3-Substituted 1-Naphthamidomethyl-C-galactosyls Interact with Two Unique Sub-Sites for High-Affinity and High-Selectivity Inhibition of Galectin-3

Abstract

The galectins are a family of galactose-binding proteins playing key roles in inflammatory processes and cancer. However, they are structurally very closely related, and discovery of highly selective inhibitors is challenging. In this work, we report the design of novel inhibitors binding to a subsite unique to galectin-3, which confers both high selectivity and affinity towards galectin-3. Olefin cross metathesis between allyl β-C-galactopyranosyl and 1-vinylnaphthalenes or acylation of aminomethyl β-C-galactopyranosyl with 1-naphthoic acid derivatives gave C-galactopyranosyls carrying 1-naphthamide structural elements that interacted favorably with a galectin-3 unique subsite according to molecular modeling and X-ray structural analysis of two inhibitor-galectin-3 complexes. Affinities were down to sub-µM and selectivities over galectin-1, 2, 4 N-terminal domain, 4 C-terminal domain, 7, 8 N-terminal domain, 9 N-terminal domain, and 9 C-terminal domain were high. These results show that high affinity and selectivity for a single galectin can be achieved by targeting unique subsites, which holds promise for further development of small and selective galectin inhibitors.

1. Introduction

Efforts towards the discovery of selective and drug-like inhibitors of the galectin family of proteins have escalated during the last decade as a consequence of an increasing awareness of the important roles of galectins in e.g., cell adhesion, tumor growth and metastasis, inflammation, and immune regulation [1,2]. Galectins may dimerize (e.g., galectin-1, 2, and 7), oligomerize (galectin-3), or have two ligand-binding sites within one polypeptide sequence (galectins-4, 6, 8, 9, and 12). These structural features enable crosslinking of endogenous glycoconjugates (e.g., glycoproteins) via oligosaccharides that carry the key ligand monosaccharide galactose [3,4,5]. Such cross-linking may in turn influence for example glycoprotein localization (on cell membrane or in intracellular vesicles) and cellular trafficking, which may influence glycoprotein function [6]. Endogenous oligosaccharide ligands and a majority of synthetic galactose-based inhibitors display at best partial selectivity among the galectins, which is why further development of inhibitors with enhanced selectivity is of significance. O-galactosyl aldoximes [7,8] and C1-galactopyranosyl oxazoles [9] carrying bicyclic aromatic N-substituents (Figure 1A) were earlier reported to be comparatively potent and highly selective galectin-3 inhibitors. A molecular modeling-based explanation was proposed in which the selectivity was due to anomeric aldoxime or oxazole substituents occupying and interacting favorably with a subsite unique to galectin-3 located perpendicular to the core galactoside subsite C conserved within galectins [4], with bicyclic aromatic substituents, such as the 1-naphthyl derivatives 1 and 2 (Figure 1A), being the most potent ones. Although these O-galactosyl aldoximes (1) and C1-galactosyl oxazoles (2) were selective and had improved affinity for galectin-3 over e.g., galactose, lactose, and N-acetyl-lactosamine, even further enhanced affinity is needed to improve pharmacological properties. Furthermore, the anomeric aldoxime functionality should preferably be replaced with functionalities less prone to undergo equilibrating reactions under physiological conditions, as an oxime ligand that exchanges with for example protein lysine side chains or other amino groups will likely possess altered galectin binding properties. Herein, we report on the synthesis and evaluation of anomeric aromatic amides and trans-alkenes as aldoxime and oxazole bioisosteres (Figure 1), leading to the discovery that amides possess galectin-3 affinities that reach the nanomolar range, while retaining the excellent selectivity earlier demonstrated for the known aldoximes and oxazoles. Furthermore, structural analysis of two anomeric 1-naphthyl amides in complex with galectin-3 confirmed that the naphthyl moiety occupied the site perpendicular to the conserved subsite C, while the amide N and O atoms were engaged in a hydrogen bond network with Q182-R186-R162 side chains, a buried water molecule, and the ligand galactose HO2, which is suggested to explain why the amide proved to be an ideal anomeric functional group over the aldoxime, oxazole, and trans-alkene.

2. Results and Discussion

The alkene isostere 3 was synthesized via an olefin metathesis between allyl β-C-galactopyranosyl 5 [10] and 1-vinylnaphthalene, followed by deacetylation, and the amide 4 was synthesized by acylation of the aminomethyl β-C-galactopyranosyl 6 [11] (Scheme 1). The yields of the amides 4 and 7–11 were apparently low due to the presence of residual inseparable aluminum salts in C-galactosyl 5 resulting from its synthesis by LiAlH₄ reduction of the preceding nitrile, as originally also reported by Coxon and co-workers for similar reactions [11].

Evaluation of 3 and 4 as inhibitors of galectin-3 in a competitive fluorescence polarization assay [12,13], revealed that while the alkene 3 was a less efficient inhibitor than the parent aldoxime 1 and oxazole 2, the amide 4 was significantly more potent (

Table 1. Dissociation constants (K_d, µM) of methyl β-d-galactopyranosyls [14], aldoxime 1, oxazole 2, alkene 3, and amides 4, and 7–11 against galectin-3 determined in a competitive fluorescence anisotropy assay [12,13].

Me β-gal	1	2	3	4	7	8	9	10	11
4400	370 [7,8]	90 [9]	1600 ± 90 1	39 ± 2.3	22 ± 0.75	79 ± 7.8	43 ± 0.38	28 ± 0.81	22 ± 0.79

¹ Dissociation constant averages and SEM were calculated from two to eight single-point measurements showing between 20% and 80% inhibition.

). Inspired by the discovery that amides were superior in binding to galectin-3, we synthesized a series of substituted bicyclic amides 7–11 by acylating 6 with different acyl chlorides following the same procedure as in the synthesis of 4 (Scheme 2). Substituent effects are very minor, as all substituted 1-naphthoyl derivatives 7–11 were all inhibitors in the low µM range (Table 1). The 6-fluoro derivative 8 was significantly weaker, while the 4-fluoro and 4-methoxy derivatives 7 and 11 were somewhat better than the other 1-naphthoyl derivatives.

With the 4-fluoro-1-naphthamide 7 being identified as one of two naphthamide ligands showing the highest affinity, the hypothesis that combining this C1-naphthamido moiety with known affinity-enhancing 5,6-difluorocoumarin- [15,16] and 3,4,5-trifluorophenyltriazolyl moieties [17,18] in the same manner as described for the aldoxime derivatives [8] to give the doubly derivatized C-galactosyls 14 and 19 (Scheme 3) appeared plausible. Stannylidene-mediated propargylation at HO3 of 7 gave the propargyl ether 12. The remaining hydroxyl groups of 12 were acetylated to give 13 to be used as the alkyne source together with 5,6-difluorosalicylic aldehyde and tosyl azide for Cu(I)-mediated multicomponent coumarin synthesis [15,19] followed by de-acetylation to give the difluoro-coumarin 14.

