Engineering a Glucose‐Responsive Glucagon Prodrug Through Arginine–Phenylboronic Acid Pendant Modification
Emily L. DeWolf, Weike Chen, Bernice Webber, Elizabeth M. Power, Rory Kilmer, Pradeep Kadu, Sijie Xian, Matthew J. Webber
TL;DR
A glucose-sensitive glucagon prodrug was developed to automatically release the drug when blood sugar is low, offering protection against hypoglycemia.
Contribution
A modular, glucose-responsive prodrug design using arginine–phenylboronic acid pendants for self-regulated glucagon delivery.
Findings
The prodrug forms aggregates at high glucose and dissolves when glucose levels drop, triggering glucagon release.
The lead design with five arginine–PBA repeats showed optimal glucose-responsive behavior and rescued mice from hypoglycemia.
The system provides prophylactic protection in a diabetic mouse model of insulin overdose without external carrier systems.
Abstract
Activatable prodrug strategies offer powerful means to control therapeutic presentation in space and time. Here, we report a single‐molecule prodrug design that enables glucose‐responsive activation of a glucagon analog for hypoglycemia protection. The system conjugates dasiglucagon with a synthetic pendant comprised of alternating arginine and phenylboronic acid (PBA) units, designed to couple peptide solubility to glucose concentration. The pendant modulates net charge through glucose‐dependent PBA–diol complexation, driving aggregation under normoglycemia and solubilization under hypoglycemia. The lead pendant contains five arginine–PBA repeats and exhibits optimal glucose‐responsive solubility and charge modulation, forming aggregates at high glucose and dissolving as glucose levels decline. Despite a modest reduction in receptor potency relative to native dasiglucagon, this…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
FIGURE 1
FIGURE 2
FIGURE 3
FIGURE 4
FIGURE 5Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsSupramolecular Self-Assembly in Materials · Supramolecular Chemistry and Complexes · Chemical Synthesis and Analysis
Introduction
1
Biopharmaceutical therapies are often limited by short circulating half‐lives, off‐target effects, or a lack of mechanisms to dynamically modulate activity in response to changing physiological needs [1, 2, 3, 4]. Incorporating activatable prodrug designs into peptide and protein therapeutics offers a strategy to control when, where, and how a therapeutic is presented [5]. Through prodrug engineering, peptides and proteins can be coupled to relevant biomarkers or disease indicators, thereby enabling site‐ or condition‐specific activity [6]. For example, tissue‐targeting prodrugs have been designed to remain inactive in systemic circulation and then activate within the disease environment, such as a tumor, reducing off‐target toxicity [7, 8, 9]. Other designs achieve temporal control by altering absorption, pharmacokinetics, or clearance mechanisms through prosthetic modifications [10, 11, 12]. More recently, metabolite‐responsive prodrugs have been developed to sense biomarkers such as oxygen, pH, or glucose, thereby providing on‐demand and self‐regulating therapeutic function [13, 14, 15, 16].
Among metabolite‐responsive drug delivery systems [17], glucose‐responsive strategies have been especially well studied due to their relevance in diabetes management [18, 19, 20, 21]. This work has largely focused on materials and formulations that sense glucose and trigger insulin release. Phenylboronic acid (PBA) chemistry has been the dominant synthetic approach: these Lewis acids form reversible, dynamic‐covalent bonds with cis−1,2 diols such as glucose, stabilizing the anionic tetrahedral boronate state [22]. Leveraging this property, insulins have been modified with PBA moieties to alter hydrophobicity upon glucose binding [23], while PBA–diol crosslinks have been incorporated into insulin‐encapsulating materials for glucose–triggered bond disruption and drug release [24, 25, 26, 27, 28]. Depot systems prepared by coupling electrostatic interactions with glucose‐responsive PBA–diol bonds have recently emerged, wherein hyperglycemic glucose levels disrupt the complexes and promote insulin solubilization and release [29, 30]. PBA chemistry has also been used in glucagon delivery systems, where glucose dissociation from PBA motifs at low concentrations triggers carrier solubilization and glucagon release [31, 32, 33]. Other work has explored the use of PBA interactions for glucose‐dependent delivery of glucagon transdermally using a hydrogel microneedle patch [34]. Although these systems show promise, they often rely on multi‐component gels or carriers, complicating therapeutic design and carrying drawbacks from slow response to drops in glucose and leakage of the therapeutic when not needed.
