Insulin Glulisine—A Comprehensive Preclinical Evaluation

Abstract

Receptor binding and signaling and the mitogenic potential of insulin glulisine (glulisine), regular human insulin (RHI), and Asp(B10) were compared in vivo and in vitro. Insulin and insulin-like growth factor 1 (IGF-1) receptor binding was studied with human insulin receptors (293HEK cells) and the human osteosarcoma-derived cell line B10. Insulin receptor–mediated signaling was assessed in rat-1 fibroblasts overexpressing insulin receptors. Activation of insulin receptor substrates 1 and 2 (IRS-1/IRS-2) was studied in rat and human myoblasts and rat cardiomyocytes. DNA synthesis induction was assessed by [³H] thymidine incorporation in the human epithelial breast cell line MCF10. Interaction with the IGF-1 receptor, DNA synthesis, and intracellular signal transduction were assessed in cardiac K6 myoblasts. Immunohistochemical examination of Sprague-Dawley rat tissue treated with glulisine for 6 months (n = 40), and glulisine and RHI for 12 months (n = 60), was performed. Steady-state insulin receptor binding affinity was slightly lower for glulisine versus RHI (~0.70). IGF-1 receptor binding affinity was lower (four-to fivefold) for glulisine, but significantly higher (four-fold) for Asp(B10) versus RHI. Glulisine, Asp(B10), and RHI showed similar insulin receptor–association kinetics; however, Asp(B10) revealed increased insulin receptor affinity. Glulisine and RHI showed similar insulin receptor–mediated phosphorylation and IRS-2 activation. Activation of IRS-1 was 6- to 10-fold lower with glulisine; glulisine was less potent and Asp(B10) slightly more potent in stimulating DNA synthesis versus RHI. Stimulation of DNA synthesis was comparable for glulisine and RHI in K6 myoblasts. At 12 months, there was no significant difference between glulisine and RHI in proliferative activity. This preclinical evaluation suggests that structural changes in glulisine versus RHI are not associated with any safety issues.

The goal of insulin replacement therapy in diabetes is to mimic the complex patterns of endogenous insulin secretion as closely as possible. In intensive insulin therapy, this requires a combination of long-acting (basal) insulin and short-acting (prandial) insulin. In recent times, rapid-acting insulin analogs have taken the place of regular human insulin (RHI) as the prandial insulin of choice for the management of diabetes, as they more closely resemble physiologic mealtime insulin secretion. Insulin glulisine is a new rapid-acting insulin analog, which differs from RHI in that the amino acid asparagine at position B3 is replaced by lysine and the lysine at position B29 is replaced by glutamic acid (Figure 1). Phase I studies support the classification of insulin glulisine as a rapid-acting insulin (Becker et al. 2003, 2004). Furthermore, Phase III studies have demonstrated the suitability of insulin glulisine for prandial insulin replacement (Dailey et al. 2004; Dreyer et al. 2004; Garg et al. 2004a, 2004b; Hanaire-Broutin et al. 2004). Here we present a comprehensive preclinical evaluation of insulin glulisine; these studies were undertaken to confirm that the structural changes in insulin glulisine compared with RHI are not associated with any safety issues.