Access to the triazole group in 19 requires double inversion reactions at C3 of 7 to introduce an azido group for a Cu(I)-catalyzed cycloaddition reaction. Regioselective sulfonylation of 7 installed a bis-3,5-trifluoromethylbenzenethanolsulfonate at O3 (15), a leaving group recently demonstrated to possess optimal balance between reactivity and stability for inversions at galactose C3 [20]. Inversion of 15 with CsOAc in DMSO at 90 °C for 3 days gave the guloside 16. De-acetylation of 16, regioselective 2-O-acetylation, 3-O-triflation, and treatment with sodium azide gave the 3-azido derivative 18. Acetate and benzylidene removal followed by Cu(I)-catalyzed cycloaddition [21,22] with 3,4,5-trifluorophenylacetylene gave 19 in 42% yield. Evaluation of 14 and 19 indeed revealed low to sub-µM affinities for galectin-3 (

Table 2. Dissociation constants (K_d, µM) of compounds 14 and 19 against galectin-1, 3, 4N, 4C, 7, 8N, 9N, and 9C determined in a competitive fluorescence anisotropy assay.

	Galectin
	1	3	4N 1	4C 2	7	8N 1	9N 1	9C 2
14	220 ± 19 3	2.0 ± 0.18	870 ± 170	100 ± 43	ND 4	1300 ± 150	>2000	ND
19	23 ± 1.4	0.54 ± 0.060	31 ± 0.80	1.3 ± 0.039	>500	1000 ± 270	14 ± 0.89	130 ± 6
Me β-gal	>10,000 [14]	4400 [14]	6600 [23]	10,000 [23]	4800 [14]	6300 [14]	3300 [14]	8600 [24]

¹N-terminal domain ² C-terminal domain. ³ Dissociation constant averages and SEM were calculated from two to eight single-point measurements showing between 20% and 80% inhibition. ⁴ Not determined.

). While the coumarin 14 bound four-fold weaker to galectin-3 than 19, it displayed a better selectivity over the other galectins evaluated, as 19 also inhibited galectin-4C (C-terminal domain) with low µM affinity.

In order to understand the structural basis for the affinities and selectivities of compounds 14 and 19, the structures of their complexes with galectin-3 were determined with X-ray crystallography (

Table 3. Data processing and refinement statistics for the X-ray crystal structures. Values in parentheses are for the highest resolution shell, unless noted otherwise. rmsd = root mean square deviation; $R_{merge}= \frac{{{\displaystyle \sum }}_{hkl}{{\displaystyle \sum }}_i\left|\overline{|F_{hkl}}\right|-\left|F_{hkl,i}\right||}{{{\displaystyle \sum }}_{hkl}{{\displaystyle \sum }}_i\left|F_{hkl,i}\right|}$, where $\left|\overline{F_{hkl}}\right|$ is the mean of all observations $\left|F_{hkl,i}\right|$ of a given reflection hkl and the sum is over all reflections and observations; $R_{model}=\frac{{{\displaystyle \sum }}_{hkl}|\left|F_{o,hkl}\right|-\left|F_{c,hkl}\right||}{{{\displaystyle \sum }}_{hkl}\left|F_{o,hkl}\right|}$, where $\left|F_{o,hkl}\right|$ and $\left|F_{c,hkl}\right|$ are respectively the observed structure factors and the ones calculated from the model for a given reflection hkl and the sum is over all reflections hkl. The bias-free R_free was calculated on 5% of the data that were excluded from the refinement calculations.

Compound	14	19
PDB code	6TF6	6TF7
station	I911-3	I911-2
wavelength (Å)	1.000	1.03841
unit cell (Å)	a = 35.85 b = 57.34 c = 62.04	a = 36.14 b = 58.16 c = 62.84
Space group	P212121	P212121
resolution range (Å)	31.0–1.5 (1.54–1.50)	29.1–1.4 (1.44–1.4)
completeness (%)	98.8 (96.3)	99.6 (96.8)
Total reflections unique reflections	74,804 (6175) 20,755 (1522)	155,192 (7866) 26,601 (1923)
multiplicity	3.6 (3.4)	5.8 (4.2)
Rmerge (%)	0.073 (0.725)	0.048 (0.519)
mean I/σ(I)	12.6 (1.8)	24.0 (2.8)
Wilson B-factor (Å2)	15.6	14.6
refinement program	Refmac5	Refmac5
Rmodel (F; %)	0.116 (0.173)	0.135 (0.240)
Rfree (F; %)	0.174 (0.275)	0.187 (0.324)
reflections used in refinement (for Rfree)	19,717 (1374) 1038 (72)	25,272 (1767) 1330 (93)
average B-factors (Å2)	protein: 14.2 ligand: 21.5 solvent: 33.4	protein: 13.5 ligand: 19.4 solvent: 28.9
Ramachandran outliers (%)	1	0
rotamer outliers (no. and %) MolProbity clash score	1, 0.6% 2.48	0, 0.0% 2.12
bond length rmsd from ideal (Å) bond angle rmsd from ideal (°)	0.020 2.1	0.019 2.2

and Figure 2). Compounds 14 and 19 were soluble at 20 mg/mL, which allowed for co-crystallizations to be performed to generate crystals that diffracted down to 1.5 and 1.4 Å resolution, respectively, with 14 and 19. The good solubility of 14 and 19 in the crystallization buffer is noteworthy as this suggests that both 14 and 19 can be dissolved in buffer at concentrations needed in cell or in vivo experiments involving selective galectin-3 inhibition. Both complexes revealed that the galactose ring of 14 and 19 bound the conserved galactose site of galectin-3 [25], which suggests that both 14 and 19 are specific for the glycan binding site (Figure 2). Furthermore, the naphthyl moieties of 14 and 19 indeed bound the pocket perpendicularly above the β-face of the galactoside site, which provides an explanation for the high galectin-3 selectivity of 14 and 19 as this pocket is unique to galectin-3. Furthermore, the naphthyl rings of 14 and 19 stack on top of and along the R162 side chain methylenes and guanidinium ion, which possibly accounts for some of the affinity enhancements of 14 and 19 via guanidinium-π/π-π and CH-π stacking interactions. The 4-fluoro atom of the naphthyls of 14 and 19 does not form any specific interactions with galectin-3, which may explain its marginal effects on affinity (cf. 4 and 7). Given that compounds 4 and 7–11 binds with the same complex geometry as 14 and 19, then any substituent influence on the naphthyl ring electron densities of 7–11 apparently has minor influence on the stacking interaction of the ligand naphthyl with R162 alkyl side chain atoms or guanidinium ion. The trifluorophenyl ring of 19 is placed in between the sidechain or R144 and the S3 β-strand, which is the same as recently described for thiodisaccharide- and galactose-based ligands [17,18,26]. The coumaryl moiety of 14 is similarly positioned between the R144 side chain and the S3 β-strand, which is different from that observed for earlier coumaryl-derivatized galactosides in which R144 is aligned along the protein surface and below the coumaryl moiety [15]. This is most probably caused by the interaction of the coumarin fluorine atoms of 14 forming multipolar C–F···H–N, C–F···C=O, and C-F···H–Cα interactions [27] (Figure 2B). The closest interactions are with the Cα of G238 (3.0 Å) and with ND2 of N160 (3.2 Å). In addition, longer fluorine-amide carbonyl interactions with residues R144, I145, and S237 are present (3.6–3.9 Å). Fluorine-amide carbonyl interactions have been observed for fluorophenyl-carrying ligands rings binding galectin-3 [17,18,28,29,30], but the magnitude of such interactions’ contribution to free energy of galectin-3 binding has been questioned [26].

