Biocompatible Nanogels Synthesized via RAFT Polymerization for Long-Acting Onset-Controllable Glucose-Responsive Insulin Delivery in Diabetic Mice

Abstract

Glucose-responsive insulin delivery (GRID) nanocarriers that can respond to glucose fluctuations rapidly and deliver insulin autonomously for long-acting normoglycemic control with low risk of hypoglycemia are highly desirable but still unattained for diabetic care. Herein, we report a class of biocompatible nanogels with controllable onset of glucose response for self-regulated insulin delivery. The narrowly distributed nanogels are prepared via controlled polymerization using the poly[oligo(ethylene glycol) methyl ether methacrylate (M_n = 500)] (pMEO₉MA) as macro-chain transfer agents copolymerized with 4-vinylphenylboronic acid (VPBA) and a crosslinker. The nanogels can be freeze-dried and easily redispersed in an aqueous phase with excellent colloidal stability. The pMEO₉MA chain length and pMEO₉MA/VPBA ratio enable tuning both particle size and the onset of glucose responsiveness of the nanogels. A single injection of the insulin-loaded nanogels can maintain normoglycemia in diabetic mice for up to 18 h. Importantly, the dose increase in the nanogel-insulin prolongs the duration of normoglycemia but does not cause hypoglycemia in diabetic mice, benefiting from the reversible on-off insulin delivery at the desirable glucose threshold. The nanogels exhibit no cytotoxicity or organ toxicity after repeated dosing in vivo. These results highlight great promise for developing swelling-based nanogels for long-acting GRID systems.

1 |. Introduction

Diabetes mellitus, a chronic metabolic disorder characterized by persistent hyperglycemia, remains one of the biggest health threats due to its high prevalence and related complications such as heart diseases, kidney failure, and neuropathy. The number of diabetic patients is projected to increase from 589 million in 2024 to 853 million by 2050 [1]. Current treatment typically involves finger-prick blood glucose level (BGL) tests followed by insulin injections multiple times daily. This procedure not only severely limits patient compliance and quality of life, but also impairs dosing accuracy due to variability in diet, activity, and metabolic state, and thus increases the risk of life-threatening hypoglycemia [2–4]. To address these challenges, significant efforts have been devoted to developing glucose-responsive insulin delivery (GRID) systems that can mimic pancreatic β-cell functionality by sensing real-time BGLs and autonomously delivering the right amount of insulin [5–12]. Glucose oxidase (GOx) or glucose-binding concanavalin A have been encapsulated within gels, vesicles, or microneedle patches for GRID platforms [13–26]. However, protein-based systems face significant drawbacks, including high cost, limited stability, and immunogenicity. As a result, synthetic phenylboronic acid (PBA) moieties with high stability, durable glucose-sensing ability, and modular design potential have been extensively explored as alternatives for designing GRID systems [4–6, 27–44].

PBA exists in an equilibrium between the undissociated trigonal-planar form and the dissociated anionic tetrahedral boronate form, but only the latter can stably bind cis-diol molecules. Thus, fluorinated-PBA moieties or neighboring B^δ−⋯N^δ+ interactions have been applied to lower the pK_a of PBA to enhance the glucose-PBA binding affinity at physiological pH [38–47]. The glucose-PBA binding specificity can be improved by engineering a diboronate motif or glucose molecular imprinting in gels [35, 36, 48, 49]. The insulin molecules embedded in the PBA-based GRID systems are released primarily via two mechanisms. One is the competitive glucose-PBA binding to displace the original diol-PBA bindings or reverse the electrostatic attractions, disintegrating the GRID macrostructures to release insulin [35–45, 50–55]. Based on this mechanism, some long-acting GRID systems have been recently developed with insulin-complexed polymer precipitates or gels serving as subcutaneous insulin reservoirs or depots [35–38, 44]. Complex particle-based GRID systems were also developed. However, most of these complex particle nanocarriers could only regulate normoglycemia in diabetic mice for 6–8 h, possibly due to limited lifetime in the circulation system related to large size or glucose-induced fast irreversible disassembly [23, 24, 39, 40, 42, 54, 55]. Only a few colloidally stable complex nanocarriers with particle size ~100 nm or below displayed long-acting times of 13 and 20 h at an insulin dose of 80 and 115 IU/kg, respectively [41, 43]. Another mechanism involves glucose-PBA binding on polymer gel network chains, producing negative charges to swell the gels for insulin to diffuse out [4, 30–34, 46, 47]. Porous PBA-polymer gels offer several advantages, including high loading capacity and reversible glucose-responsive structure [29–34, 47]. In addition, no chemical modification of insulin is required in gel carriers. Small gel particles with real-time glucose sensing ability and injectability into the circulatory system are particularly appealing. To achieve a long circulation lifetime, a particle size of 10–150 nm is required to avoid rapid clearance by the kidneys and the reticuloendothelial system (RES) [56, 57]. However, previous PBA-based gel particles synthesized by the surfactant-stabilized precipitation polymerization method typically have a hydrodynamic size of 200–400 nm [47–49, 58–62], which limits their potential for in vivo study. As a result, few in vivo GRID test results are available for the swelling-based PBA-gel nanoparticle systems. Furthermore, most of these PBA-based GRID systems begin to swell at low glucose levels, increasing the risk of premature insulin release and hypoglycemia events.

The reversible addition-fragmentation chain-transfer (RAFT) polymerization method can produce polymers with controlled chain length [63]. On the other hand, the nonlinear poly(ethylene glycol) (pMEO_nMA) polymerized from macromonomers of oligo(ethylene glycol) methyl ether methacrylate (MEO_nMA, with n being the number of repeating ethylene oxide (EO) units) has gained great interest as an alternative to linear poly(ethylene glycol) (PEG) for biomedical applications. pMEO_nMA has tunable hydrophilicity depending on the EO side chain length. pMEO_nMA not only maintains the excellent cytocompatibility, nontoxicity, and protein-repellent properties of PEG, but also provides significantly more synthetic versatility and less immunogenicity than PEG [64]. We recently demonstrated that the amounts of MEO₅MA randomly copolymerized into the PBA-based microgels can shift the onset of glucose response [65]. Combining the precise control afforded by RAFT polymerization with the unique advantages of pMEO_nMA, we herein report the development of a new class of PBA-based polymer nanogels with desirable particle size and tunable onset of glucose responsiveness, enabling long-acting normoglycemic control with a low risk of hypoglycemia in vivo. Specifically, we designed a series of pMEO₉MA@pVPBA nanogels synthesized via RAFT polymerization using pMEO₉MA as macro-chain transfer agents copolymerized with vinylphenylboronic acid (VPBA) in the presence of a crosslinker (Figure 1). Such prepared nanogels are surfactant-free and highly stable in aqueous phase due to the high hydrophilicity of pMEO₉MA chains. A small fraction of 2-(dimethylamino)ethyl acrylate (DMAEA) was used to lower the pKa of VPBA, enhancing the glucose sensitivity at physiological pH. The pMEO₉MA chain dimension and the feeding ratio of pMEO₉MA/VPBA enable fine-tuning of both particle size and onset of glucose response to the physiologically desirable range. The nanogels display narrow size distribution, high loading capacity, and optimal insulin retention and release ability. In vivo studies indicate that diabetic mice treated with the insulin-loaded pMEO₉MA@pVPBA nanogel with hydrodynamic size ~100 nm and onset of glucose response ~5 mM show prolonged normal BGLs and excellent glucose tolerance ability. The increase in the dose of nanogel-formulated insulin prolonged the duration of normoglycemia in diabetic mice but prevented hypoglycemia. The pMEO₉MA-shelled nanogels exhibit no cytotoxicity or organ toxicity after repeated dosing in healthy mice. These results demonstrated that the developed nanogels hold great potential for long-acting GRID systems.

