Dynamic Combinatorial Chemistry Affords Sugar-Based Molecules Targeting Bacterial Glucosyltranseferase as Potential Dental Biofilm Inhibitors

We applied dynamic combinatorial chemistry (DCC) to find novel ligands of the bacterial virulence factor glucosyltransferase (GTF) 180. GTFs are the major producers of extracellular polysaccharides, which are important factors in the initiation and development of cariogenic dental biofilms. Following a structurebased strategy, we designed a series of 36 glucoseand maltosebased acylhydrazones as substrate mimics. Synthesis of the required monoand disaccharide-based aldehydes set the stage for DCC experiments. Analysis of the dynamic combinatorial libraries (DCLs) via UPLC-MS revealed major amplification of four compounds in the presence of GTF180. Moreover, we found that derivatives of the glucose-acceptor maltose at the C1-hydroxyl group are acting as glucose-donors and are cleaved by GTF180. The synthesized hits display medium to low binding affinity (KD values of 0.4–10.0 mM) according to surface plasmon resonance (SPR). In addition, they were investigated for the inhibitory activity using the GTF-activity assays. The early stage DCC study reveals that careful design of DCLs opens up easy access to a broad class of novel compounds that can be developed further as potential inhibitors.


Introduction
Cariogenic dental biofilm, also known as dental plaque, is a causative agent for dental caries. An important factor for the initiation and development of this oral disease is the fermentation of dietary carbohydrates, of which sucrose is considered the most cariogenic. It acts as a substrate for the synthesis of extracellular (EPS) and intracellular (IPS) polysaccharides, which are involved in the formation of the biofilm, having α-glucan as one of the main components. The biofilm hosts bacteria and can promote their adhesion to the tooth enamel. Glucosyltransferases (GTFs) are the major producers of EPS, and are secreted by different strains of bacteria. These GTFs, also known as glucansucrases (GSs), therefore are potential targets in order to inhibit biofilm formation and therefore prevent dental caries. [1][2][3] Glucansucrases are enzymes which are part of the glycoside hydrolase family GH70, consisting of four catalytically essential conserved sequences. To the superfamily of GH-H also belong the glycoside hydrolase families 13 and 77. [4] Cocrystal structures, containing the catalytic and C-terminal domains of glucansucrase of Lactobacillus reuteri 180 were previously reported and provided evidence for an α-retaining double displacement mechanism using one nucleophilic residue. [5] Briefly, the α(1→2)glycosidic linkage of a donor substrate (e.g., sucrose) is cleaved, resulting in the release of fructose and formation of a β-glucosylenzyme intermediate. Subsequently, an acceptor substrate attacks the β-glucosyl-enzyme intermediate, and the glucosyl moiety is transferred to the acceptor (e.g., a glucan chain, maltose or water molecule) with restoration of the α-anomeric configuration ( Figure 1).

