Satoshi Kitaokaa,
Minori Kojimaa,
Miho Koitaa,
Hiroki Koyamaa,
Chisato Moria,
Mako Okabea,
Ryusei Andoa,
Kaede Kobayashia,
Ryo Watanabea,
Yuki Takanoa,
Tony D. James
bc and
Yuya Egawa
*a
aFaculty of Pharmacy and Pharmaceutical Sciences, Josai University, 1-1 Keyakidai, Sakado, Saitama 350-0295, Japan. E-mail: yegawa@josai.ac.jp
bDepartment of Chemistry, University of Bath, Bath, BA2 7AY, UK
cSchool of Chemistry and Chemical Engineering, Henan Normal University, Xinxiang 453007, P. R. China
First published on 11th August 2025
Among various treatment options for diabetes, insulin therapy remains an important approach, but it inevitably carries the risk of hypoglycaemia, particularly due to dosing errors or unexpected glucose fluctuations. To address this challenge, glucose-responsive insulin delivery systems that release insulin based on blood glucose levels have emerged as a promising solution. In this study, we developed a fully dissolved glucose-responsive insulin delivery system using p-borono-phenylmethoxycarbonyl-modified insulin aspart (BPmoc-Ins-Asp) and glucose oxidase (GOx). This system uses BPmoc-Ins-Asp as a prodrug that remains inactive until activated by hydrogen peroxide (H2O2), generated through GOx-mediated glucose oxidation. The BPmoc group undergoes a self-immolative reaction in response to H2O2, decomposing into boric acid, p-quinone methide, and CO2, thereby restoring the amino group to its original state. High-performance liquid chromatography (HPLC) confirmed the conversion of BPmoc-Ins-Asp into active Ins-Asp in the presence of GOx and glucose. Cell-based assays demonstrated that BPmoc-Ins-Asp effectively masks insulin activity until activation. Once activated, the released Ins-Asp promoted glucose transporter 4 (GLUT4) translocation, mimicking the physiological effects of insulin. In vivo studies further validated the system's glucose responsiveness, demonstrating glucose-lowering effects specifically under hyperglycaemic conditions, with no effect during normoglycaemic states. Unlike gel- or particle-based systems, this fully dissolved liquid formulation enables the use of thin needles for self-administration, simplifies manufacturing, and ensures consistent production quality, making it particularly suitable for clinical applications. These advantages underscore its potential for precise glucose control while minimizing the risk of hypoglycaemia.
Regular blood glucose monitoring is essential for preventing hypoglycaemia and recognizing early symptoms. In recent years, continuous glucose monitoring (CGM) and flash glucose monitoring systems have been widely adopted to track glucose fluctuations, demonstrating improved glycaemic control when combined with insulin pumps.5–7 However, a recent study revealed that, among its participants, 25.6% of type 1 diabetes patients with CGM systems experienced at least one severe hypoglycaemic episode in the previous six months.8 Another study shows that, despite increasing adoption of insulin pumps and CGM, the incidence of severe hypoglycaemia remains considerable, with 7.0% of self-injectors and 5.1% of insulin pump users reporting at least one episode per year.9 Hypoglycaemia prevention thus remains a critical challenge.
One promising approach to reduce hypoglycaemia risk involves developing glucose-responsive insulin delivery systems.10 These systems regulate insulin release based on blood glucose levels, enhancing release during hyperglycaemia and reducing it during hypoglycaemia. This strategy has the potential to optimise glycaemic control and reduce the risk of hypoglycaemia, addressing a key challenge in diabetes management.
Among several strategies, boronic acids have been extensively studied for their glucose-responsive properties.11–17 Two main mechanisms have been explored: reversible covalent bond formation with polyols, including glucose, and an irreversible reaction with hydrogen peroxide (H2O2). This study focuses on the latter, where glucose oxidase (GOx) plays a key role by converting glucose concentration into H2O2 concentration, which subsequently triggers the degradation of boronic acid derivatives.18–22 Specifically, among the boronic acid derivatives, p-borono-phenylmethoxycarbonyl (BPmoc) group undergoes a reaction with H2O2, decomposing into boric acid, p-quinone methide, and CO2.23–26 This self-immolative reaction allows BPmoc-modified amino groups to revert to their original state upon exposure to H2O2 (Scheme 1).