Interestingly, the amide NH of both 14 and 19 formed a hydrogen bond to a water molecule short-cutting a salt-bridge between R162 and E184 (Figure 2B,D and Figure 3A). In natural ligands, for example lactose, this salt bridge is short-cut by the HO3 of the glucose subunit of lactose (Figure 3B). Hence, the amide NH of 14 and 19 hydrogen bonding to water mimics the similar hydrogen bonding pattern formed by, e.g., lactose or lacNAc-containing natural glycoconjugate ligands (Figure 3C). Furthermore, the hydrogen bond network with R162, E184, the water molecule, and 14 or 19 extends via the ligand amide oxygen to accept a hydrogen bond from the galactose HO2 (Figure 3A). This extended hydrogen bond network represents adds to the high 14- and 19-galectin-3 complex complementarity, which possibly contributes to the good affinity of 14 and 19 for galectin-3.

3. Conclusions

In conclusion, 1-naphthamido C-β-galactosyls have been synthesized and discovered to bind galectin-3 with high affinity and selectivity. The affinity enhancements and high selectivity stems from three factors. First, the 1-naphthamido moieties binding a pocket unique to galectin-3 and perpendicular to the core galactose binding site, in which it stacks onto the side chain and the guanidinium group of R162. Second, the trifluorophenyltriazolyl- and difluorocoumaryl-moieties at the galactose C3/O3 of 14 and 19, respectively, are known to form affinity-enhancing stacking interactions with R144, as well as to confer selectivity for galectin-3. Finally, the amide NH-water mediated hydrogen bonds to R162 and E184 may not necessarily induce selectivity as is mimics a conserved hydrogen bonding motif involving the glucose or N-acetylglucosamine 3-OH of lactose or lacNAc as parts of endogenous ligands binding to several different galectins, but it apparently enhances affinity as seen when the amide 4 is compared to the oxime 1, oxazole 2, and alkene 3.

4. Materials and Methods

General

NMR spectra were recorded on a Bruker Avance II 400 MHz spectrometer (Fällanden, Switzerland) at ambient temperature. ¹H-NMR spectra were assigned using 2D-methods (COSY). Chemical shifts are given in ppm downfield from the signal for Me₄Si, with reference to residual CHCl₃ or CD₂HOD. High resolution mass spectrometry (HRMS) was determined by direct infusion on a Waters XEVO-G2 QTOF mass spectrometer (Waters, Milfor, MA, USA) using electrospray ionization (ESI). Reactions were monitored by TLC using aluminum-backed silica gel plates (Merck 60F_254, Darmstadt, Germany) and visualized using UV light and by charring with ethanolic H₂SO₄ (7%). Preparative chromatography was performed using silica gel (40–60 μm, 60 Å) columns. Solvents were dried by storing over activated M.S. Reagents were supplied by Sigma-Aldrich and used as is.

(E) 3-(β-d-galactopyranosyl)-1-(naphthalen-1-yl)-1-propene 3:

To a solution of 5 (30 mg, 0.081 mmol) and Grubbs catalyst 2nd generation (3.4 mg, 0.0041 mmol) in CH₂Cl₂ (5 mL) was added 1-vinylaphthalene (62.5 mg, 0.405 mmol, 5 eq) and the resulting mixture was refluxed overnight. The solvent was evaporated and the residue dissolved in MeOH (5 mL), and NaOMe (1 M, 1 mL) was added. The mixture was stirred overnight, the pH was adjusted to 7 by addition of DOWEX 50W H⁺, and the solvent was evaporated. The residue was purified by flash chromatography (CH₂Cl₂:MeOH 9:1) to give 3 (20 mg, 75%). [α]_D²⁰ – 3.0 (c 1.32, MeOH). ¹H NMR (CDCl₃): δ 8.17 (d, J 8.0, 1H, Ph), 7.83 (d, J 8.0, 1H, Ph), 7.73 (d, J 8.1, 1H, Ph), 7.61 (d, J 7.1, 1H, Ph), 7.50–7.39 (m, 3H, Ph), 7.25 (d, J 15.5, 1H, CH), 6.44 (dt, J 15.6, 7.0, 1H, CH), 3.90 (d, J 3.3, 1H, H-4), 3.77 (dd, J 11.3, 6.8, 1H, H-6), 3.71 (dd, J 11.3, 5.4, 1H, H-6), 3.56 (t, J 9.3, 1H, H-2), 3.51 (t, J 6.1, 1H, H-5), 3.47 (dd, J 9.3, 3.3, 1H, H-3), 3.33 (obscured by CD₂HOD, H-1), 2.88 (dd, J 14.9, 7.0, 1H, CH₂), 2.57 (m, 1H, CH₂). ¹³C NMR (CDCl₃): δ 131.5, 130.0, 129.4, 128.3, 126.8, 126.6, 125.0, 124.5, 81.5, 80.3, 76.5, 72.3, 71.0, 62.9, 36.6. HRMS calcd for C₁₉H₂₂O₅Na (M + Na)⁺: 353.1365: found: 353.1368.

