How Structural Modifications of Insulin Icodec Contributes to Its Prolonged Duration of Action: A Structural and Biophysical Perspective

1. Introduction

Ligand-receptor binding affinity is an essential parameter in both computational and experimental drug discovery & design [1]. Thanks to the continued development of experimental structural biology and the fifty-three-year old Protein Data Bank (PDB) [2,3], a high-throughput comprehensive structural biophysical analysis becomes possible [4] for specific ligand-receptor complex structures deposited in PDB, such that our understanding of the structural and biophysical basis of their interfacial stability is able to help us modify the binding affinity of certain drug target and its interacting partners.

Take insulin icodec for example [5,6,7,8]. Insulin icodec of Novo Nordisk is a long-acting insulin analogue for better management of blood sugar levels in people with diabetes [9,10,11]. Insulin icodec is designed to provide a steady release of insulin throughout the day, mimicking the natural insulin production in the body [12,13,14]. Insulin icodec is typically administered through injection once a week, which helps lower blood sugar levels by allowing glucose to enter the body’s cells, where it is subsequently used for energy production [15,16]. Moreover, insulin icodec has a distinct pharmacokinetic profile compared to other long-acting insulin analogs. It exhibits a long duration of action, with a half-life of approximately 196 hours, leading to improved glycemic control and reduced hypoglycemia risk [17,18]. Interestingly, Icosema (of Novo Nordisk, too) [19,20,21,22] represents a combination medication that consists of insulin icodec and semaglutide, which belongs to a class of medications called glucagon-like peptide-1 (GLP-1) receptor agonists and helps regulate blood sugar levels by stimulating the release of insulin, reducing the production of glucagon (a hormone that increases blood sugar levels), and slowing down the digestion process for weight reduction [23,24,25].

Given the two-dimensional structures of native human insulin and insulin icodec as shown in Figure 1, this article summarizes below a list of structural modifications of insulin icodec compared to native human insulin [26,27,28]:

• addition of a fatty acid (K50B_C20, Figure 1 and Figure 2);
• deletion of threonine at position B30 (T51B_del, Figure 1 and Figure 2);
• a site-specific missense mutation, Y14A_E (Figure 1 and Figure 2);
• a site-specific missense mutation, Y37B_H (Figure 1 and Figure 2);
• a site-specific missense mutation, F46B_H (Figure 1 and Figure 2).

Overall, these structural modifications (Figure 2) in insulin icodec result in a more stable and longer-acting insulin analogue compared to regular insulin, providing a more consistent and sustained blood glucose-lowering effect:

• insulin icodec is able to form aggregates or clusters at the subcutaneous injection site, gradually releasing into the bloodstream over an extended period.
• insulin icodec undergoes structural modifications that increases its stability and solubility and preventing enzyme-mediated degradation and rapid clearance [36].
• insulin icodec is conjugated with a fatty acid at position B30 (K50B_C20, Figure 2). After injection, the fatty acid chain in insulin icodec interacts with albumin in the subcutaneous tissue, forming reversible albumin-insulin complexes. These complexes act as a reservoir, gradually releasing insulin icodec into the bloodstream, increasing its fat solubility and allowing it to bind to fatty acid-binding proteins, forming a depot of the insulin icodec reversibly bound to albumin.
• the incorporation of fatty acid chains facilitate the formation of stable hexameric structures, thereby delaying insulin absorption and promoting sustained release.

With respect to the addition of a fatty acid (K50B_C20, Figure 2), it is conceivable that the deletion of threonine at position B30 (T51B_del, Figure 2) is for K50B_C20 to take place easier and more efficiently than without the deletion of threonine at position B30. Nonetheless, the underlying biophysics remains elusive of the roles of the three site-specific mutations (Y14A_E, Y37B_H, F46B_H) of insulin icodec in its prolonged activity. Therefore, through a comprehensive structural and biophysical analysis of the insulin (both native and icodec) structures bound to its receptor, this article delves deep into the biophysics underlying the molecular design of insulin icodec, and identified a delicate biophysical mechanism through which two missense mutations of insulin icodec (Y37B_H and F46B_H) contribute to its prolonged duration of action.

2. Materials and Methods

2.1. A summary of insulin receptor-related structures in PDB

As listed in Table 1, as of November 16, 2023, there are a variety of experimental complex structures of insulin (analogues) bound to its receptor (IR), such as PDB entries 7PG3 (insulin receptor bound to 3 insulins), 7PG4 (insulin receptor bound to 2 insulins), 6SOF (insulin receptor bound to 4 insulins).

