A Simplified and Robust Activation Procedure of Glass Surfaces for Printing Proteins and Subcellular Micropatterning Experiments

Depositing biomolecule micropatterns on solid substrates via microcontact printing (µCP) usually requires complex chemical substrate modifications to initially create reactive surface groups. Here, we present a simplified activation procedure for untreated solid substrates based on a commercial polymer metal ion coating (AnteoBindTM Biosensor reagent) that allows for direct µCP and the strong attachment of proteins via avidity binding. In proof-of-concept experiments, we identified the optimum working concentrations of the surface coating, characterized the specificity of protein binding and demonstrated the suitability of this approach by subcellular micropatterning experiments in living cells. Altogether, this method represents a significant enhancement and simplification of existing µCP procedures and further increases the accessibility of protein micropatterning for cell biological research questions.


Introduction
Depositing biomolecules onto solid substrates in regular 2D patterns with micrometer resolution, also known as molecular printing or, more commonly, as protein micropatterning, has found widespread use in academic laboratories, and multiple applications for biomedical and cell biological research have emerged [1][2][3][4][5][6][7][8]. In this regard, many different methodologies have been developed in the recent years, whereas the fabrication of biomolecule micropatterned (and even nanopatterned) solid substrates is mainly depending on the intended application and available lab infrastructure. Basically, those techniques can be classified into direct and indirect deposition strategies.

Cell Culture and Transfection
HeLa cells were obtained from ATCC, cultured in RPMI medium supplemented with 10% FBS and 1% penicillin/streptomycin (all PAN-Biotech GmbH, Aidenbach, Germany) and grown at 37 • C in a humidified incubator with 5% CO 2 . Cells were transiently transfected with fluorescent-fusion constructs using jetOPTIMUS DNA transfection reagent (Polyplus transfection, Illkirch, France), according to the manufacturer's protocol. A total of 24-48 h after transfection, cells were used for subcellular micropatterning experiments.

Subcellular Micropatterning Experiments in Living Cells
For live cell micropatterning experiments, streptavidin-patterned surfaces were further modified by incubating biotinylated bait antibodies (10 µg/mL, as indicated) for another 20 min at room temperature. Antibody-functionalized chambers were finally washed with PBS (three times) and bait-protein expressing cells were grown on the patterned surfaces for at least 3-4 h prior to fluorescence imaging.

Fluorescence Microscopy
TIRF microscopy was carried out on a microscopy set-up as used in a previous study [36].

Image Analysis and Statistical Evaluation
Initial imaging recording was supported by the Olympus XcellenceRT software package (version 1.1). Images were exported as TIFF frames and fluorescence contrast analysis was performed using the Spotty framework (version 3.7, https://bioinformatics.fhhagenberg.at/bin_typo3/htdocs/fileadmin/user_upload/Downloads/spotty.html) as described previously [36,38].
For significance testing, an unpaired t-test was used to compare two experimental groups, whereas comparison of more than two different groups was performed using one-way ANOVA, which was followed by Tukey's multiple comparisons test. All data transformation and statistical comparisons were carried out in GraphPad Prism software (version 9).