![]() | ||
Scheme 1 Activation of BPmoc-Ins-Asp through self-immolative elimination to produce Ins-Asp, triggered by H2O2 generated during glucose oxidation. |
Previous studies have demonstrated the feasibility of using BPmoc-modified compounds for glucose-responsive insulin delivery. Ikeda et al. developed hydrogels by self-assembly of BPmoc-phenylalanine dipeptide (BPmoc-FF) through π–π interactions and hydrogen bonding.23 These hydrogels encapsulated GOx and insulin, enabling glucose-responsive insulin release. Hu et al. expanded this concept by creating polymer vesicles that incorporated BPmoc groups and polyethylene glycol (PEG).25 These vesicles encapsulated GOx and insulin and were designed for use as microneedle patches. Their innovative designs showcased the potential of a glucose-triggered insulin delivery platform.
Despite their innovative nature, both hydrogels and particle-based formulations face practical challenges. Hydrogels are highly viscous and require high injection pressure. This makes them unsuitable for practical use with thin needles commonly employed in self-injection devices.27,28 Particle-based systems, such as polymer vesicles, struggle to maintain uniformity in critical parameters like particle size, drug loading amount, and glucose permeability—key factors for consistent therapeutic efficacy and glucose responsiveness. These difficulties become even more pronounced during large-scale production, limiting their viability for widespread clinical use.
To overcome the limitations of gel- and particle-based formulations, this study introduces a fully dissolved liquid formulation that leverages the superior solubility of insulin aspart (Ins-Asp) over native insulin.29 To achieve glucose-responsive activation, Ins-Asp was modified with BPmoc groups to create a prodrug, BPmoc-Ins-Asp, which remains inactive until activated by H2O2 generated via GOx-catalysed glucose oxidation. This strategy, involving a BPmoc-modified drug and GOx, was originally developed and reported by our team in a previous study.30 At that time, the BPmoc modification was applied to native human insulin, which has relatively low solubility. As a result, BPmoc-modified insulin displayed poor solubility, necessitating its formulation as a suspension. With this suspension formulation, the particles of BPmoc-modified insulin had to dissolve before activation, and the dissolution process led to variations in the activation time. This posed a significant challenge for accurate kinetic evaluation of the prodrug's activation process, as the dissolution step interfered with the precise measurement of activation kinetics and glucose responsiveness.
Our fully dissolved liquid-type glucose-responsive insulin formulation differs in its usage from the so-called artificial pancreas-type glucose-responsive insulin formulations. An artificial pancreas stores multiple doses of insulin and releases it in response to glucose fluctuations over a long period. Despite over 30 years of effort, the development of artificial pancreases has been hindered by technical difficulties, limiting practical implementation. Notably, administering multiple doses of insulin carries the risk of inducing hypoglycaemia, which is a significant concern.
By contrast, our fully dissolved liquid formulation is designed for single-dose administration. Since it does not require multiple insulin doses like an artificial pancreas, the risk of hypoglycaemia is minimized. This single-dose glucose-responsive formulation offers a practical approach toward the clinical implementation of glucose-responsive insulin. It has the potential to serve as a safer alternative to currently available rapid-acting insulin formulations.
The primary theoretical advantage of our system lies in its carrier-free design, which enables straightforward and precise kinetic tuning of glucose-responsive behaviour. In conventional carrier-based systems, such as those employing gels or particles, the overall response kinetics are affected by multiple factors, including the rate of glucose diffusion into the carrier and the subsequent release of insulin. As a result, simply varying the amount of GOx offers limited flexibility in modulating system responsiveness. In contrast, our carrier-free approach allows direct and efficient control of the response rate solely by adjusting the amount of GOx. This aspect is discussed in greater detail in the main text.
From a practical standpoint, the absence of carriers offers several significant benefits. In conventional carrier-based formulations, achieving uniform loading of each component onto the carrier and maintaining consistent carrier sizes—such as particle dimensions—are essential, yet they pose significant challenges for reproducible manufacturing and scale-up. In contrast, our system is a carrier-free liquid formulation, which inherently eliminates concerns related to formulation uniformity and particle size control. Furthermore, as a liquid formulation, it can be administered using a fine needle suitable for self-injection by patients, offering superior usability.
