A Synthetic Galectin Mimic

Abstract

Galectins are a galactoside specific subclass of carbohydrate binding proteins (lectins) involved in various cellular activities, certain cancers, infections, inflammations, and many other biological processes. The molecular basis for the selectivity of galectins is well-documented and revolves around appropriate interaction complementarity: an aromatic residue for C-H⋅⋅⋅π interactions and polar residues for (charge assisted) hydrogen bonds with the axial hydroxyl group of a galactoside. However, no synthetic mimics are currently available. We now report on the design and synthesis of the first galectin mimic (6), and show that it has a higher than 65-fold preference for n-octyl-β-galactoside (8) over n-octyl-β-glucoside (7) in CD₂ Cl₂ containing 5 % [D₆ ]DMSO (with K_a ≥4500 M^-1 for 6:8). Molecular modeling informed by nOe studies reveal a high degree of interaction complementarity between 6 and galactoside 8, which is very similar to the interaction complementarity found in natural galectins.

Keywords: carbohydrate recognition; carbohydrates; galactosides; galectin mimic; molecular recognition.

Figure 1

a) Galactoside binding domain of human galectin‐3 (1A3K) bound to the galactose residue of N‐acetyl‐D‐lactosamine (H‐bonded water molecules are shown as red spheres). b) The structures of O‐linked β‐D‐glucoside/galactoside with OH‐4 highlighted in blue. c) Covalent cage as lectin mimic for all‐equatorial carbohydrates. d) New design presented here aimed to mimic galectins. R=solubility handle.

Scheme 1

Synthesis of tetrapyridyl ligand 5. PFP=pentafluorophenyl, DIPEA=N,N‐diisopropylethylamine, THF=tetrahydrofuran. The given yields in percentages (bottom) are non‐optimized isolated yields. Compound 1 can be prepared on multiple gram scale in >30 % isolated yield from p‐bromo‐t‐butylbenzene as detailed in the Supporting Information.

Figure 2

a) Formation of 6 (bottom) from 5 (top) by addition of the indicated equivalent of Pd(BArF)₂, as followed by ¹H NMR (spectra partially assigned). Solvent is 5 % [D₆]DMSO in CD₂Cl₂. b) Representation and labeling of 6; c) partial NOESY spectrum of 6 zoomed‐in on the nOe's between b3α and b2/b4. See the Supporting Information, Figures S28–S33 for complete spectra and full assignments.

Figure 3

Molecular model of 6 with partial assignment as calculated with DFT (ωB97X‐D/ 6‐31G*) and viewed: a) facing the smaller p2/s4/b4 portal; b) facing the larger p2/s2/b2 portal; c) from the interior looking down at the flat biphenyl; d) from the interior looking at the uneven surface of the [Pd(pyridyl)₄]²⁺ complex. The solubilizing groups are omitted and the distances are corrected for the van der Waals radii of the atoms. See also the Supporting Information, Section S5a for details.

Figure 4

a) Partial ¹H NMR spectra and HypNMR curve fitting analysis of 1.9 mM [6][BArF]₂ titrated with glucoside 7. Fitting was done on protons s3‐NH, p3, s4, and p4, giving K _a=67 M⁻¹ with r ²=0.9963 over all 52 data points. b) Partial ¹H NMR spectra of 3.3 mM [6][BArF]₂ titrated with galactoside 7 up to about three equivalents. The top spectrum is taken at the end of the titration at −30 °C. The solvent is in CD₂Cl₂ with 5 % [D₆]DMSO. See also the Supporting Information, Figure S39 and Figure S40.

Figure 5

Partial ¹H NMR spectrum of [4⋅7][BArF]₂ at 25 °C and [4⋅8][BArF]₂ at −30 °C together with selective 1D nOe's recorded at the same temperature after excitation of p3, b2 or s4 (t _m=350 ms). The large signal at 3.16 p.p.m. in the sample with 8 at −30 °C is from water.

Figure 6

a) Truncated view of a molecular model of [6⋅7]²⁺ in space‐filling mode (left) and as capped sticks visualizing four of the seven HB interactions found. b) idem for [6⋅8]²⁺ with five out of ten HBs. Both models were DFT optimized (ωB97X‐D/ 6‐31G*) and the complex with 7 was calculated to be 12 kcal mol⁻¹ less stable. Only polar hydrogens and hydrogens involved in CH–π interactions are shown for simplicity. See the Supporting Information, Section S5b for full details.