We compared mitogenicity and intracellular signalling by human insulin and the AspB10 (X-10) human insulin analogue in MCF-7 human mammary adenocarcinoma cells. By flow analysis of phosphorylated histone H3 or cell cycle distributions, insulin and X-10 were mitogenic at physiologically relevant concentrations (2 nm to 74 pm range), with X-10 being approximately 3-fold more mitogenic than insulin. By western blotting with phospho-specific antibodies, insulin induced phosphorylation of IRS-1, Akt, p70S6K, S6 ribosomal protein, 4E-BP1, FoxO3a, FoxO1, p44/42 MAPK and the EGFR. Blocking with wortmannin, rapamycin and U0126 showed that these signalling events conformed to the canonical PI3K pathway. IRS-1 (Ser302) phosphorylation was abolished by wortmannin and rapamycin, suggesting a feedback from the PI3K pathway on insulin signalling. Compared with equimolar insulin, X-10 caused up to 2-fold higher phosphorylation of all proteins examined in this study. The phosphorylation sites that responded most strongly to insulin were not generally the same as those responding most strongly to X-10. In the PI3K pathway, the most X-10-sensitive protein localized to the translation-regulating arm (p70S6K), with FoxO3a and FoxO1 transcription factors showing a more comparable response to insulin and X-10. Using flow analysis, we confirmed the correlation between insulin-triggered translational activation in G0/G1 (S6 phosphorylation) and S-phase entry by MCF-7 cells. In summary, our findings implicate asymmetrical PI3K pathway activation and specifically stimulation of protein translation in the hypermitogenic effect of insulin analogues such as X-10. It remains to be shown whether these findings are relevant to other human mammary cancer cell types.

Human mammary cell lines are extensively used for preclinical safety assessment of insulin analogs. However, it is essentially unknown how mitogenic responses can be optimized in mammary cell-based systems. We developed an insulin mitogenicity assay in MCF-7 human mammary adenocarcinoma cells, under low serum (0.1% FCS) and phenol red-free conditions, with 3H thymidine incorporation as endpoint. Based on EC50 values determined from 10-fold dilution series, beta-estradiol was the most potent mitogen, followed by human IGF-1, human AspB10 insulin and native human insulin. AspB10 insulin was significantly more mitogenic than native insulin, validating the ability of the assay to identify hypermitogenic human insulin analogs. With MCF-7 cells on a collagen IV surface, the ranking of mitogens was maintained, but fold mitogenic responses and dynamic range and steepness of dose-response curves were increased. Also, PI3K pathway activation by insulin was enhanced on a collagen IV surface. This study provided the first determination and ranking of the mitogenic potencies of standard reference compounds in an optimized MCF-7 assay. The optimized MCF-7 assay described here is of relevance for in vitro toxicological testing and carcinogenicity safety assessment of new insulin compounds.

Insulin-exposed rat mammary cancer cells were flow sorted based on a c-myc reporter plasmid encoding a destabilized green fluorescent protein. Sorted cells exhibited gradual increases in c-myc levels. Cells overexpressing c-myc by only 10% exhibited phenotypic changes attributable to c-myc overexpression, such as cell cycle disturbances, increased cell size, and overexpression of the S6 ribosomal protein. Cells overexpressing c-myc by 70% exhibited additional phenotypic changes typical of c-myc overexpression, such as increased histone H3 phosphorylation, and reduced adherence. Sorted cells also exhibited overexpression of the IGF-1R, and slightly elevated expression of the IR. Increased susceptibility to the mitogenic effect of insulin was seen in a small proportion of the sorted cells, and insulin was more effective in activating the p44/42 MAPK pathway, but not the PI3K pathway, in the sorted cells than in the nonsorted cell population. To our knowledge, this is the first in vitro system allowing functional coupling between mitogenic signaling by a well-defined growth factor and gradual overexpression of the normal, endogenous c-myc gene. Thus, our flow-sorting approach provides an alternative modeling of the receptor-mediated carcinogenic process, compared to the currently used approaches of recombinant constitutive or conditional overexpression of oncogenic transmembrane receptor tyrosine kinases or oncogenic transcription factors.