Similar to our formulation, MK-2640—developed by Merck & Co., Inc.—represents a carrier-free glucose-responsive insulin system.31,32 To the best of our knowledge, it remains the only glucose-responsive insulin system—excluding those involving insulin pumps—that has advanced to a Phase 1 clinical trial, underscoring the practical and clinical relevance of carrier-free formulations. MK-2640 is an insulin analogue chemically modified with mannose-based carbohydrate moieties to enable binding to mannose receptor C-type 1 (MRC1), a lectin receptor predominantly expressed on macrophages and liver Kupffer cells. MRC1 plays a key role in the innate immune system by recognizing and internalizing pathogen-associated high-mannose glycoproteins. MK-2640 was designed to exploit this clearance mechanism to achieve glucose responsiveness: under hypoglycaemic conditions, MK-2640 is rapidly removed from circulation through MRC1-mediated endocytosis, while under hyperglycaemic conditions, elevated glucose levels competitively inhibit MRC1 binding, slowing MK-2640 clearance and prolonging its glucose-lowering activity.
Although preclinical studies in dogs and minipigs demonstrated promising glucose-dependent pharmacokinetics, MK-2640 failed to show comparable efficacy in humans. This failure was attributed to multiple factors, including reduced insulin receptor binding affinity due to the carbohydrate modification—resulting in diminished insulin bioactivity—and the limited translatability of the MRC1-mediated clearance mechanism due to species-specific differences and receptor saturation. Despite encouraging preclinical data from other investigational approaches, no candidate has progressed beyond the preclinical stage, emphasizing the substantial hurdles facing clinical translation in this field. In this context, our proposal of a carrier-free glucose-responsive insulin formulation based on a fundamentally different mechanism represents a significant advance.
Unlike MK-2640, our system does not rely on receptor-mediated clearance pathways that are susceptible to species-dependent variability, thereby offering improved translational reliability. Furthermore, by employing a prodrug that is converted into pharmacologically active insulin (Ins-Asp) under glucose-responsive conditions, our approach overcomes the issue of reduced insulin potency observed in MK-2640, addressing a key limitation of previous designs.
In this study, BPmoc-Ins-Asp was successfully evaluated in its dissolved state, owing to its high solubility. The activation of BPmoc-Ins-Asp was analysed using HPLC, while the insulin activity of Ins-Asp generated from the prodrug was assessed through in vitro cell-based experiments. Animal studies confirmed the glucose responsiveness of the formulation via continuous blood glucose monitoring. Furthermore, toxicity studies were conducted to evaluate its safety. Detailed results are presented in the following sections.
The mass spectrometer was a 4000 QTRAP (SCIEX, Framingham, MA) and detection was performed in Q3 MS mode. The ion source used was Turbo V Spray. Ionisation was performed in negative ion mode and the various other parameters were set as follows: curtain gas: 10 psi; Turbo V ionspray voltage: −4.5 kV; ion source gas 1:
30 psi; ion source gas 2
:
80 psi; source temperature: 500 °C.
HPLC analysis was conducted using an octadecylsilyl (ODS) column (5 μm, 150 × 4.6 mm i.d., CAPCELL PAK C18, Shiseido Fine Chemicals, Tokyo, Japan) and gradient elution at a flow rate of 1.0 mL min−1. Mobile phase A consisted of AcCN/water/TFA (300:
700
:
1, v/v/v) and mobile phase B consisted of AcCN/water/TFA (950
:
50
:
1, v/v/v). The HPLC system (LC-20AT, SPD-20A, Shimadzu Corporation, Kyoto, Japan) was operated in gradient mode with a linear increase from 0 to 33.3% for mobile phase B over 20 min. Elution was monitored by absorbance at 271 nm.
Saline was used to prepare the injection formulations: BPmoc-Ins-Asp (0.10 mg mL−1), BPmoc-Ins-Asp (0.10 mg mL−1) + GOx (4.0 μg mL−1) and Ins-Asp (0.10 mg mL−1). These formulations were administered to rats restrained in a Ballman cage (s.c., 1.0 mL kg−1 body weight).