Figure 1.
Reaction scheme of the proposed catalytic mechanism of GTF180 via α-retaining double displacement, leading to retention of the α-configuration. [5] Reportedly, glucansucrases can be inhibited by natural as well as synthetic compounds. Natural inhibitors can for example be found in culture broths of bacteria, as was the case for acarbose (Table  1). In 1977, researchers from Bayer discovered α-amylase inhibitors from broths of Actinoplanes strains SE 50, SE 82 and SB 18, of which BAY g 5421 (acarbose) was the most potent. They postulated that acarbose could be a transition-state analogue. [6] Since then, acarbose has been used as an antidiabetic drug throughout the world, and was found to have cardiovascular benefits. [7][8][9] Newbrun et al. showed that acarbose also inhibits GSs, [10] and its mode of action was confirmed by Figure 2. Schematic illustration of target-directed dynamic combinatorial chemistry (tdDCC) using the acylhydrazone linkage, formed by the reversible reaction between a glucose-or maltose-linked arylaldehyde and three representative hydrazides. Binding to the target protein causes a change in equilibrium, which, in turn, leads to the amplification of the hit compounds. cocrystal structure. [11] Compounds from plant sources such as polyphenols in green tea extracts, [12] theaflavins in black tea extracts, [13] curcumin [14] and oxyresveratrol [15] showed marked inhibitory effects on the biofilm of the cariogenic pathogen Streptococcus mutans, however, high concentrations and sufficient time were required. Their antibiofilm activities were mainly exerted through direct inhibition of the bacterial GTFs or down-regulation of GTF expression. [12][13][14][15] On the other hand, small molecule GTF inhibitors can be categorized into only two groups, hydroxychalcones and compounds with a high number of heteroatoms, especially nitrogen. [16] In view of this limited number of inhibitors, we were interested in discovering a novel chemical class that addresses the promising yet underexploited antibiofilm target GTF180. To identify new ligands of a protein, dynamic combinatorial chemistry (DCC) has become an attractive strategy. DCC allows a target protein to alter the equilibrium of a mixture of products, also known as dynamic combinatorial library (DCL). Due to the change in equilibrium in presence of a protein, good binders get amplified and will therefore be selected as hits ( Figure 2). The conditions, reactions, protocols, analysis and applications of DCC were reviewed before. [17][18][19][20] For example, Lehn et al identified inhibitors of the plant lectin concanavalin A using a carbohydrate DCL and dynamic deconvolution. [21] The group of Beau probed DCC to discover binders for the glycosidase hen egg-white lysozyme (HEWL). [22] Ernst and coworkers used DCC to identify submicromolar binders of the bacterial adhesin FimH via adjusting the ratio of building blocks and establishing a proteincapturing protocol. [23] Furthermore, we demonstrated that DCC can be applied to challenging targets involved in protein-protein interactions by discovering stabilizers of 14-3-3(ζ)-synaptopodin complex. [24] In this work, we exploited the power of DCC to identify new inhibitors of the sugar-modifying enzyme GS using the wellestablished acylhydrazone formation as a reversible reaction. Mono-or disaccharide motifs as substrate mimics constitute the building blocks of four DCLs.

Designing of Dynamic Combinatorial Library
We adopted structure-based design using the co-crystal structures of Lactobacillus reuteri 180 GTF180-ΔN with two disaccharides, namely sucrose as donor substrate (PDB: 3HZ3) and maltose as an acceptor substrate (PDB: 3KLL). [5] The active site of GS is relatively wide and it can be divided into subsites -1 (the catalytic pocket), +1, and +2 ( Figure 3A). The donor substrate such as sucrose binds at the catalytic site -1 with its glycosidic linkage in near proximity to the catalytic triad Asp1025, Glu1063, and Asp1136 (highlighted in orange in Figure 3A), and is cleaved by GS. On the other hand, the acceptor substrate, which accepts a glucose molecule such as maltose, binds at subsites +1 and +2 ( Figure 3A). Inspired by these substrates, we designed the basic scaffold of acylhydrazone using glucose and maltose linked to benzaldehyde at para position (A1 and A2, respectively, Scheme 1), which would be elongated by a hydrophobic hydrazide moiety for full occupation of the active site. We excluded sucrose (a glucosedonor) due to its liability to cleavage by GTFs and opted for glucose instead. On the other hand, p-hydroxybenzaldehyde was chosen for glycosylation as it fits perfectly into subsite +2 permitting π-π interactions with Trp1065 ( Figure 3A). Moreover, the para position holds the optimum distance between the sugar moiety and the hydrazide, mimicking the α(1→4) glycosidic connection of α-glucans. Subsequently, we selected 18 chemically different and commercially available hydrazides (H1-H18) that were divided into two groups (Hydrazides I and II, Scheme 1). Each aldehyde (A1 or A2) was allowed to react separately with both sets of hydrazides (I and II), resulting in four series of substrate mimetics (DCLs 1-4, Scheme 1). Docking of all possible 36 acylhydrazones into GTF180-ΔN active site indicated a favorable binding with high-energy scores (-7.2 to -10.4 kcal mol -1 ) similar to that of acarbose (Table S1). Compound A2H2 for instance, is anchored to subsite -1 through the maltose moiety, establishing six H-bonds with the catalytic residues Asp 1025 and Glu 1063 as well as Arg 1023, His 1135, Asp 1458, and Gln 1509. The aromatic ring of the aldehyde part is located at subsite +2 and involved in a pi-pi interaction with Trp 1065. An additional hydrogen bond is formed between C=O of the acylhydrazone linker as H-acceptor and NH 2 of Arg 1088 as Hdonor. The piperidine and tert-butyl motifs of the hydrazide in part occupy the distal region of the active site through hydrophobic interactions with Met 1092 and Asn 1138 and are partially exposed to the solvent ( Figure 3). Worth mentioning, the maltose moiety of the A2-derived acylhydrazones is forced into a different binding site (subsites -1 and +1) and a new binding mode different to that of free maltose. These modifications could affect the binding energy and recognition by the enzyme as an acceptor substrate.