Among the insulin receptor-related structures listed in Table 1, PDB entry 6SOF [37] is the only experimental complex structure of insulin bound to IR, where all four distinct binding sites of the IR dimer are saturated by four insulin molecules. Therefore, PDB entry 6SOF [37] is chosen here as the structural template for subsequent structural modelling of insulin icodec bound to IR.

2.2. Construction of a complex structural model of insulin icodec bound to IR

Towards the construction of a complex structural model of insulin icodec bound to IR, first, the amino acid sequences of the two chains of native human insulin and IR (according to PDB entry 6SOF [37], supplementary file nati.pdb) are listed in italics in fasta format as below,

>InsulinA

GIVEQCCTSICSLYQLENYCN

>InsulinB

FVNQHLCGSHLVEALYLVCGERGFFYTPKT

>IRpartOne

HLYPGEVCPGMDIRNNLTRLHELENCSVIEGHLQILLMFKTRPEDFRDLSFPKLIMITDYLLLFRVYGLESLKDLFPNLTVIRGSRLFFNYALVIFEMVHLKELGLYNLMNITRGSVRIEKNNELCYLATIDWSRILDSVEDNYIVLNKDDNEECGDICPGTAKGKTNCPATVINGQFVERCWTHSHCQKVCPTICKSHGCTAEGLCCHSECLGNCSQPDDPTKCVACRNFYLDGRCVETCPPPYYHFQDWRCVNFSFCQDLHHKCKNSRRQGCHQYVIHNNKCIPECPSGYTMNSSNLLCTPCLGPCPKVCHLLEGEKTIDSVTSAQELRGCTVINGSLIINIRGGNNLAAELEANLGLIEEISGYLKIRRSYALVSLSFFRKLRLIRGETLEIGNYSFYALDNQNLRQLWDWSKHNLTITQGKLFFHYNPKLCLSEIHKMEEVSGTKGRQERNDIALKTNGDQASCENELLKFSYIRTSFDKILLRWEPYWPPDFRDLLGFMLFYKEAPYQNVTEFDGQDACGSNSWTVVDIDPPLRSNDPKSQNHPGWLMRGLKPWTQYAIFVKTLVTFSDERRTYGAKSDIIYVQTDATNPSVPLDPISVSNSSSQIILKWKPPSDPNGNITHYLVFWERQAEDSELFELDYCLKGLKLPSRTWSPPFESEDSQKHNQSEYEDSAGECCSCPKTDSQILKELEESSFRKTFEDYLHNVVFVPRPS

>IRpartTwo

HRPFEKVVNKESLVISGLRHFTGYRIELQACNQDTPEERCSVAAYVSARTMPEAKADDIVGPVTHEIFENNVVHLMWQEPKEPNGLIVLYEVSYRRYGDEELHLCVSRKHFALERGCRLRGLSPGNYSVRIRATSLAGNGSWTEPTYFYVTDYLDVPSNIAK

Second, the amino acid sequences of the two chains of insulin icodec and IR are listed in italics in fasta format as below,

>InsulinIcodecA

GIVEQCCTSICSLEQLENYCN

>InsulinIcodecB

FVNQHLCGSHLVEALHLVCGERGFFHTPKT

>IRpartOne

HLYPGEVCPGMDIRNNLTRLHELENCSVIEGHLQILLMFKTRPEDFRDLSFPKLIMITDYLLLFRVYGLESLKDLFPNLTVIRGSRLFFNYALVIFEMVHLKELGLYNLMNITRGSVRIEKNNELCYLATIDWSRILDSVEDNYIVLNKDDNEECGDICPGTAKGKTNCPATVINGQFVERCWTHSHCQKVCPTICKSHGCTAEGLCCHSECLGNCSQPDDPTKCVACRNFYLDGRCVETCPPPYYHFQDWRCVNFSFCQDLHHKCKNSRRQGCHQYVIHNNKCIPECPSGYTMNSSNLLCTPCLGPCPKVCHLLEGEKTIDSVTSAQELRGCTVINGSLIINIRGGNNLAAELEANLGLIEEISGYLKIRRSYALVSLSFFRKLRLIRGETLEIGNYSFYALDNQNLRQLWDWSKHNLTITQGKLFFHYNPKLCLSEIHKMEEVSGTKGRQERNDIALKTNGDQASCENELLKFSYIRTSFDKILLRWEPYWPPDFRDLLGFMLFYKEAPYQNVTEFDGQDACGSNSWTVVDIDPPLRSNDPKSQNHPGWLMRGLKPWTQYAIFVKTLVTFSDERRTYGAKSDIIYVQTDATNPSVPLDPISVSNSSSQIILKWKPPSDPNGNITHYLVFWERQAEDSELFELDYCLKGLKLPSRTWSPPFESEDSQKHNQSEYEDSAGECCSCPKTDSQILKELEESSFRKTFEDYLHNVVFVPRPS