FIGURE 1 |.

Schematic representation of the fabrication of insulin-loaded pMEO₉MA@pVPBA polymer nanogels and their onset-controllable glucose-responsive insulin delivery in diabetic mice.

2 |. Results and Discussion

2.1 |. Composition and Molecular Weight of the pMEO₉MA Macro-RAFT Agents

To obtain pMEO₉MA polymers with a controlled amount of EO units, two macro-RAFT agents of p(MEO₉MA)₁₅ and p(MEO₉MA)₂₀ with degrees of polymerization of 15 and 20 were respectively prepared using the RAFT polymerization technique. To study the EO side (or brush) chain length effect on the glucose-responsive behavior of designed nanogels, two additional macro-RAFT agents of p(MEO₅MA)₂₀ and p(MEO₅MA)₃₀ with shorter side chains of EO₅ on each monomer) were also synthesized, respectively. The FT-IR analysis of the p(MEO₉MA)₁₅ macro-RAFT agent shows the characteristic absorptions at 1037 and 1107 cm⁻¹ from C—O—C stretching, 1727 cm⁻¹ from the C=O stretch in ester, 1349 and 1458 cm⁻¹ from the CH₂/CH₃ bending on the backbone and side chains, and 2875–2950 cm⁻¹ from the C—H stretching of CH₂/CH₃ (Figure S1a), indicating the presence of these functional groups in the p(MEO₉MA)₁₅ chains. The synthesized macro-RAFT polymers were further analyzed by ¹H NMR, which shows the characteristic resonance signals from all the protons in the MEO₉MA and MEO₅MA repeating units and the chain transfer agent (Figure S2), confirming the successful synthesis of the pMEO₉MA and pMEO₅MA macro-RAFT agents. The number-average molecular weight (M_n) of the resulting polymers can be calculated from the comparison of the integral area of the characteristic —OCH₂— proton (labeled as 6) in the MEO₉MA repeating unit at δ = 4.08 ppm and the characteristic —SCH₂— proton (labeled as 1) of the terminal chain transfer agent at δ = 3.18 ppm. The M_n results obtained from NMR analysis for p(MEO₉MA)₁₅, p(MEO₉MA)₂₀, p(MEO₅MA)₂₀ and p(MEO₅MA)₃₀ polymers are not only comparable with the M_n results obtained from the gel permeation chromatography (GPC) analysis (Figure S3), but also close to the theoretical M_n values calculated from the number of monomers fed in the RAFT polymerization reactions (Table S1). The GPC results show that all the prepared macro-RAFT agents have very narrow molecular weight distributions. These results confirm the well-controlled polymer chain length from the RAFT polymerization method.

2.2 |. Compositions of the pMEO₉MA@pVPBA Nanogels

The four pMEO₉MA@pVPBA polymer nanogels were synthesized via RAFT-mediated copolymerization using the pre-synthesized p(MEO₉MA)₁₅ or p(MEO₉MA)₂₀ macro-RAFT agent as polymeric chain transfer agents in the presence of a designed amount of VPBA and DMAEA comonomers and crosslinker in a 95:5 DMF/water mixture. Following the same procedure, two additional nanogels were synthesized from the p(MEO₅MA)₂₀ and p(MEO₅MA)₃₀ macro-RAFT agents to study the influence of EO side chain length on the properties of the nanogels. The sample code and feeding compositions of all six nanogels are listed in Table 1. Unlike the most commonly used precipitation polymerization method that requires a large amount of surfactants as stabilizers for the synthesis of microgels [66], the RAFT polymerization method for the synthesis of nanogels was conducted under controlled conditions with no surfactants involved. It is expected that well-defined polymer nanogels can be produced with the hydrophilic brush-like pMEO₉MA chains anchored on the surface of a crosslinked core composed of pVPBA network chains (Figure 1). To confirm the composition of the resultant nanogels, both FT-IR and ¹H NMR spectra of the synthesized nanogels were collected. Compared to the FT-IR spectrum of the free p(MEO₉MA)₁₅ homopolymer chain precursor (Figure S1a), two features should be noted in the IR spectra of the pMEO₉MA@pVPBA nanogels (Figure S1b,c). First, all the absorption peaks from the pMEO₉MA chains become sharper and stronger in the nanogels, especially in C—O—C (1107 cm⁻¹), C=O (1727 cm⁻¹), and C—H (2875 cm⁻¹) stretches, which can be attributed to the increased vibrational coherence induced by the crosslinked structural rigidity. The crosslinked core of the nanogel reduces the segmental motion and conformational freedom of the pMEO₉MA chains covalently anchored on its surface, which lowers vibrational broadening. Second, a new absorption peak at 1608 cm⁻¹ from the C=C stretching of the phenyl ring was observed, indicating the presence of PBA groups in all the tested nanogels. This peak was more obvious in the nanogel prepared from the pMEO₅MA chains with shorter EO₅ side chains (Figure S1d), possibly because the PBA units in the core were not blocked heavily by the less bulky pMEO₅MA chains. The NMR spectra of the nanogels further confirm the successful incorporation of pVPBA units. Compared to the NMR spectra of pMEO₉MA (or pMEO₅MA) macro-RAFT precursors (Figure S2), the new signals at δ = 6.57 and 7.38 ppm from the nanogels can be assigned to the characteristic resonances of phenyl protons of the PBA component (labeled as 17 and 18, Figure 2a and Figure S4). In addition, a small new peak at δ = 3.98 ppm originating from the characteristic resonances of the —COOCH₂— protons of DMAEA (labeled as 12, Figure 2a and Figure S4) appeared, confirming the presence of a small amount of DMAEA in nanogels. The comparison of the integral area of the characteristic —COOCH₂— proton 6 in the MEO₉MA (or MEO₅MA) units and the phenyl proton 18 in the VPBA units enables us to calculate the actual molar ratio of MEO₉MA/VPBA (and MEO₅MA/VPBA) in the nanogels. By comparing the integral areas of the –COOCH₂– proton 12 in the DMAEA units and the phenyl proton 18 in the VPBA units, the actual molar ratio of DMAEA to VPBA in the nanogels can be calculated. The ¹H NMR analysis indicated that the actual contents of incorporated VPBA and DMAEA in the nanogels were close to the feeding compositions, benefiting from the excellent control afforded by the RAFT polymerization technique (Table 1).

TABLE 1 |.

Composition of the pMEO_nMA@pVPBA nanogels.^a

Nanogel sample code xEOn-yPBAb	pMEOnM A macro-RAFT agent	Feeding Ratio MEOnMA/VPBA	Ratio from NMR tests M EOnMA/VPBA	Feeding Ratio DMAEA/VPBA	Ratio from NMR tests DMAEA/VPBA
15EO9-30PBA	p(MEO9MA)15	15/30	15/26	5/100	6/100
20EO9-25PBA	p(MEO9MA)20	20/25	20/22	5/100	6/100
20EO9-30PBA	p(MEO9MA)20	20/30	20/26	5/100	7/100
20EO9-35PBA	p(MEO9MA)20	20/35	20/31	5/100	7/100
20EO5-30PBA	p(MEO5MA)20	20/30	20/25	5/100	6/100
30EO5-30PBA	p(MEO5MA)30	30/30	30/24	5/100	4/100

^aAll ratios are molar ratios; n = 9 or 5.

^bx: degree of polymerization in p(MEOnMA)x macro-RAFT chains; y: feeding number of VPBA monomers per p(MEOnMA)x chain.

FIGURE 2 |.

