Sulfonylurea Bioisosteres of Sulfonic Acid in JFD 00458‐Based ST6GAL1 Inhibitors Enhance Membrane Permeability Authors: Natan Koraj, Sushmaa Dangudubiyyam, Adelyn M Betances‐Mora, Ema Faganeli, Anne Harduin‐Lepers, Jesús Angulo, Tan Khanh Nguyen, María Bort‐Griñó, Ramon Hurtado‐Guerrero, Marko Anderluh Published: 2026 ABSTRACT AbstractSialyltransferases catalyze the transfer of sialic acid to glycoconjugates within the Golgi apparatus, generating cell‐surface sialoglycans that regulate cell–cell and immune communication. This process can be modulated by inhibiting α−2,6‐sialyltransferase 1 (ST6GAL1), one of the most important sialyltransferases. Most β‐galactoside ST6GAL1 inhibitors are polar CMP‐Neu5Ac derivatives with limited membrane permeability, a limitation shared by the noncarbohydrate inhibitor JFD 00458. Here, we designed JFD 00458 analogs to improve permeability while maintaining ST6GAL1 inhibition and direct target engagement. Incorporation of sulfonic‐acid bioisosteres, particularly sulfonylureas, improved passive permeability, as measured by parallel artificial membrane permeability assay, with the best analog reaching log(P_app [cm/s]) = −4.3. Potency was also improved in an expanded series of sulfonylureas, with the best inhibitor displaying an IC50 of 1.9 µM. Microscale thermophoresis confirmed direct ST6GAL1 binding, with similar apparent K d values of approximately 22–25 µM for JFD 00458 and the optimized sulfonylureas. Saturation transfer difference NMR identified the substituted phenoxy phenyl core as a common ST6GAL1‐contacting epitope, while closer contacts from the sulfonylurea‐linked aromatic substituents in 6d and 6e were associated with the highest inhibitory potency across the series. These results establish sulfonylureas as membrane‐permeable sulfonic‐acid bioisosteres and provide a structure–activity framework for further ST6GAL1 inhibitor optimization.Keywords: α−2,6‐sialylatransferase; enzyme inhibitor; glycosyltransferase; sialyltransferase; ST6GAL1 inhibition Introduction Aberrant sialylation is one of the most common glycosylation changes observed in cancer [1, 2]. In particular, terminal β‐galactoside α−2,6‐sialylation mediated by β‐galactoside α−2,6‐sialyltransferase (ST6GAL1) has been strongly implicated in cancer progression [3], playing a role in cancer hallmarks such as PI3K/AKT signaling [4], HER2/ERK activation [5], and Siglec‐mediated immune regulation [6]. This ultimately leads to cancer progression and metastasis, immune evasion, and chemoresistance; most notably in colorectal [7], gastric [8], and prostate cancer [9]. Modulation of ST6GAL1 activity through enzymatic reversal [10], inhibitor treatment [11], or gene knockout [9] has resulted in reduced disease progression, establishing ST6GAL1 as a promising target for cancer drug development. However, potent and drug‐like ST6GAL1 inhibitors are still lacking. Reflecting the challenging nature of the bisubstrate active site, shallow hydrophobic pockets and numerous polar contacts revealed in crystal structures with natural donor and acceptor [12, 13], potent ST6GAL1 inhibitors have been predominantely CMP‐Neu5Ac mimetics [14, 15]. A fundamental flaw of CMP‐Neu5Ac mimetics is their lack of cell membrane permeability due to their high polarity, and consequently, the inability to reach their target in the Golgi apparatus. The first cell‐based inhibition studies of sialyltransferases (STs) were only possible after the development of a cell‐permeable prodrug of the pan‐ST metabolic inhibitor, 3FAx‐Neu5Ac [16]. The requirement for cell‐membrane permeability and affinity for the target of hydrophilic substrates is the fundamental limitation that underscores the unmet need for ST6GAL1 inhibitors. In this work, we used the previously reported noncarbohydrate ST inhibitor JFD 00458 as a starting point for the development of membrane‐permeable ST6GAL1 inhibitors [17]. Because the ionizable sulfonic acid of JFD 00458 contributes to its high polarity and poor permeability, we designed analogs in which this group