Targeting the liver via the asialoglycoprotein-receptor : synthesis of directed small molecule libraries for the H1-CRD

The ASGP-R, exclusively located on the cell surfaces of hepatocytes, is well known
for removing desialylated glycoproteins with terminal galactose or N -
acetylgalactosamine residues from circulation. However, small high affinity ligands
with more drug-like properties than the natural glycoproteins, that can be used for
specific targeting of the liver, were not yet investigated for the human H1-CRD.
Various small-directed libraries of galactose derivatives and mimics thereof were
synthesized in this work. The design of the ligands was based on the crystal structure
of the H1-CRD[10], and on data from in silico studies[40] as well as from previous SAR
studies for similar receptors. The influence of different types of modifications at the
galacto moiety on the binding affinity, were analyzed by a competitive target-based
assay and Biacore experiments, both performed with immobilized H1-CRD.
Monosaccharide mimics with slightly improved binding affinities for the H1-CRD
relative to GalNAc were found. This work provides experimental data for simple
synthetic ligands of the H1-CRD and therefore gives guidelines for the rational design
of a novel generation of improved ligands.
1.13 Galactose mimics modified at their 6-position.
Galactose mimics based on the 1,5-anhydro-2-deoxy-D-lyxo-hexitol core were
modified at their 6-position by conjugated additions, nucleophilic substitutions and
1,3-dipolar cycloadditions. Compound 3 was used as scaffold. By
substituting its 6-position it was attempted to establish hydrophobic interactions with
the Trp 243. Due to the reduced number of sites with similar reactivity, short synthetic
pathways provided fast access to these ligands.
All the derivatives competed approximately two to three-fold better for H1-CRD than
Gal. Binding affinities comparable to the one of methyl β-Dgalactopyranoside
were obtained. Since for derivatives modified with 1,4-
disubstituted-[1,2,3]-triazoles (45, 50), the behavior was independent of the 4-
substituent at the triazole, the interaction with Trp 243 is possibly restricted to the
triazole ring. The 6-position is directed towards the surrounding water and bulky
substituents could be accommodated without interfering with the binding.
Compound 39, obtained as secondary product from Michael additions and
nucleophilic substitutions, showed a comparable binding affinity as GalNAc (figure
22) in the competitive target-based assay. However, Biacore monitoring of the direct
interaction between 39 and the immobilized H1-CRD did not show a comparable
increase of the binding affinity. Since hexitol 39 has two pairs of free 3,4-hydroxyls in
equatorial and axial arrangement, the difference may come from a local
concentration effect which could convert two weak binders into one highly improved
binder.
In general, galactose mimics modified at the 6-position by hydrophobic groups
showed only a modest improvement of the binding affinity for H1-CRD relative to Gal.
However, bulky groups in the 6-position did not interfere with the binding indicating
that this position can be used to link the targeting device with a possible drug.
1.14 Galactose derivatives and mimics thereof modified only at their 2-
position.
To investigate the role that modifications at the 2-position of the galactose moiety
may play, two families of derivatives were synthesized.
Compounds 66 to 75 were obtained by 1,3-dipolar cycloadditions performed on an
azido group at the C-2’ linked to the 2-OH of the sugar scaffold. They do not show
improved binding affinity in the competitive assay and the Biacore experiments.
Derivative 72 was the only member of this group to show a binding affinity (IC50 (μM)
= 110 ± 13) comparable to the one of GalNAc. However, a different binding mode
was found by biosensor studies of the direct interaction between this compound and
the immobilized monomer of H1-CRD.
2-N-acyl derivatives of a methyl 6-azido-β-D-galactopyranoside were synthesized on
solid phase by amino acid coupling at the N-2-position of the sugar moiety.
Compounds 98-100 showed improved binding affinities for the H1-CRD relative to
methyl β-D-galactopyranoside. Whereas derivatives acylated with glycine
(98) or valine (100) experienced only a modest improvement of their binding
affinities, the 2-(2-amino propionamido) derivative 99 achieved a binding affinity
approximately 1.5 to 2-fold better than GalNAc and 12-fold better than methyl β-Dgalactopyranoside.
Even though the galactose mimic 46 containing an azido
functionality at the 6-position, competed 2 to 3 times better for binding to the H1-CRD
relative to 3, the 6-azido group present also in derivatives 98-100 does not contribute
to the binding.
N-acyl coupled chains appear to improve the binding affinity for the receptor
depending on the size of the acyl group. The propionamido derivative displayed the
lowest IC50 value and therefore the optimal chain length studied. It is possible that
such N-acyl derivatives allow establishing an optimal interaction with His 256,
through H-bond formation with the hydrogen of the nitrogen of the N-acyl group. This
interaction is known for being responsible for the preferential binding of GalNAc with
respect to Gal. 2-N-acyl derivatives are an excellent starting point for the
development of novel high affinity ligands for the H1-CRD.
1.15 Monosaccharides mimics modified simultaneously at their 2- and 6-
positions.
Methyl β-D-galactosamine-derivatives modified at the 2- and 6-positions were
synthesized to elucidate the hypothesis of an added effect on the binding affinity. A
collection of ten di-substituted galactose mimics was synthesized on solid phase with
excellent yields. Amino acids were first coupled with the N-2-position of a
galactosamine mimic and modified Huisgen 1,3-dipolar cycloadditions were
performed at the 6-position of the monosaccharide mimic.
Results from the competitive target-based assay indicated that all the derivatives
synthesized were better binders for the H1-CRD than methyl β-D-galactopyranoside.
However, all di-substituted compounds showed very similar IC50 values to their
parent derivatives N-acyl monosubstituted (98-100). Improved binding due to the
additional 6-substitution could not be observed. The better-defined orientation of the
monosaccharide in the binding pocket, after establishing additional interactions
through its 2-N-substituent, could be the reason for the loss of a complementary 6-
interaction. In addition, this hypothesis may also explain the behavior observed in the
preliminary IC50 determination for compound 69 relative to 68 (figure 24). In the
presence of the substitution in 2-position, compound 68 did not showed an improved
binding affinity compared to Gal. Lacking good interactions with the protein in the
region of the 2-position of Gal, the substituent at the 6-position in 69 could improve
binding.
Since none of the substituents employed at the 6-position interfered with the binding, a collection of eight derivatives obtained by amino acid coupling with alanine (99 and 104-110), with similar or slightly improved binding affinities than GalNAc was
obtained for the H1-CRD.
Outlook.
The results obtained in this thesis may be applied to produce a new generation of
ligands for the H1-CRD.
First, the synthesis of three 2-substituted galactose derivatives (figure 26) may be
envisaged.
The influence of the one-carbon extension of the N-acyl chain of 99 on the binding
affinity could be studied with compound 112. Between the propionamido derivative
99 and the valine-derivatized compound 100 exists a space to be analyzed for
determining the real optimal length preferred for the N-acyl chain. Moreover, the
absence or replacement of the α-amino group on the N-acyl chains (113) would allow
further exploration of the binding site and the type of established interactions.
Substituents as H, CH3 and OH are suggested in order to determine whether
hydrophobic interactions or H-bond formation are preferred. Additionally, compound
114 will help to establish if the N-acyl derivatives have a preferred binding relative to
their O-counterparts due to specific interactions with His 256.
Second, compound 99 could be employed to synthesize a triantennary ligand[36] for
the H1-CRD. Determination of the IC50 value of the novel multivalent ligand in a
competitive assay and comparison with a triantennary galactose ligand would be of
high interest for the development of liver targeting strategies. Additionally, a
comparative study of its cellular uptake by the hepatocytes would also be required.