(a–c) Typical ¹H NMR spectra of pMEO₉MA@pVPBA polymer nanogels: (a) 20EO₉-35PBA, (b) 20EO₉-30PBA, and (c) 20EO₉-25PBA in CD₃ OD; (d–f) Particle size distribution in terms of hydrodynamic diameter (D_h) of the polymer nanogels with (d) same feeding number of VPBA per pMEO₉MA chain of different length, (e) same feeding number of VPBA per pMEO_nMA chain of same length but bearing different EO side chain length (EO₉ vs. EO₅), (f) different feeding number of VPBA per p(MEO₉MA)₂₀ chain. All measurements were carried out in 5 mM PBS at pH = 7.4 and 37°C. (g–i) Typical TEM images of the polymer nanogels: (g) 20EO₉-30PBA, (h) 20EO₉-25PBA, and (i) 30EO₅-30PBA.

2.3 |. Size and Morphology of the Nanogels

All six synthesized polymer nanogels exhibited narrow size distributions, with relative variance (μ₂/< Γ >²) below 0.025 in terms of the hydrodynamic diameter (D_h) measured by dynamic light scattering (DLS) in 5 mM phosphate buffer solution (PBS) at pH = 7.4 and 37°C (Figure 2d–f and Figure S5). The size of the nanogel particles can be well-controlled by varying the feeding compositions. When the feeding number of VPBA monomers per pMEO₉MA chain was fixed, increasing the chain length from p(MEO₉MA)₁₅ to p(MEO₉MA)₂₀ resulted in smaller nanogels. For example, the average D_h (< D_h >) decreased from 135 nm for the 15EO₉-30PBA nanogels to 127 nm for the 20EO₉-30PBA nanogels (Figure 2d). Similarly, when the degree of polymerization of pMEO₅MA chains increased from 20 to 30, < D_h > decreased from 143 nm for the 20EO₅-30PBA nanogels to 114 nm for the 30EO₅-30PBA nanogels (Figure S5). In addition, the length of the EO side chain grafted on each repeating unit of the main chains could also control the size of nanogels. When the degree of polymerization of pMEO_nMA main chains and the feeding amount of VPBA were fixed, the increase in the side chain length from EO₅ to EO₉ significantly decreased the size of nanogels, with < D_h > of 143 nm for 20EO₅-30PBA nanogel reduced to 127 nm for 20EO₉-30PBA nanogel (Figure 2e). The size of nanogels made from the p(MEO₉MA)₂₀ macro-RAFT chains but varying the feeding amount of VPBA monomer, was also studied. The decrease in the feeding number of VPBA monomers per p(MEO₉MA)₂₀ chain reduced the particle size gradually (Figure 2f). The < D_h > of 156 nm of the 20EO₉-35PBA nanogel decreased noticeably to 107 nm for the 20EO₉-25PBA nanogel. All these results indicate that the increase in the EO/PBA ratios in the feeding components can reduce the size of the synthesized nanogels, which is understandable because the solvation of the more hydrophilic pMEO₉MA chains relative to the VPBA core monomers can increase the surface-to-volume ratio during the polymerization growth of the nanogel particles. It was also observed that the EO/PBA molar ratios in the nanogels significantly influenced the colloidal stability of nanogels. The 20EO₉-25PBA, 20EO₉-30PBA, and 30EO₅-30PBA nanogels with relatively high EO/PBA ratios exhibited excellent colloidal stability, while a small degree of aggregation was observed for the 15EO₉-30PBA and 20EO₉-35PBA nanogels with low EO/PBA ratios after suspending in PBS for a few months. It is well known that good colloidal stability and suitable size (about 10–150 nm) of polymer nanoparticle carriers are critical to achieve long circulation time by escaping renal and RES clearance in the body [56, 57]. The RAFT polymerization approach developed for the synthesis of these nanogels not only avoided the use of surfactants but also enabled precise control over nanogel size, making the particles suitable for in vivo drug delivery.

The three nanogels of 20EO₉-30PBA, 20EO₉-25PBA, and 30EO₅-30PBA with high EO/PBA ratios and excellent colloidal stability were selected for TEM characterization. The typical TEM images show that the nanogel particles have a spherical morphology without aggregation (Figure 2g–i). The sizes of nanogels observed in the TEM images are smaller than the < D_h > obtained from the DLS measurements. This discrepancy in particle size is attributed to the fact that DLS analysis was performed on the hydrated nanogels in PBS, whereas TEM images were taken from the dried particles cast on the surface of a carbon film. No clear core–shell structure was observed for the dried nanogels due to the low electron density contrast of the non-crosslinked pMEO₉MA and pMEO₅MA shell chains.

2.4 |. Tuning the Onset of Glucose Response of the Nanogels

It has been demonstrated that reversible glucose-responsive swelling/shrinking transitions of crosslinked polymer gels can be used to trigger the on-off insulin release from the gels [30–32, 46, 47, 58–61]. However, one major challenge of these PBA-based gels as insulin carriers is that they start to swell and release insulin at low glucose levels, thereby increasing the risk of hypoglycemia if used in vivo. To minimize the premature insulin release in the hypo- and normoglycemic range, Matsumoto et al. developed a thermo-responsive PBA-containing hydrogel with a metastable dehydrated surface “skin” that prevents insulin release at glucose levels below a threshold of 5.6 mM [33, 34]. However, these bulky gels required subcutaneous surgical implantation in diabetic mice for long-acting regulation of the BGLs. In addition, bulky gels often exhibit slow response rates and are non-injectable into the circulatory system for real-time glucose sensing. Our results show that the onset of glucose response can be simply tuned by varying the EO/PBA molar ratios incorporated in the pMEO₉MA@pVPBA nanogels. First, when the feeding number of VPBA monomers per pMEO₉MA chain was fixed, increasing the chain length from p(MEO₉MA)₁₅ to p(MEO₉MA)₂₀ shifted the onset of glucose response from 1 to 4 mM to initiate the swelling of the synthesized nanogels (Figure 3a). In a similar trend, the increase in the chain length from p(MEO₅MA)₂₀ to p(MEO₅MA)₃₀ at a fixed amount of VPBA shifted the onset of glucose response from 0 to 2 mM in the prepared nanogels (Figure S6). Second, when both the feeding amount of VPBA and degree of polymerization of the pMEO_nMA main chain were fixed, a simple increase in the EO side chain length from EO₅ to EO₉ shifted the onset of glucose response from 0 to 4 mM for the resultant nanogels (Figure 3b). Based on these trends, the third series of nanogels were synthesized by simply varying the feeding amount of VPBA monomer but using the same amount of long p(MEO₉MA)₂₀ macro-RAFT chains to achieve a desirable onset of glucose response. As expected, the increase in the p(MEO₉MA)₂₀/VPBA ratio gradually shifted the onset of glucose-induced swelling toward higher glucose levels (Figure 3c). For example, the onset of glucose response increased from 0 mM for the 20EO₉-35PBA nanogel to 4 mM for the 20EO₉-30PBA nanogel and 5 mM for the 20EO₉-25PBA nanogel, respectively. The glucose-induced swelling of these pMEO₉MA@pVPBA nanogels is driven by the ionization of the crosslinked pVPBA network chains, resulting from the formation of negatively charged cis-diol–PBA complexes by glucose-PBA binding (Figure 1). The minimal swelling observed at glucose concentrations below 5 mM for the 20EO₉-25PBA nanogel and below 4 mM for the 20EO₉-30PBA nanogel implies that a limited number of anionic glucose-PBA complexes were formed under these conditions. The zeta potential analysis of the 20EO₉-25PBA, 20EO₉-30PBA, and 20EO₉-35PBA nanogels at different glucose concentrations confirmed this interpretation (Figure 3d). In the absence of glucose in PBS, all three nanogels exhibit slightly positive zeta potentials, which originate from the ionization of a very small fraction of the DMAEA moieties in the nanogels (Table 1). After the addition of glucose to the PBS dispersion solution, the zeta potentials of the three nanogels showed different trends with increasing glucose concentration. The zeta potential of the 20EO₉-35PBA nanogel dropped sharply to −5 mV even at 1 mM glucose and became progressively more negative as the glucose concentration increased. In contrast, the 20EO₉-30PBA and 20EO₉-25PBA nanogels started to exhibit negative zeta potentials only when the glucose concentrations reached 4.0 and 5.0 mM, respectively. With further increases in the glucose concentration, the zeta potential of all three nanogels continued to decrease to more negative values. It is also noteworthy that the 20EO₉-35PBA nanogel with the lowest EO/PBA ratio (highest content of PBA) displayed the most negative zeta potential at all glucose concentrations, followed by the 20EO₉-30PBA nanogel. The 20EO₉-25PBA nanogel with the highest EO/PBA ratio exhibited the least negative zeta potential under all the tested glucose concentrations. This trend of glucose-responsive zeta potential changes observed in the three nanogels is consistent with their glucose-responsive swelling behavior, confirming that the swelling of the nanogels is driven by the formation of anionic glucose–PBA complexes within the pVPBA core network chains. Such dynamic glucose-PBA complexation is highly reversible, leading to the reversible glucose-responsive swelling and deswelling of nanogels. As shown in Figure S7, the < D_h > values of the 20EO₉-25PBA nanogels reproducibly increased upon exposure to 20 mM glucose and returned to their original values after glucose removal by dialysis and subsequent exposure to 5 mM glucose. This reversible swelling–deswelling behavior was maintained over five consecutive cycles without appreciable changes in < D_h > at either glucose concentration, demonstrating the excellent reversibility and structural stability of the nanogels in response to repeated changes in glucose concentration. Such highly reversible glucose responsiveness is particularly important for self-regulated insulin delivery, as it enables the nanogels to repeatedly adjust their swelling state to switch insulin release on and off in response to glucose fluctuations.