was replaced by common and noncanonical sulfonic‐acid bioisosteres selected to retain acidic functionality while modulating pKa, hydrogen‐bonding geometry, lipophilicity, and predicted logD [18]. In the absence of a ST6GAL1–JFD 00458 cocrystal structure, despite our attempts and previous efforts [19], analog design relied on ligand‐based bioisosteric replacement and systematic substituent variation. The initial bioisostere set, comprising compounds 5a–6a, was used to assess the tolerance of the scaffold to different acid replacements (Scheme 1). Among these, the sulfonylurea derivative 6a retained ST6GAL1 inhibition and provided a synthetically tractable handle for further diversification. Given the limited synthetic accessibility of broader modifications to the pyrazolone‐containing scaffold, we therefore expanded this series to compounds 6b–6f, introducing aliphatic and aromatic substituents on the sulfonylurea to explore their effects on inhibitory potency and passive permeability (Scheme 1). The resulting compounds were evaluated using the recently developed microplate sialyltransferase assay (MPSA), which enables determination of inhibitory potency and inhibition mode [20], and by parallel artificial membrane permeability assay (PAMPA) [21]. Because ST6GAL1 is a bisubstrate enzyme and the enzymatic assay includes a glycoprotein acceptor, selected compounds were further characterized by microscale thermophoresis (MST) and saturation transfer difference (STD) NMR to confirm direct target engagement, determine apparent binding affinities, and map ligand epitopes. Chemical structures of JFD 00458 and developed analogs highlighting the bioisosteric replacement strategy and scaffold expansion. Results and Discussion Inhibitory Activity Against ST6GAL1 and Inhibition Mode of JFD 00458 Analogs The inhibitory activity of the synthesized compounds on ST6GAL1 mediated transfer of sialic acid from donor to acceptor was determined using a recently published MPSA [20]. Initial derivatives were first screened at a single concentration (50 µM), and the compounds that showed significant inhibition were characterized further by determining their IC50 and analyzing their inhibition mode (Table 1). The full IC50 data with corresponding confidence intervals are reported in Table 1. Among the first series, compounds 5c and 6a, both containing acidic groups, exhibited significant ST6GAL1 inhibition. The remaining neutral derivatives, 5a and 5b, showed no significant activity at the screening concentration. This is in line with the observation that natural substrates and potent inhibitors contain anionic or ionizable acidic groups, which may help mimic electrostatic interactions made by the phosphate moiety or carboxylic acid of CMP‐Neu5Ac [15]. Compound 5d, however, did not demonstrate significant inhibition despite containing a carboxylic acid. This confirms that simple replacement of the sulfonic acid moiety with another acidic group is insufficient and that additional interactions are required for retaining potency. TABLE 1Enzymatic inhibition of ST6GAL1 by JFD 00458 analogs. IC50 values represent the mean of n = 3; 95% CI reported in brackets; outliers excluded. Modes of inhibition determined from Michaelis–Menten and Lineweaver–Burk plots (Figures 1 and S1–S3).CompoundIC50, µM 95% CI, µMMode of inhibition JFD 00458 4.0[2.6–6.5]Noncompetitive/ mixed 5a >50.0// 5b >50.0// 5c 23.3[17.3–31.5]Competitive 5d a >100.0// 6a 30.7[18.3–55.0]Competitive 6b a 651.1[232.2–N.D.]Competitive 6c 8.3[3.0–20.2]Noncompetitive 6d 1.9[1.6–2.3]Noncompetitive 6e 2.6[1.8–3.7]Noncompetitive 6f b 22.7[14.7–35.3]Competitiveb a Assay performed with a different batch of enzyme. Enzyme concentration increased (4–30 ng) to achieve comparable V max between batches.b Compound shows competitive character only at the top concentration (50 µM) in Michaelis–Menten plot. Enzymatic inhibition of ST6GAL1 by JFD 00458 analogs. IC50 values represent the mean of n = 3; 95% CI reported in brackets; outliers excluded. Modes of inhibition determined