Typical procedure for the synthesis of the compounds 4 and 7–11: l-(4-Fluoro-naphth-1-amido)-2,6-anhydro-l-deoxy-d-glycero-l-manno-heptitol 7:

To a solution of 1-C-aminomethyl-galactose 6 [11] (0.47 g, 2.43 mmol) in water (15 mL) was added a solution of 4-fluoro-1-naphthoyl chloride (0.76 g, 3.64 mmol) in THF (4 mL) followed by the addition of Na₂CO₃ (390 mg, 3.64 mmol). The reaction mixture was left overnight at room temperature. The solvent was evaporated and purified by flash chromatography (SiO₂, CH₂Cl₂:MeOH 10:1) to afford pure 7 (0.102 g, 12%) as a colorless foam. The yield of 7 is apparently low due to the presence of residual inseparable aluminum salts in C-galactosyl 5 resulting from its synthesis by LiAlH₄ reduction of the preceding nitrile, as originally also reported by Coxon and co-workers [11]. [α]_D²⁵ + 10.5 (c 1.1 in MeOH), ¹H NMR (400 MHz, CD₃OD) δ: 8.29–8.27 (m, 1H, ArH), 8.14–8.12 (m, 1H, ArH), 7.67–7.61 (m, 3H, ArH), 7.25 (t, 1H, J 8.0 Hz, ArH), 4.05 (t, 1H, J 2.8 Hz, H-1), 3.85 (d, 1H, J 2.4 Hz, H-4), 3.78 (dd, 1H, J 8.0, 7.5 Hz, H-6a), 3.66–3.48 (m, 5H, H-6b, H-2, H-3, CH₂NH, H-5), 3.41–3.36 (m, 1H, CH₂NH). ¹³C NMR (100 MHz, CD₃OD) δ: 172.0, 133.3, 132.2, 129.1, 128.0, 127.1, 127.0, 126.7, 125.0, 121.4, 109.6, 80.5, 80.2, 76.08, 71.06, 70.2, 63.2, 42.5. HRMS calcd for C₁₈H₂₀FNO₆Na (M + Na)⁺: 388.1172, found: 388.1176.

Compounds 4, 8–11 were synthesized following the same procedure:

l-(Naphth-1-amido)-2,6-anhydro-l-deoxy-d-glycero-l-manno-heptitol 4:

1-Naphthoyl chloride gave 30 mg (11%) of 4. [α]_D²⁵ + 12.5 (c 1.3 in MeOH), ¹H NMR (400 MHz, CD₃OD) δ: 8.22 (dd, 1H, J 1.6, 0.4 Hz, ArH), 7.98 (d, 1H, J 8.4 Hz, ArH), 7.92 (dd, 1H, J 2.8, 2.0 Hz, ArH), 7.65 (dd, 1H, J 1.2 Hz, ArH), 7.58–7.49 (m, 3H, ArH), 4.07 (dd, 1H, J 2.4 Hz, H-1), 3.85 (d, 1H, J 2.4 Hz, H-4), 3.78 (dd, 1H, J 8.0, 7.6 Hz, H-6a), 3.65–3.48 (m, 5H, H-6b, H-2, H-3, CH₂NH, H-5), 3.41–3.36 (m, 1H, CH₂NH). ¹³C NMR (100 MHz, CD₃OD) δ: 172.8, 135.8, 135.1, 131.5, 131.4, 129.4, 128.0, 127.4, 126.4, 126.4, 126.0, 80.5, 80.3, 76.1, 71.1, 70.2, 63.2, 42.5. HRMS calcd for C₁₈H₂₁NO₆Na (M + Na)⁺: 370.1267, found: 370.1270.

l-(5-Fluoro-naphth-1-amido)-2,6-anhydro-l-deoxy-d-glycero-l-manno-heptitol 8:

5-Fluoro-1-naphthoyl chloride gave 28 mg (11%) of 8. [α]_D²⁵ + 9.7 (c 1.2 in MeOH), ¹H NMR (400 MHz, CD₃OD) δ: 8.76 (m, 1H, NH), 8.20 (d, 1H, J 8.4 Hz, ArH), 8.03 (d, 1H, J 8.4 Hz, ArH), 7.73 (t, 1H, J 1.2 Hz, ArH), 7.63 (t, 1H, J 6.8 Hz, ArH), 7.55–7.50 (m, 1H, ArH), 7.27–7.23 (m, 1H, ArH), 4.07–4.02 (m, 1H, H-1), 3.85 (d, 1H, J 2.8 Hz, H-4), 3.78 (dd, 1H, J 8.0, 7.6 Hz, H-6a), 3.66–3.48 (m, 5H, H-6b, H-2, H-4, CH₂NH, H-5), 3.41–3.36 (m, 1H, CH₂NH). ¹³C NMR (100 MHz, CD₃OD) δ: 172.3, 135.9, 132.9, 128.0, 127.5, 126.6, 125.2, 125.0, 123.3, 122.7, 111.0, 80.5, 80.2, 76.1, 71.1, 70.2, 63.2, 42.5. HRMS calcd for C₁₈H₂₀FNO₆Na (M + Na)⁺: 388.1172, found: 388.1173.

l-(6-Fluoro-naphth-1-amido)-2,6-anhydro-l-deoxy-d-glycero-l-manno-heptitol 9:

6-Fluoro-1-naphthoyl chloride gave 57 mg (8%) of 9. [α]_D²⁵ + 10.7 (c 1.4 in MeOH), ¹H NMR (400 MHz, CD₃OD) δ: 8.29 (t, 1H, J 5.6 Hz, ArH), 7.96 (d, 1H, J 8.0 Hz, ArH), 7.63–7.53 (m, 3H, ArH), 7.41–7.35 (m, 1H, ArH), 4.05 (t, 1H, J 2.8 Hz, H-1), 3.86 (t, 1H, J 0.8 Hz, H-4), 3.78 (dd, 1H, J 8.0, 7.6 Hz, H-6a), 3.66–3.48 (m, 5H, H-6b, H-2, H-5, CH₂NH, H-3), 3.41–3.36 (m, 1H, CH₂NH). ¹³C NMR (100 MHz, CD₃OD) δ: 172.5, 136.2, 136.1, 135.9, 130.8, 129.5, 128.5, 127.2, 125.8, 118.1, 112.3, 80.5, 80.2, 76.1, 71.1, 70.2, 63.2, 42.5. HRMS calcd for C₁₈H₂₀FNO₆Na (M + Na)⁺: 388.1172, found: 388.1174.

l-(5-Bromo-naphth-1-amido)-2,6-anhydro-l-deoxy-d-glycero-L-manno-heptitol 10:

6-Bromo-1-naphthoyl chloride gave 18 mg (3%) of 10. [α]_D²⁵ + 7.7 (c 1.4 in MeOH), ¹H NMR (400 MHz, CD₃OD) δ: 8.37 (d, 1H, J 8.4 Hz, ArH), 8.22 (d, 1H, J 8.4 Hz, ArH), 7.88 (dd, 1H, J 1.2, 0.8 Hz, ArH), 7.72–7.64 (m, 2H, ArH), 7.46 (dd, 1H, J 7.6, 7.2 Hz, ArH), 4.06 (t, 1H, J 2.4 Hz, H-1), 3.85 (d, 1H, J 2.4 Hz, H-4), 3.78 (t, 1H, J 8.0 Hz, H-6a), 3.66–3.48 (m, 5H, H-6b, H-2, H-3, CH₂NH, H-5), 3.41–3.36 (m, 1H, CH₂NH). ¹³C NMR (100 MHz, CD₃OD) δ: 172.2, 136.7, 133.2, 132.8, 131.7, 130.0, 128.4, 127.6, 127.3, 126.7, 123.8, 80.5, 80.2, 76.1, 71.1, 70.2, 63.2, 42.5. HRMS calcd for C₁₈H₂₀BrNO₆Na (M + Na)⁺: 448.0372, found: 448.0371.

l-(4-Methoxy-naphth-1-amido)-2,6-anhydro-l-deoxy-d-glycero-l-manno-heptitol 11:

4-Methoxy-1-naphthoyl chloride gave 142 mg (14%) of 11. [α]_D²⁵ + 11.4 (c 1.7 in MeOH), ¹H NMR (400 MHz, CD₃OD) δ: 8.56 (m, 1H, NH), 8.28 (2d, 2H, J 9.6 and 8.8 Hz, ArH), 7.64 (d, 1H, J 8.0 Hz, ArH), 7.58–7.47 (m, 2H, ArH), 6.92 (d, 1H, J 8.0 Hz, ArH), 4.03 (s, 3H, –OCH₃), 4.01-3.99 (m, 1H, H-1), 3.84 (d, 1H, J 2.4 Hz, H-4), 3.78 (t, 1H, J 7.6 Hz, H-6a), 3.65–3.48 (m, 5H, H-6b, H-2, H-3, CH₂NH, H-5), 3.39–3.34 (m, 1H, CH₂NH). ¹³C NMR (100 MHz, CD₃OD) δ: 173.0, 158.5, 132.7, 128.4, 127.9, 127.8, 126.8, 126.6, 126.3, 123.1, 103.7, 80.5, 80.3, 76.1, 71.1, 70.2, 63.2, 56.3, 42.5. HRMS calcd for C₁₉H₂₃NO₇Na (M + Na)⁺: 400.1372, found: 400.1377.

l-(4-Fluoro-naphth-1-amido)-2,6-anhydro-4-O-propargyl-l-deoxy-d-glycero-l-manno-heptitol 12:

Compound 6 (0.5 g, 1.36 mmol) was dissolved in dry MeOH (10 mL) and n-Bu₂SnO (0.405 g, 1.63 mmol) was added. The mixture was refluxed at 70 °C for 3 h, where after the solvent was evaporated under reduced pressure. To the residue were added 1,4-dioxane (20 mL), propargyl bromide (365 μL, 4.08 mmol) and n-Bu₄NBr (0.52 g, 1.63 mmol). The mixture was sonicated for 10 min and then refluxed overnight at 100 °C. Completion of reaction was checked by TLC (CH₂Cl₂:MeOH 6:1). The solvent was evaporated under reduced pressure and the sticky residue was dissolved in MeOH to obtain a slurry and purified by flash chromatography (SiO₂, CH₂Cl₂: MeOH 17:1) to afford compound 12 (0.434 g, 79%). [α]_D²⁵ + 9.1 (c 0.54 in MeOH), ¹H NMR (400 MHz, CD₃OD) δ: 8.28 (dd, 1H, J 2.0, 1.6 Hz, ArH), 8.14 (dd, 1H, J 2.8, 2.0 Hz, ArH), 7.67–7.61 (m, 3H, ArH), 7.25 (t, 1H, J 8.0 Hz, ArH), 4.40 (dd, 2H, J 3.9, 2.4 Hz, CH₂CCH), 4.10 (d, 1H, J 2.8 Hz, H-4), 4.05 (dd, 1H, J 13.6, 2.8 Hz, CH₂NH), 3.78–3.51 (m, 6H, H-6a, H-5, H-6b, CH₂NH, H-2, H-3), 3.46–3.41 (m, 1H, H-1), 2.89 (t, 1H, J 2.4 Hz, HCCCH₂). ¹³C NMR (100 MHz, CD₃OD) δ: 172.0, 133.2, 132.2, 129.1, 127.97, 127.95, 127.1, 126.7, 125.0, 121.4, 109.6, 83.0, 81.0, 80.3, 80.2, 76.1, 69.1, 67.9, 63.1, 57.7, 42.5. HRMS calcd for C₂₁H₂₂FNO₆Na (M + Na)⁺: 426.1329, found: 426.1331.

l-(4-Fluoro-naphth-1-amido)-3,5,7-tri-O-acetyl-2,6-anhydro-4-O-propargyl-l-deoxy-d-glycero-l-manno-heptitol 13:

Compound 12 (0.315 g, 0.78 mmol) was dissolved in pyridine (10 mL) and acetic anhydride (10 mL), stirred for 2 h, evaporated, and the residue was purified by column chromatography (SiO₂, heptane:EtOAc 1:1) to give 13 (0.388 g, 94%). [α]_D²⁵ + 16.2 (c 1.1 in CHCl₃), ¹H NMR (400 MHz, CDCl₃) δ: 8.38 (d, 1H, J 8.0 Hz, ArH), 8.15 (d, 1H, J 7.6, ArH), 7.65–7.57 (m, 3H, ArH), 7.14 (t, 1H, J 7.6 Hz, ArH), 6.37 (m, 1H, NH), 5.46 (dd, 1H, J 3.5, 1.0 Hz, H-4), 5.11 (t, 1H, J 10.0 Hz, H-3), 4.20 (d, 2H, J 2.4 Hz, OCH₂CCH), 4.11 (dd, 2H, J 2.4, 1.6 Hz, H-6a, H-6b), 4.05–3.99 (m, 1H, CH₂NH), 3.89–3.86 (m, 2H, H-3, H-5), 3.73–3.68 (m, 1H, H-1), 3.47–3.41 (m, 1H, OCH₂CCH), 2.46 (t, 1H, J 2.4 Hz, HCCCH₂), 2.16, 2.08, 1.96 (s, 9H, 3× COCH₃). ¹³C NMR (100 MHz, CDCl₃) δ: 170.6, 170.5, 170.4, 168.9, 132.1, 130.7, 128.3, 127.0, 125.63, 125.57, 125.5, 124.1, 121.0, 108.6, 79.2, 76.9, 76.7, 75.2, 74.8, 68.1, 66.3, 62.3, 56.6, 40.5, 21.1, 20.9, 20.8. HRMS calcd for C₂₇H₂₈FNO₉Na (M + Na)⁺: 552.1646, found: 552.1650.