FIGURE 3 |.

(a–c) Glucose-induced swelling ratios of the nanogels with different amounts of EO and PBA components, based on the < D_h >_[Glu]/< D_h >0.0 mM, with < D_h >0.0 mM and < D_h >[Glu] being the < D_h > values measured in PBS (pH 7.4) at 0.0 glucose and a series of different glucose concentrations, respectively. (a) same amount of pVPBA core but different pMEO₉MA chain length, (b) same amount of pVPBA core and degree of polymerization of pMEO_nMA shell chains bearing different lengths of EO brush chain (EO₉ vs.EO₅), and (c) different amount of pVPBA core with the same amount of p(MEO₉MA)₂₀ shell chains. All measurements were made at 37°C and a scattering angle θ = 90°. (d) Zeta-potential changes of the 20EO₉-25PBA, 20EO₉-30PBA, and 20EO₉-35PBA nanogels dispersed in 5 mM PBS at pH = 7.4 with different glucose concentrations at 25°C.

Clearly, both the length and amount of the pMEO₉MA chains relative to the pVPBA core are critical to tune the onset of glucose response and swelling degree of the resultant pMEO₉MA@pVPBA nanogels, which should be related to the accessibility of glucose molecules to the PBA moieties in the core area. These nanogels synthesized by RAFT-mediated polymerization should have a nanostructure with the free pMEO₉MA chains anchored on the surface of the crosslinked pVPBA core. While the p(MEO₉MA)₁₅ shell chains are shorter than the p(MEO₉MA)₂₀ shell chains, the p(MEO₅MA)₂₀ shell chains with shorter EO₅ side chains are “thinner” than the p(MEO₉MA)₂₀ shell chains. At a fixed amount of pVPBA core, both the p(MEO₉MA)₁₅ and p(MEO₅MA)₂₀ shell chains cannot block the glucose molecules diffusing into the pVPBA core region effectively; thus, the 15EO₉-30PBA and 20EO₅-30PBA nanogels begin to swell at very low glucose concentrations (Figure 3a, b). On the other hand, when the long, bulky p(MEO₉MA)₂₀ chains are used as the shell, the increase in the pVPBA core size can change the density of the shell chains and thus also influence the glucose accessibility to the pVPBA core. For example, the 20EO₉-35PBA nanogel with the largest pVPBA core will have the lowest density of the p(MEO₉MA)₂₀ shell chains, allowing glucose molecules to readily access the PBA groups in the core region to swell the nanogel at very low glucose levels. In contrast, the 20EO₉-30PBA and 20EO₉-25PBA nanogels were prepared with a reduced number of VPBA monomers per p(MEO₉MA)₂₀ shell chain, which increases the density of p(MEO₉MA)₂₀ chains anchored on the crosslinked pVPBA core. When these nanogels were dispersed in glucose solutions at low concentrations, glucose molecules could not readily penetrate the dense p(MEO₉MA)₂₀ shell to access the PBA moieties in the core; thus, not enough anionic glucose-PBA complexes were formed to induce nanogel swelling. However, it has been shown that glucose molecules can interact with EO units in PEG and modulate the phase behavior of PEG in water [67–69]. When glucose concentration increases to certain levels, the glucose-EO interactions can shrink the chain conformation of the p(MEO₉MA)₂₀ shell chains, which allows glucose molecules to cross the shell and access the PBA moieties in the core region, producing anionic glucose-PBA complexes to swell the nanogels. The denser the p(MEO₉MA)₂₀ chains on the surface of the pVPBA core, the more glucose molecules are required to modulate the chain conformation of the p(MEO₉MA)₂₀ shell chains. Therefore, the relative amount of pMEO₉MA shell chains anchored on the pVPBA core can determine the glucose level necessary to initiate the swelling of nanogels. This feature of the pMEO₉MA@pVPBA nanogels is critical to prevent swelling and premature insulin release at low glucose levels, thereby minimizing the risk of hypoglycemia.

2.5 |. Insulin Loading and In Vitro Glucose-Regulated Insulin Release From Nanogels

Three nanogel samples, 20EO₉-25PBA, 20EO₉-30PBA, and 20EO₉-35PBA, were chosen to load insulin after confirming their onsets of glucose-responsive swelling. The porous network structure together with the interactions between the insulin and the polymer chains via hydrogen bonding and hydrophobic associations makes these nanogels suitable for encapsulating insulin. The loading capacity of FITC-insulin in the nanogels was determined to be 35.2, 36.1, and 39.3 wt.% for the 20EO₉-25PBA, 20EO₉-30PBA, and 20EO₉-35PBA nanogels, respectively. For the smallest 20EO₉-25PBA nanogel selected for in vivo studies, the < D_h > increased from 107 to 127 nm after insulin loading, consistent with the incorporation of insulin into the nanogels. The loading capacity of these porous nanogels is much higher than the typical insulin loading capacities of ~7–20 wt.% found in other PBA-containing nanocarriers such as micelles, emulsions, interpolymer complexes formed by association mechanism [52–55, 59–61]. A high insulin loading capability is advantageous for long-acting insulin carriers to reduce the frequency of injections when administered in vivo.