Here, a single‐molecule prodrug design strategy is explored for glucose‐responsive glucagon analogs to provide hypoglycemia protection. This approach employs a synthetic pendant chain of repeating arginine and PBA units, enabling the net charge of the peptide to shift according to glucose concentration (Figure 1). Under low glucose, PBA groups favor their unbound neutral state, imparting a net‐positive charge that enhances solubility and functional activity. In diabetic mice, this design conferred protection against hypoglycemia in an insulin overdose model. These findings establish proof‐of‐concept for a molecularly engineered, glucose‐responsive glucagon prodrug that provides on‐demand activation under low blood glucose (BG). More broadly, this work highlights the potential of single‐molecule prodrug strategies for spatiotemporal control and metabolite‐responsive regulation of therapeutic peptides.
A glucose‐responsive glucagon prodrug is engineered with alternating arginine and phenylboronic acid (PBA) pendants that couple aggregation and bioavailability to blood glucose levels. At normoglycemia, PBA–glucose bonding promotes molecular aggregation and charge neutrality, stabilizing the complex as a subcutaneous depot. Under hypoglycemia, glucose dissociation from PBA increases net charge, leading to complex dissolution and glucagon release. In diabetic mice, this dynamic equilibrium provides glucose‐responsive protection, enabling on‐demand drug absorption to mitigate hypoglycemia.
Experimental Methods
2
Peptide Synthesis and Purification
2.1
Peptides were synthesized on 0.1–0.25 mmol scales using Fmoc‐based solid‐phase peptide synthesis through both manual coupling steps and automated additions using a CEM Liberty Blue microwave synthesizer. TentaGel R‐RAM resin (0.2 mmol/g, 170 mesh) was swollen in N,*N‐*dimethylformamide (DMF) before initial Fmoc‐deprotection with 20% (v/v) 4‐methylpiperidine in DMF. Coupling steps were performed under microwave conditions with Fmoc‐protected amino acids (0.2 M), diisopropylcarbodiimide (DIC), and Oxyma in DMF, with double couplings performed as needed. The main dasiglucagon sequence was synthesized first, with a Boc‐protected histidine residue added at the N‐terminal position. Following this, selective deprotection was performed on an Mtt‐protected lysine residue substituted for the tryptophan normally in position 25 of dasiglucagon using trifluoroacetic acid (TFA, 3% v/v) in dichloromethane (DCM). An Fmoc‐PEG_5_‐COOH was then added to the ϵ‐amine, followed by repeats of arginine and Fmoc‐Lysine(Mtt)‐OH repeats. A final arginine was then added to the N‐terminus of the pendant, after which the N‐terminal amine was acetylated with acetic anhydride. The acid‐labile Mtt protecting groups were then removed from the pendant Lys(Mtt) residues, and 4‐carboxyphenylboronic acid was added under standard coupling conditions. This step was performed at the end of the synthesis to ensure PBA groups were not exposed to degrading conditions from the 4‐methylpiperidine used in Fmoc removal. After completing the synthesis, the peptide was cleaved from the resin using a mixture of water, triisopropylsilane (TIPS), and TFA (2.5:2.5:95, v/v/v) for 3 h at room temperature. The resin was washed with DCM and combined with the TFA mixture in a round‐bottom flask before being concentrated under vacuum. The concentrated oil was added dropwise to cold diethyl ether, then centrifuged. The resulting pellet was washed three times with diethyl ether, then left to dry overnight. The crude product was dissolved in deionized (DI) water with 0.1% TFA and purified on reversed‐phase Biotage Isolera with a Sfär Bio Duo C_18_ flash cartridge (25 g) with a gradient of DI water to acetonitrile, each containing 0.1% TFA. Absorbance was monitored at 254 and 280 nm for elution collection, and fraction purity was determined via electrospray ionization mass spectrometry (Advion), high‐performance liquid chromatography, and ^1^H NMR (Figures S1–S7). The final peptides were labeled as DA‐PBA[n], where [n] denoted the number of arginine–Lys(PBA) repeats in the pendant.