Insulin and IGF-1 Receptor Binding and Signaling

Insulin acts on two receptor systems: its metabolic effects are mediated by the insulin receptor, and the growth maintaining effects are mediated by the insulin receptor and, to a lesser extent, by the insulin-like growth factor 1 (IGF-1) receptor (Hennige et al. 2005a). Although the underlying pathophysiology of diabetes is diverse, the defects are closely linked at the molecular level through this insulin/IGF-1 signaling pathway. Because the changes in the molecular structure of insulin may alter its interaction with the insulin and IGF-1 receptors, safety is always a potential concern with any insulin analog. Indeed, the European Agency for the Evaluation of Medicinal Products (EMEA) recently produced a report that detailed points to consider when evaluating the carcinogenic potential of insulin analogs (EMEA 2001). This concern has arisen from past experience where amino acid changes, particularly in the B-chain of the molecule, have affected insulin and IGF-1 receptor binding properties. For example, the rapid-acting insulin analog Asp(B10) (Figure 1) has been shown to be mitogenic (Berti et al. 1998; Bolli et al. 1999) and tumorigenic (Jorgensen, Dideriksen, and Drejer 1992). However, there is conflicting evidence as to whether the mitogenic effects observed at high insulin doses are mediated via insulin or IGF-1 receptor signaling (Moses and Tsusaki 1991; Pillemer et al. 1992). In addition, precisely how altered activation of the receptors may lead to tumor promotion is as yet unknown; this could be due to kinetic aspects of insulin receptor interaction or additionally increased affinity for the IGF-1 receptor. It is, therefore, important to investigate extensively all aspects of insulin and IGF-1 receptor binding with any new analog of human insulin. This study investigated the effects of supraphysiologic doses of insulin glulisine versus RHI and Asp(B10) in terms of insulin and IGF-1 binding and signaling in various in vitro systems.

Mitogenic Potential of Insulin Glulisine

The insulin analog Asp(B10) has been shown to have a mitogenic effect; additionally, in a 1-year study in rats, Asp(B10) demonstrated the potential to induce mammary tumors. Both of these effects were initially attributed to the increased IGF-1 receptor binding affinity of this analog with respect to RHI (Jorgensen, Dideriksen, and Drejer 1992). However, the mitogenic effects of Asp(B10) were later correlated with prolonged insulin receptor occupancy, which occurs due to a slow dissociation rate from the receptor (Hansen et al. 1996); this leads to an increased half-life of the ligand-receptor complex, which in turn leads to prolonged activation of the insulin signaling pathway via phosphorylation of intracellular proteins (Hansen et al. 1996; Shymko, De Meyts, and Thomas 1997). In light of this, the mitogenic and proliferative potential of supraphysiologic doses of insulin glulisine was studied in vitro in the nonmalignant human epithelial breast cell line MCF10 and in differentiating muscle tissue, as well as in vivo in mammary glands in rats.

MATERIALS AND METHODS

Insulin Receptor Binding

Human insulin receptors were purified from transient over-expressing 293HEK cells by fractional centrifugation, insulin affinity chromatography, concentration by wheat germ agglutinin chromatography, and sucrose density gradient centrifugation. Displacement of ¹²⁵I-insulin resulting from binding of insulin glulisine or RHI to the human insulin receptor was measured in terms of residual radioactivity in the precipitates collected on membrane filters.

Statistical Analysis

Data were corrected for nonspecific binding determined at a high insulin concentration of 1 μM. Relative amounts of bound tracer were calculated as a percentage of maximal binding and plotted against the molar concentration of competing ligands in a logarithmic scale. Competitive binding curves were evaluated with GraphPad Prism 2.01 (GraphPad, San Diego, USA) using a single-class binding site model. In order to compare insulin glulisine and RHI, IC₅₀ values were taken from the graphs by the computer program at 50% of maximal residual tracer binding.

Insulin Receptor–Mediated Signaling

Rat-1 fibroblasts overexpressing approximately 1,250,000 insulin receptors per cell (based on previous work by Hennige et al. [2005b]) were used to:

Estimate specific binding of insulin glulisine, RHI, and Asp(B10) to the human insulin receptor (using cell-associated radioactivity with a residual bound ¹²⁵I-labeled tracer)

Investigate the kinetics of intracellular signaling (insulin receptor and insulin receptor substrate [IRS] phosphorylation, using immunoblotting of phosphorylated proteins of the signaling cascade)

Statistical Analysis

Statistical analysis was performed using student’s t test.

IGF-1 Receptor Binding

The human osteosarcoma-derived cell line B10, which expresses ≥30 times more IGF-1 than insulin receptors, was used to assess the IGF-1 receptor binding capabilities of insulin glulisine, RHI, and Asp(B10) using a competition assay with ¹²⁵I-labeled IGF-1 as tracer.