The in vitro glucose-responsive insulin release profiles from the 20EO₉-35PBA and 20EO₉-25PBA nanogels were evaluated using the dialysis method for 48 h in PBS at pH 7.4 and 37°C (Figure 4a,b). In the absence of glucose, 19% and 14% of the loaded insulin were released from the 20EO₉-35PBA and 20EO₉-25PBA nanogels after 48 h, respectively. Passive diffusion of insulin molecules located near the peripheral regions of the nanogels, driven by the concentration gradient between the nanogels and the surrounding medium, should primarily account for the baseline insulin release at 0 mM glucose. Gradual desorption of insulin from the polymer network due to dissociation of noncovalent interactions, including hydrogen-bonding and hydrophobic interactions between insulin and the nanogel matrix, may also contribute to the baseline release, although the relative contributions of these mechanisms were not independently quantified in this study. As the glucose level increased gradually in the releasing media, glucose molecules diffused into the crosslinked nanogel core area and bound with the PBA groups therein to form negative charges, which can swell the nanogel with increased mesh size and thus allow the embedded insulin molecules to diffuse out more freely. The 20EO₉-35PBA nanogel with less dense shell chains exhibited a significant swelling degree at 5 mM of glucose (Figure 3c) and thus released 32% of the loaded insulin after 48 h (Figure 4a). In contrast, the 20EO₉-25PBA nanogel with a small degree of swelling at glucose concentrations below 5 mM only released 24% of the loaded insulin at 5 mM of glucose after 48 h (Figure 4b). This demonstrates the insulin retention ability of the 20EO₉-25PBA nanogel in the hypo- and normo-glycemic ranges. On the other hand, the insulin release rate from this same 20EO₉-25PBA nanogel was significantly enhanced in response to the further increase in glucose levels in the hyperglycemic range. For example, 39%, 51%, and 62% of the loaded insulin were released from this nanogel at 10, 15, and 20 mM of glucose, respectively. The 20EO₉-25PBA nanogel, with good insulin retention in the hypo-/normoglycemic range and fast glucose-responsive insulin release in the hyperglycemic range, is highly desirable for effective GRID systems. It should be mentioned that the released insulin from the 20EO₉-25PBA nanogel retained its active α-helical structure. The CD spectrum of the released insulin displayed the same peak positions as that of the native insulin at 208 nm and 222 nm, which are the characteristic peaks associated with the α-helical structure (Figure 4c). It should also be noted that the nanogels can be freeze-dried and readily redispersed in aqueous media without an appreciable change in hydrodynamic size. For example, the measured < D_h > values of the 20EO₉-25PBA nanogels before and after an additional lyophilization/redispersion cycle were 107 and 108 nm, respectively, indicating retention of particle size after processing. This behavior can be attributed to the structural stability of the covalently crosslinked nanogel network and the high hydrophilicity of the pMEO₉MA surface chains, which facilitate rehydration and redispersion. This observation is consistent with a previous report showing that covalently crosslinked microgels can be lyophilized and redispersed without appreciable changes in particle size or colloidal stability [70]. Importantly, all glucose-responsive swelling/deswelling and insulin loading/release experiments described above were performed using lyophilized and redispersed nanogels, demonstrating that the nanogels retain their glucose-responsive properties and insulin loading/release functionality after the lyophilization/redispersion process.

FIGURE 4 |.

(a,b) Release profiles of FITC-insulin from (a) 20EO₉-35PBA and (b) 20EO₉-25PBA nanogels, respectively, in the presence of 0, 3, 5, 10, 15, 20, and 80 mM glucose in PBS at pH 7.4. (c) Circular dichroism (CD) spectra of native insulin and released insulin from the 20EO₉-25PBA nanogel. All experiments were carried out at 37°C. (d,e) In vitro cytotoxicity of the nanogels with different EO/PBA ratios on (d) HEK293 cells and (e) Panc02 cells after incubation for 96 h, respectively.

2.6 |. In Vitro Cytotoxicity of the Nanogels

Five nanogel samples of 30EO₅-30PBA, 15EO₉-30PBA, 20EO₉-30PBA, 20EO₉-35PBA, and 20EO₉-25PBA were selected for in vitro cytotoxicity study. Cell viability was assessed in both a normal cell line (HEK293) and a cancerous cell line (Panc02) after being incubated with the nanogel samples for 96 h at concentrations up to 300 μg/mL (Figure 4d,e). The viability of the cells receiving no treatment with nanogels was set to 100% as the control. The results indicated that the nanogels of 30EO₅-30PBA, 20EO₉-30PBA, and 20EO₉-25PBA with high EO/PBA ratios displayed no or minimal toxicity to the HEK 293 cells even at concentrations up to 300 μg/mL (Figure 4d), suggesting good cytocompatibility. However, the other two nanogels, 15EO₉-30PBA and 20EO₉-35PBA, with lower EO/PBA ratios, show some toxicity to the HEK 293 cells at high concentrations (200–300 μg/mL). This toxicity might be attributed to the small degree of aggregation of the nanogels in the cell culture medium during the 96 h treatment, because these two nanogels either have short hydrophilic p(MEO₉MA)₁₅ chains on the particle surface or have a low density of hydrophilic p(MEO₉MA)₂₀ chains on the particle surface. Interestingly, all five nanogels displayed no or minimal cytotoxicity on the Panc02 cancer cells at concentrations up to 300 μg/mL (Figure 4e), implying that the cancer cells can survive better than the normal cells even in the medium containing the relatively less hydrophilic 15EO₉-30PBA and 20EO₉-35PBA nanogels at high concentrations. Nevertheless, the 20EO₉-25PBA nanogel sample with the highest EO/PBA ratio demonstrated the best cytocompatibility for both cell lines even at the high concentration of 300 μg/mL.

