*Article* **A Galactoside-Binding Protein Tricked into Binding Unnatural Pyranose Derivatives: 3-Deoxy-3-Methyl-Gulosides Selectively Inhibit Galectin-1**

#### **Kumar Bhaskar Pal <sup>1</sup> , Mukul Mahanti <sup>1</sup> , Hakon Le**ffl**er <sup>2</sup> and Ulf J. Nilsson 1,\***


Received: 23 June 2019; Accepted: 22 July 2019; Published: 2 August 2019

**Abstract:** Galectins are a family of galactoside-recognizing proteins involved in different galectinsubtype-specific inflammatory and tumor-promoting processes, which motivates the development of inhibitors that are more selective galectin inhibitors than natural ligand fragments. Here, we describe the synthesis and evaluation of 3-*C*-methyl-gulopyranoside derivatives and their evaluation as galectin inhibitors. Methyl 3-deoxy-3-*C*-(hydroxymethyl)-β-d-gulopyranoside showed 7-fold better affinity for galectin-1 than the natural monosaccharide fragment analog methyl β-d-galactopyranoside, as well as a high selectivity over galectin-2, 3, 4, 7, 8, and 9. Derivatization of the 3-*C*-hydroxymethyl into amides gave gulosides with improved selectivities and affinities; methyl 3-deoxy-3-*C*-(methyl-2,3,4,5,6-pentafluorobenzamide)-β-d-gulopyranoside had K<sup>d</sup> 700 µM for galectin-1, while not binding any other galectin.

**Keywords:** galectin-1; gulopyranosides; fluorescence polarization; benzamide; selective

### **1. Introduction**

Galectins are an evolutionary ancient family of small soluble proteins with affinity for β-d-galactopyranoside-containing glycoconjugates and a conserved amino acid sequence motif [1,2]. By their carbohydrate-binding activity they can cross-link glycoproteins, resulting in a variety of effects, such as regulation of cell adhesion, intracellular glycoprotein traffic, and cell signaling [3–5]. These effects in turn affect cell behavior in inflammation, immunity and cancer, and galectins appear to be rate limiting in some such pathophysiological conditions, e.g., based on effects in null mutant mice and other model systems [6–10]. This has stimulated development of galectin inhibitors as potential drug candidates, but different galectins have a different tissue distribution and function. Although all bind glycoconjugates containing β-galactose residues, each galectin may have a different affinities for larger natural glycans and for artificial small molecule ligands. Hence, there is an important need for selective galectin-inhibitors, that, for example, distinguish between the two most studied galectins in humans, galectin-1 and galectin-3.

The carbohydrate binding site of galectins is a concave groove and long enough to hold about a tetrasaccharide and based on this the carbohydrate binding site of galectins has been described as a combination of four subsites (A–D) together with an additional one less defined fifth subsite E [3]. Within this groove, subsite C is conserved among galectins, made up of the defining amino acid sequence motif and binds β-galactopyranosides by H-bond interaction with 4-OH, 6-OH and the ring galectin-1.

5-O, and CH-π interaction of the α-side of the pyranose ring with a Trp residue. The neighbouring sites, however, vary among galectins, and can be targeted for selective inhibitor development. To do this, previous inhibitor design has derivatized the positions on galactose not engaged by subsite C, namely C1, C2, and C3 [11]. Gulose is a rare saccharide not found in mammals, but can potentially bind galectins because it is structurally similar to galactose with the only difference being the stereoconfiguration at C3. Hence, the C3 is epimeric with the OH axial instead of equatorial in the galectin bound pyranose form. Here, we show that derivatization at C3 in gulose offers a new space for galectin inhibitor design and surprisingly selective inhibitors of galectin-1. In particular, amide-functionalised C3-methyl gulopyranosides are shown to be apparently selective towards human galectin-1. *2.1. Synthesis of methyl 3-deoxy-3-C-(methyl)-β-D-gulopyranosides and galectin inhibition evaluation*  The synthesis of the 3-*C-*methyl-gulo derivatives was initiated by Dess–Martin periodinane oxidation [12] of the known methyl 2,4,6-tri-*O-*benzyl-β-D-galactopyranoside **11** to afford the corresponding keto derivative **12** in 84% yield (Scheme 1). Methylenation of **12** with Petasis reagent gave the olefin **13** in 79% yield. Next, the olefin **13** was subjected to hydroboration with 9-borabicyclo- [3.3.1]nonane (9-BBN) [12], followed by oxidative cleavage of the carbon-boron bond with alkaline hydrogen peroxide to afford the corresponding gulo and galacto isomers **14a** (36%) and **14b** (24%), which were separated by flash column chromatography at a ratio 3:2. Both the gulo and galacto derivatives **14a** and **14b** were separately subjected to hydrogenation [13] in the presence of Pd(OH)2- *Int. J. Mol. Sci.* **2017**, *18*, x 2 of 23

C to give the desired methyl 3-deoxy-3-*C*-(hydroxymethyl)-*β*-D-gulopyranoside **1a** and methyl 3-

namely C1, C2, and C3 [11]. Gulose is a rare saccharide not found in mammals, but can potentially