from Michaelis–Menten and Lineweaver–Burk plots (Figures 1 and S1–S3). Assay performed with a different batch of enzyme. Enzyme concentration increased (4–30 ng) to achieve comparable V max between batches. Compound shows competitive character only at the top concentration (50 µM) in Michaelis–Menten plot. The initial active compounds showed reduced potency compared to the hit compound, JFD 00458, and the IC50 of JFD 00458 was lower than previously reported (4.0 vs. 10.8 µM, expected across differing assays) [22]. JFD 00458 was previously reported to be a competitive inhibitor of ST3GAL1 with respect to CMP‐Neu5Ac [17]. However, in our experiments using CMP‐SiaNAl as a donor substrate, JFD 00458 did not exhibit a competitive inhibition pattern, as evidenced by a decrease in apparent Vmax with increasing inhibitor concentration. (Table 1, Figure 1A). Instead, the kinetic behavior was more consistent with noncompetitive or mixed inhibition. A similar kinetic profile was observed for the most potent analogs, 6c–6e, which were best described by a noncompetitive inhibition model. In contrast, compounds 5c, 6a, 6b, and 6f displayed kinetic profiles more consistent with competitive inhibition. These differences may reflect distinct binding orientations or different degrees of overlap with the donor‐substrate binding region. Michaelis–Menten (left) and Lineweaver–Burk (right) plots of ST6GAL1 kinetics in the presence of compounds: (A) JFD 00458 displaying noncompetitive/mixed inhibition type, the trendline corresponding to 8 μM JFD 00458 was omitted from the Lineweaver–Burk plot, as near‐complete enzyme inhibition precluded meaningful linear fitting; (B) 6a displaying competitive inhibition with directional size increase of molecule; and (C) 6d reverting to noncompetitive inhibition type, indicating different binding modes (see Figure S3 for the remaining compounds). Data are mean ± SD, n = 3; outliers excluded. Due to the potential for derivatization and modulation of lipophilicity within the sulfonylurea (SU) derivatives, we designed a second series of SU‐containing derivatives (6b–6f) based on compound 6a. This series features systematic variation of the SU substituent to probe binding site properties. Compounds 6d and 6e displayed IC50 values of 1.9 and 2.6 µM, respectively, making them the most potent ST6GAL1 inhibitors in the series and showing greater or comparable potency than the parent hit JFD 00458 (Table 1). Notably, increasing substituent size alone did not substantially affect activity, as reflected by the similar IC50 values of compounds 6a and 6f (30.7 and 22.7 µM, respectively). However, the enhanced potency of 6d and 6e, which both bear aromatic groups, suggests a meaningful contribution from π–π or cation–π interactions. Additionally, kinetic characterization of the starting compound JFD 00458, initial lead 6a, and most potent inhibitor 6d illustrates a different inhibition mode across the optimization campaign (Figure 1). Michaelis–Menten and Lineweaver–Burk analysis revealed noncompetitive and competitive inhibition modes, respectively, with noncompetitive inhibitors displaying greater potency across the series. This provides the first glimpse of a structure–activity relationship within the SU series. Artificial Membrane Permeability Assay As previously established, membrane permeability is an essential property for effective ST6GAL1 inhibitors in a cellular environment. All inhibitors 5c, 6a–6f were therefore evaluated using the well‐known PAMPA‐GIT model, yielding applicable permeability values (log(P_app [cm/s]) (Table 2) [21]. In the PAMPA assay, passive permeability is estimated by measuring the diffusion of molecules from an aqueous donor well to an acceptor compartment separated by a lipid membrane mimicking the cell membrane and serves as a high‐throughput tool for estimating cell permeability [21]. Prior to synthesis, logP, logD at pH 7.4, and pKa values were calculated using the online tool Chemicalize (Chemaxon) to guide analog design toward increased predicted logD (Table 2) [23]. TABLE 2Predicted logP,a pKa,a and logDa in comparison with passive permeability log(P_app [cm/s]) (n = 3, ± SD) measured by PAMPA (Bienta) for synthesized compounds, JFD 00458, and reference compounds.Compoundlog(P_app, cm/s)Predicted logPPredicted pKaPredicted logD 7.4 pH Quanidine −4.2 ± 0.030.0 / / / Ranitidine −6.5 ± 0.12 / / / Verapamil −4.1 ± 0.05 / / / JFD 00458 N/Ab 0.7−2.2−1.3 5c N/Ab 2.25.70.8 6a N/Ab 0.85.6−0.1 6b −5.6c 0.95.6−0.0 6c −5.3 ± 0.132.25.61.3 6d −5.8d ± 0.072.54.81.5 6e −5.4d ± 0.032.25.51.3 6f −4.3 ± 0.032.55.61.6 Note: Reference compounds shown in italics.a Physicochemical properties were calculated using Chemicalize (ChemAxon Ltd., Budapest, Hungary; https://chemicalize.com).b N/A – apparent permeability is not reported; the compound was not detected in the acceptor compartment under the assay conditions, indicating negligible membrane permeability.c Reported is apparent permeability from one quantifiable measurement; the compound was not detected in the acceptor compartment in the remaining replicates.d Reported is apparent permeability from two quantifiable measurements; the compound was not detected in the acceptor compartment in one replicate. Predicted logP,a pKa,a and logDa in comparison with passive permeability log(P_app [cm/s]) (n = 3, ± SD) measured by PAMPA (Bienta) for synthesized compounds, JFD 00458, and reference compounds. Note: Reference compounds shown in italics. Physicochemical properties were calculated using Chemicalize (ChemAxon Ltd., Budapest, Hungary; https://chemicalize.com). N/A – apparent permeability is not reported; the compound was not detected in the acceptor compartment under the assay conditions, indicating negligible membrane permeability. Reported is apparent permeability from one quantifiable measurement; the compound was not detected in the acceptor compartment in the remaining replicates. Reported is apparent permeability from two quantifiable measurements; the compound was not detected in the acceptor compartment in one replicate. The commonly used thresholds for log(P_app [cm/s]), as reported by Bienta, are in the range of >−5.0 and >−6.0, which correspond to high and low permeability, respectively [24]. Evaluation of the synthesized compounds revealed moderate membrane permeability for compounds 6c–6e, and high permeability for the compound 6f (logP_app of −4.3) (Table 2) [24]. The generally accepted optimal logD for permeable compounds is between 1 and 3 [25]. In line with expectations, JFD 00458 was not detected in the acceptor compartment under the PAMPA assay conditions, indicating negligible membrane permeability. Similarly, compound 6a, which has a predicted logD well below 1, showed no detectable permeability. Compound 5c, which bears the lipophilic acidic tetrazole group and has a predicted logD close to the threshold, was likewise not detected in the acceptor compartment. Compound 6b, which also has a predicted logD below 1, exhibited only limited permeability, with apparent permeability being quantifiable in a single replicate. In contrast, compounds 6c–6e, with predicted logD at pH 7.4 above the threshold of 1, demonstrated measurable membrane permeability. Although one replicate for both 6d and 6e was below the detection limit, quantifiable transport was observed in the remaining replicates, supporting the conclusion that these compounds possess measurable permeability. Compound 6f, possessing the highest predicted logD in the series, was the most permeable overall. These findings confirm that predicted logD can reliably guide the design of membrane‐permeable inhibitors within the SU series. Taken together with the initial structure–activity relationship (SAR) findings, the compounds 6c–6e, which incorporate the SU moiety into the JFD 00458 structure, represent a promising starting point for hit‐to‐lead development. Binding Affinity Determined by MST To confirm genuine hits, enzyme inhibition should be validated by an orthogonal or secondary assay [26]. This is particularly important for