A combined high-throughput and high-content platform for unified on-chip synthesis, characterization and biological screening

Benz, Maximilian; Asperger, Arndt; Hamester, Meike; Welle, Alexander; Heissler, Stefan; Levkin, Pavel A.

doi:10.1038/s41467-020-19040-0

Download PDF

Article
Open access
Published: 26 October 2020

A combined high-throughput and high-content platform for unified on-chip synthesis, characterization and biological screening

Nature Communications volume 11, Article number: 5391 (2020) Cite this article

10k Accesses
39 Citations
30 Altmetric
Metrics details

Subjects

Abstract

Acceleration and unification of drug discovery is important to reduce the effort and cost of new drug development. Diverse chemical and biological conditions, specialized infrastructure and incompatibility between existing analytical methods with high-throughput, nanoliter scale chemistry make the whole drug discovery process lengthy and expensive. Here, we demonstrate a chemBIOS platform combining on-chip chemical synthesis, characterization and biological screening. We developed a dendrimer-based surface patterning that enables the generation of high-density nanodroplet arrays for both organic and aqueous liquids. Each droplet (among > 50,000 droplets per plate) functions as an individual, spatially separated nanovessel, that can be used for solution-based synthesis or analytical assays. An additional indium-tin oxide coating enables ultra-fast on-chip detection down to the attomole per droplet by matrix-assisted laser desorption/ionization mass spectrometry. The excellent optical properties of the chemBIOS platform allow for on-chip characterization and in-situ reaction monitoring in the ultraviolet, visible (on-chip UV-Vis spectroscopy and optical microscopy) and infrared (on-chip IR spectroscopy) regions. The platform is compatible with various cell-biological screenings, which opens new avenues in the fields of high-throughput synthesis and drug discovery.

Marrying chemistry with biology by combining on-chip solution-based combinatorial synthesis and cellular screening

Article Open access 28 June 2019

A microfluidic Braille valve platform for on-demand production, combinatorial screening and sorting of chemically distinct droplets

Article 19 October 2022

High diversity droplet microfluidic libraries generated with a commercial liquid spotter

Article Open access 23 February 2021

Introduction

Drug discovery is a lengthy and highly expensive process. The development of a single new drug often takes over 20 years and costs on average $2–4 billion^1,2,3,4. Chemists in pharmaceutical and biotechnological companies synthesize thousands of compounds per year, requiring tons of expensive and valuable chemicals and consumables. Each individual compound must be characterized and screened for its biological activity. Automated high-throughput screening of compound libraries using microtiter plates is well-established in biology^5,6,7, although the physicochemical properties of polystyrene microtiter plates limits their use in chemical applications and specialized analytical techniques. Thus, a high-throughput approach is required for diverse, solution-based chemistry enabling simultaneous synthesis of hundreds to thousands of compounds. Miniaturization and parallelization of organic synthesis is crucial to enlarge the capacity of compound library design and reduce the effort, chemical consumption and associated costs. We previously reported the chemBIOS platform in a proof-of-principle study demonstrating on-chip miniaturized solution-based synthesis of 75 transfection agents and subsequent on-chip cell-biological screening in a process that was completed in only 3 days and required only 1 mL of solutions⁸. However, these improvements have been hindered because the existing analytical methods are not designed for nanoliter scale, on-chip high-throughput approaches. Liquid handling of nanoliter-sized volumes and sample transfer of thousands of compounds between synthesis, characterization, and screening platforms slows down the process of drug development and, thus, makes it expensive. There is a clear need to unify early-stage drug discovery on a multi-functional platform that enables sequential nano-scale synthesis of structurally diverse compounds, highly sensitive chemical characterization, and high-throughput biological screening on the same chip.

In this work, we develop a dendrimer-based surface-patterning method, evolved from the existing chemBIOS platform to expand its capabilities⁸, that can be used to handle high-density nanodroplet arrays of both low (organic solvents) and high (aqueous solutions) surface-tension liquids, thus, enabling a broad range of chemical, analytical, and biological applications (chemBIOS) (Fig. 1). An indium–tin oxide (ITO) coating makes the platform conductive and therefore, compatible with on-chip high-throughput compound characterization by matrix-assisted laser desorption/ionization mass spectrometry (MALDI-TOF MS). We demonstrate highly sensitive on-chip approaches for infrared (IR) spectroscopy and high-content reaction monitoring by UV–Vis spectroscopy using this platform (Fig. 1). The open infrastructure and standardized format make chemBIOS adaptable for well-established assays and commercial devices and for both high-throughput (>50,000 droplets/plate) and high-content (>50,000 results/experiment) chemical, analytical, and biological screening (Fig. 1).