>IRpartTwo

HRPFEKVVNKESLVISGLRHFTGYRIELQACNQDTPEERCSVAAYVSARTMPEAKADDIVGPVTHEIFENNVVHLMWQEPKEPNGLIVLYEVSYRRYGDEELHLCVSRKHFALERGCRLRGLSPGNYSVRIRATSLAGNGSWTEPTYFYVTDYLDVPSNIAK

Lastly, to build a reliable structural model of the insulin icodec bound to IR, the amino acid sequences above (>InsulinIcodecA, >InsulinIcodecB, >IRpartOne and >IRpartTwo) were plugged into the SWISS-MODEL homology modeling server [38,39] and the Modeller software [40] to build a homology structural model with reasonable accuracy, supplementary file icod.pdb, which was subsequently subject to a set of comprehensive structural biophysical analysis as described previously in [4].

2.3. A comprehensive structural and biophysical analysis of insulin icodec bound to IR

As described previously in [4], a comprehensive structural and biophysical analysis was conducted for the structural model of the insulin icodec bound to IR (supplementary file icod.pdb). Specifically, the salt bridge analysis was conducted with an in-house python script only for titrateable residues (Asp, Glu, Lys, Arg and His), 4.0 Å was used as the cutoff distance for the two oppositely charged groups [4]. The hydrogen bond analysis was also conducted for only side chain nuclei with an in-house python script, and employed two geometric criteria: (a) a cutoff value of the angle formed by acceptor (A), donor (D) and hydrogen (H) ($\angle ADH$) of 30°; (b) a cutoff value of donor-acceptor distance at 3.0 Å. That is, a hydrogen bond is only considered to be formed if $\angle ADH$ is no larger than 30° and the donor-acceptor distance is not larger than 3.0 Å [4].

3. Results

As described above, the addition of a fatty acid to insulin icodec (K50B_C20, Figure 2) is for it to interact with albumin and form reversible albumin-insulin complexes, i.e., a reservoir for gradually releasing insulin icodec into the bloodstream. Moreover, it is highly likely that the deletion of threonine at position B30 (T51B_del, Figure 2) is for K50B_C20 to take place easier and more efficiently. As a result, the comprehensive structural and biophysical analysis here focuses on the three site-specific mutations (Y14A_E, Y37B_H, F46B_H) of insulin icodec.

Coincidentally, the three mutations (Y14A_E, Y37B_H, F46B_H) invariably fall into the same category of the substitution of an amino acid residue with an amino acid residue with a hydrophilic side chain. Briefly, with a set of comprehensive structural and biophysical analysis [4] for supplementary files icod.pdb and nati.pdb, this article here for the first time reports that

• Y14A_E (Figure 1 and Figure 2) is not involved in any salt bridge or hydrogen bond within insulin icodec itself.
• Y14A_E (Figure 1 and Figure 2) is not involved in any salt bridge or hydrogen bond at the binding interface between insulin icodec and IR.
• Y37B_H (Figure 1 and Figure 2) is involved in one salt bridge within insulin icodec itself (Figure 3, Table 2 and Table 4).
• Y37B_H (Figure 1 and Figure 2) is not involved in any hydrogen bond within insulin icodec itself.
• Y37B_H (Figure 1 and Figure 2) is not involved in any salt bridge or hydrogen bond at the binding interface between insulin icodec and IR.
• F46B_H (Figure 1 and Figure 2) is not involved in any salt bridge or hydrogen bond within insulin icodec itself.
• F46B_H (Figure 1 and Figure 2) is not involved in any salt bridge or hydrogen bond at the binding interface between insulin icodec and IR.

Of the three mutations (Y14A_E, Y37B_H, F46B_H), a total of three electrostatic interactions (including salt bridges and hydrogen bond) were identified and listed as below in Table 2.

