Minimizing Mitogenic Potency of Insulin Analogues Through Modification of a Disulfide Bond

Abstract

The mechanisms by which insulin activates the insulin receptor to promote metabolic processes and cellular growth are still not clear. Significant advances have been gained from recent structural studies in understanding how insulin binds to its receptor. However, the way in which specific interactions lead to either metabolic or mitogenic signalling remains unknown. Currently there are only a few examples of insulin receptor agonists that have biased signalling properties. Here we use novel insulin analogues that differ only in the chemical composition at the A6-A11 bond, as it has been changed to a rigid, non-reducible C=C linkage (dicarba bond), to reveal mechanisms underlying signaling bias. We show that introduction of an A6-A11 cis-dicarba bond into either native insulin or the basal/long acting insulin glargine results in biased signalling analogues with low mitogenic potency. This can be attributed to reduced insulin receptor activation that prevents effective receptor internalization and mitogenic signalling. Insight gained into the receptor interactions affected by insertion of an A6-A11 cis-dicarba bond will ultimately assist in the development of new insulin analogues for the treatment of diabetes that confer low mitogenic activity and therefore pose minimal risk of promoting cancer with long term use.

Keywords: biased signalling agonists; cell signalling; dicarba insulin; extracellular-signal-regulated kinase (ERK); glucose metabolism; insulin; insulin receptor; mitogenic.

Figure 1

Insulin, insulin glargine and dicarba insulin analogues. (A) Primary sequence comparison of human insulin (top) and insulin glargine (bottom). Both consist of A (blue) and B (grey) chains stabilized by three disulfide bridges (yellow). Underlined are site 1-binding residues and bold are site 2-binding residues (8). Long-acting insulin glargine has a substitution of AsnA21 ➛ GlyA21 residue and an addition of ArgB31 ArgB32 residues on the C-terminal end of B chain (highlighted in green). (B) Ribbon diagram of human insulin (2-Zn-coordinated T₆ conformation PDB entry 1MSO) showing the location of the three α-helices (A chain: blue; B chain: grey) and the three disulfide bonds (yellow). (C) Schematic diagram of native cysteine and isomeric cis- and trans-dicarba bridges. RP-HPLC chromatograms of (D) insulin, cis-dicarba insulin and trans-dicarba insulin (9); (E) insulin glargine, cis-dicarba dicarba glargine and trans-dicarba glargine. Analysis of the final synthesis products by mass spectrometry confirmed the correct masses of dicarba insulin (5766 Da) and dicarba insulin glargine (6021 Da).

Figure 2

Receptor binding and activation of insulin, glargine and dicarba glargine analogues. (A) Competition binding of insulin, glargine and dicarba glargine analogues with europium-labelled insulin for the IR-B and (B) with europium-labelled IGF-I for the IGF-1R. Results are expressed as a percentage of binding in the absence of competing ligand (%B/B₀). (C) Activation of IR-B and (D) IGF-1R by increasing concentrations of each insulin analogue (10 min stimulation) is expressed as a percentage of the maximal receptor phosphorylation induced by insulin. All data are the mean ± S.E.M. n = at least 3 independent experiments. Error bars are shown when greater than the size of the symbols. Competition binding of insulin, glargine and dicarba glargine analogues with europium-labelled insulin for the IR-A were also performed (see Figure S2 ).

Figure 3

In vitro metabolic and mitogenic studies of insulin, glargine and cis-dicarba analogues. (A) Glucose uptake stimulated by increasing concentrations of insulin, cis-dicarba insulin or cis-dicarba glargine is expressed as fold 2-deoxyglucose (2-DG) uptake (pmol/min/mg) above basal. Insulin vs cis-dicarba insulin vs cis-dicarba glargine (ns) (paired T-test). (B) DNA synthesis in response to increasing concentrations of stimulating insulin analogue is shown as percentage incorporation of 5-Ethynyl-2’-uridine (EdU) above basal. All data in (A) and (B) are the mean ± S.E.M. n = at least 3 independent experiments. (C) Insulin tolerance tests were conducted in mice fed on a normal diet (chow). Insulin, glargine, cis-dicarba insulin or cis-dicarba glargine (0.75 IU/kg) were administered through intraperitoneal injection (ip) under non-fasting conditions and tail vein blood glucose was measured via glucose meter at indicated times. n = 5-6 per group. Blood glucose levels are expressed as change over basal levels (mmol/L). AUC, area under the curves. Chow diet: insulin vs cis-dicarba insulin^ns; glargine vs cis-dicarba glargine^ns (paired T-test). Error bars for all graphs are shown when greater than the size of the symbols.