Statistical Analysis

To obtain the dose-response curves, the relative amounts of IGF-1 receptor–associated tracer from all three independent experiments were plotted against the concentrations of the competing substances in a logarithmic scale and a fitted sigmoid curve was obtained by a four parameter equation. To compare the relative binding affinities of tested compounds, IC₅₀ values were taken from this graph at 50% of maximal residual tracer binding IC.

Activation of Insulin Receptor Substrates 1 and 2

Activation of the IRS-1 and IRS-2 by insulin glulisine compared with RHI and Asp(B10) was assessed in three different muscle cells:

Rat K6 myoblasts (a Glut4-overexpressing cell line derived from H9C2 cardiac myoblasts)

Human myoblasts obtained from satellite cells isolated from muscle rectus abdominus

Adult rat cardiomyocytes isolated from the hearts of male Wistar rats

Activation of IRS-1 and IRS-2 was measured in terms of tyrosine phosphorylation by an immunoblotting method. Stimulation of 3-O-methylglucose transport of insulin glulisine and Asp(B10) was measured in comparison to RHI at a concentration of 5 × 10^–9 M and 5 × 10^–7 M using the adult cardiomyocyte system.

Statistical Analysis

All data analysis was performed using Prism or t-ease (ISI, Philadelphia, USA) statistical software.

Mitogenic Potential In Vitro

Exponentially growing cardiac K6 myoblasts were used to investigate the effect of insulin glulisine on DNA synthesis compared with RHI and Asp(B10) using the incorporation of bromodeoxyuridine (BrdU) and a highly sensitive chemiluminescence immunoassay. The human epithelial breast cell line MCF10 was also used to investigate the induction of DNA synthesis by insulin glulisine in comparison with RHI and Asp(B10) using thymidine incorporation. After stimulation with insulin for 16 h, [³H]thymidine (0.5 μCi/ml) (Amersham Buchler, Braunschweig, Germany) was added for 4 h. Cell-associated radioactivity was then determined by liquid scintillation counting after cell lysis. RHI and insulin analogues were studied in the concentration range 0.01 to 100 nM.

Exponentially growing cardiac K6 myoblasts were used to investigate the interaction of insulin glulisine with the IGF-1 receptor and its effect on DNA synthesis and intracellular signaling compared with RHI and Asp(B10). Binding studies were carried out by determining the cell-associated radioactivity of ¹²⁵I-labeled insulins using a gamma counter. Nonspecific binding was measured in parallel incubations performed in the presence of an excess of the corresponding unlabeled insulin (10^–5 M). All assays were performed in triplicate. The signaling potency of the different insulin analogues to the mitogen-activated protein (MAP) kinase pathway was determined by assessing the autophosphorylation of the IGF-1 receptor, the phosphorylation of Shc proteins, and the activation of p44/42 (ERK1/ERK2) MAP kinase. The interaction between the IGF-1 receptor and the Shc proteins was investigated by determining tyrosine phosphorylation of these intracellular substrates after stimulation of K6 myoblasts with RHI or the insulin analogues. Cellular proliferation in response to the study treatments was determined by monitoring DNA synthesis using the incorporation of bromodeoxyuridine (BrdU) and a highly sensitive chemiluminescence immunoassay.

Statistical Analysis

Data from the study in human cell line MCF10 are expressed as a percentage of the response seen in the presence of 0.5% fetal calf serum (set to 100%) used as the control in each experiment. Data from the study in cardiac K6 myoblasts was analyzed using Prism (GraphPad) or t-ease (ISI) statistical software. Significance of reported differences was evaluated using the null hypothesis and t statistics for paired data.