Synthesis of Glucose and Maltose-Based Building Blocks
The aldehydes A1 and A2 were synthesized according to the routes shown in Scheme 2. α-D-Glucose penta-acetate (1) was reacted with a solution of HBr in acetic acid, resulting in 1-bromo glucose tetra-acetate (2). The crude product 2 was coupled to phydroxybenzaldehyde using silver (I) oxide, yielding compound 10.1002/cmdc.202000222

Accepted Manuscript
ChemMedChem This article is protected by copyright. All rights reserved. hydrophobic surface (green), polar surface (magenta), exposed (red). The maltose moiety of A2H2 occupies the catalytic subsite -1 with a glyosidic bond accessible to the catalytic residues Asp1025, Glu1063 and Asp1136 (orange); (B) 2D ligand interactions.

Accepted Manuscript
ChemMedChem This article is protected by copyright. All rights reserved. 3. [25] Deprotection of the acetate groups by sodium methoxide using the classical Zemplen deacetylation [26] gave aldehyde A1 in a quantitative yield. The same route was used for the synthesis of A2, however, the starting material β-maltose first had to be acetylated [27] (Scheme 2).

DCL Formation
Each library consisted of one aldehyde (300 µM), one group of hydrazides (300 µM each), aniline (10 mM) and DMSO (10% v/v) in sodium acetate buffer (pH 5.2). Aniline enhances the rate at which the acylhydrazone formation reaches equilibrium, as it serves as nucleophilic catalyst to form Schiff bases with the corresponding aldehydes. [28] The use of 10% DMSO as a cosolvent was feasible thanks to the stability of GTFs at a DMSO concentration of up to 20%. [29] It assures the solubility of building blocks and products, preventing any undesired shift in equilibrium due to precipitation. A desired shift in the equilibrium, also known as the template effect, was achieved by the addition of the target protein GTF180 (30 µM). The protein was added following a preequilibrated approach, i.e., after an equilibrium was reached in the blank library (3 h for these building blocks). A blank reaction (DCL without protein) was prepared in parallel for monitoring the amplification.