l-(4-Fluoro-naphth-1-amido)-2,6-anhydro-4-O-(5,6-difluoro-2H-1-benzopyran-2-on-3-yl)-l-deoxy-d-glycero-l-manno-heptitol 14:

Compound 13 (0.117 g, 0.22 mmol) was dissolved in dry THF (10 mL) to which TsN₃ (0.052 g, 0.264 mmol), 5,6-difluorosalicylaldehyde (0.042 g, 0.264 mmol), and CuI (0.0042 g, 0.022 mmol) was added. After 5 min, Et₃N (62 μL, 0.44 mmol) was added to it resulting in the change of the color of the solution. The reaction mixture was left overnight and TLC was checked. After consumption of the starting material, the solution was evaporated and diluted with CH₂Cl₂. Work up was done with NH₄Cl solution. The organic layer collected was evaporated and directly proceeded to the next step. The compound was dissolved in MeOH (5 mL) and freshly prepared NaOMe (0.5 M in MeOH, 100 mL) was added to it and left for 2 days. After complete conversion of the starting material, the solution was neutralized with DOWEX 50W H+ resin, filtered, evaporated and purified by flash chromatography (SiO₂, CH₂Cl₂: MeOH 20:1) to afford colorless pure compound 14 (0.050 g, 40%). [α]_D²⁵ + 3.7 (c 0.33 in MeOH), ¹H NMR (400 MHz, CD₃OD) δ: 8.37 (s, 1H, ArH), 8.30–8.27 (m, 1H, ArH), 8.13–8.10 (m, 1H, ArH), 7.67–7.60 (m, 3H, ArH), 7.52–7.45 (m, 1H, ArH), 7.25–7.17 (m, 2H, ArH), 4.74 (dd, 1H, J 1.6 Hz, CH₂Ph), 4.61 (t, 1H, J 1.2 Hz, CH₂Ph), 4.18 (d, 1H, J 2.4 Hz, H-3), 4.08 (dd, 1H, J 2.8, 2.4 Hz, CH₂NH), 3.83–3.77 (m, 2H, H-6a, H-5), 3.70–3.55 (m, 3H, H-6b, CH₂NH, H-4), 3.52–3.41 (m, 2H, H-2, H-1). ¹³C NMR (100 MHz, CD₃OD) δ: 172.0, 161.1, 159.7 133.2, 132.2, 132.0, 129.1, 128.0, 127.14, 127.05, 126.7, 125.0, 121.41, 121.35, 120.5, 120.3, 113.4, 109.6, 109.4, 84.7, 80.3, 80.2, 69.2, 67.6, 66.8, 63.2, 42.5. HRMS calcd for C₂₈H₂₄F₃NO₈Na (M + Na)⁺: 582.1352, found: 582.1350.

l-(4-Fluoro-naphth-1-amido)-2,6-anhydro-5,7-O-benzylidene-4-O-[3,5-bis-(trifluoromethyl)-benzenesulfonyl]-l-deoxy-d-glycero-l-manno-heptitol 15:

Compound 7 (0.332 g, 0.908 mmol) was dissolved in DMF (10 mL) under nitrogen, followed by adding benzaldehyde dimethyl acetal (204 μL, 1.36 mmol) and a catalytic amount of p-toluene sulfonic acid. The reaction mixture was stirred for 2.5 h at room temperature and reaction completion was confirmed by TLC. Then the solution was diluted with ethyl acetate (20 mL) and washed successively with H₂O (2 × 10 mL) and brine (20 mL). The organic layer was separated, dried (Na₂SO₄) and evaporated as pale yellow syrup, which was pure enough to proceed to the next step. The compound (260 mg, 0.57 mmol) was dissolved in THF (30 mL) to which Bu₂SnCl₂ (17 mg, 0.057 mmol) was added under nitrogen atmosphere and the mixture was stirred at room temperature for 15 min. A mixture of 1,1,2,2,6-pentamethylpiperidine (260 μL, 1.14 mmol) and 3,5-bis-(trifluoromethyl)benzenesulfonyl chloride (197 mg, 0.63 mmol) in THF, was added to it. The mixture was stirred overnight at room temperature until the starting material was completely consumed (TLC). The solution was quenched by NH₄Cl solution and the solvents were evaporated to give the crude product. The solution was then diluted with ethyl acetate (20 mL) and washed successively with H₂O (2 × 10 mL) and brine (10 mL). The organic layer was separated, dried (Na₂SO₄) and evaporated in vacuo. The crude product was purified by flash chromatography (SiO₂, heptane:EtOAc 1:1) to afford pure 15 (216 mg, 52%) as white foam. [α]_D²⁵ + 6.2 (c 0.56 in CHCl₃), ¹H NMR (400 MHz, CDCl₃) δ: 8.45 (s, 2H, ArH), 8.24 (d, 1H, J 8.0 Hz, ArH), 8.10–8.07 (m, 2H, ArH), 7.54 (t, 1H, J 7.2 Hz, ArH), 7.45–7.33 (m, 5H, ArH), 7.24 (dd, 2H, J 8.0, 7.2 Hz, ArH), 6.67–6.61 (m, 2H, ArH), 5.58 (s, 1H, CHPh), 4.78 (dd, 1H, J 3.2 Hz, H-3), 4.56 (d, 1H, J 3.2 Hz, H-4), 4.36–4.25 (m, 3H, CH₂NH, H-2, H-6a), 4.08–4.02 (m, 2H, H-5, H-6b), 3.58 (s, 2H, OH), 3.46 (d, 1H, J 9.6 Hz, H-1), 3.33–3.27 (m, 1H, CH₂NH). ¹³C NMR (100 MHz, CDCl₃) δ: 171.4, 161.6, 159.1, 139.6, 137.4, 132.9, 132.6, 131.82, 131.8, 129.2, 128.7, 128.4, 127.1, 126.4, 126.3, 126.0, 125.1, 124.1, 123.88, 123.85, 121.1, 121.0, 120.9, 108.6, 108.4, 100.5, 83.0, 79.5, 75.3, 69.6, 69.2, 64.5, 40.2. HRMS calcd for C₃₃H₂₆F₇NO₈SNa (M + Na)⁺: 752.1165, found: 752.1170.

l-(4-Fluoro-naphth-1-amido)-3,4-di-O-acetyl-2,6-anhydro-5,7-O-benzylidene-l-deoxy-d-glycero-l-ido-heptitol 16:

