European Molecular Biology Laboratory Meverhofstrasse 1, Postfach 10.2209 69012 Heidelberg, Germany
Three-dimensional structures of lenograstim have been generated by X-ray
[ I ] and nuclear magnetic resonance (NMR) studies. Lenograstim is a member
of the 4-helix-bundle family of cytokines which also includes human growth
hormone (HuGH), GM-CSF and the interleukins (ILs). Lenograstim comprises
174 amino acids and 4% carbohydrate arranged as four long helices of approximately
equal length (A, B. C and D helices) as well as one short helix. Five cysteine
residues form two interchain disulphide bonds (cysteine 36-42 and cysteine
64-74), while residue 17 is left free. One O-linked carbohydrate chain
is attached to threonine 133. The topology of lenograstim derived from
NMR studies is depicted in Fig. 1. Through NMR analysis it is possible,
not only to determine the structure of the rHuG-CSF molecule, but also
to assess the mobility of different regions of the polypeptide chain as
well as the orientation of the sugar moiety in relation to the surface
of the protein.
Molecular modelling studies have been performed in an attempt to provide
a molecular explanation for the stabilising effects of glycosylation on
the rHuG-CSF molecule. A second important consideration is why glycosylation
does not seem to affect rHuG-CSF binding with the G-CSF receptor.
Hasegawa proposed that the greater stability of glycosylated rHuG-CSF
is due to protection of the only free cysteine residue (position 17) by
the sugar moiety. This hypothesis, however, appears unlikely on the basis
of structural data. Although the cysteine 17 residue is at the beginning
of the A helix, the distance between it and the site of glycosylation (threonine
133) is about 40 A. This distance is too far for a sugar group of about
3 subunits to stretch. Moreover, a sugar group positioned in this way would
pass over the B and C helices and, therefore, be expected to influence
receptor binding.
There is another possible mechanism. The sugar moiety may confer rigidity
to the long flexible C-D loop that contains the threonine 133 residue,
making it less susceptible to unfolding or proteolysis. Indeed, preliminary
data (V. Gervais, unpublished data) suggest that the C-D loop of glycosylated
rG-CSF is not very mobile. These initial studies have produced NMR spectra
that indicate a reduction in the flexibility of the amino acid chain at
threonine 133 and its neighbouring residues compared with non-glycosylated
rG-CSF.
Proteases exhibit a tendency to attack exposed mobile regions of proteins.
It is possible, therefore, that through glycosylation, the rHuG-CSF molecule
is less vulnerable to proteolytic attack.
Experimental evidence suggests that the G-CSF receptor is a
homodimer, although little is known about the mode of G-CSF receptor binding.
Modelling and mutational studies with other 4-helix-bundle cytokines (HuGH,
GM-CSF, IL-2and IL-4) indicate that the C, A and D helices contact the
receptor during binding. However, according to electrostatic analysis of
G-CSF receptor binding, the CAD face is not suited to binding a homodimeric
receptor because it lacks 2-fold symmetry. In contrast, the BCA face is
homogeneous and better suited to binding a homodimer. These results are
supported by mutational data which suggest that the B helix is more important
for biological activity than the D helix. In conclusion, regardless of
which face of the rHuG-CSF molecule is essential for receptor binding,
neither the CAD or BCA faces contain the threonine glycosylation site.
Glycosylation therefore, is not expected to markedly impair G-CSF receptor
binding. However, it is expected that glycosylation has a favourable impact
on the stability of the molecule and its resistance to proteolytic degradation.