Lithography of polymer nanodots
Structured nanodots of various sizes were fabricated using TPP or STED-lithography. Polymer ‘nanoanchors’ with sizes as small as 65 nm in diameter can be achieved using STED–lithography [15]. The TPP fabricated dots described in this paper are up to ~300 nm in diameter.
We use a (80/20) mixture of the two acrylate monomers SR499 (Sartomer, Colombes Cedex, France) and pentaerythritol triacrylate (PETA, Sigma-Aldrich, Vienna, Austria) including 300-400 ppm monomethyl ether hydroquinone with either 0.25 wt% 7-diethylamino-3-thenoylcoumarin (DETC, Acros Organics, Geel, Belgium) as a depletable photoinitiator [26] for STED-TPP or Irgacure 819 (BASF, Ludwigshafen, Germany), an efficient and nonfluorescent photosensitive initiator for ordinary TPP. This photoresist was shown in a previous study to have antibody affinity [15]. For fabrication of nanodots with a diameter of ~70 nm (surface areas < 104 nm2), we used STED-lithography with 5.5 mW excitation and 28.8 mW depletion power. Without depletion beam, two-photon excitation powers between 5.5 and 6.5 mW are applied to create nanodots with surface areas of 1-2 × 104 nm2, 1-2 × 105 nm2 and > 2 × 105 nm2. Hence, dots of four different sizes are written. They are arranged in arrays of 30 × 30 or 40 × 40 dots with 2 μm dot spacing each. Figure 2 shows a schematic drawing of four arrays of nanodots with various sizes and various streptavidin loading. The smallest dots, shown in Figure 2a, are fabricated with STED-lithography and show the lowest loading with streptavidin (~10% only, vide infra). TPP fabricated dots have a size dependent streptavidin loading reaching up to 100% for the largest dots (sketched in Figure 2d).
The quality of the arrays is analyzed via SEM (Figure 1a) and fluorescence microscopy using an excitation wavelength of 492 nm. The STED-lithography fabricated dots contain DETC and are hence visible due to their intrinsic fluorescence with an emission maximum at 525 nm. Irgacure 819 exhibits only a very weak broadband fluorescence when excited with 492 nm and 532 nm light. Depending on the photo-initiator, we can either use red fluorophores for labeling (Atto655 (Atto-tec GmbH, Siegen, Deutschland)) on the DETC containing STED-lithography fabricated dots, or use green and red fluorophores for labeling (e.g. Alexa555 and Atto655) when TPP with Irgacure 819 is performed. The low autofluorescence of Irgacure 819 does not disturb the quantitative signal analysis.
The streptavidin is either labeled with Atto655 or Alexa555 fluorophores (see Materials: Labeling of Streptavidin). For incubation, the streptavidin is dissolved in phosphate buffered saline (PBS) (Sigma-Aldrich, Vienna, Austria) and dropcast on to the samples. To avoid unspecific streptavidin binding to the substrate, we use two different coating strategies, either 5000-PEG-silane or a supported lipid bilayer [27] for glass surface passivation. For the first strategy, we coat the surface with PEG, which is known to reduce protein adsorption due to its hydrophilic nature [28,29], prior to writing the polymer dots. For the second strategy, the dots are written on a bare glass and subsequently, a bilayer coating is formed by spreading of palmitoyloleoylphosphatidylcholine (POPC) lipid vesicles [30]. The main advantage of lipid passivation is the short incubation time, high bilayer homogeneity and the ability of self-recovery. However, bilayer formation only works in physiological buffers. In comparison, passivation with PEG is less dependent on environmental factors but more vulnerable to surface defects [31,32].
Fluorescence images are taken after incubation of the STED-lithographically fabricated nanoanchors or the TPP dot arrays with Atto655 or Alexa555 labeled streptavidin, respectively. We performed a stepwise increase of the amount of streptavidin (2 μl of the stock solution per 5 min incubation-step each) until the nanodots are saturated.
Figure 3a shows an image of Atto655 labeled streptavidin loosely attached to a lipid-coated glass slide. Since a lipid supported bilayer exhibits strong repulsive forces against proteins [27], we used high concentrations of streptavidin without washing of the sample to introduce at least some unspecific binding. High probability to find single streptavidin on the lipid coated glass is used to quantify the fluorescence strength of single streptavidin in order to achieve a statistic of single molecule emission strength shown in Figure 3c (red histogram).
Single molecule fluorescence microscopy is used to analyze the streptavidin loading of the polymer nanodots. For an estimation of the loading rates, the fluorescence signal from occupied nanodots (Figure 3b) is compared with signals of single streptavidin molecules sparsely distributed on lipid passivated glass (Figure 3a). To quantify the fluorescence signals of the single streptavidin molecules on glass as well as streptavidin incubated nanodots, isotropic undecimated wavelet transformation [33] is used for the recognition of individual fluorescent spots [34]. Further parameterization of the fluorescence signals is performed by Gaussian fitting.
