The Bicyclic Form of galacto-Noeurostegine Is a Potent Inhibitor of β-Galactocerebrosidase

Abstract

Competitive inhibitors of galactocerebrosidase (GALC) could be candidates for pharmacological chaperone therapy of patients with Krabbe disease. The known and selective nortropane-type iminosugar galacto-noeurostegine has been found to competitively inhibit GALC with K_i = 7 μM at pH 4.6, which is 330-fold more potent than the analogous deoxynoeurostegine. It was shown through X-ray protein crystallography that galacto-noeurostegine binds to the active site of GALC in its bicyclic form.

Krabbe disease is a rare and fatal neurodegenerative disease for which there is neither a cure nor a successful general mode of treatment.^1,2 This disease is caused by the loss or severe reduction in catalytic activity of the lysosomal enzyme galactocerebrosidase (GALC).³ GALC functions in lysosomal compartments of the cell to hydrolyze the galactosylated sphingolipid substrates galactocerebroside and psychosine. Reduced enzyme activity is caused by a range of different mutations in the gene that encodes the GALC enzyme.^4,5 Several of these mutations are predicted to cause misfolding of GALC on the basis of structural data.⁶ Although these mutations result in enzyme misfolding, they do not necessarily prevent the enzyme from remaining catalytically active. Therefore, in the case of these mutations, if GALC can be allowed to reach the lysosome, the enzyme will still function and thus prevent the pathologies of Krabbe disease from occurring. One way of targeting this is through the use of pharmacological chaperone therapy (PCT).^7,8 This method targets the disease-causing enzyme with a small-molecule inhibitor that facilitates the correct folding of the enzyme upon binding. This conformational change stabilizes the protein and allows it to evade the quality control mechanisms that lead to protein degradation and reach its final biological destination. We have studied and characterized a series of ligands and their binding to native GALC and thereby their potential as pharmacological chaperones against Krabbe disease.^9,10

Noeurostegines¹¹⁻¹³ are a class of synthetic compounds conceived by us to be structural hybrids of the naturally occurring calystegines (e.g., calystegine B₂ (1))¹⁴ and the synthetic iso-iminosugar noeuromycin (2) and its family members with varying hydroxylation patterns (Figure 1).¹⁵ The ethylene bridge contained in the calystegine family was installed to obstruct Amadori rearrangement of the otherwise unstable hemiaminal function of noeuromycin.¹⁶

Figure 1.

Structures of some imino- and iso-iminosugars and GALC inhibition at pH 4.6 by GNS (5) and DGN (6).

The reason for our interest in these molecules is their ability to potently inhibit glycosidase activity. Both noeurostegine (3)¹¹ and its uronic and galacto congeners (4 and 5, respectively)¹² were found to be highly potent enzyme inhibitors, that possess an increased level of selectivity compared with the noeuromycin analogues not having the stabilizing ethylene bridge. Additionally, 3 has previously been found by us to potently inhibit native glucocerebrosidase and possess characteristics suggesting that this molecule could be active as a pharmacological chaperone in the treatment of the most common lysosomal storage disorder, Gaucher’s disease.¹⁷ In this communication, we report our findings regarding the interactions between GALC and galacto-noeurostegine (GNS, 5).

We previously explored the inhibition of three galactosidases (α-galactosidase from green coffee beans, β-galactosidase from Escherichia coli, and β-galactosidase from Aspergillus oryzae) by 5 and its 2-deoxy analogue 2-deoxy-galacto-noeurostegine (DGN, 6) and found inhibition below 1000 μM only for the last of these enzymes.¹³ The inhibition constant (K_i) toward A. oryzae β-galactosidase was measured to be competitive at 31 and 130 nM, with 5 being a factor of 4 more potent than its deoxy analogue 6.¹³ In the context of Krabbe disease and native GALC inhibition, only 6 and not 5 had been evaluated and established to be inactive (K_i = 2300 μM, pH 4.6).¹³ Despite its high degree of resemblance to 6, we can here report that 5 is a competitive inhibitor of GALC at 7 μM (pH 4.6). This makes it 330-fold more potent against GALC than its 2-deoxy analogue 6 and only 18-fold less potent than the broad-spectrum galactosidase inhibitor iso-galacto-fagomine (7) (Figure 1).

Because of the hemiaminal functionality of the noeurostegines (3–5), they can potentially exist in both monocyclic (aminocycloheptanone) and bicyclic (hemiaminal) form. Both 3 and its uronic congener 4 were found to exist exclusively as their bicyclic forms, while both forms were found in solution of 5. We ascribe this to the existence of destabilizing 1,3-diaxial interactions between the ethylene bridge and the axial hydroxyl group (Scheme 1).¹³

Scheme 1. Structural Equilibrium for GNS (5) in Water.

Although the 2-OH group of a monosaccharide substrate is known to be highly important for glycosidase binding,^18,19 the degree of difference in inhibition of GALC by 5 and 6 still seemed to be rather large. This led us to speculate whether enzyme binding still took place between GALC and the bicyclic version of 5 or whether the ligand had undergone a structural change and the increased potency originated from the monocyclic constitutional isomer (Scheme 1). In order to investigate this, the X-ray crystal structures of GALC in complex with 5 were determined at pH 4.6 and pH 6.8, corresponding to the pH values of the lysosomal and secretory compartments, respectively (see the Supporting Information).

