PREFERRED SITES OF GLYCOSYLPHOSPHATIDYLINOSITOL MODIFICATION IN FOLATE RECEPTORS AND CONSTRAINTS IN THE PRIMARY STRUCTURE OF THE HYDROPHOBIC PORTION OF THE SIGNAL

The divergent carboxyl-terminal signal peptides for glycosylphosphatid ylinositol (GPI) membrane anchor attachment in folate receptor (FR) ty pes alpha and beta were characterized. All of the candidate amino acid residues for GPI modification were identified and tested by substitut ing individually and in combination with amino acids that cannot be mo dified by GPI. Thus the GPI modification in FR-alpha was decreased to 22% by mutation of Ser234 to Thr but unaltered by changing the other c andidate, Gly235, to Met. However, the double mutant FR-alpha(Ser233-T hr,Gly235-Met) showed half of the GPI modification seen in FR-alpha(Se r234-Thr). This result suggests that Ser234 is the preferred GPI modif ication site, while Gly235 is a minor, alternate GPI modification site . Similarly, in FR-beta, mutation of Asn230 to Gin decreased GPI modif ication to 32%, while mutation of the other candidate site, Gly237, to Met had no effect. However, mutation at both sites further reduced th e GPI modification by a half. A five amino acid carboxyl-terminal dele tion (FR-beta Delta 5) caused no decrease in the extent of GPI modific ation. However, the same deletion in FR beta(Asn230-Gln) decreased the residual GPI modification by 66%. These results suggest that Asn230 i s the preferred GPI modification site in FR-beta, while Gly235 offers a minor alternate modification site; consistent with this conclusion i s the fact that modification at the downstream site is hindered by its proximity to the carboxyl terminus in FR-beta Delta 5. Further, the s uggestion that the hydrophobic portion of the GPI signal is a random s equence of neutral amino acids with overall moderate hydrophobicity wa s tested. Mutation of Ala246, which is near the carboxyl terminus of F R-beta Delta 5, to Asp or Pro reduced GPI modification of FR-beta Delt a 5 to 0% and 12%, respectively. This is the first demonstration of th e influence of a neutral amino acid (proline) substitution in the hydr ophobic region of the GPI signal even though the occurrence of proline is not uncommon in GPI signal peptides. The results suggest that (i) although candidate sites for GPI modification may occur adjacently or in close proximity on the linear sequence of the GPI signal, a single residue, Ser234 in FR-alpha and Asn230 in FR-beta, is the preferred si te of proteolysis/GPI attachment; however, (ii) the stringency of sele ction of the GPI modification site is limited because alternate sites may also be modified albeit to a limited extent; and (iii) the efficie ncy of recognition of the hydrophobic portion of the GPI signal may be constrained to a limited extent by its amino acid sequence. The uniqu eness and efficiency of a GPI signal peptide may thus be determined by a combination of limited constraints involving various parts of the p eptide.