*Int. J. Mol. Sci.* **2017**, *18*, x 2 of 23

namely C1, C2, and C3 [11]. Gulose is a rare saccharide not found in mammals, but can potentially bind galectins because it is structurally similar to galactose with the only difference being the stereoconfiguration at C3. Hence, the C3 is epimeric with the OH axial instead of equatorial in the galectin bound pyranose form. Here, we show that derivatization at C3 in gulose offers a new space for galectin inhibitor design and surprisingly selective inhibitors of galectin-1. In particular, amidefunctionalised C3-methyl gulopyranosides are shown to be apparently selective towards human

#### **2. Results and Discussion** deoxy-3-*C*-(hydroxymethyl)-*β*-D-galactopyranoside **1b** in yields of 51% and 63%, respectively. bind galectins because it is structurally similar to galactose with the only difference being the

#### *2.1. Synthesis of Methyl 3-Deoxy-3-C-(methyl)-*β*-*d*-gulopyranosides and galectin inhibition evaluation* Evaluation of **1a** and **1b** as inhibitors of human galectin-1, 2, 3, 4N (N-terminal domain) 4C (Cterminal domain), 7, 8N, 8C, 9N, and 9C in a reported competitive fluorescence anisotropy assay stereoconfiguration at C3. Hence, the C3 is epimeric with the OH axial instead of equatorial in the galectin bound pyranose form. Here, we show that derivatization at C3 in gulose offers a new space

The synthesis of the 3-*C*-methyl-gulo derivatives was initiated by Dess–Martin periodinane oxidation [12] of the known methyl 2,4,6-tri-*O-*benzyl-β-d-galactopyranoside **11** to afford the corresponding keto derivative **12** in 84% yield (Scheme 1). Methylenation of **12** with Petasis reagent gave the olefin **13** in 79% yield. Next, the olefin **13** was subjected to hydroboration with 9-borabicyclo-[3.3.1]nonane (9-BBN) [12], followed by oxidative cleavage of the carbon-boron bond with alkaline hydrogen peroxide to afford the corresponding gulo and galacto isomers **14a** (36%) and **14b** (24%), which were separated by flash column chromatography at a ratio 3:2. Both the gulo and galacto derivatives **14a** and **14b** were separately subjected to hydrogenation [13] in the presence of Pd(OH)2-C to give the desired methyl 3-deoxy-3-*C*-(hydroxymethyl)-β-d-gulopyranoside **1a** and methyl 3-deoxy-3-*C*-(hydroxymethyl)-β-d-galactopyranoside **1b** in yields of 51% and 63%, respectively. Evaluation of **1a** and **1b** as inhibitors of human galectin-1, 2, 3, 4N (N-terminal domain) 4C (C-terminal domain), 7, 8N, 8C, 9N, and 9C in a reported competitive fluorescence anisotropy assay [14,15] revealed that the gulo derivative **1a** was selective for galectin-1 with a dissociation constant of 1300 µM, which is about an order of magnitude better than for the virtually unselective reference compound methyl β-d-galactopyranoside **32** (Figure 1, Table 1). [14,15] revealed that the gulo derivative **1a** was selective for galectin-1 with a dissociation constant of 1300 µM, which is about an order of magnitude better than for the virtually unselective reference compound methyl β-D-galactopyranoside **32** (Figure 1, Table 1). **Figure 1.** Structures of the tested compounds **1–8** and reference compound **32**. for galectin inhibitor design and surprisingly selective inhibitors of galectin-1. In particular, amidefunctionalised C3-methyl gulopyranosides are shown to be apparently selective towards human galectin-1. **2. Results and Discussion**  *2.1. Synthesis of methyl 3-deoxy-3-C-(methyl)-β-D-gulopyranosides and galectin inhibition evaluation*  The synthesis of the 3-*C-*methyl-gulo derivatives was initiated by Dess–Martin periodinane oxidation [12] of the known methyl 2,4,6-tri-*O-*benzyl-β-D-galactopyranoside **11** to afford the corresponding keto derivative **12** in 84% yield (Scheme 1). Methylenation of **12** with Petasis reagent gave the olefin **13** in 79% yield. Next, the olefin **13** was subjected to hydroboration with 9-borabicyclo- [3.3.1]nonane (9-BBN) [12], followed by oxidative cleavage of the carbon-boron bond with alkaline hydrogen peroxide to afford the corresponding gulo and galacto isomers **14a** (36%) and **14b** (24%), which were separated by flash column chromatography at a ratio 3:2. Both the gulo and galacto derivatives **14a** and **14b** were separately subjected to hydrogenation [13] in the presence of Pd(OH)2-

**Scheme 1.** Synthesis of methyl 3-deoxy-3-*C*-(hydroxymethyl)-*β*-D-gulopyranoside **1a**, methyl 3 deoxy-3-*C*-(hydroxymethyl)-*β*-D-galactopyranoside **1b**. **Scheme 1.** Synthesis of methyl 3-deoxy-3-*C*-(hydroxymethyl)-β-d-gulopyranoside **1a**, methyl 3-deoxy-3-*C*-(hydroxymethyl)-β-d-galactopyranoside **1b**. 1300 µM, which is about an order of magnitude better than for the virtually unselective reference compound methyl β-D-galactopyranoside **32** (Figure 1, Table 1).