Water for injection was used to prepare the glucose solution (0.20 g mL−1) for the oral glucose tolerance test (OGTT). The rats were fasted for 16 h, after which the glucose solution (p.o., 10 mL kg−1 body weight) was administered.33
The area under the curve from 0 to 2 h (AUC0–2) of blood glucose levels was calculated using pharmacokinetic analysis software, Numeric Analysis Program for Pharmacokinetics (Napp) (version 3.071β).34
The protein structures in Scheme 1 and the Table of Contents figure were visualized using protein structure data from the PDBj database via the molecular viewer Molmil.35 The data for glucose oxidase from Aspergillus niger (PDB ID: 1CF3) was used as is,36 while BPmoc-Ins-Asp and Ins-Asp was visualized using human insulin (PDB ID: 3I3Z).37
BPmoc-modified glycine (BPmoc-Gly) used in a toxicity study was prepared by reacting the pinacol ester of 4-(hydroxymethyl)phenylboronic acid with ethyl isocyanatoacetate to form a urethane bond, which was then hydrolysed with LiOH.
Detailed synthetic procedures and synthetic schemes (Scheme S1 and S2) are given in the SI.
Fig. 1(A) illustrates the structure of Ins-Asp and its expected fragments 1–4 (Frag. 1–4) upon digestion with V8 protease and subsequent reduction with DTT.
Fig. 1(B-1) shows the LC-MS chromatogram of Ins-Asp digested with V8 protease, revealing four major peaks. Based on comparison with a previous study, these peaks were identified as Frag. 1 to 4.39 The MS spectra of each peak are presented in Fig. S1, and the theoretical and observed m/z values are summarised in Table S1.
Fig. 1(C-1) shows the chromatogram of BPmoc-Ins-Asp digested in a similar manner, where the peak corresponding to Frag. 2 of Ins-Asp was found at 41 min. Meanwhile, Frag. 1, Frag. 3, and Frag. 4 were not detected, suggesting complete chemical modification of these three fragments. Three new peaks were observed in Fig. 1(C-1), and the presence or absence of BPmoc modification was investigated from the MS spectra of each peak (Fig. S3 and Table S3). It should be noted that previous studies have shown that MS spectra of compounds containing phenylboronic acid typically exhibit characteristic dehydrated ions rather than the intact molecular weight.40 In Fig. 1(C-1), the peaks detected at 46, 51, and 56 min included m/z values that were close to theoretical m/z values for the expected ions (Fig. S3 and Table S3). The peak at 46 min corresponded to [MFrag. 3+BPmoc–2H2O–2H]2–, with a theoretical m/z of 637.10 and an observed m/z of 638.00. Similarly, the peak at 51 min corresponded to [MFrag. 4+BPmoc–H2O–H]–, with a theoretical m/z of 575.43 and an observed m/z of 576.40. The peak at 56 min corresponded to [MFrag. 1+BPmoc–H2O–2H]2–, with a theoretical m/z of 1564.67 and an observed m/z of 1564.10. These results showed that each BPmoc group was incorporated as a single modification into Frag. 3, Frag. 4, and Frag. 1.
To further investigate the position of the BPmoc group in Frag.1, DTT was used to cleave the disulphide bond. In Fig. 1(B-2) for Ins-Asp reduced with DTT, Frag. 1A and Frag. 1B were observed (Fig. S2 and Table S2). In Fig. 1(C-2) for DTT-reduced BPmoc-Ins-Asp, Frag. 1B was not present, but a new peak was seen at 50.67 min. The MS spectrum from this peak included an m/z value of 812.00, which was close to the expected m/z of 811.32 for [MFrag. 1B+BPmoc–2H2O–2H]2−, indicating that Frag. 1B was modified (Fig. S4 and Table S4).
In summary, BPmoc groups were introduced at three specific sites: Frag. 1B, Frag. 3, and Frag. 4 of Ins-Asp. Each fragment contains a single amino group: Frag. 1B features an amino group at the N-terminus of the insulin B chain, Frag. 3 has one at the 29th lysine residue of the B chain, and Frag. 4 has one at the N-terminus of the insulin A chain. These results strongly suggest that BPmoc groups were specifically introduced at these amino groups.
It has been reported that acylation of the N-terminus of Gly of the A-chain markedly reduces insulin activity.41 Furthermore, acylation at the N-terminus of Phe of the B-chain has also been shown to result in insufficient insulin activity.42 In the case of the BPmoc-Ins-Asp obtained in this study, chemical modifications are introduced at both of these N-terminal residues, and this is considered to be the main factor contributing to the loss of insulin activity.
