Microscope glass slides (Marienfeld Superior, Germany; size 76 × 26 × 1 mm, ground edges, pure white glass) were covered on one side with self-adhesive polyimide foil (Kapton, DuPont, United States, CMC Klebetechnik GmbH, Germany; thickness of polyimide layer approximately 25 μm, thickness of glue layer approximately 45 μm). A thin layer of the transfer material was placed on top of the polyimide foil by spin coating (80 rps, Schaefer Technologie GmbH, Germany; KLM Spin-Coater SCC-200). Two different spin coating solutions were prepared. Pentafluorophenyl (OPfp)-activated 9-fluorenylmethoxycarbonyl (Fmoc) protected l-glycine, (Fmoc-Gly-OPfp) 1 (3.00 mg), was pre-dissolved in dimethylformamide (DMF) (50 µL), while inert polymer matrix (27 mg) (SLEC PLT 7552, Sekisui Chemical GmbH, Germany) was dissolved in dichloromethane (DCM) (450 µL), resulting in the final spin coating solution (500 µL). The non-activated amino acid, Fmoc-propargyl-glycine (Fmoc-Pra-OH) (3 mg) was pre-dissolved in DMF (50 μL), followed by addition of N,N′-diisopropylcarbodiimide (DIC) (1.4 µL) and pentafluorophenol (PfpOH) (1.7 mg) consecutively, while the inert polymer matrix (27 mg) was pre-dissolved in DCM (450 μL), forming the desired Fmoc-Pra-OPfp 2 in situ (see Supplementary Material).

Fmoc-NH-β-Ala-PEGMA-co-MMA glass slides (∼20 nm thick coating, loading of functional groups according to vendor 1 nmol cm⁻², estimated functional group spacing of 7–10 nm) were acquired from PEPperPRINT GmbH (Germany) and the 3D-Amino glass slides (according to vendor 1–5 nmol cm⁻²) from PolyAn GmbH (Germany). On PolyAn and PEPperPRINT slides, a hydrophilic PEG ((EG)₃) -based spacer (≈17 Å length) was attached (see Supplementary Material, Section 3.2), before the synthesis of the desired tetrapeptides. In a variant process, PolyAn slides without PEG-spacer were used directly, without prior spacer functionalization.

For the array synthesis, a spot pitch of 250 μm was used. A laser scanning system with 488 nm wavelength and 120 mW maximum output power was used (Mende et al., 2020), with a laser focus diameter of ∼20 µm. PEPperPRINT slides: A laser power of 80 mW and a pulse duration of 6 ms per spot was applied. PolyAn slides: A laser power of 60 mW with a pulse duration of 6 ms was applied. The final spot diameter was about 150 μm.

General laser-based synthesis process: The laser transfer and peptide synthesis were conducted as reported previously (Eickelmann et al., 2019; Mende et al., 2020; Paris et al., 2020). The process begins with the preparation of different donor slides (Donor Slide Preparation) that are easily prepared by spin-coating a solution of polymer matrix and activated amino acid building block onto polyimide foil (Kapton) bearing glass slides. The polymer and amino acid mixture forms an about 200 nm thin layer on the polyimide. For the patterning process, an amino acid containing donor slide is placed on top of an acceptor slide (Acceptor Slide Preparation) and a focused laser (Laser Transfer Parameters) transfers solid polymer material spotwise from the donor to the acceptor (one pulse of 6 ms transfers one spot). The laser is absorbed by the polyimide foil, which heats up and expands. Eventually, the expanding polyimide contacts the acceptor slide, causing the transfer of nanometer thin and about 150 µm wide polymer material spots. The transfer is repeated with different donor slides until the desired amino acid pattern is completed. Afterwards, the acceptor slide is placed into an oven at 95°C under nitrogen for several minutes to initiate the coupling reaction. In the oven, the polymer spots “melt” while retaining their shape, enabling the reaction of the building blocks according to the transferred pattern. The activated amino acid building blocks couple to the amino groups on the acceptor slide. Next, the acceptor slide is washed, removing unreacted amino acids and residual polymer. Each amino acid coupling step is repeated three times to increase the coupling yield and to minimize deletion sequences. Then, remaining free amino groups on the acceptor surface are acetylated and the Fmoc protecting groups are removed before the next synthesis cycle. Peptide synthesis steps are repeated, until the final peptide length is reached.