Mitogenic Potential In Vivo

In order to complement the in vitro mitogenicity data, a retrospective immunohistochemical examination of all tissues, particularly the mammary glands, from Sprague-Dawley rats treated with insulin glulisine for 6 months (n = 40) or with insulin glulisine and RHI for 12 months (n = 60) was performed in vivo to identify any proliferative activity:

6-Month study: insulin glulisine was injected subcutaneously to 20 male and 20 female rats once daily for 181 days at dose levels of 5, 20, or 80 U/kg body weight over a period of 6 months. A further group of the same composition serving as control received vehicle only. At study start, the animals were approximately 6 to 7 weeks of age and had a mean body weight of 183.7 g (males) and 149.5 g (females). Examined parameters included clinical examinations, body weight, food consumption, urinalysis, hematology, clinical chemistry, organ weights, gross pathology, histopathology, serum levels, antibody formation, and Ki-67 immunocytochemistry.

12-Month study: insulin glulisine or RHI were injected subcutaneously to 30 male and 30 female rats twice daily (~8-h interval; maximum 381 days) at dose levels of 2.5 (insulin glulisine only), 5, 20, or 50 U/kg body weight (resulting in total daily doses of 5 [insulin glulisine only], 10, 40, or 100 U/kg). At study start the animals were approximately 6 to 7 weeks of age and had a mean body weight of 235 g (males) and 170 g (females). Examined parameters included clinical examinations, body weight, food consumption, hematology, clinical chemistry, antibody determinations, organ weights, gross pathology, histopathology, and Ki-67 immunocytochemistry.

Proliferative cells were detected using the mouse anti-rat Ki-67-antibody, clone MIB-5 (DAKO), the biotinylated rabbit anti-mouse antibody (DAKO), and the streptavidin-biotin peroxidase complex (DAKO). Enzyme activity was visualized by a substrate/chromogen mix of H₂O₂/AEC (amino ethyl carbazol). The sections were counterstained with hemalaun.

Statistical Analysis

In the 6-month retrospective study, proliferative activity in mammary gland tissue was analyzed using the Wilcoxon two-sample Z test. In the 12-month retrospective study, Ki-67–labeled cells in mammary glands were detected within a total number of ≥1000 counted glandular cells, and the labeling index was calculated for each individual rat. The proliferation indices in female rats were compared pairwise with each other and evaluated using the Wilcoxon two-sample Z test.

RESULTS

Insulin and IGF-1 Receptor Binding and Signaling

Insulin Glulisine Receptor Binding

In in vitro receptor binding studies with isolated insulin receptors, purified from 293HEK cells transiently overexpressing the insulin receptor, steady-state insulin receptor binding affinity was slightly less for insulin glulisine compared with RHI (IC₅₀ 8.7 to 14.4 versus 6.2 to 9.8 pM [nmol · L^–1]; relative binding affinity for insulin glulisine versus RHI was ~0.70; Table 1). In addition, IGF-1 receptor binding affinity assessed in B10 osteosarcoma cells overexpressing the IGF-1 receptor demonstrated that insulin glulisine bound to the IGF-1 receptor with lower (four- to five-fold) affinity relative to RHI. In contrast, Asp(B10) bound to the IGF-1 receptor with a fourfold greater affinity.

Insulin Glulisine Receptor Activation and Signaling

Insulin glulisine, RHI, and Asp(B10) all showed similar insulin receptor–association kinetics, reaching a maximum after approximately 5 min. Insulin glulisine and RHI were similar with respect to association kinetics, but Asp(B10) revealed a markedly increased binding affinity (Figure 2). Although insulin glulisine and RHI were similar with respect to insulin receptor–mediated activation of phosphorylation, activation with Asp(B10) differed with a prolonged phosphorylation state of the insulin receptor and presence of receptor substrates (Figure 3, Table 1).