F-actin, shown in cyan, was used as a counterstain to clearly delineate the cell outline and to provide spatial context for assessing the intracellular localization of GLUT4. Moreover, since GLUT4 translocates to the plasma membrane along F-actin filaments, co-staining of GLUT4 and F-actin enabled a more detailed evaluation of the pharmacological effect of insulin.45
Under the 5 mM glucose condition, GLUT4 translocation to the plasma membrane was detected exclusively in cells treated with Ins-Asp. In these cells, the red fluorescence signal derived from GLUT4 was distributed along intracellular F-actin filaments and extended to the cell periphery, indicating successful translocation. In contrast, cells treated with BPmoc-Ins-Asp alone or in combination with GOx showed cytoplasmic retention of the GLUT4 signal, similar to untreated controls. Notably, strong punctate red signals were observed in the perinuclear region, likely representing vesicle-trapped GLUT4. These results demonstrate that under low-glucose conditions (5 mM), BPmoc-Ins-Asp did not exert insulin activity, regardless of GOx co-treatment.
Under the 20 mM glucose condition, cells treated with either Ins-Asp or BPmoc-Ins-Asp in combination with GOx exhibited a diffuse red GLUT4 signal throughout the cytoplasm, outlining the cell shape, consistent with plasma membrane localisation. In contrast, the control and BPmoc-Ins-Asp alone groups displayed centrally clustered GLUT4 signals with poorly defined cell boundaries. Without the F-actin (cyan) counterstain, it was difficult to identify the plasma membrane in these groups. Thus, under high-glucose conditions, BPmoc-Ins-Asp elicited insulin-like activity only in the presence of GOx. Importantly, BPmoc-Ins-Asp alone did not induce GLUT4 translocation, even under hyperglycaemic conditions.
Collectively, these findings validate the functionality of our glucose-responsive activation system: the insulin activity of Ins-Asp is masked by BPmoc conjugation and reactivated in the presence of GOx and 20 mM glucose. Moreover, the lack of activity under 5 mM glucose further highlights the glucose-dependent nature of this mechanism.
First, we observed the conversion of BPmoc-Ins-Asp to Ins-Asp in the presence of H2O2. In the HPLC chromatogram, BPmoc-Ins-Asp appeared as a single peak at 12 min (Fig. S5(A)). Upon addition of H2O2, a new peak corresponding to Ins-Asp appeared at 8 min (Fig. S5(B)), and its intensity increased with higher H2O2 concentrations (Fig. S5(C)). At an H2O2 concentration of 5.0 mM, the amount of Ins-Asp produced reached a plateau after 30 min, indicating that the conversion was complete. These results confirm the H2O2 responsiveness of BPmoc-Ins-Asp.
The next step was to investigate the effect of GOx concentration on the activation of BPmoc-Ins-Asp. The concentration of BPmoc-Ins-Asp was kept constant at 0.10 mg mL−1, while the GOx concentration (1.0, 4.0, 10 and 16 μg mL−1) and the glucose concentration (5.0, 20, 40 and 80 mM) were varied. Under all conditions, the conversion to Ins-Asp increased with increasing glucose concentration (Fig. 3(A)), confirming that the proposed glucose-responsive mechanism was functional. Furthermore, the results show that the activation rate of BPmoc-Ins-Asp could be modulated by adjusting the concentration of GOx.
The average reaction rate was calculated by dividing the peak area at each time point by the corresponding time period, and then plotted against the glucose concentration (Fig. 3(B)). The graph revealed that under the condition of a GOx concentration of 16 μg mL−1, the reaction rate increased rapidly even when the glucose concentration was below 10 mM. This suggests that at 16 μg mL−1 GOx, BPmoc-Ins-Asp is converted to Ins-Asp even at normal blood glucose levels, potentially causing hypoglycaemia. In contrast, at GOx concentrations of 1.0 to 10 μg mL−1, the reaction rate increased gradually within the glucose concentration range of 0–10 mM. Based on these results, GOx concentrations below 16 μg mL−1 were considered suitable for injectable solutions.
The group receiving the formulation (0.10 mg mL−1 BPmoc-Ins-Asp + 10 μg mL−1 GOx) tended to have lower blood glucose levels than the saline-treated control group (Fig. S6(A)). The AUC0–2h of this group was significantly lower than that of the control group (Fig. S6(B)). This result is meaningful in the context that BPmoc-Ins-Asp can be activated to Ins-Asp in the body. However, it is not consistent with our concept of a glucose-responsive formulation that does not lower blood glucose levels under normoglycaemic conditions. The group treated with the formulation (0.10 mg mL−1 BPmoc-Ins-Asp + 4.0 μg mL−1 GOx) did not reduce blood glucose levels in normoglycaemic rats, and the AUC0–2h was not significantly different from that of the control group (Fig. S6(B)). Thus, it was decided to further investigate the potential of this formulation.