Synthesis of tetrapeptide scaffolds: Commercially available slides from PEPperPRINT or PolyAn were used. Before the synthesis of the tetrapeptides, a PEG-based spacer was attached if not indicated otherwise, (see Supplementary Material). PEPperPRINT slides require a spacer due to the high protein resistance of the surface. The first layer of OPfp-activated and Fmoc-protected amino acids was transferred via laser transfer, using two different donor slides sequentially to create the desired combinatorial pattern on the acceptor slide. The coupling reaction was accomplished under heat in an oven under nitrogen atmosphere at 95°C for 10 min. Subsequently, the slides were washed with acetone twice, initially for 2 min in an ultrasonic bath, and then for another 2 min in a petri dish on a shaker (450 rpm). Then, slides were dried in a jet of air. The laser transfer of the same amino acid pattern, the coupling, and the acetone washing steps were repeated twice, to increase the coupling efficiency. Each time, a new donor slide was used for every transfer and coupling cycle. Free unreacted amino groups on the slides were acetylated with a capping solution twice for 30 min (see Supplementary Material). The slides were washed with DMF (3 × 5 min), methanol (MeOH) (1 × 2 min), DCM (1 × 1 min), and dried in a jet of air. Deprotection of the terminal Fmoc-groups was achieved for 20 min with Piperidine (see Supplementary Material) on a shaker (450 rpm). The slides were washed with DMF (3 × 5 min), MeOH (1 × 2 min), DCM (1 × 1 min), consecutively, and dried in a jet of air. The whole process was repeated, as needed, for each pattern to synthesize the desired peptides. In the case of terminal amino acids within the peptide chain, the Fmoc removal was accomplished before the acetylation step, capping of the free amino groups.

Each sugar azide 3–7 was obtained according to known literature procedures (see Supplementary Material, Section 2.1). Two sugar azides, 8 and 9, were obtained from Conju-Probe.

CuSO₄ (530 μg, 3.36 μmol, 2.00 equiv) was dissolved in a mixture of dimethyl sulfoxide (DMSO) and water (1:1, 200 μL). Sodium ascorbate (998 μg, 5.04 μmol, 3.00 equiv) was added and the mixture was thoroughly vortexed. The precipitate was centrifuged for 1 min. The remaining solution was passed through a polypropylene syringe filter (0.2 µm polypropylene filter media with polypropylene housing, 25 mm diameter, Whatman, Global Life Sciences Solutions Operations United Kingdom). The sugar azide (1.68 μmol, 1.00 equiv) was dissolved in this solution and then applied on the acceptor surface (c = 8.4 μmol/ml). For the incubation, a 16-well format incubation chamber was used. The prepared solution (200 μL) was poured in one of the wells and then shaken overnight in the dark. The next day, the slide was washed with water three times for 5 min inside the well and one time for 30 min in a petri dish on a shaker (450 rpm). Finally, the slide was dried in a jet of air.

To avoid unspecific binding, the acceptor slides were incubated with a blocking buffer for 40 min (Rockland, United States; blocking buffer for fluorescent western blotting MB-070). Fluorescently labeled plant lectins, concanavalin A (i.e., ConA; CF^®633 ConA, Biotium, Inc., United States) was diluted to 100 μg/ml in lectin buffer (50 mM HEPES, 100 mM NaCl, 1 mM CaCl₂, 1 mM MnCl₂, 10% blocking buffer, 0.05% Tween 20, pH 7.5), ricinus communis agglutinin I, (RCA-I), peanut agglutinin (PNA), soybean agglutinin (SBA), dolichos biflorus agglutinin (DBA), and wheat germ agglutinin (WGA) (Rhodamine labeled, Lectin kit 1, Vector laboratories, United States) were diluted to 10 μg/ml in lectin buffer and incubated for 1 h at room temperature. Subsequently, each stained well was washed with PBS-T (3 × 5 min). Then, the acceptor slide was rinsed with Tris buffer (1 mM Tris-HCl buffer, pH = 7.4) to remove all the remaining salt residues, and dried in a jet of air. Fluorescence scanning was used to detect the lectin binding on the corresponding sugar moieties.

All fluorescence scans were carried out in a high-resolution microarray GenePix 4000B scanner. CF^®ConA labeled glycopeptides were screened with an excitation wavelength of 635 nm and PMT gain of 600. Rhodamine RCA-I, PNA, SBA, DBA, WGA labeled glycopeptides were scanned at an excitation wavelength of 532 nm and PMT gain of 500. Carboxytetramethylrhodamine (TAMRA) labeled tetrapeptides were detected at an excitation wavelength of 532 nm and PMT gain of 400. The laser power was always set to 33% and the pixel resolution to 5 μm. For the analysis of the fluorescence images, the analysis software GenePix Pro 6.0 (Molecular Devices, Sunnyvale/California, United States) was used.

For each sugar azide, the reaction was performed in a separate cavity of a 16-well format incubation chamber (PEPperPRINT GmbH, Germany). Each well contained three sets of quadruplicates of the same single sugar azide and tetrapeptide, giving twelve glycopeptide replicas of each synthesized structure. The median of the fluorescence intensity of the scanned area was determined with the microarray analysis software GenePix Pro 6.0. For the analysis, the mean value of the twelve spot medians was calculated. Spots (i.e., outlier/artifacts) with more than 40% standard deviation from the mean were excluded from calculations.