ST6GAL1, where an apparent inhibitory effect could result from compound interaction with the glycoprotein acceptor rather than the intended target. MST has recently been applied to study protein–ligand interactions in glycosyltransferases [27], and is well‐suited for this purpose as it directly confirms inhibitor binding and determines binding affinity while being sensitive and material‐efficient [27, 28]. The most promising candidates with respect to inhibitory activity and membrane permeability—compounds 6c–6e and JFD 00458 as control, were evaluated by MST. The His‐tagged GFP‐ST6GAL1 fusion protein was used as the target, and the His‐tagged GFP was used as a control. This His‐tagged GFP‐fused construct was used exclusively for MST measurements. Initial attempts to measure the MST signal via GFP fluorescence were unsuccessful due to excessive fluorescence quenching, which rendered the system unsuitable. The presence of histidine tags on both constructs provided an alternative labeling strategy. Following confirmation of the absence of ligand–dye interference, the assay was performed using RED‐tris‐NTA 2nd Generation dye as the fluorophore [28]. Binding affinities (K d) were determined by plotting the fraction bound against ligand concentration. The results demonstrated that compounds 6c, 6d, 6e, and JFD 00458 exhibit comparable binding affinities for the His‐tagged GFP‐ST6GAL1, with K d values of 26.5, 25.4, 22.3, and 25.3 µM, respectively (Figure 2). As expected, compounds 6c–6e did not elicit an MST response in the His‐tagged GFP control measurements. JFD 00458, however, produced a response in the GFP control, with a binding affinity of 38.8 µM, which is indicative of nonspecific binding. Notably, this promiscuity was not observed for the optimized compounds 6c–6e, suggesting that the structural modifications introduced during optimization successfully eliminated nonspecific interactions. Dose‐response curves (Fraction bound against ligand concentration) for binding to His‐tagged GFP‐ST6GAL1 (main graph) and His‐tagged GFP control (inset) acquired by MST: (A) JFD 00458, K d = 25.3 µM (95% CI: 17.6–36.3 µM); (B) 6c, K d = 26.5 µM (95% CI: 19.6–35.9 µM); (C) 6d, K d = 25.4 µM (95% CI: 18.5–35.0 µM); (D) 6e, K d = 22.3 µM (95% CI: 16.9–29.4 µM). No specific binding to GFP control detected in (B–D); GFP K d for A = 38.8 µM (95% CI: 24.8–60.6 µM). Top concentrations were outliers according to Nano Temper guide (±20% of the mean initial fluorescence) [29], saturation could not be reached (mean of n = 3 independent experiments performed in duplicate (Tables S1–S4)). No apparent irregular thermophoretic traces or trace crossover effects were observed in K d measurements. Binding Epitope Determination by STD NMR Because of the absence of structural information on the binding of the parent inhibitor JFD 00458, we first characterized its interaction with ST6GAL1 in solution by STD NMR spectroscopy (Figure 3). STD NMR is a ligand‐based technique in which selective irradiation of the protein leads to transfer of saturation to the closest protons in any bound ligand; subtraction of the resulting spectrum from a reference spectrum reveals the regions of the ligand molecule that are in contact with the protein, thereby confirming binding and enabling epitope mapping [30]. The selected set comprised JFD 00458 and compounds 5c, 6a, and 6c–6e, covering the parent inhibitor, competitive inhibitors, and the most potent noncompetitive sulfonylurea derivatives, with the final aim of relating the inhibitory profiles of selected analogs to their experimental ligand–enzyme contacts (Figure 3). Compound 6f, the most lipophilic analog in the series bearing an adamantane substituent, was excluded from the measurements due to limited aqueous solubility. Within this STD NMR subset, compounds 6d and 6e were the most potent ST6GAL1 inhibitors, with comparable IC50 values, followed by JFD 00458, 6c, and the weaker inhibitors 5c and 6a. Compounds 5c and 6a were classified as competitive inhibitors with respect to CMP‐Neu5Ac, whereas