RHI exerted a prominent and equal activation of IRS-1 and IRS-2 (assessed by tyrosine phosphorylation) in both rat myoblasts and human skeletal muscle cells, as well as in adult rat cardiomyocytes (Table 1). Insulin glulisine produced only a marginal activation of IRS-1 in all three cell systems, but a prominent activation of IRS-2, which was at least equal to that induced by RHI. The strongly reduced activation of IRS-1 by insulin glulisine was not limited to myoblastic cells with a high number of IGF-1 receptors, but was also observed in adult rat cardiomyocytes expressing a high level of insulin receptors.

Mitogenic Potential of Insulin Glulisine In Vitro

In rat-1 fibroblasts overexpressing approximately 1,250,000 insulin receptors per cell, it was demonstrated that insulin glulisine did not display any enhanced promotion of thymidine incorporation into DNA compared with RHI and this was in contrast to Asp(B10) (data not shown). The human cell line MCF10 was used to investigate the induction of in vitro DNA synthesis by insulin glulisine in comparison to RHI. Results showed that insulin glulisine is less potent in stimulating DNA synthesis than RHI in the nonmalignant human epithelial breast cell line MCF10 at concentrations between 10 and 100 nM (nmol · L^–1) (Figure 4). Exponentially growing cardiac K6 myoblasts were used to investigate the interaction of insulin glulisine with the IGF-1 receptor and the effect of insulin glulisine on DNA synthesis and intracellular signal transduction. Results are summarized in Table 2. Similar to the previously discussed results in isolated insulin receptors, insulin glulisine showed a reduced binding affinity relative to RHI in cardiac K6 myoblasts (data not shown). This did not correlate with IGF-1 receptor autophosphorylation, which was higher than with RHI; however, the activation of Shc proteins was similar and the phosphorylation of MAP kinases (ERK1 and ERK2), which has been related to mitogenic activity, was even lower compared with RHI. The stimulation of DNA synthesis was shown to be comparable for insulin glulisine and RHI, which is consistent with the insulin-like activation of the Shc/MAP kinase pathway by insulin glulisine.

Mitogenic Potential of Insulin Glulisine In Vivo

Results from the 6-month study in Sprague-Dawley rats showed that the no observable adverse effect level (NOAEL) was 5 U/kg body weight of insulin glulisine. The full histopathologic evaluation of all tissues, as well as Ki-67 immunohistochemistry of the mammary glands, did not demonstrate a mitogenic effect of insulin glulisine as compared with the untreated controls. This outcome was confirmed in the 12-month study. In addition, results from the 12-month study showed that insulin glulisine is not different from RHI in terms of both mitogenic activity and histopathologic results. Statistical analysis of the results revealed no significant differences between the insulin glulisine, RHI, and control group with regard to proliferative activity in non-neoplastic mammary gland tissue, as indicated by the pairwise comparison using the Wilcoxon two-sample Z test (Table 3).

In both the 6-month and 12-month studies, the toxicological profile of insulin glulisine was characterized by excessive hypoglycemia after high doses. In the 6-month study, body weight development was generally comparable in all study groups. However, in the 12-month study, mean body weights in 2 × 50 U/kg insulin glulisine males were statistically significantly higher than other groups from study day 50 onwards. Mean body weights in 2 × 5, 2 × 20, and 2 × 50 U/kg insulin glulisine-treated females were dose dependently and statistically significantly elevated from study day 8 onwards; mean body weights in all RHI female groups were dose dependently and statistically significantly elevated from study day 8 onwards. Absolute and relative organ weights did not reveal toxicologically relevant differences in both the 6- and 12-month studies.