Monitoring the DCLs
The DCLs were allowed to stir at room temperature and were regularly monitored via UPLC-MS on hourly basis along with the zero hour sample. Samples were prepared by taking 100 µL of the corresponding library and raising pH to > 8 by the addition of NaOH (2 M, 8 µL) to freeze the equilibrium, followed by acetonitrile (100 µL) for protein denaturation and liberation of protein-bound ligands. The mixture was centrifuged at 9,720 g for 2 min, and the supernatant was subjected to UPLC-MS analysis. Samples of the blank reaction were treated in the same manner. The formation of the acylhydrazones reached equilibrium within three hours. It was at this time point that we added the GTF180-∆N and continued the analysis via UPLC-MS. The distribution of the products in the DCLs of the blank library versus the protein library can be compared by the relative peak areas from the UVchromatograms ( Figure 4, S1-S3 and Tables S2-S5). We selected the most amplified ligation product of each library for synthesis and biological evaluation (Table 1).
Since aniline could cause false-positive hits from the reaction with the aldehydes in DCC, [30] we screened the LCMS chromatograms for the possible imine products or hemiaminal intermediates and no corresponding peaks were detected. Other factors could result in false-positive or false-negative hits such as concentration and purity of the protein template, buffer composition, and binding interaction (aggregation) between the DCL members. [31] To avoid such pitfalls, we used low concentration of purified GTF180 (0.1 equivalent to the building blocks) and selected the optimum buffer and pH for DCC according to our systematic stability monitoring. [20] Furthermore, we utilized cheminformatics to exclude potential aggregators from the DCLs via the aggregator advisor. [32] Interestingly, analysis of DCL2, featuring the maltose-derived aldehyde component, in the presence of GTF after 6 h revealed the formation of two compounds that show molecular weights corresponding to acylhydrazones of the glucose-based aldehyde A1 (A1H2 and A1H8). These compounds can result from cleavage of the α-glycosidic bond between the two sugar units of maltose for the DCL2 members (A2H2 and A2H8) ( Figure 4 and Table S3). Both compounds possess a non-planar hydrophobic alicyclic substituent on the hydrazide moiety. This finding indicates that these compounds can indeed bind to the active site, act as a substrate and get cleaved by the enzyme in agreement with our docking study ( Figure 3). Accordingly, we also synthesized an example of the substrate molecules (A2H2, featuring the hydrazide H2 that showed favorable amplification in DCL1) and the hydrolysis products (A1H2 and A1H8) in order to investigate their affinity and inhibitory activity toward GTF.
Analysis of the NMR spectra of the prepared acylhydrazones revealed two sets of signals. This can be ascribed to the presence of cis/trans-amide CO-NH conformers of the energetically more stable E imine C=N configuration as the Z isomer is usually disfavored by steric repulsion. [33][34][35] Owing to the reversible nature of the acylhydrazone linkage, we investigated the chemical stability of the compounds under the conditions of SPR and GTF180 activity assays. Encouragingly, all six acylhydrazones showed marked stability in physiological pH 7.4 at room temperature for 24 h as well as in acidic pH 4.7 at 37 °C up to 3 h with less than 5% degradation as indicated by the UPLC chromatograms ( Figures S4-S15). These results are in agreement with previous findings using UV/VIS spectroscopy and 1 H NMR techniques. [34,36]

Binding Studies by Surface Plasmon Resonance (SPR)
We evaluated the binding affinities of the DCC hits, the substrate compound A2H2 and the cleavage products using SPR. The Lactobacillus reuteri GTF180-ΔN was immobilized covalently to a carboxymethyldextran-coated sensor chip via amine coupling. In order to ensure that the active site was accessible after immobilization, we used acarbose as a positive control acting via a competitive inhibition mechanism. [10] Binding affinity determination for acarbose showed a submillimolar dissociation constant (K D ) of 0.18 mM in line with the reported values for GTFs from other bacterial strains. [10,37] Subsequently, we determined the binding affinities of our hit compounds using 6-9 different concentrations in the range of 0.0076-2.0 mM according to their solubility. Except for compound A2H6, all compounds showed concentration-dependent binding responses to GTF180-ΔN, yet with K D values mainly about one order of magnitude higher than acarbose (Table 1 and Figures S17-S23). Generally, the glucosebearing acylhydrazones exhibit higher affinity than the maltose derivatives with compound A1H12 displaying the most promising K D value of 0.4 mM. On the other hand, the sensorgrams of the least amplified hit A2H6 showed low responses up to a concentration of 1 mM and therefore its affinity could not be determined ( Figure S22). These unexpectedly weak affinities of the new acylhydrazones may be attributed to the intrinsic weak affinity and very weak inhibitory activity of their sugar moieties especially maltose for GTF. [10] Moreover, the hydrophobic hydrazide moieties of the compounds seem to be unfavorably accommodated in the mainly hydrophilic active site of GTF180. Furthermore, the rather rigid and planar acylhydrazone scaffold possibly hinders a proper fit into the binding cavity.