The Atto655 labeled streptavidin molecules which were bound to the nanodots are shown in Figure 3b. Due to the PEG passivation, almost no unspecific binding of the streptavidin is observed after washing. The distribution of fluorescence counts from nanodots loaded with streptavidin is shown in Figure 3c (blue histogram, > 500 single molecule signals). The intensity distributions of streptavidin on lipid passivated glass (red) and on nanodots (blue) differ, which indicates that quite a number of dots have more than one SA molecule attached.
Streptavidin loading of nanodots
We find that only ~10% of all STED-lithographically fabricated nanoanchors with ~70 nm average diameter (surface areas <104 nm2) are loaded with at least one streptavidin. This result was rather unexpected since we were able to load 98% of STED-lithographically fabricated nanoanchors with antibodies in a previous study [15]; i.e. the affinity of proteins to nanoanchors seems to strongly depend on the specific type of protein. Since only ~10% of all patches were loaded by streptavidin on average, the STED-lithographically fabricated nanoanchors were not considered for further analysis.
In order to achieve a more detailed characterization of the streptavidin binding properties, TPP nanodots with different surface areas are polymerized. We classified the dot area into three groups; the first groups surface area is ~2 × 104 nm2, the second group covers around 1 × 105-2 × 105 nm2, and the third category comprises areas above 2 × 105 nm2. The dot area was approximated by a spheroid which was parameterized by atomic force microscopy. These area classes correspond roughly to dots with diameters of <150 nm, 200–300 nm and slightly >300 nm, respectively. The nanodot surface is not only given by the diameter of the nanodot, but also by the nanodots height. This height can be adjusted by positioning the excitation point spread function axially with ~15 nm precision.
Figure 4 shows the loading of TPP nanodots with Atto655-streptavidin with respect to the three nanodot surface areas and for lipid (Figure 4a) or PEG (Figure 4b) passivated sample surfaces. For instance, in case of the PEG passivated TPP nanodots with surface areas of <2 × 104 nm2, only 40 ± 9% of all dots carry at least one streptavidin molecule. These 40 ± 9% of SA carrying dots carry on average one streptavidin complex (1.3 = σ+; 0.96 = σ−). In case of the lipid bilayer passivated surface, 28±9% of all measured dots are labeled with an average of 1.1 streptavidin/dot (1 = σ+; 0.7 = σ−).
For PEG passivated arrays with larger TPP dots (i.e. 1 × 105-2 × 105 nm2 dot surface area), 32 ± 11.6% of all TPP dots carry 1.27 streptavidin complexes (1.2 = σ+; 0.91 = σ−). When the substrate is passivated with lipids, a much better total dot occupancy of 84 ± 22.4% is achieved with 1.22 streptavidin molecules per labeled dot (1.2 = σ+; 0.73 = σ−). The improved streptavidin coverage of the dots on the lipid coated slides is not fully understood. An explanation may be the difference in the thickness of the two passivation layers. The POPC lipid bilayer has usually a thickness of 5–6 nm [35], whereas 5000-PEG molecule form a ~27 nm (if fully stretched [36]) high ‘mesh’ if exposed to an aqueous solution [28,37]. Hence, due to the adhesion of the PEG backbone to the nanodot, the PEG layer may screen a part of the TPP nanodot surface from proteins [28,37].
For the largest TPP dot arrays of >2×105 nm2 surface area, fabricated on a PEGylated surface, we observe that 70-93% of all dots are occupied with an average of 1.94 streptavidin molecules (1.6 = σ+; 1.1 = σ−). In case of the lipid bilayer passivated substrate with >2×105 nm2 dot surface area, 100% of all measured dots are loaded with at least one streptavidin and 1.87 streptavidins (1.51 = σ+; 1 = σ−) on average. In order to verify the influence of the fluorophores on the streptavidin binding properties, intermediately sized TPP dots with 1×105-2×105 nm2 area are incubated with Alexa555-streptavidin. The Alexa555-streptavidin incubated arrays, fabricated on PEGylated slides, have 36 ± 7% of all dots loaded with at least one streptavidin with an average of 1.2 streptavidin per loaded dot (1.36 = σ+; 1 = σ−). In case of lipid passivated substrates, 53 ± 13% of all dots are occupied by at least one SA and these occupied dots carry 1.3 streptavidin molecules on average (1.35 the upper error; 1 the lower error).
Binding of biotin
To prove whether the streptavidin is still biochemically active, the TPP dots were incubated with a fluorophore–biotin conjugate. For these experiments we used Atto655 labeled biotin and Alexa555 labeled streptavidin. Figure 1 depicts a section of a nanodot array with an average dot surface area between 1×105 and 2×105 nm2, sequentially incubated with Alexa555-streptavidin and Atto655-biotin (Figure 1b, c). We determined an average binding of 0.7 biotin molecules per Alexa555-conjugated streptavidin. This result clearly illustrates that binding pocket accessibility is influenced by the adhesion to the nanodot, since a native streptavidin possesses four biotin binding pockets [21]. However, we find that on TPP dots with 200–300 nm in diameter, statistically at least one biotin molecule is attached.