Insulin-like growth factor binding protein-3 (IGFBP-3) is unique among the IGF binding proteins in its extensive glycosylation in the native state. To determine the functional significance of carbohydrate moieties on IGFBP-3, we examined the effects of nonglycosylated Escherichia coli-derived recombinant human IGFBP-3 (hIGFBP-3E. coli) and glycosylated Chinese hamster ovary cell-derived hIGFBP-3 (hIGFBP-3CHO) on IGF-I action in cultured bovine fibroblasts. Both hIGFBP-3 preparations bound IGF-I with high affinity and were approximately 5-fold more potent than unlabeled IGF-I in inhibiting [125I]IGF-I binding to bovine fibroblasts. Coincubation of IGF-I and hIGFBP-3E. coli or hIGFBP-3CHO produced a dose-dependent inhibition of IGF-I but not insulin-stimulated [3H]aminoisobutyric acid (AIB) uptake. In contrast, preincubation of bovine fibroblasts with hIGAFBP-3E. coli or hIGFBP-3CHO potentiated subsequent IGF-I-stimulated [3H]AIB uptake. When cells were preincubated with 50 nM hIGFBP-3E. coli for 24 h, [125I]IGF-I binding to bovine fibroblasts increased 2.4-fold, whereas responsiveness to IGF-I was increased only 25%. After a 72-h preincubation, IGF-I cell binding remained increased 2-fold with commensurate enhancement of IGF-I-stimulated [3H]AIB uptake. The increase in [125I]IGF-I binding to bovine fibroblast monolayers was primarily due to association of hIGFBP-3E. coli with the cell surface; there was no significant change in IGF-I receptor number or affinity under these conditions. Affinity cross-linking experiments indicated that intense binding of [125I]IGF-I to cell-associated 29,000 Mr hIGFBP-3E. coli seen after 24 h of incubation was reduced approximately 70% after 72 h, concomitant with the appearance of smaller bands indicating hIGFBP-3E. coli forms of 12,000-27,000 Mr. Cell-associated IGFBP-3E. coli (72 h preincubation conditions) had a 10-fold lower affinity for IGF-I compared to hIGFBP-3E. coli in solution and a 2-fold lower affinity compared to the IGF-I receptor. These data demonstrate that glycosylation is not obligatory for biologically functional IGFBP-3. Furthermore, they suggest that processing of cell-associated IGFBP-3 to forms with altered affinity for IGF-I peptide may underly the potentiating effect of IGFBP-3 on IGF-I action.