Wistar rats with normoglycaemia were treated according to the protocol shown in Fig. 4(A-1), and four formulations were administered for evaluation: (a) control (saline), (b) 0.10 mg mL−1 BPmoc-Ins-Asp, (c) 0.10 mg mL−1 BPmoc-Ins-Asp + 4.0 μg mL−1 GOx, and (d) 0.10 mg mL−1 Ins-Asp. In the control group, blood glucose levels remained stable at approximately 150 mg dL−1 (8.3 mM) for 2 h (Fig. 4(B-1)(a)). Similar results were observed in the BPmoc-Ins-Asp group (Fig. 4(B-1)(b)), consistent with in vitro findings indicating that the insulin activity of BPmoc-Ins-Asp was masked. The data shown in Fig. 4(B-1)(c) is the same as the 0.10 mg mL−1 BPmoc-Ins-Asp + 4.0 μg mL−1 GOx data in Fig. S6(A) described previously, and it did not reduce blood glucose levels. The Ins-Asp-treated group exhibited a rapid drop in blood glucose levels, resulting in hypoglycaemia (Fig. 4(B-1)(d)); hypoglycaemia in rats is reported as a blood glucose level below 3.5 mM.47
For conditions with high blood glucose levels, Goto-Kakizaki (GK) rats were used, which are known as a model of type 2 diabetes.48 In the control group with GK rats (Fig. 4(B-2)(a)), blood glucose levels were high, ranging from 300 to 500 mg dL−1 (approximately 17–28 mM). Subsequently, blood glucose levels gradually decreased. This slow decline is presumed to result from endogenous insulin secretion, as GK rats partially respond to hyperglycaemia by producing their own insulin.49 The glucose profile of the BPmoc-Ins-Asp group was similar to that of the control group (Fig. 4(B-2)(b)). Notably, the BPmoc-Ins-Asp + GOx group showed a trend towards lower blood glucose (Fig. 4(B-2)(c)).
To simulate an increase in postprandial blood glucose levels, the formulations were administered to GK rats immediately after OGTT, according to the protocol shown in Fig. 4(A-2). In the control group, blood glucose levels peaked approximately 1 h later, followed by a gradual decline, reflecting the endogenous insulin activity of GK rats (Fig. 4(B-3)(a)). The BPmoc-Ins-Asp group showed a similar glucose profile (Fig. 4(B-3)(b)). In contrast, the BPmoc-Ins-Asp + GOx group showed a clear glucose-lowering effect, although there was a time lag (Fig. 4(B-3)(c)).
The absence of a glucose-lowering effect in normoglycemic rats is likely due to limited conversion of the prodrug to Ins-Asp under low glucose conditions. This is consistent with our in vitro cell-based assays, which demonstrated a marked difference in prodrug activation between 5 mM and 20 mM glucose (Fig. 2).
To further compare the above results, AUC0–2h was calculated, as conceptualised in Fig. 5(A). In Fig. 5(B-1), which shows AUC0–2h in normal Wistar rats, only the Ins-Asp group showed a significant decrease compared to the control group (***p < 0.001). The BPmoc-Ins-Asp group was also significantly different from the control group; however, this was a slight increase rather than a decrease (*p < 0.05). Fig. 5 (B-2) shows the AUC0–2h in GK rats without OGTT, where the BPmoc-Ins-Asp + GOx group showed a significant reduction compared to the control group (*p < 0.05). Fig. 5(B-3) shows the AUC0–2h in GK rats with OGTT, in which the difference between the control and BPmoc-Ins-Asp + GOx groups was more pronounced (**p < 0.01).
As schematically shown in Fig. 5(A), the rate of blood glucose lowering was calculated as the slope from the maximum blood glucose level (tmax) to 1 h later. In both normoglycaemic Wistar rats and GK rats without OGTT, the BPmoc-Ins-Asp + GOx group did not differ from the control group (Fig. 5(C-1 and 2)). However, in the GK rat study with OGTT (Fig. 5(C-3)), the BPmoc-Ins-Asp + GOx group had a significantly greater negative slope than the control group (***p < 0.001).