To a solution of compound 15 (216 mg, 0.29 mmol) in dry pyridine (10 mL) was added Ac₂O (10 mL) and stirring was continued for 2 h at room temperature until the starting material was completely consumed according to TLC analysis. The solvents were evaporated in vacuo and the residue (198 mg, 0.256 mmol) was dissolved in DMSO (7 mL) followed by addition of CsOAc (147 mg, 0.769 mmol). The reaction mixture was stirred at 90 °C for 3 days until the starting material was completely consumed as confirmed by TLC. Then the reaction mixture was diluted by water (10 mL) and worked up with EtOAc (3 × 10 mL). The organic layer was separated and dried over Na₂SO₄ and evaporated. It was purified by flash chromatography (SiO₂, hexane-EtOAc 1:1) to afford pure compound 16 (41 mg, 35%). [α]_D²⁵ +5.9 (c 1.1 in CHCl₃), ¹H NMR (400 MHz, CDCl₃) δ: 8.36–8.34 (m, 1H, ArH), 8.12–8.09 (m, 1H, ArH), 7.58–7.52 (m, 3H, ArH), 7.40–7.32 (m, 3H, ArH), 7.27–7.23 (m, 2H, ArH), 6.93 (dd, 1H, J 8.0, 7.6 Hz, ArH), 6.56 (t, 1H, J 6.0 Hz, NH), 5.53 (m, 2H, CHPh, H-3), 5.29 (dd, 1H, J 10.4, 3.2 Hz, H-2), 4.28 (dd, 1H, J 12.6, 1.7 Hz, H-6a), 4.13–4.02 (m, 2H, H-4, H-1), 4.05 (dd, 1H, J 12.6, 1.8 Hz, H-6b), 3.81–3.78 (m, 2H, CH₂NH), 3.73 (s, 1H, H-5), 2.18, 2.09 (s, 6H, 2× COCH₃). ¹³C NMR (100 MHz, CDCl₃) δ: 169.8, 169.6, 169.1, 161.3, 158.7, 137.3, 129.3, 128.4, 128.2, 126.9, 126.1, 125.73, 125.69, 125.67, 125.6, 120.8, 120.7, 108.6, 108.4, 101.1, 74.5, 72.1, 69.4, 67.9, 67.0, 65.9, 40.0, 21.0, 20.9. HRMS calcd for C₂₉H₂₈FNO₈Na (M + Na)⁺: 560.1697, found: 560.1699.

l-(4-Fluoro-naphth-1-amido)-2,6-anhydro-5,7-O-benzylidene-l-deoxy-d-glycero-l-ido-heptitol 17:

Compound 16 (290 mg, 0.539 mmol) were dissolved in dry MeOH (10 mL) to which NaOMe (0.5 M in MeOH, 50 mL) was added and the solution was allowed to stir at room temperature for 2 h. The solution was neutralized with DOWEX 50W H⁺, filtered, and evaporated in vacuo to furnish pure compound 17 (240 mg, 98%) as amorphous white powder. [α]_D²⁵ + 11.9 (c 1.4 in CHCl₃), ¹H NMR (400 MHz, CDCl₃) δ: 8.32 (d, 1H, J 8.4 Hz, ArH), 8.11 (d, 1H, J 8.0 Hz, ArH), 7.55–7.46 (m, 3H, ArH), 7.36–7.29 (m, 3H, ArH), 7.21 (t, 2H, J 7.6 Hz, ArH), 6.97–6.88 (m, 1H, NH), 6.82 (t, 1H, J 8.0 Hz, ArH), 5.51 (s, 1H, CHPh), 4.30–4.21 (m, 3H, CH₂NH, H-6a, H-2), 4.07–3.99 (m, 2H, H-1, H-6b), 3.93–3.86 (m, 2H, H-3, H-4), 3.73 (s, 1H, H-5), 3.43–3.37 (m, 1H, CH₂NH). ¹³C NMR (100 MHz, CDCl₃) δ: 171.5, 161.6, 159.0, 138.0, 131.9, 129.1, 128.6, 128.3, 127.0, 126.3, 126.2, 126.1, 125.3, 120.94, 120.88, 108.6, 108.4, 100.7, 76.3, 74.1, 69.8, 69.2, 66.7, 65.0, 40.9. HRMS calcd for C₂₅H₂₄FNO₆Na (M + Na)⁺: 476.1485, found: 476.1489.

l-(4-Fluoro-naphth-1-amido)-3-O-acetyl-2,6-anhydro-4-azido-5,7-O-benzylidene-l,4-dideoxy-d-glycero-l-manno-heptitol 18:

Compound 17 (250 mg, 0.551 mmol) was dissolved in in dry CH₂Cl₂ (10 mL) followed by addition of pyridine (89 µL, 1.10 mmol) and AcCl (39 µL, 0.551 mmol) to a stirred solution at room temperature under N₂ atmosphere. After 2 h, the mixture was washed with NaHCO₃ (2 × 10 mL) and brine (10 mL). The combined aqueous phases were extracted with CH₂Cl₂ (2 × 10 mL). The organic phases were collected and dried over Na₂SO₄ and evaporated. The reaction mixture was directly dissolved in dry CH₂Cl₂ (10 mL) and cooled to −20 °C followed by dropwise addition of Tf₂O (182 µL, 1.10 mmol). The reaction was stirred for 2 hr and then allowed to reach room temperature. The mixture was then washed with 5% HCl (2 × 10 mL) and brine (10 mL). The combined aqueous phases were extracted with CH₂Cl₂ (2 × 10 mL). The collected organic phases were dried over Na₂SO₄ and concentrated to give crude product. The crude product (157 mg, 0.250 mmol) was dissolved in DMSO (10 mL), followed by the addition of NaN₃ (48 mg, 0.75 mmol). The solution was stirred at 60 °C for 8 h. Then the solution was diluted with water and washed with ethyl acetate (2 × 10 mL). The collected organic layer was dried over Na₂SO₄ followed by filtration and evaporation in vacuo to furnish crude compound 18, which was then purified by column chromatography using (SiO₂, hexane: EtOAc 1:1) to afford pure compound 18 (19 mg, 15%). [α]_D²⁵ + 4.9 (c 0.44 in CHCl₃), ¹H NMR (400 MHz, CDCl₃) δ: 8.32–8.36 (m, 1H, ArH), 8.12–8.10 (m, 1H, ArH), 7.56–7.52 (m, 3H, ArH), 7.44–7.41 (m, 2H, ArH), 7.36–7.26 (m, 3H, ArH), 6.95 (t, 1H, J 8.0 Hz, ArH), 6.52 (t, 1H, J 5.6 Hz, NH), 5.60 (s, 1H, CHPh), 5.48 (t, 1H, J 10.0 Hz, H-2), 4.39 (d, 1H, J 2.8 Hz, H-4), 4.31 (dd, 1H, J 12.5, 1.5 Hz, H-6a), 4.08 (dd, 1H, J 12.5, 1.6 Hz, H-6b), 3.97–3.92 (m, 1H, CH₂NH), 3.75–3.62 (m, 2H, H-1, CH₂NH), 3.52 (d, 1H, J 1.0 Hz H-5), 3.41 (dd, 1H, J 10.3, 2.3 Hz, H-3), 2.21 (s, 3H, CH₃). ¹³C NMR (100 MHz, CDCl₃) δ: 170.1, 169.0, 137.1, 132.0, 129.2, 128.4, 128.3, 126.9, 126.1, 125.8, 125.70, 125.67, 125.65, 120.82, 120.77, 108.7, 108.5, 101.2, 76.1, 70.1, 69.4, 66.8, 62.5, 41.1, 40.0. HRMS calcd for C₂₇H₂₅FN₄O₆Na (M + Na)⁺: 543.1656, found: 543.1658.

