Carbohydrate-targeted optimization of therapeutic gonadotropins - Carbohydrate Chemistry: Chemical and Biological Approaches - page 478

are not able to sialylate in a a2,6-linkage because they lack the needed
ST6Gal enzyme and thus exclusively express glycoproteins with a2,3-
linked sialic acids. 162 Considering that 70% of the currently marketed
proteins are glycosylated (among them 25% are therapeutic antibodies),
the availability of drug approved cells expressing ecient sialylation
activity has long been known as a major goal to achieve. More especially,
a2,6 sialic cid is often regarded as the essential human-like glycosylation
because it is the preferred outer glycosylation of circulatory human
glycoproteins. Most mammals instead use terminal a-linked galactose
and accessorily a2,3-linked sialic acid. It should be noted, that sialylation
is of special interest in antibody based products since addition of
a2,6-sialic acid to Fc glycans increases anti-inflammatory activity of
antibodies. Indeed, sialylated Fc specifically bind the C-type lectin
receptor DC-SIGN on dendritic cells 179 and the Siglec family receptor
CD22 on B cells. 180
However, ecient engineering of sialylation in cells producing a
protein of therapeutic use is not a straightforward process. Several
groups have successfully equipped CHO cells with the ST6Gal enzyme
but in all cases, the content in sialic acid did not rise tremendously and
the ratio of a2,6-to a2,3-linked sialic acid never exceeded 50% 181-184
suggesting that the recombinant ST6 enzymes poorly competes with the
endogenous ST3Gal. Indeed, earlier work from our group has shown
that sialic acid transfer may also be limited because of branch
specificity of the ST6Gal towards its glycan acceptor. 177 As a result,
recombinant proteins often display a monosialylated glycosylation
detrimental to their lifespan as it leaves several terminal galactose
residues accessible for clearance through the liver asialoglycoprotein
receptor. It therefore appeared that STs require optimization of their
catalytic properties together with an appropriate targeting to the Golgi
secretory compartments of cultured cells to display enhanced activity.
Our company therefore engineered a panel of ST6Gal variants to cor-
rectly address them to the Golgi compartments as well as to abolish
branch specificity and thereby obtain enzymes with high in vivo catalytic
activity. 178 To that purpose, a panel of optimized ST minigenes has been
designed, combining optimized catalytic domain with variable N-
terminal regions to provide non-natural membrane anchors. 185-186
Since the cytoplasmic tail and transmembrane domain proved to govern
Golgi localization 176 we carefully screened our ST variants for their
ability to locate in the Golgi compartment, in which they can meet
protein acceptors tracking to the cell surface and compete with the
endogenous ST3Gal enzyme. 187 Such optimized enzymes can be further
adapted to any protein drug family to achieve optimal sialic acid
transfer under the best cell culture conditions. 162
Figure 8 shows that such chimeric enzymes are all fully active
when expressed in CHO cells. All variants are able to be active on
cell glycoproteins without disturbing cell growth and adhesion.
When stably transfected, they were shown to be highly active on cell
surface proteins, indicating that they may facilitate protein secretion
as well.