2.7 |. In Vivo Treatment With the Insulin-Loaded Nanogel in Diabetic Mice

The 20EO₉-25PBA nanogel with the smallest size, best colloidal stability and cytocompatibility, and suitable onset of glucose response was selected to load the unmodified native insulin for in vivo study in the STZ-induced type-1 diabetic mice. Only mice with BGL above 400 mg/dL were selected for the diabetic mouse model. The diabetic mice were randomly grouped and treated with the insulin-loaded nanogels (NG-insulin) at three different equivalent insulin doses of 40, 50, and 60 IU/kg, respectively. The empty nanogel dispersed in PBS at a nanogel dose equivalent to that used for the NG-insulin treatment at an insulin-equivalent dose of 50 IU/kg was used as a control. The BGLs remained at hyperglycemic levels without apparent change throughout the 48-h observation period (Figure 5a), indicating that the 20EO₉-25PBA nanogel itself does not measurably affect the BGLs of diabetic mice. The naked insulin at a dose of 3 IU/kg was used as a comparison control. The BGLs of the treated groups dropped to the normoglycemic range (BGL < 200 mg/dL) within 30 min (Figure 5a), indicating that the subcutaneously injected nanogels can enter the circulatory stream and release insulin quickly. While treatment with naked insulin at 3 IU/kg maintained normal BGLs for 3.3 h and lowered the BGL to a minimum of 128 mg/dL, increasing the naked-insulin dose to 8 IU/kg extended the normoglycemic duration to 5.0 h and further lowered the minimum BGL to 90.8 mg/dL. At a dose of 13 IU/kg, naked-insulin treatment further extended the normoglycemic duration to 7.4 h, but the minimum BGL decreased to 57.4 mg/dL, entering the hypoglycemic range (Figure S8). These results indicate that increasing the dose of naked insulin can prolong the duration of glycemic control, but at the expense of progressively greater glucose lowering, eventually resulting in hypoglycemia. In contrast, treatment with NG-insulin at the equivalent insulin doses of 40, 50, and 60 IU/kg maintained normoglycemia in diabetic mice for 13.1, 15.3, and 18.1 h, respectively (Figure 5a,b). Statistical significance was calculated between the different treatment groups (Figure 5c). While no in vivo results are available for comparison from the swelling-based PBA-nanogel GRID systems, our results obtained from the 20EO₉-25PBA NG-insulin outperformed most of the currently developed disintegration-based nanoparticle GRID systems, which typically regulate the normoglycemia of diabetic mice for 6–8 h [23, 24, 39, 40, 42, 54, 55]. The 20EO₉-25PBA NG-insulin also outperformed the commercial long-acting insulin Glargine and insulin Detemir that regulated the normoglycemia of type-1 diabetic mice for 9–10 h at the same dose of 40 IU/kg [38]. Compared to the two long-acting disintegration-based complex nanoparticle GRID systems that maintained normal BGLs of diabetic mice for 20 h at an insulin dose of 115 IU/kg and 13.5 h at an insulin dose of 80 IU/kg [41, 43], our NG-insulin achieved similar long-acting duration but used much lower insulin doses. It is expected that treatment with our NG-insulin GRID system would maintain normal BGLs of diabetic mice for longer hours if similar high doses of insulin (80–115 IU/kg) were used. As shown in Figure 5a,b, the gradual increase in the treatment dose of NG-insulin from 40 to 60 IU/kg significantly prolonged the duration of normoglycemia of diabetic mice from 13.1 to 18.1 h, yet still maintained the normoglycemic plateaus around 100 mg/dL very well. Notably, this BGL plateau of ~100 mg/dL (~5.6 mM) is close to the glucose-response onset of ~5 mM determined for the nanogel in PBS (Figure 3c). The small difference may arise from the discrete glucose concentrations used to determine the in vitro onset, as well as the more complex physiological environment and dynamic glycemic regulation in vivo. This result is critical for the development of long-acting GRID formulations because the dose increase in the NG-insulin did not cause hypoglycemia but simply increased the normoglycemic duration. These in vivo results also indicate that the limited glucose-independent baseline insulin release (14% over 48 h) observed for the 20EO₉-25PBA nanogels in vitro does not lead to excessive glucose lowering under the conditions investigated. The stable normoglycemic plateau without hypoglycemia, even at a high insulin-equivalent dose of 60 IU/kg, suggests that the limited baseline release does not compromise the glycemic safety of the NG-insulin formulation over the treatment period investigated. The reversible on-off insulin release ability of the nanogels at the desirable onset of glucose response was likely the central reason for this crucial advantage. The 20EO₉-25PBA nanogel swells little at glucose levels below 5 mM (Figure 3c). When the BGLs of treated diabetic mice return to the normoglycemic plateau around 100 mg/dL, the nanogels will reversibly shrink back and minimize the insulin release and thus prevent hypoglycemia. The long-acting hours of the NG-insulin also benefited from the small particle size and excellent colloidal stability of the nanogels, which extends the lifetime of the particles in circulatory systems.

FIGURE 5 |.

(a) BGLs of type-1 diabetic mice treated with subcutaneous injection of naked insulin (3 IU/kg) and insulin-loaded 20EO₉-25PBA nanogel (NG-insulin) in PBS solution with the insulin-equivalent dose of 40, 50, and 60 IU/kg, respectively. The empty nanogel dispersed in PBS at a nanogel dose equivalent to that used for the 50 IU/kg NG-insulin treatment was used as a control. Data are presented as mean ± SD (n = 5). (b) Magnified view of the data in the normoglycemic range (BGL < 200 mg/dL) of (a). (c) The normoglycemic duration of the diabetic mice after receiving treatments of naked insulin and NG-insulin at different doses, respectively. Results are shown as mean ± SD (n = 5), with statistical significance calculated between different groups using two-tailed Welch’s t-test. (d) Intraperitoneal glucose tolerance test (IPGTT) results of the diabetic mice at 2 h after treatment with naked insulin (3 IU/kg) and NG-insulin at a dose of 40 IU/kg. The glucose dose was set at 1.5 g/kg. Healthy mice were used as controls. Data are presented as mean ± SD (n = 4). (e) Response to IPGTT in terms of the area under the curve (AUC) from 0 to 150 min from (d). Results are shown as mean ± SD (n = 4). Statistical significance was calculated using a two-tailed Welch’s t-test. (f) IPGTT results of the diabetic mice with two consecutive doses of glucose (1.5 g/kg for each dose) injected respectively at 2 and 3 h after treatment with NG-insulin of 40 IU/kg. Data are presented as mean ± SD (n = 4). (g) Serum chemistry to test liver and kidney toxicity of the 20EO₉-25PBA nanogel dosed serially every two days for one week in healthy mice, with values expressed as mean ± SD (n = 4) and healthy mice as the control group. (h) H&E-stained kidney, liver, lung, and spleen tissues of healthy mice receiving no treatment and injections of the 20EO₉-25PBA nanogel every two days for one week, respectively. Scale bar: 100 μm.

To verify the treatment efficacy of the NG-insulin formulation in response to a glucose challenge such as after a meal, an intraperitoneal glucose tolerance test (IPGTT) was performed in the STZ-induced type-1 diabetic mice. A glucose solution at a dose of 1.5 g/kg was injected intraperitoneally into the diabetic mice at 2 h after the treatment with the NG-insulin of 40 IU/kg, which caused a small spike in BGL. The BGL quickly returned to the original normal BGL values within 1 h after the IPGTT due to the fast response to BGL elevation and timely insulin release from the nanogels. The BGL remained within normal values till the end of the test (2.5 h since the initial IPGTT) (Figure 5d). In contrast, the mice group treated with naked insulin could not tolerate a sudden rise in glucose level. The glucose injection (1.5 g/kg) at 2 h after treatment with naked insulin caused a dramatic elevation of BGLs within 30 minutes, and the BGLs did not return to normoglycemia even at the end of the test. On the other hand, the healthy mice tolerated the sudden increase in BGL. The glucose injection at the dose of 1.5 g/kg caused a spike in the BGL of healthy mice, which then returned to the original normal BGL value within 1 h after the initial IPGTT and maintained normoglycemia to the end of the test. The comparison of the area under the curve (AUC) of BGL responsiveness to IPGTT indicates that the diabetic mice treated with NG-insulin of 40 IU/kg show a similar glucose tolerance ability to that of healthy mice but much better than the group treated with the naked insulin (Figure 5e). To test the glucose tolerance ability of the NG-insulin in response to a more intensive glucose challenge, two consecutive doses of glucose (1.5 g/kg for each dose) were injected into the diabetic mice at 2 and 3 h, respectively, after treatment with NG-insulin of 40 IU/kg. Similarly, the first dose of glucose injection at 2 h after the treatment caused a small spike in the BGL of diabetic mice but quickly returned to the original normal BGL value within 1 h. At this time point (3 h after the initial treatment with the NG-insulin), another dose of glucose was injected into the diabetic mice, which caused a second spike in BGL. The BGL again returned to the normoglycemic range (< 200 mg/dL) within 1 h (Figure 5f). All the above results indicate that the diabetic mice treated with NG-insulin have excellent glucose tolerance.

2.8 |. In Vivo Toxicity of the Nanogels

To assess the toxicity of the nanogels, the 20EO₉-25PBA nanogel particles dispersed in PBS solution at a concentration of 0.845 mg/mL were injected into the healthy mice every two days for one week. Healthy mice were selected as a model to avoid the potential toxicity of STZ in diabetic mice. The serum indicators of aspartate transferase (AST) for the liver and blood urea nitrogen (BUN) for the kidney from the nanogel-treated mice showed no significant changes compared to those of untreated healthy mice (Figure 5g). Likewise, the histological analysis of the H&E-stained tissues from the kidney, liver, lung, and spleen organs of the mice receiving repeated doses of the 20EO₉-25PBA nanogels for one week did not show any abnormal differences compared to those of the untreated healthy mice group (Figure 5h and Figure S7). These results indicate that the tested 20EO₉-25PBA nanogel has no apparent impact on the metabolic function or key organs of healthy mice.