The comparison of blood glucose levels, AUC0–2h, and the rate of blood glucose reduction demonstrated that the combination of BPmoc-Ins-Asp and GOx effectively reduced blood glucose levels exclusively under hyperglycaemic conditions, without affecting normoglycaemic levels. These results confirm that our concept is functional; however, the glucose-lowering effect became evident only after 1.5 h, and a faster response is desirable in the future.
In this study, HPLC analysis and animal experiments revealed that BPmoc-Ins-Asp was converted to Ins-Asp under normoglycaemic conditions when the GOx concentration was high. From a clinical perspective, the potential risk of hypoglycaemia due to excessive GOx levels must be carefully considered. While the amount of H2O2 generated can be regulated by adjusting the GOx concentration, this alone is insufficient to achieve ideal glucose dependency. Therefore, to restrict insulin activation to hyperglycaemic conditions, it is necessary not only to optimize the GOx concentration but also to consider other contributing factors. In particular, adjusting the intrinsic glucose reactivity of GOx—through the selection of alternative isoforms or the application of protein engineering techniques—may represent a promising future approach.
Four injection solutions, (a) control (saline), (b) 0.10 mg mL−1 BPmoc-Ins-Asp, (c) 0.10 mg mL−1 BPmoc-Ins-Asp + 4.0 μg mL−1 GOx, and (d) 0.10 mg mL−1 Ins-Asp, were administered to Wistar rats daily for 5 days. Skin and muscle samples from the injection site were collected 2 h post-final administration and stained with haematoxylin and eosin (HE). Histological evaluation revealed no signs of inflammation, necrosis, or other structural abnormalities in skin or muscle tissue compared to the control group, as shown in Fig. 6(A).
To evaluate toxicity in in vitro systems, BPmoc-Gly, a glycine derivative modified with a BPmoc group, was synthesised to address the possibility that the Ins-Asp produced during prodrug activation could mask the potential toxic effects. Since glycine is present in the culture medium at a concentration of 400 μM, the additional glycine released by BPmoc-Gly activation (0.10 μM) was considered negligible. Analysis of the reaction of BPmoc-Gly with H2O2 using FAB-MS confirmed the production of glycine ([M+H]+ = 76) and quinone methide ([M+H]+ = 107), as evidenced by peaks corresponding to these masses in Fig. S7.
The toxicity associated with the activation of the prodrug formulation was evaluated using a WST-8 assay system with HaCaT cells, a type of keratinocyte.51,52 The results in Fig. 6(B) showed that doxorubicin, used as a positive control, inhibited cell proliferation in a concentration-dependent manner. In contrast, no significant differences in proliferation were observed in HaCaT cells treated with the prodrug formulation or other associated compounds.53
Taken together, the histological analysis and in vitro assays indicate that the formulation containing BPmoc-Ins-Asp and GOx, as well as its degradation products, does not induce significant toxicity, supporting its potential for safe therapeutic applications.
Although this study was limited to short-term evaluations, no signs of toxicity were observed within the experimental timeframe. However, comprehensive long-term safety assessments will be essential to support clinical translation. While GOx is commonly used in glucose-responsive insulin systems, its toxicological profile remains largely uncharacterized, making this study a notable contribution in that regard. Given that GOx is a non-human protein, potential immunogenicity—including allergenicity—should be carefully assessed. If immunological responses are detected, strategies such as PEGylation may be necessary to mitigate immunoreactivity. Furthermore, long-term safety concerns associated with the generation of quinone methide intermediates and the presence of boronic acid derivatives, including their potential for cumulative exposure, should be addressed in future studies.
Furthermore, histological and cell proliferation assays indicated no significant toxicity from BPmoc-Ins-Asp, GOx, or their degradation products, supporting the formulation's safety. While the formulation successfully functioned as designed, activation kinetics require further optimisation to achieve a faster glucose-lowering effect comparable to rapid-acting insulin analogues.
Overall, this study highlights the potential of a single-dose, glucose-responsive insulin formulation as a safer alternative to conventional insulin therapies. By improving activation kinetics, this approach could offer a practical solution for reducing the risk of insulin-induced hypoglycaemia, advancing diabetes management.
The supplementary information includes synthetic procedures, LC-MS analyses, kinetic studies, glucose monitoring data, and analytical profiles. See DOI: https://doi.org/10.1039/d5sc02817e.
This journal is © The Royal Society of Chemistry 2025 |