DISCUSSION

Insulin analogs have been developed using recombinant DNA techniques in order to provide patients with diabetes the most efficient, reproducible, and convenient therapy possible. Although the alteration of the amino acid sequence of insulin can confer therapeutic advantages, it is also possible that these changes can alter the receptor binding and signaling characteristics of the insulin analog. Rigorous investigation is therefore required to ensure the safety of these compounds. Previous studies have demonstrated that the insulin analogs insulin aspart and insulin lispro have a similar binding affinity for the insulin receptor to RHI. Furthermore, insulin aspart showed an IGF-1 receptor affinity similar or slightly lower than that of RHI, whereas insulin lispro was 1.5-fold more potent in binding to the IGF-1 receptor than RHI, although this did not appear to significantly influence mitogenic potency (Kurtzhals et al. 2000). The data presented here demonstrate that insulin glulisine is able to bind to and effectively activate both the insulin and IGF-1 receptors with similar pharmacodynamic properties to RHI. Therefore, it was concluded that insulin glulisine behaves like RHI in vitro with respect to insulin receptor binding and activation of the initial insulin signaling cascade. In contrast, Asp(B10) exhibits an increased insulin receptor binding and induces a prolonged phosphorylation state of the insulin receptor.

Once activated, insulin and IGF-1 receptors engage and phosphorylate a number of cellular proteins, including the IRS proteins, which contain several potential tyrosine phosphorylation sites in various amino acid sequence motifs. Previous studies have suggested a critical role for IRS-2–dependent signaling pathways, particularly in liver, pancreatic beta cells, adipose tissue, the hypothalamus, and the reproductive system, whereas IRS-1 appears to be mainly involved in insulin signaling in skeletal muscle (Hennige et al. 2005a). Indeed, IRS-2 has been implicated in the promotion of beta-cell survival and evidence shows that cytokine- and fatty acid–induced apoptosis is strongly inhibited by insulin glulisine compared with RHI, insulin aspart, and insulin lispro (Rakatzi et al. 2003b). Results from the current studies demonstrated that insulin glulisine induces prominent IRS-2 activation similar to that of RHI. This prominent antiapoptotic activity of insulin glulisine might serve to counteract autoimmune- and lipotoxicity-induced beta-cell destruction, although further studies are required to confirm this potential benefit of insulin glulisine. In contrast, only marginal IRS-1 activation with insulin glulisine compared with RHI was seen in vitro and similar results have been shown in a number of other in vitro studies (Rakatzi et al. 2003a, 2003b). However, it should be noted that different data may be obtained in different in vitro cellular assay systems. Further, Hennige et al. (2005a) showed that insulin glulisine activated both insulin receptor substrates to the same extent as RHI in vivo. They suggested that this difference between the in vivo and in vitro results obtained from myoblasts, cardiomyocytes, human skeletal muscle cells, and an insulinoma cell line (Rakatzi et al. 2003a, 2003b) were most likely due to the extensive amount of insulin and the nature of the cell lines used in those studies. Overall, the results demonstrated that insulin glulisine induces prominent IRS-2 activation, but only marginal IRS-1 activation. This does not affect the metabolic response of insulin glulisine relative to RHI in muscle cells. Rakatzi et al. (2003a) demonstrated that insulin glulisine, like other analogs, produced full activation of Akt and GSK-3, concluding that regular activation of IRS-2 combined with a minor activation of IRS-1 is sufficient to produce a full metabolic response.

In general, the investigation of receptor binding and proliferation in vitro and in vivo has shown that there is no difference in the mitogenic potential of insulin glulisine, RHI, or untreated controls. In this evaluation, the incidence of mammary tumors in the insulin glulisine– and RHI-treated rats was comparable with historical control data of this strain. In addition, this was similar to that seen with insulin aspart in previous studies (FDA 2000). Therefore, it is possible to conclude that the amino acid modifications in insulin glulisine have no significant influence on mitogenic potential and that the mitogenic potential of insulin glulisine is identical to that of RHI in differentiating muscle tissue in vitro.

Overall, results from this comprehensive preclinical evaluation suggest that the profile of insulin glulisine does not raise any safety concerns regarding its clinical use. Although, compared with RHI, insulin glulisine displays slightly reduced binding and activation of the insulin receptor in vitro, this does not appear to impact on the glucose-lowering activity of insulin glulisine in vivo, as shown in clinical trials (Dailey et al. 2004; Dreyer et al. 2004; Garg et al. 2004a, 2004b).