**Fig. 1: Chemical, analytical, and biological high-throughput methods compatible with the chemBIOS platform.**

Results

Manufacturing and characterization of the platform

Previously reported hydrophilic–hydrophobic functionalized droplet arrays used for high-throughput cell screenings were based on porous polymer substrates, making the surface oleophilic and incompatible with organic solvents with low surface tension^8,9,10. Feng et al. developed a flat oleophilic–oleophobic patterned substrate enabling the formation of droplet arrays of organic solvents used for miniaturized and parallelized solution-based organic synthesis^8,11. However, the low contrast between the advancing water contact angle of the oleophobic parts (θ_adv = 110.5 ± 1.2°) and the receding water contact angle of the oleophilic parts (θ_rec = 33.7 ± 0.8°) makes this substrate incompatible for handling high surface tension liquids⁸. To create arrays of droplets of both high and low surface tension liquids, the hydro-/oleophobic (omniphobic) patterns should have high advancing contact angles while the hydro-/oleophilic (omniphilic) parts should possess very low receding contact angles. We hypothesized that dendrimeric surface modification with high-density functional groups can increase the contrast of wettability properties leading to an omniphilic–omniphobic patterning. The dendrimeric surface modification is based on a poly(thioether) dendrimer synthesis developed by Killops et al. (Fig. 2a)¹². The surface of a standard glass slide was silanized with triethoxyvinylsilane to produce a reactive, vinyl group-presenting surface. The dendrimeric layer was then synthesized in a repetitive two-step reaction cycle consisting of a photochemical thiol-ene click reaction with 1-thioglycerol followed by esterification with 4-pentenoic acid (Fig. 2b and Supplementary Fig. 1); three cycles produce a dendrimerized surface decorated with high-density alkene groups. This surface can then be photochemically functionalized either with 1-thioglycerol or 1H,1H,2H,2H-perfluorodecanethiol (PFDT), to yield an omniphilic or omniphobic surface, respectively. Omniphilic–omniphobic micropatterns of defined geometry can be created by sequential functionalization through a quartz photomask (Fig. 2c).

**Fig. 2: Manufacturing of patterned dendrimer slides based on grafted poly(thioether) dendrimers.**

We assumed that an increase in functional group density should correlate with the wettability properties of the surface. To test this hypothesis, we compared the static water contact angle of different dendrimer generations (G0–G4) (Fig. 3a), starting with cysteamine hydrochloride functionalized surfaces^8,11. In all, ± values of all contact angle measurements are standard deviations based on triplicate experiments. A slight increase in the static water contact angle (8.6°) was observed between the PFDT-modified G0 (θ_stat(H₂O) 104.3 ± 1.9°) and G4 (θ_stat(H₂O) 112.9 ± 1.5°) surfaces. A much greater decrease (23.8°) was observed between the cysteamine-modified G0 (θ_stat(H₂O) 55.9 ± 3.0°) and G4 (θ_stat(H₂O) 32.1 ± 5.7°) surfaces (Fig. 3a). There were no significant differences in θ_stat(H₂O) between the G2 and G4 surfaces. Therefore, we further analyzed the advancing and receding water contact angles of the G3 surfaces (Fig. 3b). In addition to omniphobic PFDT-modified G3 surfaces (θ_adv(H₂O) 124.9 ± 2.9°, θ_stat(H₂O) 116.6 ± 3.6°, θ_rec(H₂O) 111.2 ± 2.7°), we also investigated omniphilic cysteamine-modified surface (θ_adv(H₂O) 62.1 ± 3.0°, θ_stat(H₂O) 33.0 ± 3.0°, θ_rec(H₂O) 6.7 ± 2.9°), in which omniphilic nature was further improved by modifying the G3 surface with thioglycerol, leading to a surface with extremely low θ_rec(H₂O) (θ_adv(H₂O) 32.6 ± 2.2°, θ_stat(H₂O) 24.0 ± 0.3°, θ_rec(H₂O) 1.2 ± 0.6°).

