The C-Terminal PAC Domain of a Secreted Arabinogalactan-Protein from Carrot Defines a Family of Basic Proline-Rich Proteins

Abstract

Since the first authenticated AGP core polypeptide was cloned by Chen et al (1994), a total of six cDNAs corresponding to authenticated AGPs have been reported (Nothnagel 1997, Gao et al 1999). In view of this recent molecular data, the criteria previously used to define an AGP (a Hyp/Pro, Ala and Ser rich core polypeptide; possession of an arabinogalactan Type II glycan moiety attached to the core protein; and binding to β-glucosyl Yariv reagent) needed to be amended to take account of this ‘new’ information. With this in mind, the protein sequences corresponding to the core polypeptides of the cloned AGPs have been divided into two types, “classical” and “non-classical” (Mau et al 1995, Du et al 1996). According to this convention if the predicted AGP sequence beyond the signal peptide has Pro/Hyp, Ala, and Thr/Ser as the most abundant residues, it is designated “classical”. In general, “classical” AGP sequences encode a polypeptide with at least three distinct domains: an N-terminal signal sequence, a central Pro/Hyp-rich region and a C-terminal hydrophobic region that contains a glycosylphosphatidylinositol (GPI) anchor signal (Youl et al 1998). “Non-classical” AGP genes, however, encode polypeptides which contain an N-terminal secretion signal followed by a variety of domains, none of which seem to encode a C-terminal GPI membrane anchor signal sequence. Furthermore they may contain consensus sites for N-glycosylation, which do not occur in their “classical” counterparts.