**Figure 1.** Structures of the tested compounds **1–8** and reference compound **32**. **Figure 1.** Structures of the tested compounds **1**–**8** and reference compound **32**.

O

OBn OMe

61% **<sup>13</sup>**

1. 9-BBN, THF, ∆ 2. NaOH/ H2O2 THF, rt

O

OR

OR OMe

**14b** R=Bn  **1b** R=H (63%)

RO

HO

Pd(OH)2-C H2

O

OR

OR OMe

RO

HO **14a** R=Bn **1a** R=H (51%)

Pd(OH)2-C H2

BnO

OBn Petasis reagent

Toluene 60 oC 79%

**Scheme 1.** Synthesis of methyl 3-deoxy-3-*C*-(hydroxymethyl)-*β*-D-gulopyranoside **1a**, methyl 3-

deoxy-3-*C*-(hydroxymethyl)-*β*-D-galactopyranoside **1b**.

O

OBn

OBn OMe

BnO

Dess-Martin periodinane DCM, rt 84% **11 12**

O

O

OBn

OBn OMe

BnO HO


**Table 1.** *K*d-values (mM)<sup>a</sup> of compounds **1a**–**1b**, **2**–**3**, **7a**, **8**, and the reference methyl β-d-galactopyranoside **32** against human galectin-1, 2, 3, 4N, 4C, 7, 8N, 8C, 9N, and 9C in a competitive fluorescence polarization assay [15,16].

<sup>a</sup> The data are average and SEM (standard error of mean) of 4–8 single-triple point measurements. <sup>b</sup> N-terminal domain. <sup>c</sup> C-terminal domain. <sup>d</sup> Not determined. <sup>e</sup> Not binding at the highest concentration tested: 4 mM.

In stark contrast, the galacto derivative **1b** did not bind any galectin tested, except for a weak binding to galectin-4N. This observation encouraged us to further explore the 3-*C*-methyl gulopyranoside scaffold for the discovery of galectin-1-selective inhibitors. Hence, we initiated synthetic efforts toward replacing the hydroxymethyl of **1a** with amide, ether, urea, and triazole functionalities. An aryl ether was synthesized following a recently reported iodonium-salt mediated reaction [20] to give the aryl ether **15**, which after hydrogenolysis [13] of the benzyl protecting groups gave **2** (Scheme 2). The hydroxymethyl **14a** was methylated with methyl iodide to give the methyl ether **16**, which after debenzylation gave the 3-methoxymethyl guloside **3**. Treatment of **14a** with methanesulfonyl chloride furnished the corresponding gulo mesylate, which was then directly treated with NaN<sup>3</sup> in dry DMF at 80 ◦C to provide the gulo azide, **17** in 83% yield. The gulo azide **17** was treated with 1-ethynyl-3-fluorobenzene in the presence of the CuI and DIPEA catalytic system [21] in dry dichloromethane to give the triazole **18** within 48 h in 86% yield. Debenzylation provided the desired triazole-derived methyl guloside **4**. The urea **20** was obtained via reduction of the azide **17** to give the amine **19**, followed by reaction with 3-fluorophenylisocyanate. Debenzylation [13] of **20** afforded the target gulo urea derivative **5** in 66% yield. The amine **19** was treated with benzensulfonyl chloride, benzoyl chloride, and diphenyl phosphoryl chloride in the presence of Et3N to give the protected sulfonamide **21**, amide **22a**, and diphenylphosphonamide **23**, which were subjected for hydrogenolysis [13] in the presence of Pd(OH)<sup>2</sup> to get the unprotected amides **6**, **7a**, and **8**.

Evaluation of aryl ether **2**, methyl ether **3**, triazole **4**, urea **5**, sulfonamide **6**, benzamide **7a**, and phosphonamide **8** derived methyl gulosides' affinities for the human galectin-1, 2, 3, 4N, 4C, 7, 8N, 8C, 9N, and 9C showed that most of the gulo derivatives were inactive as ligands for galectins, the benzamide **7a** displayed moderate affinity, similar to that of **1a**, for galectin-1 and with excellent selectivity (Table 1). Particularly noteworthy was that **7a** also had a significantly better affinity for galectin-1 than the simple reference monosaccharide methyl β-d-galactopyranoside **32**. Furthermore, the hydroxylmethyl group of **1a** plays an important role in the interaction with galectin-1, as the corresponding methyl ether **3** binds galectin-1 significantly worse than **1a** does.

**Scheme 2.** Synthesis of methyl 3-deoxy-3-C-methyl-*β*-D-gulopyranoside ether **2**–**3**, triazole **4**, urea **5**, sulfonamide **6**, amide **7a**, and phosphonamide **8** derivatives. **Scheme 2.** Synthesis of methyl 3-deoxy-3-*C*-methyl-β-d-gulopyranoside ether **2**–**3**, triazole **4**, urea **5**, sulfonamide **6**, amide **7a**, and phosphonamide **8** derivatives.