Specifically, before Y37B_H (Figure 1 and Figure 2), i.e., in the case of native human insulin (Table 3), only one salt bridge (3.204 Å long) was formed between C_LYS_921 (of IR) and F_GLU_1804 (of regular insulin) at the binding interface of native human insulin bound to IR (supplementary file nati.pdb). Similarly, after Y37B_H (Figure 1 and Figure 2), i.e., in the case of insulin icodec (Table 4), only one salt bridge (3.669 Å long) was formed between C_LYS_921 (of IR) and F_GLU_1804 (of insulin icodec) at the binding interface of insulin icodec bound to IR (supplementary file icod.pdb). Interestingly, apart from the interfacial salt bridge between C_LYS_921 (of IR) and F_GLU_1804 (of insulin icodec), there is also one intra-molecular salt bridge (3.659 Å long) between F_HIS_1799 (of insulin icodec) and F_GLU_1804 (of insulin icodec).

Thus, it is obvious that in the case of native human insulin (Table 3), F_GLU_1804 (of regular insulin) focuses on binding to C_LYS_921 (of IR) alone with a 3.204 Å long interfacial salt bridge between regular insulin and IR, while in the case of insulin icodec (Table 4), F_GLU_1804 (of insulin icodec) has to bind both C_LYS_921 (of IR) and F_HIS_1799 (of insulin icodec) with two salt bridges, as listed in Table 4.

Furthermore, the existence of F_HIS_1799 (of insulin icodec) is entirely due to the site-directed mutation Y37B_H (Figure 1 and Figure 2). Therefore, it is entirely due to the site-directed mutation Y37B_H that the salt bridge at the binding interface between insulin and IR becomes weaker (from 3.204 Å long to 3.669 Å long), but is still not disrupted by the site-directed mutation Y37B_H, such that the binding affinity of ligand-receptor is lowered but not eliminated by the site-directed mutation Y37B_H (Figure 1 and Figure 2), and thereby ensuring downstream signal transduction for the prolonged blood glucose-lowering effect of insulin icodec.

With respect to the site-directed mutation F46B_H (Figure 1 and Figure 2), it is not involved in any salt bridge or hydrogen bond within insulin icodec itself, nor is it involved in any salt bridge or hydrogen bond at the binding interface between insulin icodec and IR, according comprehensive structural and biophysical analysis [4] of insulin icodec bound to IR (supplementary file icod.pdb). Nonetheless, a close inspection with PyMol [41] of supplementary file icod.pdb revealed that the positively charged side chain of His1808 of insulin icodec (due to the site-specific mutation F46B_H, Figure 2) is only 6.4 Å away (Figure 4) from the positively charged side chain of R900 (Arg900) of IR. Despite the cutoff distance (4.0 Å) as used in [4], a distance 6.4 Å is not sufficient to be defined as strong an electrostatic interaction as a salt bridge, it still is able to induce a weak electrostatic repulsive force between the positively charged side chains of His1808 of insulin icodec and that of R900 of IR, and thereby reducing the binding affinity between insulin icodec and IR. Moreover, in case the site-specific mutation is not F46B_H, Figure 2), but F46B_E, i.e., it is not a histidine, but a glutamate, it is conceivable that the distance between Glu1808 of insulin icodec and R900 of IR is shorter than 6.4 Å (Figure 4), make it highly likely for a novel stable salt bridge to be formed between E1808 (Glu1808) of insulin icodec and R900 (Arg900) of IR, according to the 4 Å distance cutoff as used in [4].

Finally, with respect to the site-directed mutation Y14A_E (Figure 1 and Figure 2), , it mutation does not seem to impact the binding between insulin icodec and IR, as it is a fully solvent-exposed amino acid residue, as shown by Figure 5. Thus, as of November 16, 2023 from structural and biophysical perspective, it is not clear yet why site-directed mutation Y14A_E is included in the design of insulin icodec. Here, this article is only able to speculate that Y14A_E is for the overall structural stability of icodec insulin for it to exert its prolonged blood glucose-lowering effect for patients with diabetes.

4. Conclusion and Discussion

For the first time, through a comprehensive structural and biophysical analysis of the insulin (both native and icodec) structures bound to its receptor, this article puts forward a delicate biophysical mechanism through which two missense mutations of insulin icodec (Y37B_H and F46B_H) contribute to its prolonged duration of action. In addition to the biophysics underlying the molecular design of insulin icodec, this article also argues that this structural and biophysical understanding is of help for rational design of future long-acting insulin analogues, offering further enhanced therapeutic options for diabetic patients.

5. Ethical statement

No ethical approval is required.

6. Declaration of generative AI and AI-assisted technologies in the writing process

During the preparation of this work, the author used OpenAI’s ChatGPT in order to improve the readability of the manuscript, and to make it as concise and short as possible. After using this tool, the author reviewed and edited the content as needed and takes full responsibility for the content of the publication.