All sixteen possible variants of the peptide tetramers, containing the two derivatives Fmoc-Gly-OPfp 1 and Fmoc-Pra-OPfp 2, were synthesized in the microarray format (Figure 2). Amine functionalized glass slides from PEPperPRINT (PPP) were used with prior functionalization with a PEG-based spacer (Stadler et al., 2008). 3D-amino glass slides from PolyAn were either used with or without prior PEG-spacer functionalization. Before the synthesis, we optimized the transfer and coupling conditions for each solid support (see Supplementary Material, Section 4). Subsequently, a pre-patterning of all acceptor slides was performed with two glycines 1, to further increase the distance between the tetrapeptides and the solid support and, thereby, the accessibility of the glycopeptides. After Fmoc deprotection of the N-terminus, the free amino groups were used for peptide synthesis. Two donor slides were employed to synthesize the sixteen tetrapeptide combinations, Fmoc-Gly-OPfp 1 (G) and Fmoc-Pra-OPfp 2 (B) Figure 2A (conventional synthesis from C-terminus to N-terminus, e.g., N-GBGB-C, 1VII). Coupling and laser transfer of each amino acid layer was repeated three times to achieve high coupling efficiency and prevent deletion sequences while growing the chains. Coupling of the amino acids was conducted in an oven under nitrogen gas atmosphere at 95°C, resulting in three sets of quadruplicates on one array (n = 12 spots; binding intensity is calculated as the mean of the 12 spot replica) (Figure 2B). Quality control of the three synthesized arrays was carried out via clicking a TAMRA azide dye to the scaffolds and analyzing the fluorescence intensity. On the PEPperPRINT slides, a rather constant fluorescence intensity was observed, indicating a quenching effect for higher valencies, as reported previously (Mende et al., 2020). Comparing the results of the two PolyAn slides with and without PEG-spacer, also some quenching could be observed (see Supplementary Material for more details, Section 5).

The copper(I)-catalyzed alkyne-azide cycloaddition (CuAAC) has been widely used in the last years for the synthesis of glycoconjugates on solid support (Freichel et al., 2017; Hill et al., 2018; Camaleño de la Calle et al., 2019; Mende et al., 2020). Herein, we used this approach to attach the following collection of sugar azide monomers to our synthesized peptide scaffolds: α-mannose (α-Man) azide 3, β-galactose (β-Gal) azide 4, β-galactose PEG3-spacer (β-Gal-PEG3) azide 5, N-acetyl-β-galactosamine azide (β-GalNAc) 6, N-acetyl-β-galactosamine PEG3-spacer (β-GalNAc-PEG3) azide 7, N-acetyl-β-glucosamine (β-GlcNAc) 8, and N-acetyl-β-glucosamine PEG3-spacer (β-GlcNAc-PEG3) azide 9. The sugar azides 3–6 and 8 were synthesized based on known experimental procedures from their corresponding unmodified building blocks, while compounds 7 and 9 were commercially acquired (Figure 2C, see Supplementary Material). Each CuAAC reaction with individual sugars was performed in a separate well, reacting all peptide scaffold spots (n = 12) of one array with on sugar.

After the generation of the glycopeptides on the differently functionalized acceptor slides, we probed the synthesized structures with their corresponding fluorescently labeled lectins (Figure 3). Tetrapeptides, carrying the α-Man azide 3 were incubated with ConA (100 μg/ml, Figure 3A). Structures with β-Gal azide 4, and β-Gal-PEG3 azide 5, were probed with fluorescently labeled RCA-I, (Figure 3C, D) and PNA (10 μg/ml, Supplementary Material, Section 7.2). Tetrapeptides with attached β-GalNAc azide 6 and β-GalNAc-PEG3 azide 7 were incubated with DBA and SBA (10 μg/ml) (see Supplementary Material, Sections 7.3 and 7.4), while scaffolds with β-GlcNAc 8 (see Supplementary Material, Section 7.6) and β-GlcNAc-PEG3 azide 9 were probed with WGA (Figure 3E, F) (10 μg/ml). Since we observed an intensity plateau with WGA already for divalent structures, which was different from all other lectins, a 50-fold decreased WGA concentration (0.2 μg/ml) was screened additionally. We analyzed the spacing, density, and ligand dependent binding, and we could confirm that protein binding is surface dependent. In the case of the multivalent glycan-GBP interactions, similar intensity trends were observed for all used lectins on the microarrays (except for WGA, Figure 3E, F), with an increase in binding with an increasing number of sugars on the peptide backbone. Structures with only one attached sugar moiety, e.g., BGGG, GGBG, GBGG, GGGB, showed structure dependent binding, with higher intensity for the N-terminal propargylglycine on all used slides. This could be explained by the higher distance between the sugar and the surface, making it more accessible. The tetra-glycine scaffold (GGGG) was considered as the background control.

In terms of slide functionalization, for all detected interactions, the fluorescence intensities were higher on the PEPperPRINT slides (apart from WGA and DBA). Between the two differently functionalized PolyAn slides, some structure and lectin dependent binding differences were observed.