#### *2.2. Synthesis and Optimization of 3-deoxy-3-C-Amidomethyl-β-D-Gulopyranoside Derivatives as Galectin-1 Inhibitors 2.2. Synthesis and Optimization of 3-Deoxy-3-C-Amidomethyl-*β*-*d*-Gulopyranoside Derivatives as Galectin-1 Inhibitors*

The observation that the amide **7a** showed moderate affinity but high selectivity for galectin-1 prompted us to prepare a series **7c**–**7l** of benzamide analogs carrying selected different substituents at different positions, including four fluorbenzamide expected to possess improved metabolic stability and pharmacokinetic properties, as well as a reference acetamide analog **7b** (Scheme 3). Furthermore, in order to investigate the role of the gulo 3-*C-*methyl substituent, the 3-OH **9** and 3 benzamido **10** gulosides were synthesized (Scheme 3). Hydrolysis of the known 4,6-*O*-benzylidene gulose derivative, **24** [22] with 80% AcOH at 80 °C gave the diol **25** in 91% yield, which upon Zemplen de-*O*-acetylation [23] afforded the target methyl *β*-D-gulopyranoside **9** in 93% yield. Selective 3-*O*triflation of methyl 4,6-*O*-benzylidene-*β*-D-galactopyranoside **26** [24], followed by one-pot benzoylation of 2-*O*-hydroxyl gave **27**. The crude triflate **27** was subsequently converted into the 3 azido-3-deoxy-guloside **28** by treatment with sodium azide in DMF. De-benzylidenation with 80% AcOH at 80 °C and subsequent benzoylation afforded **29** in 43% yield over four steps from **26**. Azide hydrogenation gave **30**, which upon benzoylation and de-*O*-benzoylation gave the benzamide **10**. The observation that the amide **7a** showed moderate affinity but high selectivity for galectin-1 prompted us to prepare a series **7c**–**7l** of benzamide analogs carrying selected different substituents at different positions, including four fluorbenzamide expected to possess improved metabolic stability and pharmacokinetic properties, as well as a reference acetamide analog **7b** (Scheme 3). Furthermore, in order to investigate the role of the gulo 3-*C*-methyl substituent, the 3-OH **9** and 3-benzamido **10** gulosides were synthesized (Scheme 3). Hydrolysis of the known 4,6-*O*-benzylidene gulose derivative, **24** [22] with 80% AcOH at 80 ◦C gave the diol **25** in 91% yield, which upon Zemplen de-*O*-acetylation [23] afforded the target methyl β-d-gulopyranoside **9** in 93% yield. Selective 3-*O*-triflation of methyl 4,6-*O*-benzylidene-β-d-galactopyranoside **26 [24]**, followed by one-pot benzoylation of 2-*O*-hydroxyl gave **27**. The crude triflate **27** was subsequently converted into the 3-azido-3-deoxy-guloside **28** by treatment with sodium azide in DMF. De-benzylidenation with 80% AcOH at 80 ◦C and subsequent benzoylation afforded **29** in 43% yield over four steps from **26**. Azide hydrogenation gave **30**, which upon benzoylation and de-*O*-benzoylation gave the benzamide **10**.

An immediate observation upon evaluating the affinities of **7b**–**7l** and **9**–**10** (Figure 2, Table 2) was that the acetamide **7b** displays a similar affinity for galectin-1 as the benzamides **7a** and **7c**–**7k**. Hence, the phenyl moieties of **7a** and **7c**–**7k** do not contribute to enhancing the affinity for galectin-1. However, the phenyl moieties and substitution patterns of **7a** and **7c**–**7k** influence the selectivity over other galectins, as six substituted amides (**7a**, **7d**, and **7f**–**7i**) retained high selectivity for galectin-1 over the other galectins. The pentafluorophenyl **7g** turned out to be the best β-D-gulopyranosidebased monosaccharide inhibitor for human galectin-1 with 14-fold improved affinity over the reference methyl β-D-galactopyranoside **32**. The larger biphenyl **7l** did not bind galectin-1, which suggests that the galectin-1 site accommodating the axial gulo substituent is limited in size. Evaluation of the guloside **9** revealed that while it is similar to the reference galactoside **32** in the affinity for galectin-1, it displays a much higher selectivity in that it is inactive against the other galectins under the evaluation conditions used. Unfortunately, extensive molecular dynamics and docking analyses to explain the selective galectin-1 binding to 3-*C-*methyl-gulosides were inconclusive as such calculations cannot provide reliable relative affinities of bound ligands. Hence, it remains to find a plausible structural explanation for this selectivity. Interestingly, the benzamide **10** showed no binding to galectin-1 under the assay conditions but instead had improved binding to An immediate observation upon evaluating the affinities of **7b**–**7l** and **9**–**10** (Figure 2, Table 2) was that the acetamide **7b** displays a similar affinity for galectin-1 as the benzamides **7a** and **7c**–**7k**. Hence, the phenyl moieties of **7a** and **7c**–**7k** do not contribute to enhancing the affinity for galectin-1. However, the phenyl moieties and substitution patterns of **7a** and **7c**–**7k** influence the selectivity over other galectins, as six substituted amides (**7a**, **7d**, and **7f**–**7i**) retained high selectivity for galectin-1 over the other galectins. The pentafluorophenyl **7g** turned out to be the best β-d-gulopyranoside-based monosaccharide inhibitor for human galectin-1 with 14-fold improved affinity over the reference methyl β-d-galactopyranoside **32**. The larger biphenyl **7l** did not bind galectin-1, which suggests that the galectin-1 site accommodating the axial gulo substituent is limited in size. Evaluation of the guloside **9** revealed that while it is similar to the reference galactoside **32** in the affinity for galectin-1, it displays a much higher selectivity in that it is inactive against the other galectins under the evaluation conditions used. Unfortunately, extensive molecular dynamics and docking analyses to explain the selective galectin-1 binding to 3-*C*-methyl-gulosides were inconclusive as such calculations cannot provide reliable relative affinities of bound ligands. Hence, it remains to find a plausible structural explanation for this selectivity. Interestingly, the benzamide **10** showed no binding to galectin-1 under the assay conditions but instead had improved binding to and selectivity for galectin-4N. Hence, while