JFD 00458 and compounds 6c–6e showed noncompetitive inhibition. Thus, the IC50 values and inhibition modes suggested a nonlinear SAR that could not be explained by sulfonylurea substitution or substituent size alone. STD NMR spectroscopy was expected to clarify whether these differences were reflected in the binding epitopes and whether the increased potency of 6c relative to 6a, and of the aromatic sulfonylureas 6d and 6e relative to 6c, could be rationalized by changes in ligand–enzyme contacts. Binding epitopes for selected ST6GAL1 inhibitors determined by STD NMR spectroscopy. Relative STD enhancements identify ligand protons in closest contact with ST6GAL1 (100‐stronger contact, 0‐weaker contact). The STD NMR data showed that the diphenyl ether core of JFD 00458, 5c, and compounds 6c–6e make comparable close contacts with ST6GAL1 in the bound state whereas this core appears less engaged in compound 6a, showing lower contacts with the enzyme, and globally reduced STD intensities. This reduced core contact is consistent with the lower potency of 6a and suggests that the ethyl sulfonylurea substituent, which is making contacts with the enzyme, alters the binding orientation of the scaffold. Increasing the size of the sulfonylurea substituent in 6c with a hydrophobic cyclohexyl ring restored contacts involving the core scaffold and was accompanied by improved potency, although the inhibition mode changed from competitive to noncompetitive, suggesting a distinct binding mode. The aromatic sulfonylurea derivatives 6d and 6e showed stronger STD signals (closer contacts) in the sulfonylurea‐linked substituents than nonaromatic 6c, in line with their higher inhibitory potency. These observations suggest that additional interactions involving the aromatic substituent, potentially π–π or cation–π contacts, may contribute to enhanced binding of 6d and 6e to ST6GAL1 and to the observed enhanced inhibitory activity. However, in the absence of structural data, this interpretation remains tentative; alternatively, the weaker contacts at the cyclohexyl observed for 6c may reflect a more solvent‐exposed or less defined interaction surface for the sulfonylurea substituent. The competitive inhibition mode of 5c, despite STD NMR epitopes resembling those of the noncompetitive inhibitors, further indicates that similar ligand contact patterns do not necessarily correspond to identical binding modes or kinetic behavior. This may reflect differences in binding orientation, partial overlap with the donor‐binding region, or indirect effects on donor binding. Precise structural information on the binding pocket and the bioactive orientations of these inhibitors remain to be established and will require additional structural studies. Nonetheless, the combined STD NMR and enzymatic assay data provide a useful SAR framework for further development of the sulfonylurea series as potent inhibitors of ST6GAL1. Conclusion In summary, we report a new series of noncarbohydrate ST6GAL1 inhibitors derived from the previously reported inhibitor JFD 00458. Introduction of sulfonylurea (SU) bioisosteres of the sulfonic acid group enabled the development of inhibitors combining measurable membrane permeability with low‐micromolar ST6GAL1 inhibition, with compounds 6c–6e emerging as the most balanced analogs, and 6d/6e as the most potent inhibitors of the sulfonylurea series. We demonstrated that predicted logD can guide the design of permeable compounds within this scaffold and confirmed direct target engagement by MST. STD NMR binding epitope mapping further showed that the diphenyl ether core is a common structural feature for establishing contacts with ST6GAL1 in the bound state, whereas increased contact regions from the sulfonylurea‐linked aromatic substituents in 6d and 6e correlate with their enhanced inhibitory potency. Together with IC50 determination and inhibition‐mode analysis, these ligand‐based binding data provide a SAR framework for the further optimization of SU‐containing JFD 00458 derivatives against