Galectin-8N-Selective 4-Halophenylphthalazinone-Galactals Double π-Stack in a Unique Pocket

Galectin-8 contains two different carbohydrate recognition domains (CRDs). Selective inhibitors for at least one CRD are desirable for galectin-8 biology studies and potentially for pharmacological purposes. Structure-guided design led to the discovery of potent and selective glycomimetic–heterocycle hybrid ligands, with a 4-(p-bromophenyl)phthalazinone derivative displaying a 34 μM Kd for galectin-8N (N-terminal CRD), no binding to galectin-8C (C-terminal CRD), -1, -3, -4N, -7, -9C, or -9N, and >40-fold selectivity over galectin-4C. Selectivity was achieved with the halogenated 4-phenylphthalazinone moiety occupying a galectin-8N-specific sub-pocket. A 1.30 Å resolution X-ray structure revealed the phthalazinone moiety stacking with Arg45 and the 4-bromophenyl moiety stacking both Arg59 and Tyr141 of galectin-8N. Physicochemical and in vitro ADME studies revealed a desirable LogD, which also translated to good passive permeability. The chemical, microsome, and plasma stability support these compounds as promising tool compounds and candidates for hit-to-lead optimization.


X-ray crystallography data collection and refinement statistics S2
Superpositions of the X-ray structure with the two poses previously predicted by MD simulation S3

Physicochemical and in vitro ADME analyses S4
Table S3.Structures of tested compounds S4 Table S4.Solubility of compounds in 1% DMSO in phosphate buffer pH 7.4 S5

Chemical stability S6
Table S6.Chemical stability of compounds in acidic (pH 4) and basic (pH 9.5) condition S6

Chromatography Instrumentation S11
General procedures, synthesis, data, and copies of NMR Spectra and Chromatograms S12
Human Galectin-8N 4-158-Met was cloned as amino acid 4-158 from human galectin-8 (UniProt entry O00214-1) without fusion or tags into pET26b(+), and the plasmid was transformed into Escherichia coli TUNER(DE3).LB medium, 8 L, supplemented with 50 μg/mL kanamycin, was inoculated to OD600 = 0.1 with an overnight culture of E. coli TUNER(DE3) / pET26b(+)_hGal8N_4-158-M.The culture was grown at 25°C, 250 rpm, 1 L / flask.At OD600 = 0.46, the temperature was lowered to 18°C.At OD600 = 0.9, IPTG was added to a final concentration of 1 mM.20 hours after induction, the cells were harvested by centrifugation, resuspended in MEPBS, supplemented with three tablets of Complete Protease Inhibitor, EDTA-free (Roche) and a pinch of DNase I.The cell suspension (~100 mL) was passed two times through a French Pressure Cell at 18000 psi.The resulting lysate was ultracentrifuged, the supernatant passed through a 0.45 μm syringe filter, and then used for affinity chromatography.The 19 mL lactocyl-sepharose column was pre-equilibrated with 5 CV MEPBS before applying the sample.The run was performed with a flow rate of 2 mL/min at RT.The column was then washed with 10 CV MEPBS, and bound protein was eluted with 5 CV MEPBS+150 mM lactose.At 6 °C, 5 mL fractions were collected during elution.Pure elution fractions were pooled, concentrated, dialysed and further concentrated to give 13.1 mg/mL in 135 mL (1768 mg).The purity was >95%, estimated using SDS-PAGE analysis.

