Insulin analogues: impact of cell model characteristics on results and conclusions regarding mitogenic properties

In a publication in the June issue of the journal Diabetes, [Kurtzhals et al. (2000)] reported on experiments with different insulin analogues. The authors compared the four insulin analogues lispro, aspartat, detemir and glargine using a series of in-vitro parameters which they believed to be related to efficacy and safety. One of their findings was an increased IGF-1 receptor affinity and mitogenic potency in a human osteosarcoma cell line of glargine compared to human insulin. Although the authors inferred that these findings could have implications for the in-vivo safety of glargine, this point was not directly assessed in their study, and they concluded that the safety implications of their findings are unclear.

The study of [Kurtzhals et al. (2000)] confirmed earlier work by other groups showing that the insulin analogue glargine has an increased affinity to the IGF-1 receptor. However, in contrast to previous studies, Kurtzhals et al. reported an increased mitogenic potency of glargine. This difference can be ascribed to the different characteristics of the cell models used in the published studies, which will be discussed below in more detail.

As the concentrations of insulin or insulin analogues which might cause mitogenesis through the IGF-1 receptor of untransformed cells are unlikely to occur in vivo it is the major concern that an insulin analogue might cause mitogenic effects not through the IGF-1 receptor but through the insulin receptor. That this can indeed occur was shown for the insulin analogue AspB10 which caused tumors in animal studies ([Drejer, 1992]). This insulin analogue showed an altered insulin receptor binding behavior which seems to cause mitogenic signaling through the insulin receptor. It is believed that a prolonged occupancy of the insulin receptor allows the coupling to mitogenic signaling pathways in the cell which does not normally occur with the insulin receptor. Since AspB10 also has an increased IGF-1 receptor affinity, the relative contributions of IGF-1 and insulin receptor mediated mitogenesis to the tumourigenic potential of AspB10 has not been determined.

In our studies with glargine insulin ([Berti et al., 1998]), we used a cell system which is thought to optimally detect the mitogenic potential of glargine through binding to the insulin receptor, i.e., transfected rat-1 fibroblasts (rat-1 HIR) which express a very high number of insulin receptors (about 1,250,000 per cell) and few IGF-1 receptors. Glargine did not show increased thymidine incorporation in rat-1 HIR cells. As the cell line predominantly expressed the insulin receptor, it can indeed be assumed that predominantly mitogenic effects of the insulin analogues via the insulin receptor itself have been studied. While the effect of AspB10 was confirmed, we found no increased mitogenic effect of glargine in these cells. In another study, [Bahr et al. (1997)], using a cell type with approximately 7,000 IGF-1 and few insulin receptors per cell as well, did not observe increased mitogenic effects of glargine.