3-*C*-methyl gulosides represent an interesting structural class for the discovery of selective galectin-1 inhibitors, 3-*C*-amido gulosides may represent a novel structural class for galectin-4 inhibitor discovery. *Int. J. Mol. Sci.* **2017**, *18*, x 6 of 23

**Scheme 3.** Synthesis of 3-deoxy-3-*C-*amidomethyl-*β*-D-gulo derivatives **7b**–**7l**, methyl *β*-Dgulopyranoside **9**, and methyl 3-deoxy-3-*N*-benzamido-*β*-D-gulopyranoside **10**. **Scheme 3.** Synthesis of 3-deoxy-3-*C*-amidomethyl-β-d-gulo derivatives **7b**–**7l**, methyl β-d-gulopyranoside **9**, and methyl 3-deoxy-3-*N*-benzamido-β-d-gulopyranoside **10**. and selectivity for galectin-4N. Hence, while 3-*C*-methyl gulosides represent an interesting structural class for the discovery of selective galectin-1 inhibitors, 3-*C*-amido gulosides may represent a novel structural class for galectin-4 inhibitor discovery.

13C NMR: CDCl3 δ 77.16; CD3OD δ 49.00) with multiplicity (*b* = broad, *s* = singlet, *d* = doublet, *t* = triplet, *q* = quartet, *quin* = quintet, *sext* = sextet, *hept* = heptet, *m* = multiplet, *td* = triplet of doublets, *dt* **Figure 2.** Structures of all tested compounds **7a**–**7l** and **9–10**. **Figure 2.** Structures of all tested compounds **7a**–**7l** and **9**–**10**.

= doublet of triplets), coupling constants (in Hz) and integration. Copies of nmr spectra are provided in the supplementary information. High-resolution mass analysis was obtained using the Micromass **Table 2.** *K*d-values (mM)a of compounds **7a**–**7l**, **9**, and **10** against human galectin-1, 2, 3, 4N, 4C, 7, 8N, **Table 2.** *K*d-values (mM)<sup>a</sup> of compounds **7a**–**7l**, **9**, and **10** against human galectin-1, 2, 3, 4N, 4C, 7, 8N, 8C, 9N, and 9C in a competitive fluorescence polarization assay.


Hz, C*H*2Ph), 4.73 (d, 1H, *J* 12.0 Hz, C*H*2Ph), 4.58 (d, 1H, *J* 12.0 Hz, C*H*2Ph), 4.51 (d, 1H, *J* 12.0 Hz, **7l** NB NB NB NB NB NB NB NB NB NB <sup>a</sup> The data are average and SEM of 4–8 single-triple point measurements. <sup>b</sup> N-terminal domain. <sup>c</sup> C-terminal domain.

**9** 10 ± 0.25 10 ± 1.5 NB ND 11 ± 1.2 NB NB NB NB NB **10** NB NB NB 1.3 ± 0.2 NB ND NB NB NB NB <sup>d</sup> Not binding at the highest concentration tested: 4 mM. <sup>e</sup> Not determined.

c

determined.