**Fig. 3: Characterization of dendrimer-grafted surfaces.**

A huge difference of 123.7° between the θ_adv(H₂O) of PFDT-modified G3 surface (124.9 ± 2.9°) and the θ_rec(H₂O) of thioglycerol-modified G3 areas (θ_rec(H₂O) 1.2 ± 0.6°) enabled the generation of stable nanodroplet arrays of high surface tension liquids, such as water (γ_lv 72.8 mN m⁻¹) or cell suspensions. The difference between θ_adv of PFDT-modified and θ_rec of thioglycerol-modified G3 surface was above 80–100° even for organic solvents with a much lower surface tension (22.1–43.5 mN m⁻¹) than water (72.8 mN m⁻¹) enabling the generation of nanodroplet arrays of low surface tension liquids, including various common organic solvents such as ethanol (γ_lv = 22.1 mN m⁻¹), 1-decanol (γ_lv = 28.5 mN m⁻¹), dimethylformamide (DMF; γ_lv = 37.1 mN m⁻¹), and dimethyl sulfoxide (DMSO; γ_lv = 43.5 mN m⁻¹) on the same substrate, as well (Figs. 2d, 3c and Supplementary Table 1).

In the next step, thioglycerol-modified G0–3 surfaces were characterized by atomic force microscopy (AFM). In all, ± values of all AFM measurements are standard deviations based on triplicate experiments. An increase in surface roughness (Rq) of 90 ± 5 pm between each dendrimer generation was observed (Fig. 3d, Supplementary Fig. 2 and Supplementary Table 2). The thickness of each dendrimeric layer increased by 1.1 ± 0.2 nm per generation, which is consistent with a calculated maximum growth of 1.6 nm in height per dendrimer generation (Fig. 3e and Supplementary Tables 3 and 4). The increase in density of surface functional groups with dendrimer generation was confirmed using time-of-flight secondary ion MS (ToF-SIMS) (Supplementary Fig. 3). The decrease in the Si⁺ signal arising from the silicone glass slide measured for each dendrimer generation matched the dendrimer growth (Supplementary Fig. 3). A well-defined chemical pattern was identified for the omniphilic–omniphobic-patterned G3 surface (Fig. 3f). The lateral resolution between omniphilic patterns and surrounding omniphobic borders was ~8 µm (Fig. 3g), enabling the generation of patterns with any defined geometry down to the low micrometer range (Fig. 3h) by photolithography using highly reactive dendrimer substrates.

On-chip MALDI-TOF MS

Recent developments in on-chip miniaturized and parallelized solution-based synthesis have enabled rapid synthesis of hundreds to thousands of compounds on a single plate, although on-chip characterization of large compound libraries remains a challenge⁸. On-chip high-throughput synthesis requires novel on-chip analytical methods that are compatible with low volume, low concentration, high-density arrays, and an on-chip platform. The goal of this research was to accelerate the drug development process by developing a combined platform for on-chip diverse chemical synthesis and straightforward compound characterization on the same plate. MALDI-TOF MS enables characterization of thousands of compounds in seconds with high spatial resolution in the micrometer range using a conductive, flat, and arrayed plate format, which is therefore ideal for characterizing large compound libraries^13,14,15,16. In order to enable MALDI-TOF MS on our chemBIOS platform, we applied developed surface-patterning method to a conductive ITO slide (Fig. 4a). Compounds can be applied or synthesized on-chip in an array format on the conductive slide followed by solvent evaporation, processing with matrix solution and characterization by MS (Fig. 4a). The slides remain transparent (Fig. 3i) and are therefore suitable for use in correlative analysis by optical methods such as microscopy (Supplementary Fig. 4), which is an advantage over standard MALDI stainless steel targets. All MALDI-TOF MS measurements of this study were performed on omniphilic–omniphobic patterned G3 ITO glass slides (Fig. 4a–c and Supplementary Fig. 4).