Glucose‐Dependent Turbidity Measurements
2.2
Turbidity measurements were performed on DA‐PBA[n] peptide solutions in (4‐(2‐hydroxyethyl)‐1‐piperazineethanesulfonic acid (HEPES) buffer (0.25% w/v, pH 7.4, 20 mM) across 0, 25, 50, 100, and 200 mg/dL glucose concentrations. To prepare the samples, DA‐PBA[n] stock solutions were prepared in DI water (0.5% w/v), then diluted in HEPES buffer (pH 7.4, 40 mM) with 2x the final desired glucose concentration. Turbidity was measured by absorbance at 540 nm in triplicate on the Tecan Infinite M200 multimode plate reader in black clear‐bottom 96‐well plates over the course of 1 h at room temperature.
Zeta Potential and Size
2.3
DA‐PBA[n] samples were solubilized in DI water (0.5 mg/mL), then diluted to 0.25 mg/mL in HEPES buffer (pH 7.4, 40 mM) with 2x of the desired glucose condition (0, 50, 100, 200, or 400 mg/dL glucose). The zeta potential was measured in triplicate using a Malvern Zetasizer with a Huckel model at 25°C. Size of the insoluble aggregates was evaluated using dynamic light scattering (DLS) on the Malvern Zetasizer with the same sample preparation methods as above.
Aggregate Visualization
2.4
Aggregate visualization was performed on 0.25 mg/mL DA‐PBA5 solutions in HEPES buffer (pH 7.4, 20 mM) across 0, 25, 50, 100, and 200 mg/dL glucose concentrations with Nile Red staining, using the RFP cube of an EVOS‐Auto microscope with a 20x objective. To prepare the samples, a 0.5 mg/mL peptide stock solution was made in DI water, then diluted 1:1 in 2x HEPES buffer (pH 7.4, 40 mM) prepared with different glucose levels (0, 50, 100, 200, or 400 mg/dL glucose). The samples were incubated for 30 min before Nile Red (10 μM) was added to stain peptide aggregates. Five images from different regions of each sample were captured in both brightfield and fluorescent channels to visualize aggregate formation across glucose concentrations. Images were processed using ImageJ.
In Vitro Analog Activity
2.5
A calcium mobilization assay in HEK293/GCGR/Gα15 cells (GenScript) was used to evaluate the in vitro potency of DA‐PBA5 compared to dasiglucagon. Cells were plated in clear‐bottom black 384‐well plates at 25k cells/well in 50 μL of media with poly‐D‐lysine (2 μg/mL) to improve adherence and incubated overnight at 37°C with 5% CO_2_. After incubation, the plate was washed with 20 mM HEPES in HBSS, and 20 μL of loading dye buffer containing Fluoforte (Enzo), sulfinpyrazone, and Pluronic F‐127 was added to each well. The plate was returned to the incubator for 45 min to allow for dye uptake into the cells. During this time, the agonist plate was prepared with glucose containing buffers and sample serial dilutions at 3x the desired final concentration. Following 45 min of incubation, the cell plate was added to the reading chamber, the agonist plate was added to the source chamber, and fresh tips were added to the tips chamber on the FlexStation 3 (Molecular Devices). Using flex mode, fluorescence measurements from the wells were taken every 3.04 s over 2.5 min. At 17 s, 20 μL of 3x glucose buffer was transferred from the agonist plate to the cell plate at 8 μL/s, and a baseline was established for 52 s. At 69 s, 20 μL of 3x sample solution was transferred from the source plate to the cell plate at 8 μL/s, and the resulting fluorescence from calcium mobilization was recorded for 80 s. For data analysis, the fluorescence readings of negative control wells containing cells, the tested glucose concentration, and buffer were subtracted from experimental wells treated with DA‐PBA5 or dasiglucagon. Experimental data were reduced using (v max − v min) to give the maximum fluorescence from the calcium spikes. Reduced data were then normalized based on a positive control of wells that had been treated with Triton X‐100.
