The expression and characterization of human recombinant proinsulin-like growth factor II and a mutant that is defective in the O-glycosylation of its E domain.

In humans, newly synthesized proinsulin-like growth factor II (pro-IGF-II), i.e. IGF-II with an E domain extension of 89 amino acids, is 0-glycosylated on Thr75. As an approach to define the role that glycosylation of the E domain serves in the processing, secretion, and biological activities of IGF-II and to identify the sites of endoproteolytic processing, we constructed a mutant that encodes carbohydrate-free prepro-IGF-II. The mutant and wild-type prepro-IGF-II were expressed in NIH-3T3 cells, and the protein products were analyzed by SDS-PAGE followed by immunoblots with antipeptide antibodies to human and homologous rat E domain sequences. Transfectants that express glycosylated pro-IGF-II, i.e. xz97 and G11 cells, have intracellular forms of the growth factor with apparent Mr (appMr) of 21, 23, and 27K. NIH-3T3 xz95 cells, i.e. transfected with DNA that is missing the 0-glycosylation sequence, could also synthesize pro-IGF-II with an appMr of 21K. However, they did not accumulate the 23K and 27K forms of presumably glycosylated growth factor. None of the transfected NIH 3T3 cells processed much pro-IGF-II intracellularly, as the appMr 21K, 23K, and 27K forms had terminal E domain amino acid sequences that were recognized by antibodies to the homologous rat peptide sequence Met117 to Gln156. Subsequent to their secretion, the IGF-II in xz97 and G11 cells accumulated in the conditioned medium mostly as two partially processed species with appMr, of 17K and 14K, respectively. The IGF-II that accumulated in the conditioned medium of the xz95 cells had an appMr of 11K. As evidenced by a decrease in mass after treatment with neuraminidase and 0-glycosidase, the 17-kDa form of pro-IGF-II secreted by the NIH-3T3 xz97 cells was 0-glycosylated, whereas that secreted by the xz95 cells was oligosaccharide free. All of the pro-IGF-II forms have E domain amino acid sequences that reacted with antipeptide Ab to the Asp69 to Lys88 sequence. However, appMr 17K IGF-II, but not 14K IGF-II, also contained a larger E domain that was recognized by Ab to the sequence Phe89 to Arg101. The final step in the processing of 11- to 17-kDa IGF-II at Arg68 and the generation of mature IGF-II did not occur in the NIH-3T3 transfectants and is similar to what has been observed in human embryonic cells and mesenchymal tumors. The failure to remove the glycosylated E domain peptide from appMr, 14K and 17K IGF-II did not affect their binding to IGF-II/cation-independent mannose-6 phosphate receptors or presumably to IGF-I receptors, because in in vitro mitogenic assays they were equipotent with mature IGF-II. Unglycosylated pro-IGF-II from the NIH-3T3 xz95 cells also bound to these receptors. However, it was about 10 times more potent than IGF-II in stimulating thymidine incorporation into NIH-3T3 i24 IGF-IR cells, possibly because of the absence of negatively charged sialic acid and/or steric occlusion.