Crystallisation, X-ray data collection and atomic structure determination
Human galectin-8N was co-crystallised with lactose as described previously. 1 In brief: 12 mg/ml galectin-8N in 10 mM lactose, 10 mM Tris/HCl pH 8.0, 1 mM TCEP and 150 mM sodium chloride was mixed with 25 % (w/v) PEG 2000 monomethylether (PEG 2000 MME) and seed solution from crystals grown at the same condition (1.5 μL protein + 1.25 μL reservoir + 0.25 μL seed solution in a hanging drop over 1 ml reservoir using a NEXTAL plate).The experiment was setup at 295 K and crystals appeared within a few days.The co-crystals with lactose were used to soak in compound 10 by transferring crystals in three steps to different soaking drops.These were 2 μL drops with a soaking solution of 20% ethylene glycol, 25% PEG 2000 MME, 10 mM Tris pH 8, 1 mM TCEP, 50 mM NaCl, and 2 mM compound 10.The incubation times in the first two drops were about 10 min and in the third drop about 24 h.Then the crystals were transferred to a 2 μL drop with the same constituents and flash-frozen in liquid nitrogen.
Data were collected at Diamond Light Source beamline I24 to 1.30 Å and processed using the Xia2 pipeline 2 and Aimless. 3 The structure was determined with two molecules in the asymmetric unit in space group P212121.The structure has been refined starting from the 1.35 Å model of galectin-8N with lactose (PDB id: 7ALS) determined in the same space group.Anisotropic B-factor refinement were made using Refmac5, 4 the model building was made in Coot, 5 and the models have been analysed using Molprobity.Superpositions of the X-ray structure with the two poses previously predicted by MD simulation.

Screen solubility method
The starting materials for the assay were dissolved in 10 mM DMSO solutions.The solubility was assessed by dilution in Phosphate-buffer saline (PBS, 10 mM pH 7.4) to a concentration of 100 μM.The samples were equilibrated for 20−24 h, filtered (Millex LH 0.45 µM), and the solubility was determined by LC-UV in the reported range of 1−95 μM.LogDoctanol/water pH 7.4 Distribution coefficients between 1-octanol and 10 mM phosphate buffer pH 7.4 were determined with a shake flask method with LC−UV detection of the compound concentration in both phases.
Compound stock solutions were added to 600 μL 1-octanol.Immediately, 600 μL of buffer were added to each vial.Vortex mixing was performed after addition of buffer.The equilibrium between the biphasic system were performed during 10-20 minutes at room temperature in a Thermomixer at 500 rpm.The two phases were separated by centrifugation.Prior to LC-UV analysis, the octanol phase was diluted 10 times in methanol in a new vial.The LC sample from the buffer phase were directly taken from the bottom layer by the autosampler.

Chemical stability
Chemical stability was determined by monitoring the diminishing of the parent compound over several days in acidic and basic media.The stability was measured based on the UV response of the compound in LC-UV analysis, which diminishes with the formation of degradation products.Compound stock solutions were diluted in ammonium acetate buffers (pH 4 or pH 9.5) to a concentration up to 100 μM with 1% DMSO.The first measurement was taken immediately after dissolution of the compound.The samples were mixed in a rolling mixer for 5 days, and the samples were analysed at 3 time points (0 h, 24 h, and 120 h).Table S6 shows the remaining percentages of the parent compounds present after 5 days of agitation.

Plasma Stability
In vitro Plasma Stability Assay.
The reaction was started by the addition of the compound to blank plasma in LC vials after prewarming of the plasma at 37 °C.The initial compound concentration was 1 μM.The reaction was stopped by pipetting a sample to a new vial containing acetonitrile at seven time points (0, 0.17, 0.5, 1, 2, 4, and 8 h).Time point 0 h was about 15−20 s after spiking the plasma with the compound.The precipitated samples were centrifuged, and the supernatant were diluted and analysed by LC−MS/MS to determine the % of compound remaining.
Table S7 shows the remaining percentages of compound at each time-point and the half-life of the compounds determined by linear regression of the natural logarithm of the concentration percent over time.T½ = ln2/-k.Pooled human blood plasma (Innovative Research, Inc., IPLAK2E100ML-32215) was collected from donors via apheresis, aliquoted and shipped, and stored frozen.Each unit is tested for viral markers and found negative for HBsAg, HCV, HIV-1, HIV-2, HIV-1Ag or HIV-1 NAT, ALT, and syphilis using FDA-approved methods.To the NA Citrate-collected product was added an anticoagulant K2 EDTA.Plasma was used within one year of collection, well within the industry standard expiration of 3 years from Date of Manufacture.Stored at -20 °C.