In Vivo Hypoglycemia Protection
2.6
Prophylactic hypoglycemia protection of DA‐PBA5 was evaluated using our previously established hypoglycemia mouse model [35]. These studies were detailed in a protocol approved by the University of Notre Dame Animal Care and Use Committee (Assurance of Compliance #A3093‐01) and adhered to all relevant Institutional, State, and Federal guidelines. Briefly, diabetes was induced in male C57BL6/J mice, aged 8 weeks, with a 150 mg/kg dose of streptozotocin (STZ). After 12 d, mice with non‐fasted BG levels over 600 mg/dL were selected for study. After an 8 h fast, mice with BG > 450 mg/dL were given 0.5 IU/kg basal insulin detemir (Levemir, Novo Nordisk) via subcutaneous (s.c.) injection in a total volume of 100 μL. After 4 h, as BG levels returned to a normal range (∼180 mg/dL), mice were randomly divided into 3 groups (n = 6/group) and treated with buffer, dasiglucagon (138 nmol/kg), or DA‐PBA5 buffer (138 nmol/kg), via a 100 μL s.c. injection. All samples were administered in a 20 mM HEPES buffer with 100 mg/dL glucose. BG levels were monitored after the treatment (t = 0 min). After 2 h, hypoglycemia was triggered in the mice with an i.p. injection of AOF recombinant human insulin (Gibco) at a dose of 3 IU/kg in 100 μL of saline. BG levels were monitored for another 4 h. During the process, mice exhibiting “high” readings were noted as a BG value of 600 mg/dL, while “low” readings were noted as a BG value of 20 mg/dL (also indicated dead).
Results and Discussion
3
Molecular Design
3.1
In systems using PBA chemistry for insulin delivery, glucose binding induces competitive displacement to trigger insulin release [29, 30]. By contrast, glucagon delivery systems reported thus far have relied on binding‐dependent charge modulation to alter assembly states of a carrier material in response to glucose levels [31, 32, 33]. These designs for glucagon delivery have been limited by baseline leakage, slow responsiveness to declining glucose concentrations, and a high carrier‐to‐drug ratio, with the therapeutic comprising only a small fraction (≤10% dry weight) of the total material. To overcome these challenges, a single‐molecule prodrug strategy was explored here, conjugating an alternating Arg–PBA pendant to dasiglucagon. This design directly couples the net charge of the drug and its concomitant solubility to ambient glucose concentration (Figure 1).
Dasiglucagon was chosen as the starting scaffold due to its established stability and resistance to amyloid formation, overcoming key limitations of native glucagon [36]. A modified sequence was prepared in which Trp^25^ was replaced with Lys(Mtt), a site shown previously in native glucagon to tolerate modification without loss of function [37]. Selective deprotection of Lys(Mtt) enabled installation of a PEG_5_ spacer, intended to minimize steric interference with receptor binding [38]. Onto this spacer, repeating [n] Arg–PBA units (n = 1–6) were sequentially attached, followed by capping with a terminal Arg and N‐terminal acetylation. On‐resin Mtt deprotection and subsequent PBA coupling yielded a small library of DA‐PBA[n] prodrug conjugates (Figure 2A).
(A) Sequence modifications to dasiglucagon to create DA‐PBA[n] where [n] is the number of repeating Arginine–Lys(PBA) groups on the pendant chain conjugated via a PEG5 linker to W25K. (B) Glucose‐dependent turbidity, quantified by absorbance at 540 nm, and (C) zeta potential measurements of DA‐PBA[n] for n = 1–6 evaluated across a range of glucose concentrations from 0 to 200 mg/dL, error bars depict standard deviation (SD) ( = p < 0.05).*
Whereas unmodified dasiglucagon carries a net charge of –2, the pendant modifications were designed to render the molecule responsive to glucose through tunable charge states. PBA groups undergo an equilibrium shift from a neutral trigonal form to an anionic tetrahedral form upon glucose binding [39], while adjacent Arg residues confer positive charge to ensure solubility under glucose‐depleted conditions. In combination, these features were expected to drive a transition from a near‐neutral, insoluble peptide under normoglycemia to a net‐positive, soluble peptide under hypoglycemia. Although the amide‐linked PBA used here typically exhibits a pK _ a _ of ∼8.4 [40], limiting binding at physiological pH of 7.4, proximal Arg residues are known to lower the effective pK _ a _ and enhance glucose recognition under physiological conditions [32, 41]. Accordingly, a DA‐PBA[n] library (n = 1–6) was synthesized to systematically probe how pendant composition and charge balance influence glucose‐responsive function.