Simplified Procedure for the Fabrication of Protein Micropatterned Glass Substrates
The principle of our surface preparation is shown in Figure 1A-F, exemplified by large-area surface patterning for increased experimental throughput. A simplified scheme of the µCP process is depicted in Figure S1. Following this workflow, the micron-scale protein-patterned glass substrate was generated by printing a biocompatible and chemically stable background protein possessing good surface passivation properties onto the AnteoBind TM precoated glass. As shown in our previous work [36,39], BSA meets those demands and was used throughout this study. AFM images of the BSA grid revealed an almost defect-free surface passivation ( Figure 1H), with an average height of the printed BSA of~3-4 nm, which corresponds to a BSA monolayer formed upon µCP. After µCP, the pre-patterned glass substrate was manually bonded to a multi-well plastic casting resulting in a modular ready-to-use micropatterning platform. The unblocked 3 µm patterns were subsequently filled with streptavidin, followed by the incubation of biotinylated antibodies ( Figure 1G, TIRF microscopy images of BSA-Cy5 grid and FITC-labeled antibodies are shown for illustration) and the seeding of cells expressing fluorescence fusion proteins of interest. The antibody will bind to the extracellular domain of a membrane-anchored protein (e.g., receptor; also termed as bait protein), resulting in a rearrangement of the bait in the cell membrane into an ordered array according to the micron-scale antibody pattern. We and others have used similar assays (although with methodological variations) to investigate cell signaling and cell membrane receptor-dependent protein-protein interactions (PPIs) [40][41][42][43][44][45][46][47]. Most recently, we further developed the approach enabling the subcellular dynamic immunopatterning of cytosolic protein complexes [36]. The use of this platform is not only restricted to antibodies, rather, it enables fast and easy modular surface functionalization using different bait-capturing biomolecules such as specific ligands [46], ligand-decorated multi-scale origami structures [39] and multivalent chelator nanotools [47].
By adapting our micropatterning assay to the described polymer-metal ion coating, we attempted to create very stable and active surfaces, which are capable of binding many classes of biomolecules and serve as biointerfaces for subcellular micropatterning experiments in living cells.