GTF180 Activity Assay
The inhibitory effects of the compounds on GTF180 was assessed via monitoring the hydrolysis of sucrose into glucose A2H6 A1H8 A B A1H2 10.1002/cmdc.202000222

Accepted Manuscript
ChemMedChem This article is protected by copyright. All rights reserved.
using the glucose oxidase/peroxidase (GOPOD) analysis. Acarbose and the compounds at a concentration of 0.5 mM were incubated with the enzyme for 30 min at 37 °C, then sucrose was added, and the hydrolysis products were determined over time to calculate the GTF activity. Consistent with our affinity results, only acarbose showed about 70% inhibition of GTF180 activity at 0.5 mM, whereas none of the compounds exhibited significant inhibition at the same concentration ( Figure S24). Besides the moderate to weak affinity of the compounds, the non-observed residual inhibitory activity can be ascribed to the presence of a high concentration of the substrate sucrose in the assay (20-fold more than that of the compounds). Such large excess of the substrate concentration can diminish or abolish the effects of competitive inhibitors. [38] Altogether, these results suggest that the new GTF ligands most probably act via competitive inhibition in agreement with our design rationale.

Conclusions
We described the first application of DCC to the bacterial glucosyltransferase 180 belonging to the GS family, a potential target for combating dental caries. We designed our compounds to bear a glucose or maltose anchor targeting the active site, resembling natural substrates, as a rational starting point. These molecules were then expected to grow into the pocket by 18 different aromatic or aliphatic tails via DCC. A docking study of the 36 DCC products into the active site of GTF180 supported our rationale showing high binding energy scores. By separating the complex library into four individual DCLs, we were able to analyze the DCC experiments without overlap among the DCL components. UPLC-MS analyses of the DCLs resulted in the identification of four most amplified hit compounds, which we synthesized and evaluated for their biophysical and biochemical properties via SPR and a GTF-activity assay. Remarkably, we discovered that the maltose-derived acylhydrazones A2H2 and A2H8 with a lipophilic 5/6-membered alicyclic motif can be cleaved by GTF180, acting as glucose-donor substrates, in contrast to the parent maltose, which is known as an acceptor. This indicates that modification of maltose at the C1-hydroxyl group can alter its recognition by GTF from an acceptor substrate to a donor. Acarbose showed moderate binding affinity for GTF180 (K D 0.18 mM) and partial inhibition at 0.5 mM. In comparison to acarbose, the hit compounds showed only moderate to low affinities to GTF180. Results of the activity assay are in line with the SPR measurements, showing no pronounced inhibition at 0.5 mM, most probably due to weak affinity of the sugar units, hydrophobicity of the hydrazide tails, and overall rigidity of the scaffold due to the acylhydrazone linker. Nevertheless, our endeavor for targeting the sugar-binding site of GTF180 using DCC resulted in the identification of moderate to weak binders that are indeed capable of binding to the active site as indicated by the cleavage of the maltose derivatives A2H2 and A2H8. This work demonstrates the utility of DCC for notoriously challenging targets such as sugar-converting enzymes with inherently weak ligand interactions and millimolar affinity. [22] Besides saving time and resources, DCC can be particularly advantageous in absence of structural information or a known ligand to afford novel binders. We gave insight into some challenges encountered by using a carbohydrate-based scaffold for inhibiting GTFs. Optimization of the physicochemical parameters such as topological polar surface area, flexibility, and water solubility would be required in order to improve the affinity and inhibitory activity of this class. Alternatively, exploration of non-carbohydrate chemical scaffolds should be envisaged.