Functional Outcome of Pendant Length
3.2
Aggregation was assessed by turbidity measurements (light scattering at 540 nm) across a physiologically relevant glucose range (0–200 mg/dL) for the full DA‐PBA[n] series (Figure 2B). DA‐PBA1, DA‐PBA2, and DA‐PBA3 analogs became increasingly soluble with rising glucose, opposite to the intended design, likely due to the added arginine residues being insufficient to counterbalance the net‐negative charge of the peptide. DA‐PBA4 showed modest improvement, with enhanced solubility under low‐glucose conditions, as would be desirable for this application, though the overall glucose‐dependent changes were limited. In contrast, DA‐PBA5 displayed pronounced glucose‐dependent aggregation, with similar trends observed for DA‐PBA6. Accordingly, for pendants of n ≥ 5, the installed charges are able to effectively neutralize the inherent net‐negative charge of dasiglucagon under normoglycemic conditions, rendering the peptide insoluble, while also facilitating solubilization under low glucose conditions.
Consistent results were obtained from zeta potential analysis (Figure 2C). DA‐PBA1 through DA‐PBA4 maintained negative zeta potentials across all glucose concentrations tested (0–200 mg/dL). DA‐PBA5 exhibited a positive zeta potential in the absence of glucose that approached neutrality as glucose increased, while DA‐PBA6 shifted from positive at low glucose to negative at high glucose. Together, these results confirmed that pendants of n ≥ 5 are necessary to overcome the negative charge of the dasiglucagon scaffold and confer glucose‐dependent solubility. Importantly, DA‐PBA5 demonstrated a broader dynamic range upon changes in glucose for both turbidity and zeta potential compared to DA‐PBA6. Coupled with the synthetic advantage of a shorter pendant, these findings supported the selection of DA‐PBA5 in all subsequent studies. While it is also expected that this general approach could be applied to other glucagon analogues, such use may entail further optimization of the pendant length and/or charge to account for differences in net charge or isoelectric point of the specific glucagon analogue.
Glucose‐Dependent Aggregation
3.3
With DA‐PBA5 identified as the lead candidate for glucose‐responsive solubility, aggregation behavior was further investigated by kinetic turbidity studies, dynamic light scattering (DLS), and microscopic analysis. These experiments were designed to further probe the process by which glucose‐dependent aggregation of DA‐PBA5 arises from pendant charge modulation, with increasing glucose concentrations expected to increase the extent of PBA–glucose binding and shift the overall charge balance toward neutrality (Figure 3A). An Alizarin Red S assay performed on DA‐PBA5 confirmed the formation of PBA–glucose complexes within the pendant chain, supporting the primary mechanism for charge modulation via glucose binding (Figure S9).
(A) A cartoon schematic of the mechanism for glucose‐responsive solubility driven by charge modulation of a the DA‐PBA5 pendant group. With increasing glucose concentrations, more PBA groups are bound to glucose, stabilizing the negative tetrahedral boronate species and neutralizing positive contributions of neighboring arginines. (B) Turbidity kinetics measured at 540 nm in response to glucose levels of 0 mg/dL (black), 25 mg/dL (pink), 50 mg/dL (teal), 100 mg/dL (purple), and 200 mg/dL (lilac). (C) Final turbidity values across glucose concentrations for DA‐PBA5, with one‐way ANOVAanalysis. (D) Size of the aggregates measured using DLS across glucose concentrations. (E) Nile red staining and fluorescence microscopy of aggregates formed across glucose concentrations. Error bars depict SD ( = p < 0.05, *** = p < 0.001, **** = p < 0.0001, and n.s. is not significant).*
Turbidity kinetics (Figure 3B) were monitored for 0.25 mg/mL DA‐PBA5 solutions in 20 mM HEPES buffer with glucose concentrations of 0, 25, 50, 100, and 200 mg/dL. At high glucose (100 and 200 mg/dL), turbidity rose sharply within the first 20 min and plateaued, consistent with rapid and stable aggregate formation. At lower glucose levels (25 and 50 mg/dL), turbidity increased more gradually, reflecting slower aggregation kinetics and a lower plateau level of aggregation. At low glucose, less PBA–glucose binding and a reduced extent of equilibrium complexation likely served to limit aggregation, consistent with a more positively charged and soluble peptide. One‐way analysis of variance (ANOVA) with post hoc multiple comparisons confirmed significant differences among all glucose conditions at the 60‐min endpoint (Figure 3C).