Characterization of Protein Binding on Micropatterned Substrates
To elaborate on the applicability of the polymer-metal ion coating for protein micropatterning via µCP, we functionalized untreated glass coverslips using the procedure as described in Figure 1. In a first step, we identified optimum working concentrations of the AnteoBind TM coating reagent (Figure 2A,C). For this purpose, glass substrates were covered by different concentrations of the polymer-metal ion reagent followed by µCP of the micron-scale BSA grid (5 mg/mL) and subsequent incubation of STA-Cy5, whereas the quality of BSA transfer was evaluated by means of STA-Cy5 fluorescence signal. Figure 2A shows representative TIRF microscopy images of STA-Cy5-patterned glass surfaces that were precoated with indicated AnteoBind TM reagent concentrations. Representative line profiles of the respective STA-Cy5 signals are shown in Figure 2C. Non-specific binding of STA-Cy5 in BSA-patterned grid areas was comparably reduced to a minimum for surfaces coated with pure reagent down to a dilution step of 1:20. However, the highest levels of specific STA-Cy5 enrichment inside the non-passivated patterns were detected on substrates with pure AnteoBind TM reagent coating, followed by slightly reduced fluorescence signals down to a concentration of a 1:20 dilution. AnteoBind TM reagent concentrations below 1:20 delivered significantly lower levels of STA-Cy5 patterning, whereas the highest background signal in combination with the lowest STA-Cy5 signal was obtained for glass surfaces without polymer-metal ion coating. Indeed, biomolecule printing and binding without loss of functionality onto untreated glass coverslips has been reported in various studies [44,48,49]. However, exact mechanisms of protein binding on unmodified glass have not been fully understood so far and are mainly dependent on the surface properties themselves, as well as on the adsorption protein [50]. Most importantly, attaching proteins to rather inactive surfaces, such as plane glass, might lead to lower surface passivation, increased unspecific binding, decreased specific binding of the protein of interest, eventual loss of protein function or reduced binding capabilities, and unstable biointerfaces with shortened storage life. Furthermore, it is not entirely clear what holds the polymer-metal ion complex on the bare glass surface. However, it was previously shown that such metal complexes can bind to plain non-irradiated polystyrene surfaces, as well as to untreated cyclic olefin copolymer substrates [31,51]. What is noteworthy, is that untreated glass surfaces were also reported to possess anionic surface properties [52], therefore being able to bind cationic biomolecules, as it might also be the case for the positively charged polymermetal ion complex. Hence, we speculate that the 'adsorption forces' are a mixture of ionic interactions and hydrogen bonds due to surface charges and impurities (e.g., presence of other metal ions, surface oxygen groups, etc.). Eventually, simple diffusion into the porous glass structure might also be a possible explanation.
AnteoBind TM precoated glass. As shown in our previous work [36,39], BSA meets those demands and was used throughout this study. AFM images of the BSA grid revealed an almost defect-free surface passivation ( Figure 1H), with an average height of the printed BSA of ~3-4 nm, which corresponds to a BSA monolayer formed upon µCP. After µCP, the pre-patterned glass substrate was manually bonded to a multi-well plastic casting resulting in a modular ready-to-use micropatterning platform. The unblocked 3 µm patterns were subsequently filled with streptavidin, followed by the incubation of biotinylated antibodies ( Figure 1G, TIRF microscopy images of BSA-Cy5 grid and FITC-labeled antibodies are shown for illustration) and the seeding of cells expressing fluorescence fusion proteins of interest. The antibody will bind to the extracellular domain of a membrane-anchored protein (e.g., receptor; also termed as bait protein), resulting in a rearrangement of the bait in the cell membrane into an ordered array according to the micron-scale antibody pattern. We and others have used similar assays (although with methodological variations) to investigate cell signaling and cell membrane receptor-dependent protein-protein interactions (PPIs) [40][41][42][43][44][45][46][47]. Most recently, we further developed the approach enabling the subcellular dynamic immunopatterning of cytosolic protein complexes [36]. The use of this platform is not only restricted to antibodies, rather, it enables fast and easy modular surface functionalization using different bait-capturing biomolecules such as specific ligands [46], ligand-decorated multi-scale origami structures [39] and multivalent chelator nanotools [47]. Knowing the optimum surface coating dilution, we next investigated the impact of the surface passivation by printing a micron-scale BSA grid with varying concentrations (Figure 2B,D). In order to reduce specific costs for surface functionalization (which plays a major role, especially for the large substrates presented here), we picked the 1:20 dilution for surface coating. Figure 2B shows TIRF microscopy images of STA-Cy5-patterned glass surfaces that were precoated with 1:20 AnteoBind TM reagent concentrations and subsequently patterned with a BSA grid using concentrations between 0.001 and 10 mg/mL BSA. Representative line profiles of the respective STA-Cy5 signals are shown in Figure  2D. As a starting point, we chose 1 mg/mL BSA, which we and others have used for surface passivation on epoxysilane-coated glass slides and COP foils in previous studies [36,53]. Nevertheless, higher concentrations of BSA for improved surface passivation coatings have been recently reported [54,55]. Therefore, we increased the BSA amount for µCP up to 10 mg/mL, which indeed resulted in the highest STA-Cy5 signal inside the active patterns ( Figure 2D). A slightly lower pattern intensity was observed for 1 and 5 mg/mL Knowing the optimum surface coating dilution, we next investigated the impact of the surface passivation by printing a micron-scale BSA grid with varying concentrations ( Figure 2B,D). In order to reduce specific costs for surface functionalization (which plays a major role, especially for the large substrates presented here), we picked the 1:20 dilution for surface coating. Figure 2B shows TIRF microscopy images of STA-Cy5-patterned glass surfaces that were precoated with 1:20 AnteoBind TM reagent concentrations and subsequently patterned with a BSA grid using concentrations between 0.001 and 10 mg/mL BSA. Representative line profiles of the respective STA-Cy5 signals are shown in Figure 2D. As a starting point, we chose 1 mg/mL BSA, which we and others have used for surface passivation on epoxysilane-coated glass slides and COP foils in previous studies [36,53]. Nevertheless, higher concentrations of BSA for improved surface passivation coatings have been recently reported [54,55]. Therefore, we increased the BSA amount for µCP up to 10 mg/mL, which indeed resulted in the highest STA-Cy5 signal inside the active patterns ( Figure 2D). A slightly lower pattern intensity was observed for 1 and 5 mg/mL BSA-printed surfaces, whereas, already, a tenfold lower concentration led to a substantial reduction in STA-Cy5 pattern enrichment. Almost no specific STA-Cy5 patterning could be detected for 0.001 mg/mL BSA passivated surfaces. Unpatterned, but AnteoBind TM reagent and BSA-coated surfaces were used as a control.
To further characterize the versatility of the implemented coating, we aimed in the printing of different proteins in addition to BSA. As shown in Figure S2, we were able to deposit various proteins (streptavidin, avidin, anti-EGFR IgG antibody, fibrinogen) onto the coated glass substrates by µCP.
The general applicability of the AnteoBind TM reagent for immobilizing biomolecules, and more specifically antibodies, on various surfaces has been reported in several studies [27,28,31,33]. We could unequivocally show that this coating is also a superior and simple strategy for the surface activation, prior to µCP of proteins. Based on the intended application and pursued sensitivity of the assay, the reagent can be used in a wide concentration range, also in combination with common surface passivation steps.