Microsomal stability
Metabolic Stability in Liver Microsomes.
The assay was run in a 96-deep well format and solutions were prepared containing microsomes in phosphate buffer.The microsomes employed were from humans and mice.The reaction was initiated by addition of NADPH after preincubation at 37 °C with test compounds.The reaction was stopped by the addition of acetonitrile (MeCN) at six time points (0, 5, 10, 15, 25, and 45 min), the samples were centrifuged, and the supernatant was diluted to determine the loss of compound by LC−MS/MS.Two incubations per compound per species.For further details regarding working concentrations, volumes, and final concentrations at incubation, see below.
Table S8 (mouse) and Table S9 (human) show the intrinsic clearance (CLint)of each compound determined by linear regression of the initial disappearance, and the associated half-life of the compounds.The half-lives obtained in this method are depending on the microsomal concentration.

Parallel Artificial Membrane Permeability Assay (PAMPA)
The membrane permeability is determined using a pre-coated tri-layer PAMPA Plate System.The plate system is made up of a donor-and an acceptor-compartment separated by a porous filter coated with a lipid-oil-lipid tri-layer.Compound stock solutions were diluted in PBS (10 mM, pH 7.4) / 10% methanol to a concentration up to 100 μM with 1% DMSO.This compound solution was added to the donor-compartment, and the same solvent mixture without compound was added to the acceptorcompartment.After 5 hours of equilibration, the compound concentration in both the donor-and acceptor-compartment were compared based on the peak area obtained by LC-UV analysis.Table S10 shows the membrane permeability calculated from the average of four replicates.Because PAMPA is an artificial permeability assay, results must be interpreted by comparison to in vitro Caco-2 data; the relative permeability of known and tested compounds is a useful indicator of in vitro and in vivo permeability.

Chemistry
All reagents and solvents were dried before use.Commercial reagents were used without further purification.TLC analysis was performed on precoated Merck silica gel 60 F254 aluminium plates using UV light and charring solution (H2SO4: EtOH 1: 9) with heating.Flash column chromatography was performed on SiO2 purchased from Aldrich (technical grade, 60 Å pore size, 230-400 mesh, 40-63 μm).
Preparative HPLC was performed on an Agilent 1260 Infinity system with a SymmetryPrep C18, 5 μM, 19 mm × 100 mm column using a gradient (water with 0.1% formic acid and acetonitrile).Monitoring and collection were based on UV−vis absorbance at 210 and 254 nm.Automated reversed phase chromatography was performed on a Biotage® Isolera™ with a Sfär C18 D-Duo 100 Å 30 μm column column using a gradient (water and acetonitrile).Monitoring and collection were based on UV−vis absorbance at 254 nm.NMR spectra 1 H, 13 C, 2D COSY, and HMQC were recorded with a Bruker Avance II 400 MHz spectrometer (400 Hz for 1 H, 100 Hz for 13 C) or a Bruker Avance III 500 MHz spectrometer (500 Hz for 1 H, 125 Hz for 13 C) at ambient temperature.Chemical shifts are reported in δ parts per million (ppm), with multiplicity (b = broad, s = singlet, d = doublet, t = triplet, q = quartet, quin = quintet, hept = heptet, m = multiplet), coupling constants (in Hz) and integration.High-resolution mass analyses were performed using a Micromass Q-TOF mass spectrometer (ESI).Purities of final compounds were determined by UPLC (Waters Acquity UPLC system, column Waters Acquity CSHC18, 0.5 mL min−1 H2O-MeCN gradient 5-95% 10 min with 0.1% formic acid).Analytical data are given for tested compounds.All tested compounds were ≥95% pure according to analytical HPLC analysis.
General procedure A, hydrazine condensation 10 Functionalised 2-(benzoyl)benzoic acid (1 equiv.) was dissolved (with approx.5 mL/mmol) in ethanol.Hydrazine hydrate (3 equiv.) was added and the mixture was refluxed for 4 h.The mixture was allowed to cool, solids were filtered off, washed with cold EtOH, and dried in vacuo to give the product.
General procedure C, bromination with hydrogen bromide 7,11 N-methyl pivalate (1 equiv.) was dissolved (with approx.8 mL/mmol) in glacial acetic acid, and hydrogen bromide (1.5 equiv., 33% wt in acetic acid) was added.The mixture was heated to 75 °C for 2 h, and then allowed to cool to room temperature.The addition of ice precipitated the product, which was filtered off and dried in vacuo.Bromides that did not precipitate were diluted with water and the products extracted with EtOAc.The organic layers were combined, washed with brine, dried over Na2SO4, filtered, and concentrated in vacuo to give the products.