Accepted Manuscript
ChemMedChem This article is protected by copyright. All rights reserved.

4-(β-D-Glucopyranosyloxy)benzaldehyde (A1)
Compound A1 was synthesized by a slight modification of the reported method. [44] p-(Tetraacetyl-β-D-glucopyranosyl)benzaldehyde 3 [40,41] (225 mg, 0.5 mmol) was dissolved in 3 mL dry MeOH in a flame-dried flask under nitrogen. Next, a methanolic NaOMe solution (1.5 M, 0.5 mL) was added dropwise, and the reaction was stirred for 4 h at room temperature. The reaction was neutralized with DOWEX 50WX8 hydrogen-form ion exchange resin, filtered and passed through a pad of charcoal to remove any colored impurities and then evaporated to dryness in vacuo to obtain pure compound A1 (137 mg, quantitative). NMR spectra show that A1 is present as a hydrate form in the NMR solvent.

4-(4-O-α-D-Glucopyranosyl-β-D-glucopyranosyloxy)benzaldehyde (A2)
p-(Heptaacetyl-β-D-maltosyl)benzaldehyde 7 [40] (340.7 mg, 0.46 mmol) was dissolved in 5 mL dry MeOH in a flame-dried flask under nitrogen. Next, a methanolic NaOMe solution (1.5 M, 1 mL) was added dropwise, and the reaction was stirred for 4 h at room temperature. The reaction was neutralized with DOWEX 50WX8 hydrogen-form ion exchange resin, filtered and passed through a pad of charcoal to remove any colored impurities and then evaporated to dryness in vacuo to obtain pure compound A2 (142 mg, 70%). 1  General procedure for acylhydrazone formation (GP1): [36] To the hydrazide (1 equiv) dissolved in MeOH, the corresponding aldehyde (1.2 equiv) was added. The reaction mixture was stirred at room temperature or refluxed until completion. After cooling to room temperature, the reaction mixture was concentrated in vacuo. Purification of acetylated products was performed by column chromatography and deprotected sugars were purified by preparative HPLC, affording the corresponding acylhydrazone in 60% to quantitative yields. General procedure for the deprotection of the acetyl groups (GP2): [26] The classical Zemplén deacetylation method of the O-acetyl protecting groups with sodium methoxide in MeOH at room temperature was used.
The O-acetyl protected sugar was dissolved in MeOH (0.01 M), and a catalytic amount of sodium methoxide (0.15 equiv) was added. The reaction mixture was stirred at room temperature until complete deprotection was achieved.

Preparation of ligands and protein structure for docking
The 2D structures of 36 acylhydrazone products of DCL1-4 were sketched using ChemDraw professional 17.0 and were pasted to the MOE window. The compounds were subjected to an energy minimization up to a gradient of 0.01 kcal mol −1 Å 2 using the MMFF94x force field then they were saved as mdb file. In the database viewer window, the acylhydrazone structures were washed via compute | molecule | wash command. Deprotonation of strong acids and protonation of strong bases were performed by choosing the dominant protonation at pH 7 option in the wash panel. X-ray crystal structure of the Lactobacillus reuteri 180 GTF180-ΔN in complex with maltose (PDB code: 3KLL) [5] was used to perform the molecular docking study. Potential was set up to Amber10:EHT as a force field and R-field for solvation. Addition of hydrogen atoms, removal of water molecules farther than 4.5 Å from ligand or receptor, correction of library errors, and tethered energy minimization of binding site were performed via QuickPrep module.

Ligand-receptor docking
The binding site was set to dummy atoms, which were calculated by the site finder command, and the amino acid residues were chosen where maltose binds in the GTF180-ΔN active site. Docking placement was triangle matcher with an induced fit refinement option. The first scoring function was alpha HB with 100 poses, followed by a refinement score affinity dG with 10 poses.