l-(4-Fluoro-naphth-1-amido)-2,6-anhydro-4-[4-(3,4,5-trifluorophenyl)-1H-1,2,3-triazol-1-yl]-l,4-dideoxy-d-glycero-l-manno-heptitol 19:

A solution of compound 18 (19 mg, 0.036 mmol) in 90% aq AcOH (10 mL) was stirred at 90 ⁰C for 3 h. The solvents were evaporated and co-evaporated with toluene to remove residual AcOH. The residue was dissolved in MeOH (10 mL), to which NaOMe in MeOH (0.5 M, 1 mL) was added and the solution was stirred at room temperature for 5 h. The solution was then neutralized by DOWEX 50 W H⁺ resin, filtered, and evaporated. The residue (18 mg, 0.046 mmol) was dissolved in DMF (10 mL), followed by the successive addition of 3,4,5-trifluorophenyl acetylene (21 mg, 0.138 mmol) and CuI (8 mg, 0.1 mmol). After 5 min, Et₃N (6.3 μL, 0.046 mmol) was added and the mixture was stirred at room temperature for 48 h. The solvent was evaporated in vacuo to furnish crude 18, which was purified by column chromatography (SiO₂, hexane: EtOAc 1:12) to afford pure 19 (10.6 mg, 42%). [α]_D²⁵ + 2.5 (c 1.1 in MeOH), ¹H NMR (400 MHz, CD₃OD) δ: 8.51 (s, 1H, triazole H), 8.33–8.31 (m, 1H, ArH), 8.16–8.14 (m, 1H, ArH), 7.71–7.62 (m, 5H, ArH), 7.28 (t, 1H, J 8.0 Hz, ArH), 4.86 (d, 1H, J 3.2 Hz, H-1), 4.30 (dd, 1H, J 10.4, 9.0 Hz, H-2), 4.18 (dd, 1H, J 11.7, 1.7 Hz, CH₂NH), 4.12 (d, 1H, J 3.2 Hz, H-4), 3.84–3.77 (m, 2H, H-5, H-6a), 3.72–3.62 (m, 3H, H-6b, CH₂NH, H-1). ¹³C NMR (100 MHz, CD₃OD) δ: 172.1, 133.30, 133.25, 132.13, 132.09, 129.1, 128.0, 127.1, 126.7, 125.0, 124.9, 122.9, 121.5, 121.4, 110.9, 110.7, 109.7, 109.5, 81.2, 80.7, 70.1, 69.2, 67.2, 62.9, 42.5. HRMS calcd for C₂₆H₂₃F₄N₄O₅ (M + H)⁺: 547.1605, found: 547.1614.

Fluorescence polarization experiments:

Fluorescence polarization experiments were carried out either with a POLARStar plate reader and FLUOstar Galaxy software or with a PheraStarFS plate reader and PHERAstar Mars version 2.10 R3 software (BMG, Offenburg, Germany). The dissociation constant (Kd) values were determined in PBS as described earlier [12,13]. Specific conditions for galectin-1, 3, 4N, 4C, 7, 8N, 9N, and 9C were kept as reported [32,33]. Average Kd values and SEMs were calculated from 2–8 single-triple point measurements showing between 30% and 70% inhibition.

Co-crystallization of galectin-3 C-terminal domain with compounds 14 and 19:

Galectin-3C [34] (C-terminal domain) solution (12 µL, 20 mg/mL in 10 mM sodium phosphate buffer pH 7.5, 100 mM NaCl and 10 mM β-mercaptoethanol) was mixed with either compound 14 or 19 (1.2 µL, 50 mM in 100% DMSO) and incubated on ice for 30 min. Crystallization drops of 2 + 2 µL were set up over 1 mL reservoir solution (30% PEG 4000, 0.1 M Tris/HCl pH 7.5, 0.1 M MgCl₂, 0.4 M NaSCN, 7.9 mM β-mercaptoethanol). Immediately after setup, the drops were seeded by adding 0.25 µL of crystals of apo-galectin-3C that had been crushed using Seed Beads (Hampton Research) in a stabilizing solution (reservoir solution). Co-crystals of compound 14 measuring 0.03 × 0.05 × 0.05 mm and compound 19 measuring 0.1 × 0.15 × 0.15 mm were flash-frozen in cryo solution (15% glycerol, 25.5 w/v % PEG 4000, 250 mM NaSCN, 85 mM Tris/HCl pH 7.5, 85 mM MgCl₂, 2.5 mM 14 or 0.5 mM 19).

Data collection and structure solution of galectin-3C in complex with 13 and 18:

Data to 1.5 Å for compound 14 were collected at 100 K at station I911-3 of MAX-lab, Lund, Sweden (λ = 1.0000 Å), equipped with a marMosaic 225mm CCD detector. Data to 1.4 Å for compound 19 were collected at station I911-2 of the same synchrotron (λ = 1.0384 Å), equipped with a marCCD 165 mm detector. Data for both structures were integrated using XDS and scaled using XSCALE [35]. The structures were refined using Refmac5 [36] with PDB entry 3ZSL [31] as starting model, first by rigid-body refinement. 5% of the total reflections were randomly set aside for cross validation. The graphic software Coot [37] has been used for model building and Monomer Library Sketcher software [38] for generation of restraints for the ligands 14 and 19. The models were refined to a resolution of 1.5 Å/1.4 Å with anisotropic B-factors. Water molecules were added to positive difference density peaks more than 5 σ above the mean and present in the 2m|Fo|-d|Fc| map at 1 σ level. Refinement statistics are listed in Table S1. Molecular images were generated using PyMOL v1.7 (Schrödinger LLC, New York, NY, USA).