Recently, the human homolog of ER Mannosidase I has been cloned and characterized (45). Similar to the yeast enzyme it cleaves Man9GlcNAc2 to yield Man8GlcNAc2 in the ER of mammalian cells. A recombinant form of the human enzyme has been purified and crystallized and the structures of the enzyme in the presence and absence of small inhibitors that mimic the a1,2-mannose glycone have been determined (6) (PDB IDs 1FMI, 1FO2, and 1FO3) . The image below on the left shows the ribbon diagram of the human ER mannosidase I with coloring to show the adjoining helices in the (aa)7 barrel. The figure on the right shows the end-on view of the barrel showing the position of the bound kifunensine molecule (CPK colors), a mimetic of the mannose glycone, in the core of the barrel structure.
| Figure
1: Structure of human ER mannosidase I (Click on the figure to enlarge) 
|
Figure
2: Position of Kifunensine or 1-deoxymannojirimycin binding (Click on the figure to enlarge) 
|
There are several valuable
insights that have come from a comparison of the structures of the yeast and
human forms of ER mannosidase I. Subtle differences in the structures of the
yeast and human forms were found predominately in the loop regions of the
barrel, but the overall a-helical structures were
similar. One of the major differences between the two structures is the lack
of glycosylation in the human enzyme resulting from sequence differences between
the two enzymes at the three glycosylation sites of the yeast enzyme. Despite
the lack of glycosylation in the human enzyme, the overall folding structure
of the two enzymes are remarkably similar.
| Figure 3:
Comparison of the structure of Yeast (yellow) and Human (blue) ER Mannosidase
I: End view (Click on the figure to enlarge)  |
Figure 4:
Comparison of the structure of Yeast (yellow) and Human (blue) ER Mannosidase
I: Side view (Click on the figure to enlarge) 
|
The structure of the human ER mannosidase I in the presence and absence of KIF or dMNJ also revealed novel aspects of the protein structure and mechanism of action. The structure of the yeast enzyme implied that during the hydrolytic reaction the glycone (the residue being cleaved from the oligosaccharide) extended into the barrel cavity and that acidic residues bordering the cavity were involved in the mechanism of action. There was no direct demonstration of this hypothesis, however, since the oligosaccharide in the active site was the presumed product of the reaction (the aglycone) and was missing the glycone in the active site.
The structure of the human
enzyme in the presence of the inhibitors provided the first direct evidence
in confirmation of this hypothesis, since the inhibitors were found to be
present at the base of the pocket with the equivalent of the C1 position of
the sugar facing outward toward the opening of the pocket within an equivalent
of the bonding distance from the O2' hydroxyl position of the aglycone in
the yeast enzyme structure (see Figure 3 above). Thus, the human enzyme structure
in the presence of the inhibitor reflects the complementary part of the active
site containing the equivalent of the glycone in comparison to the yeast enzyme
structure containing the oligosaccharide aglycone product docked in the active
site.
| Figure 5: Binding of
Kifunensine to the active site of ER mannosidase I (Click on the figure to enlarge) 
|
Figure 6: Zoom into
the binding of Kifunensine to the active site of ER mannosidase I (Click on the figure to enlarge) 
|
The figure to the left above shows the ribbon structure of the enzyme in grey looking end-on into the barrel. The structure of kifunensine in the active site is in CPK colors (you can barely see it) and the atoms on the side chains that extend into the barrel cavity within 4.0 Å of the kifunensine molecule are shown in yellow. The remainder of the side chain is shown in green. The calcium in the active site is shown in red spacefill. Although it is difficult to tell from this figure, the Ca+2 is bound to the base of the crevice in the core of the barrel below the position of the inhibitor. Figure 6 above shows a zoom into the active site equivalent to the picture in Figure 5, except only the interacting atoms are shown. Water molecules are shown in white spacefill and the Ca+2 is shown in magenta spacefill. The structure and the binding of the inhibitor to the active site are better seen in the CHIME ANIMATION OF THE ER MANNOSIDASE I. As shown in the animation the Ca+2 is directly involved in mediating the interaction between the enzyme and the inhibitor (presumably equivalent to the glycone).
Examination of the interaction of the inhibitor in the active site of ER mannosidase I demonstrated the structural basis of the Ca+2 requirement for enzyme activity (see Chime animation above) (6). The Ca+2 ion is bound in the active site through an 8-coordinate pentagonal bi-pyramid geometry. The apices of one of the bi-pyramids are occupied by four water molecules that are in turn hydrogen-bonded to four conserved Glu residues at the base of the cavity (see Figure 7 below). Note the white spacefill structures are water molecules involved directly in the coordination to the Ca+2 (red large spacefill structure). The other bi-pyramid is coordinated to the peptide carbonyl and Og of a Thr residue on the COOH-terminal b-harpin at the base of the cavity. The other two points at the apex of the bi-pyramid are coordinated to the equivalents of the O2' and O3' mannose hydroxyl positions of the inhibitors. The coordination of the Ca+2 ion is indicated by the light dotted lines. Thus, the Ca+2 is directly associated with only one amino acid in the enzyme, and the ion is directly involved in substrate binding. These two characteristics are unique among all known glycosidases.
Although the octahedral
coordination of Ca+2 to ER mannosidase I is unique among glycosidases, it
can also be found in carbohydrate binding proteins (lectins), such as the
mannose binding protein. The mannose binding protein in an homo-oligomeric
protein with carbohydrate binding domains (CRDs) that bind to ligands that
have vicinyl hydroxyls on the O3’ and O4’ positions of mannose,
N-acetylglycosamine, and fucose. Similar to ER mannosidase I, the mannose
binding protein directly coordinates two ligand hydroxyls of the ligand/substrate
to the Ca+2. In contrast the lectin coordinates the other six points of coordination
directly to the protein rather than indirectly through water as is the case
for four points of coordination for ER mannosidase I.
An additional contrast is that the mannose binding protein interacts with
two equitorial hydroxyls (O3' and O4') rather than an axial and equitorial
and equitorial conformation for ER mannosidase I (O2' and O3').
| Figure 7: Coordination
of Ca+2 to the protein and glycone substrate in ER mannosidase I (Click on the figure to enlarge) 
|
Figure 8: Coordination
of Ca+2 and ligand in the mannose binding protein (Click on the figure to enlarge) 
|
The conformation of the
inhibitor within the active site is an additional unique structural characteristic.
Free mannose in solution or as a part of an oligosaccharide is most predominately
(>99%) found in the 4C1 conformation, with the O3',
O4', and O5' hydroxyls in an equatorial low energy conformation. Both KIF
and dMNJ are found in the equivalent of the energetically unfavorable 1C4
"all-axial" conformation in the active site, stabilized through a series
of interactions with the equivalents of the O2', O3', O4', and O6' hydroxyls
in mannose. Stabilization of this energetically unfavorable conformation may
be an important part of the catalytic cycle of the enzyme.
| Figure 9: Conformation
of kifunensine and deoxymannojirimycin in the active site of ER mannosidase
I (Click on the figure to enlarge)  |
Figure 10: Conformation
of kifunensine and deoxymannojirimycin in the active site of ER mannosidase
I (Click on the figure to enlarge) 
|
Part 4: Hypothesized mechanism for ER mannosidase I