Aggregate size analysis by DLS corroborated these findings, showing a positive correlation between glucose concentration and particle size (Figure 3D). Complementary Nile red staining and fluorescence microscopy further confirmed the presence of larger and more abundant aggregates at higher glucose concentrations (Figure 3E).
Together, turbidity, DLS, and microscopy established that DA‐PBA5 undergoes glucose‐dependent aggregation driven by pendant charge modulation through PBA–glucose complexation and the equilibrium extent of binding for this interaction. Notably, the rapid formation of large, stable aggregates under elevated glucose conditions suggests the potential for DA‐PBA5 to form depot‐like precipitates upon subcutaneous administration, which could resolubilize for systemic absorption as glucose levels fall and the equilibrium in PBA–glucose bonding is shifted to the unbound—and less anionic—state.
In Vitro Signaling Potency
3.4
To assess whether DA‐PBA5 retained activity despite its large pendant modification, peptide secondary structure was first evaluated with circular dichroism to assess glucose‐dependent changes in the α‐helical secondary structure of the dasiglucagon peptide (Figure S8). The modified peptide retained the typical α‐helical structure of dasiglucagon; some glucose‐induced shift was observed though these results were confounded by peptide aggregation in the presence of glucose. Bioactivity on its cognate receptor was also compared to unmodified dasiglucagon using a calcium (Ca^2+^) mobilization assay (Figure 4). In this assay, stimulation of the glucagon receptor (GCGR) on engineered cells induces Ca^2+^ release from the endoplasmic reticulum, with the magnitude of the resulting spike providing a direct readout of agonist activity (Figure 4A). Fluoforte dye was employed as the reporter due to its weak Ca^2+^ affinity, enabling rapid binding–unbinding cycles and enhanced sensitivity for kinetic measurements [42].
(A) A schematic of the signaling pathway activated by the stimulation of the glucagon receptor (GCGR) in HEK293/GCGR/Gα15 cell line. Overexpression of the Gα15 G‐protein increases calcium (Ca2+) mobilization after GCGR stimulation. Fluoforte‐AM® dye passively enters the cell where enzymes in the cytosol cleave the acetoxymethyl ester, revealing a negatively charged dye that fluoresces when bound to Ca2+. (B) Relative fluorescence normalized by Ca2+ mobilization of a Triton X‐100 positive control for dasiglucagon at 0 mg/dL (black), dasiglucagon at 0 mg/dL (pink), DA‐PBA5 at 0 mg/dL (teal), and DA‐PBA5 200 mg/dL glucose (purple). Curves represent at least 4 biological replicates with a minimum of 3 technical replicates. A 95% CI of a dose–response fit is shaded.
Activities of dasiglucagon and DA‐PBA5 were measured at 0 and 200 mg/dL glucose to evaluate potential glucose‐dependent effects. Pendant modification reduced potency by approximately ten‐fold relative to unmodified dasiglucagon (Figure 4B). Neither peptide displayed significant glucose‐dependent differences in activity, consistent with the dilute assay conditions where glucose‐directed aggregation of DA‐PBA5 is limited. Under these conditions, receptor activation is dictated primarily by peptide folding and steric effects, rather than peptide aggregation state. Importantly, these findings do not diminish the translational promise of DA‐PBA5, as its intended mechanism of action relies on glucose‐responsive subcutaneous depot formation at higher concentration, rather than glucose‐modulated receptor activation.