**Fig. 4: Indium-tin oxide substrates enable on-chip characterization by MALDI-TOF mass spectrometry and IR spectroscopy.**

Miniaturizing chemical synthesis implies reduction of volumes and concentrations, respectively. Our first aim was to investigate the sensitivity of the on-chip MALDI-TOF approach to determine the limits of miniaturization of our chemBIOS platform. Three different lipid-like structures, lipidoids 1 (m/z 643.522), 2 (m/z 686.569), and 3 (m/z 672.553), dissolved in 2-propanol were dispensed to round spots (diameter 2.83 mm; border width: 1.67 mm) to yield a final amount of substance per spot in the range of 0.05–1000 fmol (Supplementary Fig. 4). The samples were co-crystallized with α-cyano-4-hydroxycinnamic acid (CHCA) matrix (Fig. 4a). All three lipidoids were identified with a sensitivity of 100 fmol substance per spot and an average S/N ratio of 38 ± 10 (Fig. 4c and Supplementary Table 5). The sensitivity was enhanced by a factor of four (S/N(100 fmol) 148 ± 25) by additional on-target washing with the matrix buffer solution (Fig. 4c, Supplementary Figs. 5 and 6, and Supplementary Table 5). In all, ± values of all MALDI-TOF MS measurements are standard deviations based on triplicate experiments.

Our next aim was to reduce the spot size, thereby increasing the throughput, and to evaluate the effect of the spot size on the sensitivity of the on-chip MALDI-TOF MS method. Solutions of lipidoids 1, 2, and 3 were dispensed to spots with diameters of 900 and 500 µm (border widths: 225 and 250 µm, respectively) to yield an amount of substance between 1000 and 0.05 fmol per spot. An equivalent amount of CHCA matrix solution was added to all samples. Miniaturization facilitated testing all three compounds on the same slide without altering the sample preparation parameters (Supplementary Fig. 4). This improves the ease and comparability of experiments, reduces the cost of materials, and increases the throughput, thus, enhancing workflow efficiency. The limit of detection was 0.1 fmol for each lipidoid, demonstrating a 100-fold increase in the sensitivity of this method (Fig. 4b and c, Supplementary Fig. 6 and Supplementary Table 7). The average S/N ratio for all 10 fmol samples on 900 µm spots was 117 ± 37, which was 11 times higher than that of the 2.83 mm spots (S/N: 10.3 ± 0.5) (Fig. 4c). However, further reduction of the spot diameter to 500 µm did not yield a significant gain in MALDI-TOF sensitivity (S/N(10 fmol) 208 ± 92) compared to that for 900 µm spots (Fig. 4c). Further optimization of critically important parameters (aliquot size, concentration, and matrix solvent composition) might improve the sensitivity for 500 µm spots. Nevertheless, a sensitivity limit of ~100 fmol allows detection of approximately only 60 million molecules on a single spot. Compared with standard chemical synthesis in flasks, where reactions are performed in the mol range (~0.6 × 10²⁴ molecules) in milliliter-sized volumes, the chemBIOS platform enables not only practicable handling but also analysis of ultra-small amounts of substances (attomole range) and volumes (picoliters to nanoliters). Development of this ultra-sensitive chemBIOS platform for miniaturized and parallelized organic synthesis with integrated characterization is an important step toward high-throughput and high-content synthetic and analytical chemistry.

The dendrimeric surface modification and subsequent omniphilic–omniphobic patterning is not limited to silicon oxide or ITO-coated surfaces. Each hydroxy-presenting surface can be functionalized. Thus, we modified and patterned microtiter plate-sized MALDI stainless steel plates to create a 280 × 180 droplet array (330 × 330 µm² spots separated by 60 µm borders) enabling the application and characterization of 50,400 compounds on a single MALDI plate (Supplementary Fig. 8). This demonstrates the potential for single plate ultra-high-throughput characterization. A broad variety of substance classes could be analyzed on the dendrimer-based omniphilic–omniphobic patterned stainless steel plates including peptides, proteins, carbohydrates, and small molecule drugs (Supplementary Fig. 8).

On-chip IR spectroscopy

ITO coating shows high transparency (>80% in the visible light region) (Fig. 3i) and high reflectance (~90% in the far-IR region)^17,18, properties which are used in biology for conventional light microscopy and IR microscopy to yield additional chemical information^19,20. We aimed to develop an on-chip IR spectroscopy approach that enables mapping of compound libraries and thereby, obtain structural information of compounds in individual spots. An ITO-coated, omniphilic–omniphobic-patterned G3 glass slide was used as our analytical platform. Lipidoid 1 was applied at different concentrations to slides with round spots (diameter of 900 or 500 µm) (Fig. 4d, e). The lipidoid was identified down to 0.3 µg mm⁻² (94.7 fmol per 500 µm spot) (Fig. 4e). The IR spectrum for each compound in the array was used for structural analysis (Fig. 4f). Similar results were observed for spots with a diameter of 900 µm (Supplementary Fig. 9). On-chip IR imaging in an array format offers the potential for non-destructive and position-coded collection of chemical information and characterization of synthesized compound libraries. Although the sensitivity of on-chip IR spectroscopy is lower than that of MALDI-TOF MS, it remains a highly sensitive and complementary method for on-chip characterization and thus, an essential technique in miniaturized and parallelized chemistry. However, light-scattering between the droplet–air interface limits this approach to measurements in the dried state. Further investigations are needed to develop methods for on-chip IR spectroscopy of liquids, which would be useful for on-chip in situ reaction monitoring, as well as a UV–Vis spectroscopy approach, which we developed in the next step.

