Atsuko Yoneda

Engineering of an FGF–proteoglycan fusion protein with heparin-independent, mitogenic activity

In the absence of heparan sulfate (HS) on the surface of target cells, or free heparin (HP) in the vicinity of their receptors, fibroblast growth factor (FGF) family members cannot exert their biological activity and are easily damaged by proteolysis. This limits the utility of FGFs in a variety of applications including treatment of surgical, burn, and periodontal tissue wounds, gastric ulcers, segmental bony defects, ligament and spinal cord injury. Here we describe an FGF analog engineered to overcome this limitation by fusing FGF-1 with HS proteoglycan (PG) core protein. The fusion protein (PG–FGF-1), which was expressed in Chinese hamster ovary cells and collected from the conditioned medium, possessed both HS and chondroitin sulfate sugar chains. After fractionation, intact PG–FGF-1 proteins with little affinity to immobilized HP and high-level HS modification, but not their heparitinase or heparinase digests, exerted mitogenic activity independent of exogenous HP toward HS-free Ba/F3 transfectants expressing FGF receptor. Although PG–FGF-1 was resistant to tryptic digestion, its physiological degradation with a combination of heparitinase and trypsin augmented its mitogenic activity toward human endothelial cells. The same treatment abolished the activity of simple FGF-1 protein. By constructing a biologically active proteoglycan–FGF-1 fusion protein, we have demonstrated an approach that may prove effective for engineering not only FGF family members, but other HP-binding molecules as well.

Molecular Cloning of Mouse Alpha-1,6-Fucosyltransferase and Expression of Its mRNA in the Developing Cerebrum

Complementary DNA encoding mouse alpha-1,6-fuc-osyltransferase (FUT) was cloned. The deduced primary structure consisted of 575 amino acids and had 96.0%, and 93.0% identity with alpha-1,6-FUT of human and porcine origin, respectively. Quantitative analysis of alpha-1,6-FUT mRNA expression during selected developmental stages of the cerebrum showed that the expression increased during later embryonic stages and was highest in the early postnatal stages (P1 to P7), after which it declined somewhat but still remained relatively high in the mature adult. The expression profile suggests important roles of FUT in the developing central nervous system. Database. Accession No.: AB025198

Characterization of fibroblast growth factor-6 expressed by Chinese hamster ovary cells as a glycosylated mitogen for human vascular endothelial cells.

The gene for fibroblast growth factor (FGF)-6/hst-2 was originally identified by its close homology with the FGF-4/hst-1 gene. Aside from its ability to transform cultured fibroblasts, the characteristics of FGF-6 protein have only been studied using a simple preparation from E. coli. In the present study, we expressed FGF-6 cDNA in CHO cells and characterized the resultant protein. We found that CHO cells secreted several forms of the FGF-6 polypeptide, and that there were multiple N-terminal modifications. The longest form (18-kDa) contained the sequence, SerProAlaGlyAlaArg, as its N-terminus, which was consistent with the signal peptide cleavage site predicted from its primary structure. The core polypeptide was primarily modified by heterogeneous N-glycans that were sialylated to a small degree; among them, biantennary structures were found to predominate. Moreover, possible O-glycosylation was also detected. N-glycosylated FGF-6 potently induced DNA synthesis and proliferation of human vascular endothelial cells, whereas in the absence of N-glycosylation, FGF-6 mitogenicity was substantially diminished. The results clearly indicate that FGF-6 expressed by mammalian cells is a glycosylated mitogen for vascular endothelial cells and further suggests that N-glycosylation plays a key role in determining the mitogenicity of FGF-6.

The AATPAP sequence is a very efficient signal for O-glycosylation in CHO cells.

The peptide signal sequence for protein O-glycosylation is not fully characterized, although a recent in vitro study proposed that the sequence motif, XTPXP, serves as a signal for mucin-type O-glycosylation. Here, we show that the AATPAP sequence acts as an efficient O-glycosylation signal, in vivo. A secreted fibroblast growth factor (secFGF) was used as a model to analyze glycosylation and its effects on the biological activity of FGF. Two constructs encoding [AATPAP]secFGF in which AATPAP was introduced at the N- or C-terminus of secFGF were constructed in a eukaryotic expression vector. [AATPAP]secFGF proteins were then expressed in Chinese hamster ovary (CHO) cells and secreted into the surrounding medium, primarily as modified forms sensitive to sialidase but not to peptide N-glycosidase F. The modifying groups were not seen when the AATPAP sequence was converted to AAAPAP or when [AATPAP]secFGF was expressed in mutant cells incapable of UDP-GalNAc biosynthesis. The results indicate that the modifying groups were mucin-type O-glycans and that the AATPAP served as an efficient O-glycosylation signal sequence. The O-glycosylated forms of [AATPAP]secFGF were as mitogenic toward human vascular endothelial cells as unmodified secFGF, suggesting that introduction of the signal into biologically active polypeptides is a promising approach with which O-glycosylation may be achieved without affecting original activity.

Engineering neoglycoproteins with multiple O-glycans using repetitive pentapeptide glycosylation units

Controlled protein remodeling with O-linked glycans has been limited by our incomplete understanding of the process of glycosylation. Here we describe a secretable fibroblast growth factor (FGF) with multiple mucin-type O-glycans produced by introducing a minimum pentapeptide glycosylation unit in a decarepeat format at its N- or C-terminus. Expressed in Chinese hamster ovary cells, chemical and biochemical analyses of the resultant proteins (Nm10-FGF and Cm10-FGF, respectively) demonstrated that all O-glycosylation units were glycosylated and the dominant structure was sialylated Gal[β1–3]GalNAc. This indicates that minimum O-glycosylation unit in multirepeat format serves as a remarkably efficient acceptor in CHO cells. The Nm10-FGF and Cm10-FGF proteins maintained the mitogenic activity to vascular endothelial cells. In addition, intact Cm10-FGF and its desialylated form interacted with several lectins in the same way as mucin-type glycoproteins. The intact Cm10-FGF with multiple sialylated O-glycans exhibited a longer lifetime in circulating blood, whereas the Cm10-FGF with desialylated O-glycans exhibited a shorter lifetime than the deglycosylated form of Cm10-FGF. Our approach would thus appear to be highly effective for engineering neoglycoproteins, the characteristics of which are determined by their multiple mucin-type O-glycans.

Introduction of an N-Glycosylation Cassette into Proteins at Random Sites: Expression of Neoglycosylated FGF

We developed a method for introducing an N-glycosylation cassette into proteins at random sites by constructing cDNAs and expressing it in mammalian cells. The protocol entails four steps: (i) generation of cDNAs that contain single, randomly-located blunt end cuts; (ii) ligation of N-glycosylation cassettes into the blunt end cuts in three-frame formats; (iii) selection of the cDNA clones encoding N-glycosylated proteins; and (iv) subcloning into an expression vector for transfection and expression in mammalian cells. This method was evaluated using secreted fibroblast growth factor (FGF) as a model protein. Several secreted FGF cDNA clones, each containing an AsnLeuSer-coding sequence at a random site, were obtained. When these clones were expressed in mammalian cells, some of the secreted FGFs were found to be N-glycosylated. The method described here should also be applicable for random introduction of functional oligopeptide/polypeptide cassettes into virtually any protein of interest.