3 |. Conclusion

Glucose-responsive nanogels with controllable onset of glucose response have been developed for long-acting GRID. Such nanogels can be synthesized by RAFT polymerization method using the pMEO₉MA chains as macro-chain transfer agents copolymerized with VPBA monomers and crosslinker, resulting in a nanostructure with hydrophilic brush-like pMEO₉MA chains anchored on the crosslinked pVPBA core. Increasing the EO/PBA feeding ratios in the synthesis of nanogels can reduce the particle size and elevate the onset of glucose response to initiate the swelling of nanogels. The nanogels prepared at high EO/PBA ratios display a spherical shape with narrow size distributions, which can be freeze-dried and easily redispersed in an aqueous phase with excellent colloidal stability. These nanogels also display high insulin loading capacity. The optimized 20EO₉-25PBA nanogels with < D_h > ~100 nm and an onset of glucose response at 5 mM retain insulin well in the hypo/normoglycemic range, but release insulin effectively in response to elevation of glucose in the hyperglycemic range. The diabetic mice treated with a single dose of such insulin-loaded nanogel at 40, 50, and 60 IU/kg maintained normal BGLs for up to 13.1, 15.3, and 18.1 h, respectively, which outperformed the disintegration-based nanoparticle GRID systems reported so far as well as the commercial long-acting insulin Glargine and insulin Detemir when similar doses were used. More importantly, the increase in the dose of nanogel-insulin prevents hypoglycemia while significantly prolonging the duration of normoglycemia in diabetic mice, benefiting from the reversible on-off insulin release ability of the nanogels at the desirable glucose threshold. The nanogels prepared with high EO/PBA ratios exhibited no or minimal cytotoxicity on both HEK293 normal cells and Panc02 cancer cells at concentrations up to 300 μg/mL. The 20EO₉-25PBA nanogels were nontoxic to cells and major organs after repeated dosing in healthy mice. This study opens a new avenue to engineer swelling-based nanogels for long-acting GRID platforms using the RAFT polymerization method.

4 |. Experimental Section

4.1 |. Materials

4-vinylphenylboronic acid (≥ 95%, 4-VPBA), 2-(dimethylamino) ethyl acrylate (DMAEA), poly(ethylene glycol) methyl ether methacrylate (MEO₉MA with M_n = 500 g/mol and MEO₅MA with M_n = 300 g/mol, respectively), N,N′-methylenebis (acrylamide), 4-cyano-4-[(dodecylsulfanylthiocarbonyl)sulfanyl] pentanoic acid (CDTSPA) (RAFT reagent), 4,4’-azobis(4-cyanovaleric acid) (ABCVA) (initiator), D-(+)-Glucose, fluorescein isothiocyanate-labeled insulin (FITC-insulin) and native insulin from bovine pancreas (~5800 Da) were procured from Sigma–Aldrich. All the solvents were purchased from Fisher Scientific. The MEO₉MA and MEO₅MA monomers were purified by passing through a neutral Al₂O₃ column to remove the inhibitors. A stock solution of FITC-Insulin (1 mg/mL) was prepared in 5 mM phosphate buffer at pH 7.4 and stored in a refrigerator at 4°C. Streptozotocin (STZ), ARBOR ASSAY Blood Urea Nitrogen (BUN) Colorimetric Detection Kit (K024-H1), and Invitrogen Mouse Aspartate Aminotransferase (AST) ELISA Kit (EEL086) were purchased from Fisher Scientific. The Premix water-soluble tetrazolium salt (WST-1) cell proliferation assay kit was obtained from Takara Bio USA, Inc. (San Francisco, CA). Other cell culture reagents, including Roswell Park Memorial Institute (RPMI-1640) medium, 1% gentamicin, and insulin–transferrin–selenium (ITS), were purchased from Fisher Scientific, while fetal bovine serum (FBS) was sourced from ATCC.

4.2 |. Synthesis of pMEO₉MA and pMEO₅MA Macro-RAFT Agents

Typically, the required amount of MEO₉MA (or MEO₅MA), CDTSPA, ABCVA, and 2.0 mL of dry 1,4-dioxane were added to a Schlenk tube fitted with a magnetic stirring bar. After sealing the tube with a silicone septum, the solution was purged with nitrogen gas for 30 min to eliminate dissolved oxygen. The polymerization reaction was carried out in a thermostated oil bath at 75°C. After 24 h, the tube was placed on an ice bath to quench the reaction. The polymer solution was then precipitated into a large excess of cold diethyl ether. The precipitate was dissolved in a small amount of THF and re-precipitated in cold diethyl ether. This procedure was repeated twice to purify the polymer product. The resulting product was dried in a vacuum oven at room temperature. The final polymer products with different degrees of polymerization (Table 1) appeared as a viscous yellow liquid.

4.3 |. Synthesis of pMEO₉MA@pVPBA Nanogels

The nanogels with crosslinked pVPBA nanogel as core and brush-like pMEO₉MA (or pMEO₅MA) chains as shell were synthesized by using the p(MEO₉MA)₁₅, p(MEO₉MA)₂₀, p(MEO₅MA)₂₀, and p(MEO₅MA)₃₀, respectively, as the macro-chain transfer agents according to the feeding ratios listed in Table 1. The small amount of DMAEA units (feeding DMAEA/VPBA = 5/100) was incorporated into the pVPBA core nanogels to adjust the pK_a of PBA to near physiological pH. In a typical reaction procedure, p(MEO₉MA)₂₀, 4-VPBA, DMAEA, and ABCVA at specified feeding ratios were dissolved in a DMF/water mixture (95:5 v/v) in a 25 mL round-bottom flask fitted with a magnetic stirring bar. After sealing, the solution was purged with nitrogen gas for 30 mins under stirring to eliminate dissolved oxygen. The polymerization reaction was then carried out in a thermostated oil bath under an N₂ atmosphere at 75°C. After 24 h, the reaction was quenched by placing the flask in an ice bath. The resulting mixture was dialyzed (MWCO = 12–14 kDa) against frequently changed deionized water for 2 days to remove DMF and unreacted substances. The purified nanogel solution was finally freeze-dried to obtain the dried nanogel sample as a fluffy product. The compositions and sample codes for the six nanogels with different EO/PBA ratios were listed in Table 1.

4.4 |. Preparation of Nanogel Dispersion in PBS Solution

A total of 10 mg of freeze-dried pMEO₉MA@pVPBA nanogel was first dissolved in 5 mL of dilute NaOH solution (pH = 10), and then dilute HCl solution (pH = 2) was slowly added to the nanogel dispersion solution under vigorous stirring until the appearance of opalescence. The pH of the resultant solution was carefully adjusted to 7.4. The nanogel solution was finally diluted to a concentration of 1 mg/mL using PBS of pH = 7.4. To prepare the nanogel dispersion solutions with a fixed nanogel concentration but a series of different glucose concentrations in 5 mM PBS at pH = 7.4, a specific volume of concentrated glucose solution in 5 mM PBS was mixed with a certain amount of nanogel stock solution and then diluted to a certain volume using 5 mM PBS solution.