On-chip UV–Vis spectroscopy

Many methods used for quantitative analysis of different reaction parameters are end-point driven and require numerous repetitions that are time-consuming and labor-intensive. Probe-based spectroscopy enables real-time reaction monitoring²¹, although this requires specialized devices and is limited in terms of miniaturization and parallelization potential. Here we aimed to develop a workflow for on-chip solution-based synthetic chemistry that enables miniaturized and parallelized high-throughput experimentation as well as in situ reaction monitoring. We hypothesized that due to excellent optical properties, the chemBIOS platform should be compatible with on-chip UV–Vis monitoring. The transmittance of thioglycerol-modified G0–4 surfaces exceeded 90% in the visible region and 80% in the ultraviolet (UV-A/-B) region (Fig. 3i).

We designed and 3D-printed a slide adapter that meets the standards of ANSI/SLAS 1-2004 through ANSI/SLAS 4-2004, which is therefore compatible with standard UV–Vis plate readers (Supplementary Fig. 10). Droplet arrays can be trapped between two sandwiched patterned slides within a fixed distance (1 mm), allowing analysis of individual droplets using a plate reader (Fig. 5a). In a proof-of-principle study, droplets of rhodamine 6 G and methylene blue were applied in a checkerboard pattern on slide A, inserted into the slide adapter, and sandwiched and trapped with slide B before UV–Vis absorbance measurement. Spatially separated UV–Vis spectra were observed from each individual droplet (Fig. 5b).

**Fig. 5: On-chip characterization by UV–Vis spectroscopy.**

Next, we applied our on-chip UV–Vis approach to investigate and optimize a one-pot synthesis of lipid-like structures²². The three-component reaction is initiated by an amine-based ring-opening of a thiolactone followed by a disulfide exchange with a pyridyl disulfide (Fig. 6a). The resulting UV-absorbing by-product, 2-thiopyridone (λ_max = 365 nm) was used as an indicator to monitor the reaction progress. Two compound libraries were prepared: (i) slide A contained four blocks of amine A1 (1, 10, 50, and 100 eq.) in triplicate, and (ii) slide B contained mixtures with different ratios of thiolactone T14 to pyridyl disulfide PY12 (1:5, 1:3, 1:1, 3:1, and 5:1; eq./eq.) applied in rows perpendicular to the blocks of slide A (Fig. 6b). Both slides were sandwiched in the slide adapter, thus, simultaneously initiating all reactions under equal environmental conditions (Fig. 6c). All reactions were carried out at room temperature and monitored by UV–Vis spectrometry at 5-min intervals over a period of 70 h. Twenty different reaction parameters, each in triplicates, were analyzed resulting in 50,400 data points within a single experiment (Fig. 6d). The reaction half-life (t_1/2) decreased as the thiolactone component ratio increased and decreased more rapidly as the amine concentration increased. This indicated that the first step of the reaction (ring-opening of the thiolactone by the amine) is rapid and the following disulfide exchange is rate-determining in the one-pot synthesis.

**Fig. 6: On-chip high-throughput reaction monitoring.**

Biological compatibility

The ultimate goal of accelerating the screening of novel compound libraries and increasing throughput is to unify chemical processes, including compound library synthesis, characterization, and biological high-throughput screening^8,23. However, the lack of compatibility between the diverse platforms and infrastructures (flasks, tubes, bottles, polystyrene plates, plastic/glass pipettes, syringes, filters, etc.) of these methods with different material properties (chemical resistance vs. biological compatibility) limits the applicability of high-throughput approaches and slows drug and biotechnological development. Besides compatibility with a variety of organic and aqueous solutions, the platform must be cell compatible. The viability of three cell lines (HeLa, HEK293T, and Jurkat) dispensed to each spot of a thioglycerol-modified G3 slide and cultured for 24 h was evaluated by fluorescence microscopy after life/death staining and by morphology (Supplementary Fig. 11). The viability of each cell line exceeded 97%, demonstrating the compatibility of this platform for cell culturing and opened the possibility for further biological screenings in follow-up studies.