Prophylactic Hypoglycemia Protection
3.5
Having confirmed that DA‐PBA5 retains an active secondary structure and ability to activate its cognate receptor despite pendant modification, its capacity to prevent hypoglycemia was evaluated in a STZ‐induced diabetic mouse model designed to simulate prophylactic glucagon administration prior to insulin overdose [35]. Fasted mice with severe hyperglycemia (>450 mg/dL) were first administered basal insulin detemir to lower BG to ∼180 mg/dL, establishing a normoglycemic baseline before treatment (Figure 5A). Mice were then treated subcutaneously (100 μL) with buffer, dasiglucagon, or DA‐PBA5 at an equimolar dose (138 nmol/kg). Unlike prior material‐based responsive glucagon platforms, which exhibited early leakage and burst release [31, 35], all treatment groups here showed only a modest transient BG increase comparable to buffer, likely reflecting handling and injection stress. This suggests that the single‐molecule DA‐PBA5 approach may mitigate early leakage commonly observed in gel‐ and carrier‐based systems. Two hours later, hypoglycemia was induced via insulin overdose, and BG was monitored for 4 h (Figure 5B). The nadir BG values highlighted improved protection from DA‐PBA5 treatment (65 ± 10 mg/dL; mean ± standard error of the mean (SEM)) compared to dasiglucagon (38 ± 7 mg/dL). Mortality outcomes further underscored this benefit: no mice treated with DA‐PBA5 (0/6) succumbed to hypoglycemia, whereas 16.7% (1/6) of animals in both the dasiglucagon and buffer groups did. By 4 h post‐challenge (Figure 5C), BG values in the DA‐PBA5 group (158 ± 36 mg/dL) were significantly higher than in dasiglucagon (61 ± 10 mg/dL) or buffer (72 ± 11 mg/dL) groups. These results highlight how, without a mechanism to form a depot, dasiglucagon is rapidly cleared and ineffective as a prophylactic agent against a subsequent insulin challenge. Despite a ∼10‐fold reduction in in vitro potency relative to unmodified dasiglucagon, equimolar DA‐PBA5 dosing provided markedly enhanced hypoglycemia protection. Improved function in vivo may also arise from weakened affinity to the glucagon receptor, leading to lower receptor‐mediated clearance and longer circulation times [43]. These results support the mechanism of a glucose‐responsive depot, in which DA‐PBA5 aggregates under hyperglycemia and resolubilizes under falling glucose levels to deliver effective, on‐demand glucagon action.
(A) The full in vivo experimental model, depicted also in the cartoon, to assess prophylactic hypoglycemia protection with DA‐PBA5. After an 8 h fast, STZ diabetic mice were administered insulin detemir (t = −240 min) to stabilize blood glucose within a normal range. Treatments of buffer, dasiglucagon, or DA‐PBA5 were then administered (t = 0), and blood glucose was monitored. An insulin overdose was induced 2 h after treating, and blood glucose changes were monitored for an additional 4 h. A dotted line drawn at 60 mg/dL marks the threshold used for mild hypoglycemia. (B) A zoomed in view of blood glucose measurements during the period of hypoglycemia, with arrows denoting observed deaths. (C) The final blood glucose measured at 360 min compared between groups. Each treatment group was n = 6 mice, error bars denote SEM for each group, and statistical analysis was performed using ANOVA with multiple comparisons post hoc testing.
Conclusions
4
This study introduces a glucose‐responsive glucagon prodrug that achieves self‐regulated, on‐demand activation through a single‐molecule design. Pendants integrating arginines with PBA‐mediated glucose sensing achieved glucose‐dependent charge modulation and transitions between aggregated and soluble forms of the peptide therapeutic. Pendant composition was shown to govern glucose‐dependent charge balance, while in vivo evaluation demonstrated effective protection against insulin‐induced hypoglycemia in diabetic mice. Importantly, this performance was achieved without polymeric matrices, carriers, or multi‐component assemblies that may complicate translation of other previously described glucose‐responsive glucagon delivery platforms.
The simplicity and tunability of this design highlight its potential for next‐generation, metabolite‐responsive peptide and protein prodrugs. The modular pendant chain offers a blueprint for tuning sensitivity and biological compatibility, where PBA motifs could be replaced or combined with other metabolite‐sensing chemistries to couple activity to disease biomarkers. Overall, DA‐PBA5 establishes proof‐of‐concept for a glucose‐responsive prodrug capable of mitigating severe hypoglycemia hours after administration, representing a step toward safer, long‐lasting protection against sudden hypoglycemia in individuals with type 1 diabetes.