ST6GAL1. Future efforts will focus on improving potency and permeability through systematic scaffold refinement, obtaining structural insight into the binding pocket and ligand orientations, and evaluating inhibition of ST6GAL1‐dependent α2,6‐sialylation in cell‐based assays to assess activity in a cellular context. Experimental MST Binding affinities were determined by MST using NanoTemper Monolith NT.115 at 100% LED and MST power of 40%; laser time: 3s–20s–1s (initial fluorescence—MST on time—back‐diffusion). His‐tagged GFP‐ST6GAL1 fusion protein (target) and His‐tagged GFP (control) constructs (Supporting Information—Protein Expression and Purification of human ST6GAL1) were fluorescently labeled via their His‐tag using the RED‐tris‐NTA 2nd Generation dye (NanoTemper Technologies) at a 75 nM target and 25 nM dye concentrations for 30 min at room temperature in buffer (25 mM Tris, 150 mM NaCl, 0.05% Tween‐20, pH 7.4). Ligand concentrations ranged from 300 to 0.15 µM in a two‐fold serial dilution series. Samples were prepared in Corning 384‐well microplates by mixing 19.5 µL of labeled protein solution with 0.5 µL of compound from each dilution point and incubated for 30 min at room temperature prior to capillary loading. Measurements were carried out using 24‐capillary Premium Coated cassettes (2 x 12 ligand concentrations). K d values were determined using a laser‐on time of 4–5 s and derived from Fraction bound (Fraction bound = (F norm − F norm unbound)/ (F norm bound − F norm unbound), where F norm = F hot/F cold) plotted against ligand concentration in µM. Measurements in which the initial fluorescence deviated by more than 20% from the mean were excluded as outliers. Affinity measurements were performed in technical duplicates across three independent measurements (Tables S1–S4). PAMPA PAMPA‐GIT assay was performed by Bienta, Enamine Biology. All steps of the PAMPA were carried out according to the pION Inc. PAMPA Explorer Manual [31]. All compounds were tested in triplicate. Prisma HT buffer (pH 7.4) containing 50 μM test compounds (0.5% DMSO) was added to the Donor Plate wells. Acceptor Sink Buffer was added to each well of the Acceptor Plate. Incubation was done at room temperature for 4 h without stirring. After incubation, aliquots from both plates were transferred to optic UV–Vis plates, and optic plates were read on a microplate reader in absorbance mode in the range of 230–550 nm with a 4 nm step. Then the apparent permeability coefficient log(P_app [cm/s]) was calculated accordingly (Table S5). MicroPlate Sialyltransferase Assay The MPSA was performed as per [20] with the following modifications: 400 ng asialofetuin was dissolved in 100 µL of 20 mM sodium bicarbonate buffer pH 9.6 and used to coat the wells on a 96‐well microplate (Nunc‐Immuno standard strip plate, Merck Z755273) and allowed to rest overnight at room temperature [20]. The wells are washed with 125 µL 1xPBST‐0.05% two times and once with 125 µL 1x PBS and then treated with 100 µL oxidized BSA [32] for 1 h at room temperature for blocking [32]. The wells were washed again as stated above and the sialylation reaction was performed in 100 µL by the sequential addition of 1. increasing concentration of donor CMP‐SiaNAl (0–400 µM) and 2. increasing concentration of inhibitor compounds (0–100 µM), 3. 4 ng (1.8 nM) recombinant human ST6GAL1 (RnD systems 7620‐GT‐010) in Tris‐Cl buffer pH 7.4 (25 mM Tris, 150 mM NaCl, 5 mM CaCl2 pH 7.4). The reactions were performed in triplicates and the reaction mixture was added to the microplate and incubated at 37 °C for 1 h. The microplate wells were washed again, and a biorthogonal labeling reaction was performed for detection of the transferred modified sialic acid donor SiaNAl. 