In summary, we developed a covalent surface modification based on a surface-grafted poly(thioether) dendrimer resulting in an increase in the density of surface presentation of reactive alkene groups (Fig. 2a, b). Further photochemical patterning of this surface with 1-thioglycerol and PFDT sequentially through a photomask resulted in the formation of omniphilic spots surrounded by omniphobic borders (Fig. 2c). The high contrast of advancing and receding contact angles of both aqueous and organic liquids between the omniphilic and omniphobic patterns, presenting high contrast in surface wettability, enabled the formation of nanodroplet arrays (1–3000 nL depending on the spot size) of both low surface tension (organic) and high surface tension (aqueous) liquids (Figs. 2d and 3b, c). This offers the potential for use in a broad range of chemical, analytical, and biological applications, such as miniaturized and parallelized on-chip solution-based synthesis or on-chip biological screening (chemBIOS) (Fig. 1). Each nanodroplet forms a distinct vessel that can be used individually. Chemicals, agents, or cell suspension can be added at any time during an assay or whole compound libraries can be combined within a single step (Fig. 5a). The dendrimer-based omniphilic–omniphobic platform enabled generation of nanodroplet arrays of both low-surface tension liquids (organic solvents) and high-surface tension liquids (aqueous solutions) on the same substrate. This opened the possibility to combine on-chip synthesis, on-chip characterization and on-chip biological screening in an iterative cyclic workflow (Supplementary Fig. 12). We aimed to study the sensitivity of various analytical methods and evaluated their compatibility for on-chip compound characterization. An additional ITO coating made the chemBIOS surface conductive and, thus, compatible with MALDI-TOF MS (Fig. 4a). The sensitivity of the on-chip MALDI-TOF MS approach was investigated using an exemplar lipid-like compound class. An increased sensitivity down to the attomole range per spot was observed for spots of increasingly smaller size (diameter 900 and 500 µm) (Fig. 4b, c). The synergy between higher sensitivity for smaller spot sizes enable miniaturization and, thus, parallelization of the on-chip approach. We created a nanodroplet array in the microtiter plate size containing 50,400 droplets (330 × 330 µm² spots; border width: 60 µm) demonstrating the possibility for ultra-high-throughput screening (Supplementary Fig. 8). Furthermore, the IR-reflective properties of the ITO coating enabled IR imaging of a compound array and on-chip IR spectroscopy of selected compound spots (Fig. 4e, f). The sensitivity of the IR spectroscopy approach was found to be 95 fmol/500 µm spot (Fig. 4f). The dendrimer-modified chemBIOS platform showed excellent transparency in the visible and UV region (Fig. 3i) and is therefore compatible with optical analysis methods, enabling on-chip high-throughput in situ reaction monitoring using UV–Vis spectroscopy (Figs. 5 and 6). An on-chip, solution-based, combinatorial three-component reaction was carried out to demonstrate the possibility of an on-chip chemical reaction optimization with on-line reaction progress monitoring. Thus, we simultaneously screened 60 reactions resulting in the generation of 50,400 data points within a single experiment (Fig. 6). The results provided insights into the reaction mechanism that could be used for reaction optimization. The dendrimer-based surface modification combines compartmentalization, excellent optical properties (UV–Vis transparent and IR reflective), chemical resistance (glass substrate compatible with organic solvents) and conductive material properties and, thus, makes it possible to combine important characterization methods (for chemical and biological readouts), such as on-chip highly sensitive MS, on-chip spectroscopy, and on-chip (optical or potentially electron) microscopy in one multifunctional platform. ChemBIOS is compatible with various adherent and suspension cell lines (HeLa, HEK293T, and Jurkat) and is therefore suitable for high-throughput cell-based screening (Supplementary Fig. 11). ChemBIOS combines on-chip high-throughput chemical synthesis, high-content reaction monitoring, highly sensitive compound characterization and biological screening and, thus, unifies all areas of early-stage drug discovery.