#### **3. Materials and Methods**

#### *3.1. General Methods Experimental Procedures*

All reactions were carried out in oven-dried glassware. All solvents and reagents were mainly purchased from Sigma-Aldrich or Fluka and were used without further purification or synthesized via the literature protocol. TLC analysis was performed on pre-coated Merck silica gel 60 F<sup>254</sup> plates using UV light and charring solution (10 mL conc. H2SO4/90 mL EtOH). Flash column chromatography was done on SiO<sup>2</sup> purchased from Aldrich (technical grade, 60 Å pore size, 230–400 mesh, 40–63 µm). All NMR spectra were recorded with the Bruker DRX 400 MHz spectrometer (400 MHz for <sup>1</sup>H, 100 MHz for <sup>13</sup>C (125 MHz <sup>13</sup>C for compound **7k** with the Bruker Avance III 500 MHz spectrometer equipped with a broadband observe SMART probe, Fällanden, Switzerland), 376 MHz for <sup>19</sup>F, 162 MHz for <sup>31</sup>P, ESI) at ambient temperature using CDCl<sup>3</sup> or CD3OD as solvents. Chemical shifts are given in ppm relative to the residual solvent peak (1H NMR: CDCl<sup>3</sup> δ 7.26; CD3OD δ 3.31; <sup>13</sup>C NMR: CDCl<sup>3</sup> δ 77.16; CD3OD δ 49.00) with multiplicity (*b* = broad, *s* = singlet, *d* = doublet, *t* = triplet, *q* = quartet, *quin* = quintet, *sext* = sextet, *hept* = heptet, *m* = multiplet, *td* = triplet of doublets, *dt* = doublet of triplets), coupling constants (in Hz) and integration. Copies of nmr spectra are provided in the supplementary information. High-resolution mass analysis was obtained using the Micromass Q-TOF mass spectrometer. Analytical data is given if the compound is novel or not fully characterized in the literature. Final compounds were further purified via HPLC before evaluation of galectin affinity. All tested compounds were >95% pure according to the analytical HPLC analysis.

#### *3.2. Methyl 2,4,6-Tri-O-Benzyl-*β*-*d*-Xylo-Hex-3-Ulopyranoside* **12**

Into a solution of alcohol **11** (8.1 g, 17.45 mmol) in dry dichloromethane (250 mL) Dess–Martin periodinane (9.62 g, 22.68 mmol, 1.3 equiv.) was added, under nitrogen atmosphere and the reaction mixture was stirred for 4 h (TLC heptane/EtOAc, 3:1, R<sup>f</sup> 0.5). After that, a saturated NaHCO<sup>3</sup> solution (400 mL) was added and the mixture was stirred for 30 min. Then, the organic layer was collected and washed successively with the saturated Na2S2O<sup>3</sup> solution (2 × 250 mL). The organic layer was collected, dried over Na2SO4, filtered and concentrated in vacuo. Flash chromatography of the crude material (heptane/EtOAc, 7:2) afforded ketone **12** (6.45 g, 13.955 mmol, yield 80%) as a white solid. [α] 25 <sup>D</sup> <sup>−</sup>72.3 (c 1.4, CHCl3). <sup>1</sup>H NMR (CDCl3, 400 MHz): 7.47–7.21 (m, 15H, Ar*H*), 4.76 (d, 1H, *<sup>J</sup>* 12.0 Hz, C*H*2Ph), 4.73 (d, 1H, *J* 12.0 Hz, C*H*2Ph), 4.58 (d, 1H, *J* 12.0 Hz, C*H*2Ph), 4.51 (d, 1H, *J* 12.0 Hz, C*H*2Ph), 4.48 (d, 1H, *J*1,2 7.6 Hz, H-1), 4.44 (d, 1H, *J*1,2 7.6 Hz, H-2), 4.43 (d, 1H, *J* 11.6 Hz, C*H*2Ph), 4.35 (d, 1H, *J* 11.6 Hz, C*H*2Ph), 3.95 (d, 1H, *J* 1.2 Hz, H-4), 3.83–3.75 (m, 3H, H-5, H-6a, H-6b), 3.61 (s, 3H, OC*H*3). <sup>13</sup>C NMR (CDCl3, 100 MHz): 203.8, 137.7, 137.2, 136.4, 128.29, 128.27, 127.26, 128.0, 127.8, 127.6, 127.5, 104.9, 82.1, 80.7, 73.5, 73.4, 72.3, 72.1, 67.5, 57.1. HRMS calcd for C28H30O<sup>6</sup> <sup>+</sup>NH<sup>4</sup> <sup>+</sup> (M+NH4) <sup>+</sup>: 480.2386, found: 480.2378.