Applicability for Subcellular Micropatterning Experiments
Finally, we demonstrated the applicability of AnteoBind TM reagent-coated and BSAstructured glass surfaces for subcellular micropatterning experiments in living cells. For this purpose, surfaces were further functionalized with streptavidin and anti-GFP antibodies and cells expressing a GFP-fused bait protein were grown on the antibody-patterned substrate. Upon specific antibody-antigen interaction, bait proteins will be rearranged in the plasma membrane according to the micron-scale antibody pattern ( Figure 3A). To reduce fluorescence background signal, and to specifically visualize GFP-fused proteins within or near the cell membrane, the fluorescence readout was conducted in total internal reflection mode (TIRF microscopy). As a proof-of-concept bait, we overexpressed GFP-labeled ErbB2 (Erb-B2 receptor tyrosine kinase 2; GFP was fused to the extracellular domain of the receptor) in Hela cells. First, we elaborated on the specificity of bait patterning due to the antibody-antigen interaction ( Figure 3B). For the quantitation of the lateral bait distribution, the respective fluorescence signal intensities within and outside the antibody-patterned areas were compared (mean fluorescence contrast <c>) [41]. No rearrangement of GFP-ErbB2 was detected in cells grown on surfaces only functionalized with STA ( Figure 3B,C, row 1; <c> = 0.02 ± 0.06) and 'inert' anti-HA antibodies ( Figure 3B,C, row 2; <c> = −0.02 ± 0.04), respectively. On the contrary, a significant lateral redistribution of GFP-ErbB2 was detected in cells grown on anti-GFP antibody-patterned substrates ( Figure 3B,C, row 3), resulting in a mean fluorescence contrast of <c> = 0.46 ± 0.11. This result again confirms robust surface protein patterning. Furthermore, the transfer of the micron-scale antibody pattern into the plasma membrane was only visible in cells facing functional and specific anti-bait antibodies.
As quantitative TIRF microscopy requires a flat interface between the plasma membrane and the patterned substrate to avoid false-positive signals and misinterpretation, we next checked the cell contact surface by coexpressing GFP-ErbB2 and RFP-Lact-C2 (RFP fused with C2 domain of bovine lactadherin) ( Figure 3D). The inner-leaflet peripheral protein RFP-Lact-C2 turned out to be a good negative control, as it showed a homogenous membrane distribution in the central regions of GFP-ErbB2-patterned cells. Most importantly, the lack of RFP-Lact-C2 copatterning indicates that bait micropatterning has no measurable influence on plasma membrane curvature. Biosensors 2022, 12, x FOR PEER REVIEW 8 of 13  AnteoBind TM reagent-coated surfaces have been reported to possess direct antibody binding capabilities with enhanced orientation and functionality, as the polymeric metal ions chelate to available electron-donating groups on synthetic surfaces and biomolecules [31]. We therefore investigated the bait-capturing capability of patterned surfaces that were directly functionalized with anti-GFP antibodies in comparison to surfaces comprising an additional STA layer prior to antibody addition (our "classical" way of antibody patterning) ( Figure 4A,B). Indeed, we already found a remarkable GFP-ErbB2 enrichment in cells grown on solely antibody-functionalized substrates (<c> = 0.33 ± 0.10). Interestingly, antibody-induced GFP-ErbB2 patterning could be further enhanced by STA preincubation, resulting in a significantly increased fluorescence contrast value of <c> = 0.53 ± 0.12. These results prove, again, the superior biomolecule-binding properties of this surface coating. Nevertheless, at least for the presented application, an additional streptavidin layer seems to be favorable for enhanced bait-capturing. Reasons for that might be diverse; however, it is fair to speculate that the covalent streptavidin-biotin interaction leads to a more optimized antibody orientation, also preserving its native character. Furthermore, an additional incubation of streptavidin might lead to a better accessibility of bound antibodies due to the compensation of possible differences in biomolecule heights present at the micropatterned glass surface.