In the [Kurtzhals et al. study (2000)], the human osteosarcoma cell line Saos-B10 has been used; this cell line has more than 30,000 IGF-1 receptors while only a very small number of insulin receptors (< 1,000) is expressed. In this cell line, the mitogenic potency of glargine was approximately 7.8-fold higher than human insulin. In comparison, the mitogenic potency of insulin AspB10 was almost 10-fold greater than that of human insulin in these cells. Several aspects of the data in the summarizing Table 1 of the Kurtzhals et al. paper deserve further comment. The data are compiled from several different experimental systems: when investigating the HIR-off rates, data for insulin-LisPro (off-rate 100% compared to regular insulin) are taken from [Slieker et al. (1997)] (HepG2 cells), while data for AspB10 (off-rate 14%) are from [Hansen et al. (1996)] and were generated in transfected CHO cells. Interestingly, [Slieker et al. (1997)] determined off-rates for AspB10 in HepG2 cells at 52%. This threefold difference between CHO and HepG2 cells indicates that the data partly rely on the experimental system used. Similarly, when determining mitogenic potency, data for B31B32diArg are taken from [Slieker et al. (1997)] and show a 21.8-fold increase compared to regular insulin in human mammary epithelial cells. In the same study, the authors find a 3.83-fold increase in AspB10, while [Kurtzhals et al. (2000)] determined 9.75-fold increase in Saos-B10 cells. The Saos-B10 cell line is quite similar to the normal human mammary epithelial cell line HMEC which has been extensively used in mitogenicity studies on insulin analogues ([Slieker et al., 1997]). HMEC express approximately 44,000 IGF-1 receptors and 2,200 insulin receptors per cell. The Saos-B10 cells and the HMEC have other common features: while it is accepted that regular insulin binds to the IGF-1 receptor with a 1,000-fold lower affinity than IGF-1, the ED₅₀ for mitogenic stimulation, however, differs only 80-fold in Saos-B10 cells and 52-fold in HMEC. This could suggest that the IGF-1 receptor in Saos-B10 cells and HMEC is glycosylated in an irregular way that increases its affinity for insulin. Alternatively, the number of insulin receptors in Saos-B10 cells and HMEC is very low compared to the number of IGF-1 receptor making it likely that all remaining insulin receptors are found as insulin/IGF-1 receptor hybrid proteins, thus excluding any binding to endogenous insulin receptors, as is found in normal tissues. This view is supported by studies with primary fibroblast cultures from insulin receptor deficient mice (e.g., [Lamothe et al., 1998]). Further, the ED₅₀ values for stimulation of mitogenesis determined in Saos-B10 cells for IGF-1 and regular insulin are much lower than those reported by [Slieker et al., 1997] (i.e. 50 pmol/l vs. 340 pmol/l for IGF and 4 nmol/l vs. 17,6 nmol/l for insulin) (1997). These data show that due to different expression levels of receptors, dose response curves are shifted in these cell models. Therefore, any extrapolation from these in vitro systems to the in vivo situation is difficult and, these cell models should only be used in deciding which insulin analogues should be tested in animal models. There is no doubt that mitogenic potentials have to be evaluated in the in vivo situation in the animal model.

Unfortunately, the in vitro data in the paper by [Kurtzhals et al. (2000)] are currently used in discussions with the message that glargine might cause cancer (e.g., arznei-telegramm, issue 12/2000). This message is as inappropriate as it would be to conclude from the Saos-B10 cell data (see dose response curve of insulin, Fig. 2 of [Kurtzhals et al., 2000]) that the insulin levels which occur in the postprandial state cause mitogenic signaling in the normal physiologic situation.

Acknowledgement: We thank Dr. Peter Kurtzhals, Vice President and Co-Chairman of Discovery, Novo Nordisk for critical comments and suggestions to this commentary.

References

• 1 Bahr M, Kolter T, Seipke G, Eckel J. Growth promoting and metabolic activity of the human insulin analogue [GlyA21,ArgB31,ArgB32]insulin (HOE 901) in muscle cells.  Eur J Pharmacol. 320 259-265 1997; 
• 2 Berti L, Kellerer M, Bossenmaier B, Seffer E, Seipke G, Häring H U. The long acting human insulin analog HOE 901: characteristics of insulin signalling in comparison to Asp(B10) and regular insulin.  Horm Metab Res. 30 123-129 1998; 
• 3 Drejer K. The bioactivity of insulin analogues from in vitro receptor binding to in vivo glucose uptake.  Diabetes Metab Rev. 8 259-285 1992; 
• 4 Hansen B F, Danielsen G M, Drejer K, Sorensen A R, Wiberg F C, Klein H H, Lundemose A G. Sustained signalling from the insulin receptor after stimulation with insulin analogues exhibiting increased mitogenic potency.  Biochem J. 315 271-279 1996; 
• 5 Kurtzhals P, Schaffer L, Sorensen A, Kristensen C, Jonassen I, Schmid C, Trub T. Correlations of receptor binding and metabolic and mitogenic potencies of insulin analogs designed for clinical use.  Diabetes. 49 999-1005 2000; 
• 6 Lamothe B, Baudry A, Christoffersen C T, De-Meyts P, Jami J, Bucchini D, Joshi R L. Insulin receptor-deficient cells as a new tool for dissecting complex interplay in insulin and insulin-like growth factors.  FEBS Lett. 426 381-385 1998; 
• 7 Slieker L J, Brooke G S, DiMarchi R D, Flora D B, Green L K, Hoffmann J A, Long H B, Fan L, Shields J E, Sundell K L, Surface P L, Chance R E. Diabetologia. 40 (Suppl 2) S54-S61 1997;