Supporting Information
Additional supporting information can be found online in the Supporting Information section. Supporting Fig. S1: DA‐PBA[n] chemical structure. Supporting Fig. S2: Intermediate DA‐PBA5 chemical structure. Supporting Fig. S3: ESI‐MS of intermediate DA‐PBA5 (positive mode); calculated 5220 [M], observed 427.6 [M+14H+5ACN +TFA]13+, 490.6 [M+12H+ACN+TFA]11+, 555.6 [M+11H+5ACN+TFA]10+. Supporting Fig. S4: DA‐PBA5 chemical structure. Supporting Fig. S5: ESI‐MS of DA‐PBA5 (positive mode); calculated 5959.6 [M], observed 738.95 [M+8H‐3H2O]8+, 1065.2 [M+9H+2ACN+3TFA]6+, 1234.2 [M+7H‐H2O+2TFA]5+, 1969.35 [M+3H‐3H2O]3+. Supporting Fig. S6: Analytical HPLC of DA‐PBA5 with UV absorbance monitored at 280 and 254 nm. Supporting Fig. S7: (A) ^1^H NMR of DA‐PBA5 (Red) and dasiglucagon (Black) in DMSO‐d_6_ at 400 MHz. (B) New peaks in the aromatic region are attributed to PBA groups, and (C) upfield changes support increased arginine content and PEG_5_ incorporation in DA‐PBA5 as compared with the Dasiglucagon standard. Supporting Fig. S8: Alizarin Red S glucose displacement assay of DA‐PBA5. Supporting Fig. S9: Circular Dichroism of 40 μM DA‐PBA5 at 0, 25, 50, 100, and 200 mg/dL Glucose.
Author Contributions
Emily L. DeWolf, Weike Chen, and Matthew J. Webber conceived of ideas and designed the studies. Emily L. DeWolf, Weike Chen, Bernice Webber, Elizabeth M. Power, Rory Kilmer, Pradeep Kadu, and Sijie Xian conducted experiments and contributed to data collection and analysis. Emily L. DeWolf and Matthew J. Webber wrote the manuscript.
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Supplementary Material
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1R. Zaman , R. A. Islam , N. Ibnat , et al., “Current Strategies in Extending Half‐Lives of Therapeutic Proteins,” Journal of Controlled Release 301 (2019): 176–189.30849445 10.1016/j.jconrel.2019.02.016 · doi ↗ · pubmed ↗
- 2U. Binder and A. Skerra , “Strategies for Extending the Half‐Life of Biotherapeutics: Successes and Complications,” Expert Opinion on Biological Therapy 25 (2025): 93–118.39663567 10.1080/14712598.2024.2436094 · doi ↗ · pubmed ↗
- 3B. Leader , Q. J. Baca , and D. E. Golan , “Protein Therapeutics: A Summary and Pharmacological Classification,” Nature Reviews Drug Discovery 7 (2008): 21–39.18097458 10.1038/nrd 2399 · doi ↗ · pubmed ↗
- 4D. S. Dimitrov , “Therapeutic Proteins,” Methods in Molecular Biology 899 (2012): 1–26.22735943 10.1007/978-1-61779-921-1_1PMC 6988726 · doi ↗ · pubmed ↗
- 5P. Barman , S. Joshi , S. Sharma , S. Preet , S. Sharma , and A. Saini , “Strategic Approaches to Improvise Peptide Drugs as Next Generation Therapeutics,” International Journal of Peptide Research and Therapeutics 29 (2023): 61.37251528 10.1007/s 10989-023-10524-3PMC 10206374 · doi ↗ · pubmed ↗
- 6D. Wang , F. Yin , Z. Li , Y. Zhang , and C. Shi , “Current Progress and Remaining Challenges of Peptide–Drug Conjugates (PD Cs): Next Generation of Antibody‐Drug Conjugates (AD Cs)?,” Journal of Nanobiotechnology 23 (2025): 305.40259322 10.1186/s 12951-025-03277-2PMC 12013038 · doi ↗ · pubmed ↗
- 7K. R. Wong , E. Menendez , C. S. Craik , W. M. Kavanaugh , and O. Vasiljeva , “In Vivo Imaging of Protease Activity by Probody Therapeutic Activation,” Biochimie 122 (2016): 62–67.26546838 10.1016/j.biochi.2015.11.003PMC 5709043 · doi ↗ · pubmed ↗
- 8I.‐J. Chen , C.‐H. Chuang , Y.‐C. Hsieh , et al., “Selective Antibody Activation through Protease‐Activated pro‐Antibodies that Mask Binding Sites with Inhibitory Domains,” Scientific Reports 7 (2017): 1–12.28912497 10.1038/s 41598-017-11886-7PMC 5599682 · doi ↗ · pubmed ↗