Competitive fluorescence polarisation experiments
4][15] Galectin-8N was expressed and purified as described in the supplementary information.
Fluorescence polarisation experiments were performed using the PHERAstar FS plate reader (software version 2.10 R3), and the fluorescence anisotropy of fluorescein tagged probes were measured by excitation at 485 nm and emission at 520 nm.4][15] In this assay a fixed concentration of galectin (e.g., 100 nM for Gal-8N) and trace concentration of a fluorescein-tagged probe (typically 4 nM) are mixed with a series of concentration of inhibitor to be tested.The measured anisotropy will represent an average of the value for free probe (typically around 30) and the value for 100% galectin bound probe (determined in a separate experiment), linearly weighted by their relative concentration.From this the free and bound probe concentration was calculated for each data point and, together with Kd for the galectin-probe interaction (from a separate experiment without inhibitor), the concentration of free galectin was calculated by the law of mass action.Inhibitor bound galectin concentration was calculated from total galectin concentration minus concentration of free galectin and the small amount bound to the probe, and then free inhibitor from total inhibitor minus galectin-bound inhibitor.Now with known concentrations of free inhibitor, galectin-inhibitor complex, and free galectin, the Kd of the galectin inhibitor interaction was calculated from law of mass action. 15he synthesised compounds were dissolved in pure DMSO at 20 mM concentration and diluted in PBS to 3-6 different concentrations, and each concentration was tested in duplicate.The highest inhibitor concentrations tested were 1.

Figure
Figure S1 Superposition.Crystal structure of galectin 8N-galactal-phthalazinone 10 (PDB ID: 9FXZ) with the ligand presented in orange sticks, protein coloured in red-white-blue according to B-factor.In superposition, the molecular dynamics snapshot of initial binding mode of the 4-phenylphthalazinone 2 (published previously) 7 with the ligand and the protein presented in green sticks.Hydrogen bonding network is presented with yellow dotted lines, cation-π and π-π stacking are presented with red dotted lines.(Figure was generated with PyMOL v1.7, Schrodinger LLC.)

Table S1 .
Comparative fluorescence polarisation.Kd (µM ±SEM) a [a] If not stated otherwise, results represent the mean ± SEM of n = 4 to 8 globally analysed from two independent experiments.

Table S2 . X-ray crystallography data
collection and refinement statistics.

Table S5
shows the obtained peak area and the calculated LogD for each compound.The LogD values for controls (cyclobenzaprine, glyburide) were within reference values.logD = log {[compound area in octanol]*10*50/[compound area in buffer]*5}.

Table S6 .
Chemical stability of compounds in acidic (pH 4) and basic (pH 9.5) conditions.
UV detection at 254 nm.Results over 95% are considered 'stable'.UV activity and solubility of degradation products is presumed.-Area% obtained by LC-UV directly from the sample.

Table S7 .
Plasma stability of compounds at 1 μM over 8 hours.
-Compound sample solution (6 µL, 100 µM) was spiked into the respective plasma containers -Mixture was vortexed and the first (0 h) sample (60 µL) transferred to a 'stop'-vial -At each time-point, the mixture was vortexed and a sample (60 µL) transferred to a 'stop'-vial -LC-MS/MS detection by MRM scan.

Table S8 .
Liver microsomal stability (mouse) of compounds at 1 μM over 45 minutes.CLint determined by linear regression of peak area derived from LC-MS/MS chromatogram of samples at each time point.Reference clearance values: diphenhydramine 64 ± 20, formoterol 26 ± 10.
CLint determined by linear regression of peak area derived from LC-MS/MS chromatogram of samples at each time point.Reference clearance values: dextromethorphan 25 ± 10, verapamil 150 ± 25.
UV detection at 254 nm.Linear compound concentration-UV activity is presumed.
5 mM.The average values of Kd and SEM were calculated from 4 to 8 duplicate measurements, showing 10-90% inhibition.