Figure 4

Cis-dicarba insulin and cis-dicarba glargine insulin exhibit dose-dependent signaling bias via IR-A overexpressing cells. Serum starved IR-A overexpressing L6 myoblasts (hIR-A L6) were stimulated for 10 min with increasing concentrations (0, 0.1, 0.5 1, 10 and 100 nM) of (A–B) insulin or cis-dicarba insulin, (C, D) insulin glargine or cis-dicarba glargine. Pink arrow aligns to 75 kDa marker (representative blots of n = at least 3 independent experiments). (E-L) Quantitation of western blots. Phosphorylation levels are expressed as percentage of level detected when cells were stimulated with 100 nM of insulin for 10 min; normalized to β-tubulin: (E, F) Akt (phospho-Ser473), (G, H) Akt (phospho-Thr308), (I, J) ERK 1 (phospho-Thr202/Tyr204) and (K, L) ERK 2 (phospho-Thr185/Tyr187). Quantitation of other phosphorylated proteins are included in Figure S3 . All data are the mean ± S.E.M. Statistical significance of the overall difference in phosphorylation levels stimulated by cis-dicarba insulin compared to native insulin or cis-dicarba glargine compared to glargine insulin were determined via 2-way ANOVA (bars above graph); difference comparing each stimulating concentration (pink asterisks) were further analyzed using Holm-Sidak test. ns: non-significant; * (P ≤ 0.05), ** (P ≤ 0.01); *** (P ≤ 0.001); **** (P ≤ 0.0001). Dose-dependent signalling analyses were also performed in IR-B overexpressing cells (see Figures S4–S5 ).

Figure 5

Cis-dicarba insulin and cis-dicarba glargine insulin exhibit time-dependent signaling bias via IR-A overexpressing cells. Serum starved IR-A overexpressing L6 myoblasts (hIR-A L6) were stimulated with 10 nM of (A, B) insulin or cis-dicarba insulin or (C, D) glargine or cis-dicarba glargine in a time-course of t = 0, 0.33 (20 s), 0.5 (30 s), 1, 3, 5, 8, 10, 20 and 30 min. The full time-course of insulin and glargine can be seen in Figure S6 . (E-H): Quantitation of western blots (n = at least 3 independent experiments). Phosphorylation levels are expressed as percentage of level detected when cells were stimulated with 10 nM of insulin for 30 min; normalized to β-tubulin: (E) Akt (phosho-Ser473), (F) Akt (phosho-Thr308), (G) ERK 1 (phospho-Thr202/Tyr204) and (H) ERK2 (phospho-Thr185/Tyr187). All data are the mean ± S.E.M. Error bars are shown when greater than the size of the symbols. AUC, area under the curves derived from % of phosphorylation over 30 minutes. Statistical significance of the difference in total phosphorylation over 30 minutes measured as AUC when stimulated by cis-dicarba insulin compared to insulin or cis-dicarba glargine compared to glargine insulin were determined via ordinary one-way ANOVA. ns: non-significant; * (P ≤ 0.05), ** (P ≤ 0.01); *** (P ≤ 0.001); **** (P ≤ 0.0001). Quantitation of other phosphorylated proteins are included in Figure S7 .

Figure 6

Cis-dicarba analogues are incapable of promoting IR-A internalization due to inability to promote phosphorylation of IR-A, IRS-1 and Shc. (A) Serum starved IR-A overexpressing L6 myoblasts were treated with serum-free media (SFM; non-stimulated condition, grey) or 10 nM of insulin analogues in a time-course of t = 0, 5, 10, 20, 30, 60 and 120 min. Data are presented as % of surface receptor/total receptor followed by normalisation with % of surface receptor in SFM at t = 0; i.e. SFM at t = 0 is equivalent to 100%. (B, C) Serum starved IR-A overexpressing L6 myoblasts (hIR-A L6) were stimulated with 10 nM of insulin or cis-dicarba insulin in a time-course of t = 0, 0.5 (30 s), 1, 3, 5, 8 and 10 min. Phosphorylation levels are expressed as percentage of level detected when cells were stimulated with 10 nM of insulin for 10 min: (B) IR (phospho-Tyr1150/Tyr1151); normalized to β-tubulin, IRβ for loading control C) immunoprecipitated (IP) IRS-1 (pTyr20: total tyrosine phosphorylation); normalized to total IRS-1 and (D) p52 Shc (phospho-Tyr239/Tyr240); normalized to β-tubulin. All data are the mean ± S.E.M. n = at least 3 independent experiments. Error bars are shown when greater than the size of the symbols. Statistical significance were determined via one-way repeated measures ANOVA followed by Dunnets multiple test. For receptor internalization (A), statistical analyses were performed comparing the effect of insulin analogues and SFM. For (B-D), statistical analyses were performed comparing insulin and cis-dicarba insulin. ns: non-significant; **** (P ≤ 0.0001).

Figure 7

Akt and ERK requires different IR activation thresholds to achieve full activation that influence the fate of biological outputs. Schematic of the Akt-PI3K and ERK-MAPK signalling pathways following ligand binding. Increased/decreased green bars in battery symbol represent an increase or decrease in levels of phosphorylation. (A) Insulin binding to the IR with full activation of pathways leading to metabolism, growth and survival. (B) Dicarba insulin binding to the IR with decreased activation of the IR, IRS-1, Shc and ERK1/2 resulting in impaired feedback to IRS-1 and decreased IR internalization (decreased growth and survival). Akt signalling continues equivalent to the response to insulin (C) Schematic indicating the threshold levels of IR activation that are required for full ERK and/or full Akt activation (mitogenically or metabolically biased).