#### *3.3. Methyl 2,4,6-Tri-O-Benzyl-3-Deoxy-3-C-Methylene-*β*-*d*-Xylo-Hex-3-Ulopyranoside* **13**

Into a solution of ketone **12** (6.3 g, 13.63 mmol) in dry toluene (100 mL) bis (cyclopentadienyl) dimethyltitanium was added, 5 wt% in toluene (125 mL, 30 mmol, 2.2 equiv.), under nitrogen atmosphere and the reaction mixture was stirred for 48 h at 65 ◦C in the dark. After that, the reaction mixture (TLC heptane/EtOAc, 4:1, R<sup>f</sup> 0.47) was concentrated in vacuo and flash chromatography of the crude material (heptane/EtOAc, 10:1–5:1) afforded methylene derivative **13** (4.6 g, 9.99 mmol, yield 71%) as a light-yellow oil. [α] 25 <sup>D</sup> <sup>−</sup>40.3 (c 1.1, CHCl3). <sup>1</sup>H NMR (CDCl3, 400 MHz): 7.49–7.28 (m, 15H, Ar*H*), 5.61 (t, 1H, *J*2,H-7a 2.0 Hz, C*H*2), 5.20 (t, 1H, *J*2,H-7b 2.0 Hz, C*H*2), 5.00 (d, 1H, *J* 12.0 Hz, C*H*2Ph), 4.78 (d, 1H, *J* 12.0 Hz, C*H*2Ph), 4.65 (d, 1H, *J* 12.0 Hz, C*H*2Ph), 4.58 (d, 1H, *J* 12.0 Hz, C*H*2Ph), 4.56 (d, 1H, *J* 12.0 Hz, C*H*2Ph), 4.36 (d, 1H, *J* 7.6 Hz, H-1), 4.28 (d, 1H, *J* 12.0 Hz, C*H*2Ph), 4.14 (dt, 1H, *J*1,2 7.6 Hz, *J*2,H-7a, *J*2,H-7b 2.0 Hz, H-2), 4.03 (d, 1H, *J* 0.4 Hz, H-4), 3.91–3.79 (m, 3H, H-5, H-6a, H-6b), 3.65 (s, 3H, OC*H*3). <sup>13</sup>C NMR (CDCl3, 100 MHz): 142.2, 138.5, 138.2, 137.9, 128.4, 128.3, 128.2, 128.0, 127.9, 127.7, 127.62, 127.59, 127.5, 113.7, 104.9, 77.7, 77.3, 76.6, 73.7, 73.6, 69.2, 69.0, 56.7. HRMS calcd for C29H32O5+NH<sup>4</sup> <sup>+</sup> (M+NH4) <sup>+</sup>: 478.2593, found: 478.2607.

### *3.4. Methyl 2,4,6-Tri-O-Benzyl-3-Deoxy-3-C-Hydroxymethyl-*β*-*d*-Gulopyranoside* **14a** *and Methyl 2,4,6-Tri-O-Benzyl-3-Deoxy-3-C-Hydroxymethyl-*β*-*d*-Galactopyranoside* **14b**

A solution of **13** (4.6 g, 9.99 mmol) in dry THF (150 mL) was treated with a 9-BBN solution in THF (0.5 M, 125 mL) and heated to reflux for 24 h. After that, the solution was cooled to 0 ◦C and a 10% aqueous sodium hydroxide solution (100 mL) and a 30% hydrogen peroxide solution (100 mL) were added simultaneously within 5 min and stirring continued for another 30 min. Then, diethyl ether (200 mL) was added followed by careful addition of a 20% aqueous sodium hydrogen sulfite solution (7 mL). This mixture was stirred further for 60 min and extracted with diethyl ether, and the combined organic layers were dried with Na2SO4, filtered, and concentrated in vacuo (TLC heptane/EtOAc, 2:1 (double run), R<sup>f</sup> 0.48 for **14a**, R<sup>f</sup> 0.4 for **14b**). Flash chromatography (Heptane/EtOAc, 8:1 to 2:1) of the residue afforded a gulo-isomer, **14a** (1.74 g, 3.638 mmol) and galacto-isomer, **14b** (1.16 g, 2.426 mmol) to be ≈3:2 in favor of the guloisomer at an overall yield of 61% (2.9 g, 6.064 mmol). Gulo-isomer **14a**: [α] 25 <sup>D</sup> <sup>−</sup>25.7 (c 1.3, CHCl3). <sup>1</sup>H NMR (CDCl3, 400 MHz): 7.36–7.20 (m, 15H, Ar*H*), 4.82 (d, 1H, *<sup>J</sup>* 12.0 Hz, C*H*2Ph), 4.65 (d, 1H, *J* 6.4 Hz, H-1), 4.57 (d, 1H, *J* 11.6 Hz, C*H*2Ph), 4.54 (d, 1H, *J* 11.6 Hz, C*H*2Ph), 4.52 (d, 1H, *J* 11.6 Hz, C*H*2Ph), 4.47 (d, 1H, *J* 12.0 Hz, C*H*2Ph), 4.41 (d, 1H, *J* 11.6 Hz, C*H*2Ph), 3.95–3.88 (m, 2H, H-4, H-5), 3.73 (dd, 1H, *J*1,2 6.4 Hz, *J*2,3 5.2 Hz, H-2), 3.74–3.57 (m, 4H, H-6a, H-6b, C*H*2OH), 3.54 (s, 3H, OC*H*3), 2.53–2.47 (m, 1H, H-3), 2.35 (bs, 1H, CH2O*H*). <sup>13</sup>C NMR (CDCl3, 100 MHz): 138.2, 138.1, 138.0, 128.6, 128.52, 128.47, 128.2, 128.1, 128.03, 127.96, 127.9, 127.8, 101.2, 77.2, 74.8, 73.8, 73.6, 73.4, 71.9, 69.5, 62.0, 56.5, 41.6. HRMS calcd for C29H34O6+NH<sup>4</sup> <sup>+</sup> (M+NH4) <sup>+</sup>: 496.2699, found: 496.2700. Galacto-isomer **14b**: [α] 25 <sup>D</sup> <sup>−</sup>13.4 (c 0.9, CHCl3). <sup>1</sup>H NMR (CDCl3, 400 MHz): 7.39–7.28 (m, 15H, Ar*H*), 4.92 (d, 1H, *J* 11.2 Hz, C*H*2Ph), 4.65 (d, 1H, *J* 11.2 Hz, C*H*2Ph), 4.60–4.52 (m, 4H, C*H*2Ph), 4.41 (d, 1H, *J*1,2 7.6 Hz, H-1), 3.90(d, 1H, *J*3,4 2.8 Hz, H-4), 3.82 (dd, 1H, *J* 4.8 Hz, *J* 7.2 Hz, C*H*2OH), 3.73–3.55 (m, 8H, H-5, H-6a, H-6b, H-2, C*H*2OH, OC*H*3), 2.04 (bs, 1H, CH2O*H*), 1.87–1.82 (m, 1H, H-3). <sup>13</sup>C NMR (CDCl3, 100 MHz): 138.5, 138.1, 137.8, 128.6, 128.53, 128.52, 128.4, 128.3, 128.03, 127.98, 127.8, 106.4, 76.5, 76.2, 74.8, 74.7, 74.6, 73.7, 68.6, 62.2, 56.8, 47.3. HRMS calcd for C29H34O6+H<sup>+</sup> (M+H)+: 479.2434, found: 479.2434.