Methods

Preparation of dendrimeric-modified and patterned surfaces

The surface of standard microscope glass slides (25 × 75 × 1 mm, width × length × thickness, Schott Nexterion) was cleaned for 10 min using an UVO-Cleaner 42-220 (Jelight) and then silanized in the gas phase using 400 µL triethoxyvinylsilane (Sigma-Aldrich) at 80 °C for 20 h. The slide was washed with acetone and ethanol and then dried by compressed air. The solid-phase synthesis of one dendrimer generation consisted of two steps: (i) 300 µL 10% (v/v) 1-thioglycerol (Sigma-Aldrich) in ethanol/water (1:1, v/v, Merck Millipore) containing 1 wt% 2,2-dimethoxy-2-phenylacetophenone (DMPA, Sigma–Aldrich) was applied to the alkene-presenting surface. A thiol-ene click reaction was carried out by irradiating the surface with 260 nm UV light (OAI model 30) at 3 mW cm⁻² intensity for 1.5 min. The slide was washed with ethanol and then dried by compressed air. (ii) Esterification of the hydroxy-presenting surface was performed by incubating the slides in esterification mixture (112 mg 4-(dimethylamino)pyridine (Novabiochem), 250 µL pentenoic acid (Sigma-Aldrich) and 180 µL N,N’-diisopropylcarbodiimine (Alfa Aesar) in 90 mL acetone (Merck Millipore)) for 4 h while shaking at room temperature. The surface was patterned by first applying 300 µL 10% (v/v) PFDT (Sigma-Aldrich) in acetone containing 1 wt% DMPA (Sigma-Aldrich). The thiol-ene photoclick reaction was carried out by irradiating the surface through a photomask (Rose Fotomasken) with 260 nm UV light (OAI model 30) at 3 mW cm⁻² intensity for 1 min to create omniphobic borders. The slide was washed with acetone and then dried by compressed air. Omniphilic spots were formed by applying 300 µL 10% (v/v) 1-thioglycerol (Sigma-Aldrich) in ethanol/water (1:1, v/v, Merck Millipore) containing 1 wt% DMPA (Sigma-Aldrich) onto the patterned surface and irradiating the surface with 260 nm UV light (OAI model 30) at 3 mW cm⁻² intensity for 1.5 min. The slide was washed with ethanol and then dried by compressed air.

Dendrimeric modification and omniphilic–omniphobic patterning of ITO-coated slides were carried out using the protocol described above using commercial ITO slides (Bruker). Modification and patterning of stainless steel HTS sample plates (Bruker) was conducted accordingly.

Surface characterization

Advancing, static, receding contact, and sliding angles were measured using a Drop Shape Analyzer DSA25 (Krüss). Solvents were purchased without further purification from Merck Millipore.

AFM was performed using a Dimension Icon AFM (Bruker) under the following conditions: tapping mode, scan rate 1 Hz, sample/line 512 and AFM Cantilevers HQ:NSC15/AI BS-15 (Innovative Solutions Bulgaria Ltd.). Thickness was measured by scratching the surface with a tweezer and analyzing the depth of the scratch compared to the intact dendrimer surface surrounding the scratch.

ToF-SIMS analysis was performed using a ToF-SIMS-5 instrument (ION-TOF GmbH) equipped with a Bi cluster primary ion source and a reflectron-type ToF analyzer. UHV base pressure during analysis was <2 × 10⁻⁹ mbar. For high mass resolution the Bi source was operated in a bunched mode providing short Bi₃⁺ primary ion pulses at 25 keV energy, a lateral resolution of ~4 μm, and a target current of 0.4 pA. The short pulse length of 1 ns allowed for high mass resolution (8500 m/Δm). For large fields of view (Fig. 3f and h) the primary beam and the sample stage were rastered, For high lateral resolution (Fig. 3g) the primary gun was operated in a non-bunched mode combined with delayed extraction. The lateral resolution of the instrument is better than 0.5 µm in this case. For charge compensation on glass slide without ITO an electron flood gun (21 eV) was applied and the reflectron tuned accordingly. Primary ion dose was held below the static limit of 1 × 10¹¹ ions cm⁻², spectra were mass calibrated on the omnipresent signals C⁺, CH⁺, CH₃⁺, C₂H⁺ or C⁻, CH⁻, CH₂⁻ C₂⁻.

Transparency analysis was performed using a Lambda 35 UV–Vis spectrometer (PerkinElmer) equipped with a solid sample holder (PerkinElmer).