300 µM CuSO4, 600 µM BTTAA (Jena BioScience CLK‐067−100), 2.5 mM sodium ascorbate, and 250 µM azidobiotin (Sigma–Aldrich 762 024), were mixed and made up to a volume of 100 µL with 1x PBS and added to the wells. This was allowed to incubate at 37 °C for 1 h, following which the plates were washed again. Next, 100 µL of anti‐biotin antibody conjugated with HRP (1/25000) (Ozyme Jackson Immunoresearch 200‐032‐211) was made in 1x PBST was added to each well and is incubated at 37 °C for 1 h. After this, the wells were washed again and 100 µL TMB reagent (Sigma–Aldrich T8665) was added and the absorbance was measured at 620 nm after 20 min. using a spectrophotometer (SpectroStar Nano, BMG Labtech). The absorbance measured as a result of the oxidation of TMB corresponds to the amount of sialic acid transferred by the enzyme on to the acceptor in the presence of increasing concentrations of inhibitor versus donor. The apparent Km and Vmax, were computed and MichaelisMenten and Lineweaver–Burke plots were constructed using GraphPad Prism software version 11.0.0. For the determination of IC50 values, the MPSA was performed by keeping the donor concentration at 2x Km and with increasing concentrations of the inhibitor. The resulting activity was plotted against log10 of inhibitor concentration and the IC50 value was computed using GraphPad Prism software version 11.0.0. For the one‐point screening experiment, the donor CMPSiaNAl was used at the concentration equal to the Km of the donor toward the enzyme, and the inhibitors were tested at 50 µM. The resulting activity was used to plot the relative activity percentage by using the control as 100% activity. For the control the buffer with 0.5% DMSO was used in place of the inhibitors (Figures S1–S3). Saturation Transfer Difference NMR Spectroscopy 1D 1H STD NMR experiments were performed on a 600 MHz spectrometer equipped with an UltraShieldPlus superconducting magnet and a high sensitivity cryoprobe optimized for the detection of 1H, 13C, 15N, and 19F nuclei. NMR samples were prepared in a total volume of 170 μL of D2O buffer containing 150 mM NaCl and 25 mM d‐Tris (pD 7.5). ST6GAL1 (Supporting Information—Protein Expression and Purification of human ST6GAL1) was used at a final concentration of 50 μM, while ligands were added to a final concentration of 1 mM. All experiments were carried out at 25 °C. To obtain quantitative descriptors for binding epitope mapping while minimizing T1 relaxation bias and reducing experimental time, a reduced data set STD NMR approach was employed. In this way, rather than acquiring a full STD build‐up curve, the initial growth rate of the STD build‐up (STD0) was estimated from two (short and long) saturation times using the RedDat approach [33]. On‐ and off‐resonance spectra were acquired using a train of 50 ms Gaussian‐shaped selective saturation pulses, employing two saturation times (tsat) of 0.75 and 6.0 s, with a relaxation delay (D1) of 4 s. Selective saturation of the protein was achieved by on‐resonance irradiation at 0 ppm for 5c and JFD 00458, and at −1 ppm for 6a, 6c, 6d, and 6e, targeting protein aliphatic proton resonances. Off‐resonance spectra were acquired by irradiation at 40 ppm. Residual protein resonances were suppressed by applying a T1ρ filter of 30 ms. Spectra were recorded with a spectral width of 12 kHz and 33k data points, using 1000 scans for tsat = 0.75 s and 512 scans for tsat = 6.0 s. STD NMR spectra were obtained as the difference between on‐ and off‐resonance experiments. STD0 values were used as quantitative descriptors of ligand–protein contacts by normalizing them to the highest STD0 value within each ligand to generate relative binding epitope maps. Chemistry The experimental part of the compound synthesis and the analytical data are available in the Supporting Information. Funding This study was supported by HORIZON EUROPE Marie Sklodowska‐Curie Actions (101119601), Javna Agencija za Raziskovalno Dejavnost RS (P1‐0208), Spanish Ministry of Science, Innovation and Universities (PID2022‐136362NB‐I00), and Departamento de Educación, Cultura y Deporte, Gobierno de Aragón (E34_R17 and LMP58_18). Conflicts of Interest The authors declare no conflicts of interest. Supporting information The authors have cited additional references within the Supporting Information [34, 35, 36, 37, 38, 39].