4.5 |. Loading and In Vitro Release of Insulin

FITC-Insulin was loaded into nanogels via the complexation method. 10 mg of freeze-dried pMEO₉MA@pVPBA nanogels were dissolved in 5 mL dilute NaOH solution (pH 10) in a 25 mL round-bottom flask, and the solution was cooled using an ice bath. 4 mL of FITC-Insulin (1 mg/mL) was added dropwise into the nanogel solution under stirring, and the pH was slowly adjusted to 7.4 using dilute HCl solution. The immediate formation of yellowish translucence from the transparent solution indicated complexation between insulin and polymer chains in the nanogels. After stirring overnight, the suspension was centrifuged at 6000 rpm for 30 min. To remove free drug molecules, the precipitate was redispersed in 5 mL of PBS at pH 7.4 and further purified by repeated centrifugation and washing. The final insulin-loaded nanogel precipitate was resuspended in 1 mL of PBS solution at pH 7.4. All the upper clear solutions were collected, and the concentration of free insulin was determined by fluorescence spectrometry at 518 nm upon excitation at 492 nm. By calculating the decrease in drug concentration, the quantity of loaded FITC-insulin in the nanogels was determined. The loading capacity is expressed as the mass of loaded drug per unit weight of dried nanogels. The in vitro release of FITC-insulin from nanogels was assessed using the dialysis method. A dialysis bag containing 1 mL of purified insulin-loaded nanogel dispersion was placed in 50 mL of 5 mM PBS at pH 7.4 containing different glucose concentrations. The released FITC-insulin outside the dialysis bag was taken at defined time intervals and quantified by fluorescence spectrometry. Cumulative release is expressed as the total percentage of insulin released through the dialysis membrane over time.

4.6 |. In Vitro Cytotoxicity of the Nanogels

The cytotoxicity of nanogels with different EO/PBA ratios was evaluated using a WST-1 assay in both the normal HEK293 (human embryonic kidney) cell line and the cancerous Panc02 (pancreatic ductal adenocarcinoma) cell line. Prior to and during treatment, cells were maintained in serum-containing RPMI medium supplemented with 10% fetal bovine serum (FBS) and 1% gentamicin. Briefly, cells were seeded at a density of 3000 cells per well in 96-well plates and incubated overnight under standard culture conditions (37°C, 5% CO₂). For treatment, the culture medium was replaced with serum-free RPMI medium supplemented with 1% ITS and 1% gentamicin, containing various concentrations of nanogels. Cells were treated in triplicate and incubated for 96 h. Following the treatment period, the cells were gently washed three times with serum-free medium. Subsequently, cells were incubated with 10% (v/v) WST-1 reagent diluted in serum-free RPMI at 37°C for 45 min. The absorbance of the resulting solution was measured at 440 nm using a plate reader.

4.7 |. In Vivo Blood Glucose Regulation Studies in Type-1 Diabetic Mice

All animal procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of the College of Staten Island-CUNY. The male C57BL/6J mice were purchased from Jackson Laboratory and maintained under 12 h of light and dark cycles. To induce type-1 diabetic mice, 6–8 week-old mice were fasted overnight and injected with a first dose of 150 mg/kg STZ followed by an additional dose of 50 mg/kg after 48 h. Blood glucose levels were monitored for the following week using a CVS Health Advanced Glucose Meter. Only mice with BGLs consistently over 400 mg/dL were considered diabetic. Diabetic mice were arbitrarily divided into five groups with five mice in each group (n = 5). Individual groups were injected subcutaneously with PBS, naked insulin (3 IU/kg), and insulin-loaded nanogel (NG-insulin) at equivalent insulin doses of 40, 50, and 60 IU/kg, respectively. Blood samples were collected from the tail vein, and BGLs were monitored before and after treatments at different time points till returned to initial levels.

4.8 |. Intraperitoneal Glucose Tolerance Test (IPGTT)

The diabetic mice were randomly assigned to two different groups with four mice in each group (n = 4). Individual groups were treated with naked insulin solution (3 IU/kg) and NG-insulin at an equivalent insulin dose of 40 IU/kg. The BGLs were continuously monitored after treatments. Mice were then intraperitoneally administered glucose solution at a dose of 1.5 g/kg at 2 h post-treatment. The BGLs of all treated mice were monitored continuously for 150 min. The healthy mice group without treatment was used as a control group. The area under the curve (AUC) of the glucose response following IPGTT was calculated using OriginLab software. An additional group of diabetic mice (n = 4) treated with NG-insulin were tested with two consecutive glucose challenges (1.5 g/kg for each dose) at 2 and 3 h post-treatment, respectively.

4.9 |. In Vivo Toxicity Test of the Nanogels

Healthy mice (~30 g) were randomly assigned to two groups (n = 4) and treated with 150 μL of either PBS or the nanogels at a concentration of 0.845 mg/mL. Each group was then redosed with the same treatment every 2 days for a week. At the endpoint of the studies, about 800 μL of blood was collected via terminal cardiac puncture. After centrifugation for 20 min at 1000 × g at 4°C, the supernatant was collected. The BUN and AST in serum were then measured to evaluate toxicity toward the kidney and liver. In addition, organs of the kidney, liver, lung, and spleen were collected and placed in 4% paraformaldehyde and 30% sucrose for 24 and 48 h, respectively. Following this, organs were mounted in Tissue-Tek O.C.T. compound, and 30 um thick cryosections were prepared with a Vibratome cryostat. The tissues were then transferred to ethanol for histological staining using hematoxylin and eosin (H&E) stain and mounted for viewing with Acrytol mounting reagent. The images were taken using a Carl Zeiss Microimaging Gmbh 37081 with an attached AxioCam MRc digital camera.

4.10 |. Characterizations

¹H NMR spectra of the macro-RAFT polymer chains and the obtained nanogels were recorded on a Varian VNMRS 600 MHz Fourier transform nuclear magnetic resonance spectrometer in CDCl₃ and CD₃ OD, respectively. The molecular weight and molecular weight distribution (M_w/M_n) of the pMEO₉MA and pMEO₅MA macro-RAFT polymer chains were determined by gel permeation chromatography (GPC) using polystyrene standards at room temperature, using THF as eluent with a flow rate of 1.0 mL/min. The instrument was equipped with an intelligent AI12 pump, an RI 101 detector, two precolumns MZ-Gel SDplus 50 × 8 mm with 50 and 100 Å, respectively, and a column MZ-Gel SDplus 300 × 8 mm linear 5 μm. The hydrodynamic diameters (D_h) and size distributions of the nanogels were determined by a BI-200SM light scattering spectrometer equipped with a BI-9000 AT digital time correlator (Brookhaven Instruments Corp.). A Nd:YAG laser (150 mW, 532 nm) was used as the light source. All nanogel solutions were filtered through a Millipore Millex-HV filter (0.45 μm) to remove dust before the measurements, and the measurements were made at a scattering angle of 90°. The zeta potential of the nanogels were measured on a NanoBrook ZetaPALS (Brookhaven Instruments) zeta potential analyzer. The reported zeta potential was the average result of 5 runs, with 20 measurement cycles per run. The morphology and size of the dried nanogel particles were characterized by transmission electron microscopy (TEM) on a FEI TECNAI transmission electron microscope at an accelerating voltage of 120 kV. Approximately 10 μL of filtered dilute nanogel suspension was dropped onto a Formvar/carbon covered copper grid and then air-dried at room temperature for the TEM measurements. The PL spectra were obtained on a JOBIN YVON Co. FluoroMax-3 spectrofluorometer equipped with a Hamamatsu R928P photomultiplier tube, a calibrated photodiode for excitation reference correction from 200 to 980 nm, and an integration time of 1 s. The pH values were obtained on an Apera PH700-MS Benchtop Lab pH Meter. Circular dichroism (CD) spectra of the native insulin and released insulin from the nanogels were collected on a Circular Dichroism Spectrophotometer Model 410 from Biomedical, Inc. (NJ, USA).

Supplementary Material

Additional supporting information can be found online in the Supporting Information section.

Supporting File : adfm78536-sup-0001-SuppMat.pdf.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.