brane and the patterned substrate to avoid false-positive signals and misinterpretation, we next checked the cell contact surface by coexpressing GFP-ErbB2 and RFP-Lact-C2 (RFP fused with C2 domain of bovine lactadherin) ( Figure 3D). The inner-leaflet peripheral protein RFP-Lact-C2 turned out to be a good negative control, as it showed a homogenous membrane distribution in the central regions of GFP-ErbB2-patterned cells. Most importantly, the lack of RFP-Lact-C2 copatterning indicates that bait micropatterning has no measurable influence on plasma membrane curvature.
AnteoBind TM reagent-coated surfaces have been reported to possess direct antibody binding capabilities with enhanced orientation and functionality, as the polymeric metal ions chelate to available electron-donating groups on synthetic surfaces and biomolecules [31]. We therefore investigated the bait-capturing capability of patterned surfaces that were directly functionalized with anti-GFP antibodies in comparison to surfaces comprising an additional STA layer prior to antibody addition (our "classical" way of antibody patterning) ( Figure 4A,B). Indeed, we already found a remarkable GFP-ErbB2 enrichment in cells grown on solely antibody-functionalized substrates (<c> = 0.33 ± 0.10). Interestingly, antibody-induced GFP-ErbB2 patterning could be further enhanced by STA preincubation, resulting in a significantly increased fluorescence contrast value of <c> = 0.53 ± 0.12. These results prove, again, the superior biomolecule-binding properties of this surface coating. Nevertheless, at least for the presented application, an additional streptavidin layer seems to be favorable for enhanced bait-capturing. Reasons for that might be diverse; however, it is fair to speculate that the covalent streptavidin-biotin interaction leads to a more optimized antibody orientation, also preserving its native character. Furthermore, an additional incubation of streptavidin might lead to a better accessibility of bound antibodies due to the compensation of possible differences in biomolecule heights present at the micropatterned glass surface. In a final step, we intended to work out optimum anti-bait antibody concentrations for live cell experiments ( Figure 4C,D). Our standard working concentration, 10 µg/mL, was also used in previous studies under similar conditions and throughout the presented experiments. We could recently show that this concentration leads to >90% surface coverage within the pattern elements [43]. This might also explain the negligible increase in GFP-ErbB2 patterning when compared with higher antibody concentrations, such as 20 µg/mL (<c 10 > = 0.55 ± 0.10 vs. <c 20 > = 0.59 ± 0.12). Similar results were also obtained for tenfold lower antibody concentrations (1 µg/mL; <c 1 > = 0.49 ± 0.14), whereas a clear drop in GFP-ErbB2 enrichment in antibody-patterned areas was obtained for 0.1 µg/mL (<c 0.1 > = 0.19 ± 0.07). No specific bait patterning was detected below concentrations of 0.01 µg/mL antibody.

Conclusions
In this study, we describe a simple and straightforward method to produce highly condensed protein micropatterned glass substrates via µCP without the need of initial extensive chemical surface activation and modification. The presented approach has no need for clean room facilities and/or expensive equipment. PDMS stamps carrying a micron-scale array of features of interest can be easily fabricated from pre-manufactured wafers or can be directly purchased from various companies. Once the stamps are established, they can be reused a couple of times and, with the appropriate wafer size, large-area µCP can be realized. Nevertheless, the introduced method can be flexibly adapted to almost any substrate size. Most importantly, untreated substrates (e.g., glass or polymers such as COP, COC, etc.) are very cheap and, based on our experience, the AnteoBind TM reagent can even be reused when recovered adequately after substrate incubation.
Altogether, this method represents a significant enhancement and simplification of existing µCP procedures and might further increase the accessibility of protein micropatterning for cellular biological research questions.