### *3.5. Methyl 2,4,6-Tri-O-Benzyl-3-Deoxy-3-C-(3-Trifluoromethylphenoxymethyl)-*β*-*d*-Gulopyranoside* **15**

Compound **14a** (80 mg, 0.17 mmol) was stirred in a 25 mL round-bottom flask in toluene (2 mL) for 3 min. A mixture of 3-(trifluoromethyl)phenyl)(4-methoxyphenyl)iodonium tosylate (140 mg, 0.25 mmol) and potassium tert-butoxide (28.5 mg, 0.25 mmol) were added under air and the mixture turned yellow. The reaction was stirred for 3 h, when the TLC showed almost complete consumption of the starting material (TLC heptane/EtOAc, 3:1, R<sup>f</sup> 0.48). The mixture was then diluted with EtOAc (10 mL) and filtered. Then the volatiles were removed under reduced pressure, and the residue was subjected to column chromatography (heptane/EtOAc, 8:1 to 4:1) to provide the purified product **15** (92.6 mg, 0.15 mmol, 89%) as a colorless oil. [α] 25 <sup>D</sup> <sup>−</sup>70.9 (c 0.8, CHCl3). <sup>1</sup>H NMR (CDCl3, 400 MHz): 7.40–7.22 (m, 17H, Ar*H*), 7.08 (bs, 1H, Ar*H*), 7.02 (dd, 1H, *J* 8.0 Hz, *J* 2.4 Hz, Ar*H*), 4.79 (d, 1H, *J* 12.0 Hz, C*H*2Ph), 4.62 (d, 1H, *J* 6.0 Hz, H-1), 4.58 (d, 1H, *J* 12.0 Hz, C*H*2Ph), 4.57 (d, 1H, *J* 11.6 Hz, C*H*2Ph), 4.54 (d, 1H, *J* 12.4 Hz, C*H*2Ph), 4.50 (s, 2H, C*H*2Ph), 4.24 (dd, 1H, *J* 6.0 Hz, *J* 9.6 Hz, H-3a*'*), 4.19–4.15 (m, 1H, H-5), 4.08 (dd, 1H, *J* 9.6 Hz, *J* 8.0 Hz, H-3b*'*), 3.87 (dd, 1H, *J* 5.2 Hz, *J* 2.8 Hz, H-4), 3.85 (t, 1H, *J* 6.0 Hz, H-2), 3.80 (dd, 1H, *J* 10.0 Hz, *J* 6.8 Hz, H-6a), 3.72 (dd, 1H, *J* 10.0 Hz, *J* 5.2 Hz, H-6b), 3.57 (s, 1H, OC*H*3), 2.76–2.70 (m, 1H, H-3). <sup>13</sup>C NMR (CDCl3, 100 MHz): 158.8, 138.30, 138.27, 137.9, 131.9 (q, *J* 32.1 Hz), 130.0, 128.49, 128.46, 128.3, 128.0, 127.9, 127.83, 127.77, 124.1 (q, *J* 271 Hz), 118.0, 117.6 (q, *J* 3.8 Hz), 111.5 (q, *J* 3.7 Hz), 101.3, 74.7, 73.6, 73.5, 73.3, 72.8, 71.9, 69.8, 64.6, 56.4, 39.6. <sup>19</sup>F NMR (CDCl3, 376 MHz): −62.6. HRMS calcd for C36H41F3NO6+NH<sup>4</sup> <sup>+</sup> (M+NH4) <sup>+</sup>: 640.2886, found: 640.2895.