These structures show that 5 binds in its bicyclic form at both pH 4.6 and pH 6.8 and that the binding mode is very similar to that seen for 6 and 7 (Figure 2).⁹ The hemiaminal hydroxyl group of 5, which is absent in both 6 and 7, is positioned in a hydrophilic pocket and interacts with N181, W135, and E258 through hydrogen bonding. The additional hydrogen bonding that 5 possesses likely contributes to its increased potency in inhibition of GALC. The ability of bicyclic 5 to fit into this hydrophilic pocket not only allows additional hydrogen bonds to form but also enables the molecule to position itself in the active site without causing steric clashes.

Figure 2.

(A, B) X-ray crystal structures of GNS (5) bound to GALC at (A) pH 6.8 and (B) pH 4.6. Unbiased difference electron density maps (F_o – F_c, 3.0σ, green) and key active-site residues (blue sticks) are shown. (C) DGN (6) binds in a similar position but with fewer hydrogen bonds. (D) Schematic representation of hydrogen-bonding interactions (purple dashed lines) between GALC and ligand 5.

The thermal stabilization of GALC in the presence of 5 was found to be pH-dependent, with a high degree of stabilization at pH 4.6 (ΔT_m = +5.80 °C), which diminishes as the pH approaches 7.4, where a much lower degree of stabilization is observed (ΔT_m = +1.40 °C) (Figure 3). This effect most likely occurs because the strength of the electrostatic interaction between E258 and 5 decreases and a lower proportion of the ligand amine remains protonated as the pH increases. The importance of this electrostatic interaction between GALC and the inhibitor’s amine moiety is in agreement with our previous findings in which the more basic azasugars performed better in the thermal stability assays and acted as more potent inhibitors.

Figure 3.

Stabilization of GALC by GNS (5) at different pH values. (A) Thermal melt curves of GALC alone (black) and with 0.5 mM 5 (blue). (B) Values of ΔT_m. Experiments were performed in triplicate.

In both crystal structures, additional electron density was seen outside the active site, near tryptophan W597 (Figure 4A). Tryptophan side chains are generally buried within hydrophobic regions of protein structures, but in the case of GALC, W597 is exposed on the surface of the enzyme. The density near the W597 side chain may represent an additional allosteric binding site for 5. Although some features of this density appeared to be similar to 5 (Figure 4B), attempts to refine 5 into this density in either the monocyclic or bicyclic form were not convincing following refinement because of substantial negative electron density. While it is possible that this represents a lower-affinity and therefore only partially occupied binding site, it remains unclear whether this site represents a true allosteric binding site for GALC. However, the presence of density at this site opens up the possibility of designing alternative small molecules that may stabilize the enzyme without blocking the active site, thereby allowing for more efficient pharmacological chaperones.

Figure 4.

Additional difference density at a potential allosteric site. (A) Overall structure of GALC illustrating localization of the active site relative to the additional difference density (F_o – F_c) near W597 (inset). (B) Attempted modeling of GNS (5) into this density. Two orientations of the best fit are illustrated.

In conclusion, in our pursuit of a pharmacological chaperone therapy for Krabbe disease, we have found galacto-noeurostegine (GNS, 5) to be a surprisingly potent competitive inhibitor of GALC (7 μM), some 300-fold more potent than its 2-deoxy analogue. In line with our previous study with the slightly more potent yet less selective GALC inhibitor iso-galacto-fagomine (7), we investigated the interaction between 5 and the implicated GALC enzyme via X-ray crystallography. Our analysis revealed that 5 binds exclusively in the bicyclic nortropane form with the 2-hydroxyl group situated in a hydrophilic pocket of the enzyme and participating in multiple hydrogen bonds. In addition, the crystal structures also exposed a potential allosteric binding site, opening up the possibility of designing pharmacological chaperones targeting this site. The exact nature of this site is still unclear, and additional investigation will be necessary.

Glossary

Abbreviations

GALC

galactocerebrosidase

GNS

galacto-noeurostegine

DGN

2-deoxy-galacto-noeurostegine

T_m

melting temperature

N

asparaginyl

E

glutamyl

T

threoninyl

W

tryptophanyl

R

arginyl

S

serinyl

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.0c00377.

• Detailed experimental description of enzyme inhibition, protein expression/purification/crystallization, X-ray data collection, and differential scanning fluorimetry (PDF)

Author Present Address

^§ S.S.: Oxeltis, Cap Gamma, 1682 rue de la Valsière, CS 27384, 34189 Montpellier, France.

Author Present Address

^∥ J.M.: Department of Biochemistry, University of Cambridge, 80 Tennis Court Road, Cambridge CB2 1GA, U.K.

Author Contributions

The manuscript was written through contributions of all authors. All of the authors approved the final version of the manuscript.

The authors declare no competing financial interest.