Chemical stability determinations
Stability studies of the compounds in the SPR and GTF activity assays were performed as previously reported [47] with slight modifications. Two sets of samples were prepared according to the assay conditions. For SPR, 5 µL of 2 mM stock solution of the compounds in DMSO was added to 95 µL of HEPES buffer (10 mM HEPES, 150 mM NaCl, 3 mM EDTA, 0.005% v/v tween 20, pH 7.4) and vortexed to attain final concentration of 100 µM in a total volume (100 µL) containing 5% DMSO. The solutions of the compounds were allowed to stir in a shaking mixer at 10 rpm at rt. Aliquots of 5 µL were taken from the samples at time 0 In a gradient run, the percentage of B was increased from an initial concentration of 5% at 0 min to 100% at 4.2 min and kept at 100% for 0.8 min. The injection volume was 5 µL and flow rate was set to 600 µL/min. For compound A2H6, the initial concentration of B was 1% and the injection volume was 10 µL. Column temperature was 40 °C and UV tracing was acquired at wavelength of 310 nm. MS (HESI) analysis was carried out at a spray voltage of 3000 V (positive), and -2000 V (negative), and an ion transfer tube temperature of 300 °C. Spectra were acquired in positive and negative modes from 100 to 1000 m/z.

Binding studies by surface plasmon resonance (SPR)
The SPR experiments were performed using a Reichert SR7500DC surface plasmon resonance spectrometer (Reichert Technologies, Depew, NY, USA), and medium density carboxymethyl dextran hydrogel CMD500M sensor chips (XanTec Bioanalytics, Düsseldorf, Germany). Double distilled (dd) water was used as the running buffer for immobilization. HEPES buffer (10 mM HEPES, 150 mM NaCl, 3 mM EDTA, 0.005% v/v tween 20, pH 7.4) containing 5% v/v DMSO was used as the running buffer for binding study. All running buffers were filtered and degassed prior to use. GTF180 (117 kDa) was immobilized in one of the two flow cells by amine coupling according to reported procedure. [24,48] The other flow cell was left blank to serve as a reference. The system was initially primed with borate buffer 100 mM (pH 9.0), then the carboxymethyldextran matrix was activated by a 1:1 mixture of N-ethyl-N′-(3-dimethylaminopropyl)carbodiimide hydrochloride (EDC) 100 mM and Nhydroxysuccinimide (NHS) 100 mM at a flow rate of 10 µL/min for 7 min. GTF180 5 µM in 10 mM sodium acetate buffer (pH 4.0) was injected at a flow rate of 10 µL/min for 7 min. Non-reacted surface was quenched by 1 M ethanolamine hydrochloride (pH 8.5) at a flow rate of 25 µL/min for 3 min. A series of 10 buffer injections was run initially on both reference and active surfaces to equilibrate the system resulting in a stable immobilization level of approximately 2000 µ refractive index unit (μRIU) ( Figure S16). Binding experiments were performed at 20 °C. Compounds dissolved in DMSO were diluted with the running buffer (final DMSO concentration of 5% v/v) and were injected at a flow rate of 30 μL/min. Single-cycle kinetics were applied for KD determination. The association time was set to 60 s, and the dissociation phase was recorded for 120 s. Ethylene glycol 80% in the running buffer was used for regeneration of the surface. Differences in the bulk refractive index due to DMSO were corrected by a calibration curve (nine concentrations: 3-7% v/v DMSO in HEPES buffer). Data processing and analysis were performed by Scrubber software (Version 2.0c, 2008, BioLogic Software). Sensorgrams were calculated by sequential subtractions of the corresponding curves obtained from the reference flow cell and the running buffer (blank). SPR responses are expressed in resonance unit (RU). The KD values were calculated by global fitting of the kinetic curves as well as fitting of the steady state binding responses to a 1: 1 Langmuir interaction model.