phase. Importantly, multiphoton photolithography was used for easy and fast fabrication of the platform and allows, in principle, extension of its application to three dimensions. The platform, which is made up of functionalized polymer structures embedded in a mobile lipid bilayer, shows low background fluorescence and allows for mobility of arbitrary proteins.

Andor Technology Ltd., UK) with 512 x 512 active pixels and single photon sensitivity. A three axes piezo stage (TAO 3-axis sample scanning module, JPK Instruments, Germany) with 100 µm x 100 µm x 10 µm travel range in x, y and z direction was used for sample positioning. Illumination time of 5 ms is controlled by acousto-optic modulators (1205C-2, Isomet Corp., USA). The delay between two consecutive images was 100 ms. The system is controlled by a custom-made program (LabView, National Instruments Corp., USA).

Materials.
Carboxy-functional polymer structures were fabricated using a photoresist consisting of two different acrylates and a photoinitiator. The monomers are pentaerythritol triacrylate (PETA, Sigma Aldrich Co., USA) containing 300 ppm-400 ppm monomethyl ether hydroquinone (MEHQ) as inhibitor, and 20 wt% 2-carboxyethyl acrylate (CEA, Sigma Aldrich Co., USA) containing 900 ppm-1100 ppm MEHQ. 1 wt% Irgacure ® 819 (IC 819, BASF Schweiz AG, Switzerland) was added as photoinitiator. Structures without carboxy-functionality were fabricated using pure PETA with 1 wt% of IC 819. The chemical structures of the photoresist ingredients can be found in Figure S1. Polymerization is achieved by focusing a femtosecond laser beam into the photoresist. No additional pre-preparation methods (like for e.g. spin-coating) are required. The polymer anchors for immobilization of proteins were fabricated using 4.5 mW of 780 nm femtosecond pulses and an illumination time of 20 ms per anchor. The polymer grids were fabricated as follows 2 : First, a pure PETA grid was fabricated using an excitation power of 5.5 mW. The remaining pure PETA photoresist was diluted (100x amount) with a CEA containing photoresist (PETA with 20 wt% CEA and 1 wt% IC 819). The carboxy functional grids were written using an excitation power of 5 mW. All laser powers were measured directly before entering the objective lens. After polymerization, the structures were developed by rinsing with acetone (Merck KGaA, Germany) and drying with a nitrogen stream. Samples were stored in air at room temperature.
Buffers were stored at 4°C. Phosphate buffered saline (pH 7.4) was used as received (Merck KGaA, Germany).  The average number of GFP molecules for the experiment (see Figure 2 in the manuscript) was ~ 0.18 molecules/pixel (PETA + 20 wt% CEA with 1 wt% IC 819) and for the negative control (PETA with 1 wt% IC 819) the signal is ~ 0.02 molecules/pixel (pixelsize 160 nm, 5 ms illumination time, excitation intensity ~ 3.3 kW/cm 2 ). For the negative control shown in Figure S3b, the signal is 0.009 GFP molecules/pixel, for the negative control shown in Figure S4b S7 the signal is 0.02 GFP molecules/pixel. For determination of the number of molecules, we calculated the ratio of the average signal of the structure and single GFP signal determined by a gaussian fitting. 6 The unspecific binding is about 7 times lower than the specific binding.
Bilayer Quality. We determined the diffusion constants on glass slides with different cleaning procedures: on the one hand on glass slides cleaned with acetic acid (as used in the original manuscript) and on the other hand on glass slides cleaned with peroxymonosulfuric acid. In order to quantify the influence of the remaining unpolymerized photoresist left over after washing, we also placed a droplet of photoresist on the glass slides (cleaned with acetic acid or peroxymonosulfuric acid) and washed it away with acetone before spreading the lipid vesicles.
The diffusion constants were determined by manual single particle tracking (as described in 7). In order to evaluate a possible influence of crosslinking of biotin via streptavidin, we not only carried out the experiments with streptavidin-Alexa Fluor® 555 bound to biotinylated lipids, but also with the lipophilic fluorescent tracer DiD (1 mg dissolved in 1 ml ethanol, Invitrogen/Thermo Scientific, USA).

Table 1. Dependence of diffusion constants of DiD and streptavidin-Alexa Fluor ® 555 within supported lipid
bilayers on the treatment of the glass substrates. Glass slides were either cleaned using peroxymonosulfuric acid or with acetic acid. On some of the cleaned glass slides, the photoresist PETA with 20 wt% CEA and 1 wt% IC 819 was applied and subsequently removed by washing with acetone prior to the spreading of the lipid bilayer.  Overall, we observed that the cleaning procedure strongly influences the diffusion properties of the lipid bilayer. Best results are yielded on peroxymonosulfuric acid treated glass coverslips. In this case, the diffusion constant (D) is in the range between 0.2 to 0.4 µm 2 /s (see Figure S5a and b) which is consistent with published data. The application of the photoresist and washing has some, but a minor influence. The diffusion constant is decreased by less than 10%.

Substrate
For with acetic acid cleaned slides, D is in the range of 0.1 to 0.2 µm 2 /s (see Figure S5c and d). In general, the diffusion constant of streptavidin is lower than the diffusion constant of DiD.
Polymer lines as barrier for lipids. Equidistant horizontal polymer lines (height ca. 20 nm, width ca. 100 nm) have been structured and incubated with POPC vesicles. For bilayer visualization, the lipid bilayer has been incubated with DiI (see Figure S6). Figure S6a shows a DiI signal originating from a lipid bilayer which has bleached (500 ms bleach pulse, I = 3.3 kW/cm 2 ) in the center. It is clear that diffusion is stopped by the upper and lower horizontal line but DiI starts to diffuse in from the left and the right. Figure S6b depicts the same area after ~ 500 ms recovery time.

S9
The DiI signal recovery occurs only from the left and the right side, not from the top or bottom. This indicates that the polymer lines act as barriers, hindering lipid exchange between the bleached and non-bleached area.