Metal-organic Framework Encapsulation Preserves the Bioactivity of Protein Therapeutics

Abstract

Protein therapeutics are prone to lose their structure and bioactivity under various environmental stressors. Herein, we report a facile approach using a nanoporous material, zeolitic imidazolate framework-8 (ZIF-8), as an encapsulant for preserving the prototypic protein therapeutic, insulin, against different harsh conditions that may be encountered during storage, formulation and transport, including elevated temperatures, mechanical agitation and organic solvent. Both immunoassay and spectroscopy analysis demonstrate the preserved chemical stability and structural integrity of insulin offered by the ZIF-8 encapsulation. Biological activity of ZIF-8 preserved insulin after storage under accelerated degradation conditions (i.e. 40°C) was evaluated in vivo using a diabetic mouse model, and showed comparable bioactivity to refrigeration-stored insulin (−20°C). We also demonstrate that ZIF-8 preserved insulin had low cytotoxicity in vitro and did not cause side effects in vivo. Furthermore, ZIF-8 residue can be completely removed by a simple purification step before insulin administration. This biopreservation approach is potentially applicable to diverse protein therapeutics, thus extending the benefits of advanced biologics to resource-limited settings and underserved populations/regions.

1. Introduction

Therapeutic proteins have gained extensive attention in the pharmaceutical industry owing to their high specificity and therapeutic effectiveness, as well as applications in a broad range of diseases such as cancers, metabolic disorders, autoimmune diseases, chronic inflammatory diseases, cardiovascular diseases and infectious diseases.^[1–3] Unfortunately, due to the structural flexibility and susceptibility to environmental stressors, the increasing use of therapeutic proteins also poses an important challenge related to their instability, which not only leads to decreased bioactivity, but may also potentially elicit undesired immunological responses.^[4–5] At present, lyophilization or freeze-drying is the most widely used approach to stabilize these proteins, although elevated temperature and moisture (even trace amount) still need to be avoided during protein storage in dry state.^[6–9] Apart from solid forms, a large portion of therapeutic proteins today are formulated as aqueous solutions in a ready-to-use form, especially for patients in resource-limited settings and requiring rapid administration. These formulations must be stored at low temperature, which typically extends their shelf-life, but at the cost of an extensive distribution network of refrigeration - a “cold chain”- to maintain an optimal temperature during transport, storage, and handling.^[10–11] Two sophisticated alternative approaches to increase protein stability are mutagenesis and chemical modification via modifying the intrinsic structure of proteins.^[12–13] However, in addition to the complex procedure, care must be taken not to compromise protein bioactivity. Another approach to improve the stability is to add sugars into protein solutions, whereas this method still requires refrigeration or freeze-drying and in some cases lack of control of sugar crystallization can lead to protein aggregation.^{[1, 14–17]} Because of prevalent public health challenges, natural disasters, and rising global demand for protein therapeutics, the shortcomings of current preservation approaches create an urgent need for a simple and universal stabilization and preservation strategy for therapeutic proteins in solution against various stressors, preferably, without the refrigeration requirement and protein modification.

Metal-organic frameworks (MOFs) are an exciting class of nanoporous materials that are considered to be highly promising for a number of applications including catalysis,^[18–21] energy,^[22–23] environment,^[24–25] and life sciences.^[26–28] Of particular interest is the emerging biopreservation ability of the MOFs, which is believed to rival conventional porous solids and biomaterials.^[29–31] When incorporated into these nanoporous materials to form MOF biocomposites, proteins are confined within the rigid framework of MOFs, thus maintaining their structures and activities against denaturation and degradation conditions.^[32–33] As recently reviewed by Falcaro and co-authors, various approaches have been developed to incorporate proteins into MOFs.^[34] Among these approaches, a spontaneous biomineralization approach and a de novo approach are considered to offer several unique advantages in the context of protein stabilization: (i) the biocomposites are formed by simply incubating proteins with MOF precursors in mild aqueous solution, which is important to maintain protein activity; (ii) the proteins are embedded in a MOF crystal with pore size smaller than the protein size, not only preventing leaching but also taking advantage of the small pore size of MOFs for specific small molecular adsorption and separation; (iii) these approaches are universal for different types of proteins since proteins serve as nucleation sites and promote MOF crystallization. Tsung and co-workers have employed the de novo approach to encapsulate enzymes for biocatalysis applications.^[35–36] We recently demonstrated that zeolitic imidazolate framework-8 (ZIF-8) encapsulation can be highly effective in preserving biorecognition capabilities of antibodies conjugated on plasmonic nanotransducers and structural integrity of protein biomarkers in various biospecimens (urine, serum and plasma) that are exposed to elevated temperatures for extended duration.^[37–38]

In this work, insulin is selected as a model therapeutic protein because of its extensive clinical usage and well-established structure and bioactivity assays. The required storage for insulin is 2–8 °C since it exhibits a 10-fold or more increase in degradation rate for each 10 °C increment in temperature above 25 °C.^[39–41] This requirement impedes the use of temperature-sensitive insulin in pre-hospital and resource-limited settings such as disaster-struck regions and rural clinics in developing countries with low and moderate incomes, where refrigeration and electricity are not reliable or even not available. Insulin is also prone to denaturation and irreversible aggregation when subjected to organic solvents and mechanical agitation, which could be encountered during formulation of nano/microparticle delivery systems and transport.^[42–44] As with most proteins, previous stabilization methods mainly focused on mutagenesis and chemical modification of insulin. Mutagenesis can produce ultra-stable insulin analog but this requires a priori knowledge of possible degradation pathways and may not be applicable to other proteins since in some cases modification of even a single amino acid may disrupt the tertiary structure of a protein.^{[40, 45]} Conjugation of insulin with glycopolymers containing trehalose side chains can enhance both insulin stability and pharmacokinetics, while the activity of the insulin is compromised due to the steric hindrance of insulin-polymer conjugates binding to the receptor.^[46] Farha and coworkers recently demonstrated zirconium-based NU-1000 preservation of insulin against acidic stomach environment for potential oral delivery.^[47] Herein, we test the specific hypothesis that ZIF-8 (representative MOF) encapsulation would preserve insulin (representative protein therapeutic) structure and activity against various environmental stressors during formulation, transport and storage. We show that ZIF-8 encapsulation can preserve insulin against elevated temperatures, organic solvent and mechanical agitation. ZIF-8 preserved insulin is also evaluated in vivo, and shown to retain bioactivity. The ZIF-8 encapsulation approach does not require any modification to the insulin structure and the ZIF-8 residue can be completely removed by a simple purification step before insulin administration. The ensuing rapid release of encapsulated insulin within a minute enables on-demand reconstitution and usage, thus extending the benefits of advanced protein therapeutics to resource-limited settings.

2. Results and discussion

As a member of zeolitic imidazolate framework family of MOFs, ZIF-8 offers high thermal and hydrothermal stabilities,^[48] and has been demonstrated to be a biocompatible material for drug delivery.^[49–51] Here, the encapsulation of insulin into ZIF-8 crystals is achieved under mild aqueous conditions by mixing insulin solution with aqueous solutions of 2-methylimidazole and zinc acetate (Figure 1). After 12 h incubation, insulin-embedded ZIF-8 crystals were formed and could be easily collected by centrifugation. Scanning electron microscope and transmission electron microscope images showed that insulin-embedded ZIF-8 crystals exhibited a uniform size of ~1 μm (Figure 2A). The typical rhombic dodecahedral shape of the crystals (inset of Figure 2A) was similar to that of the pure ZIF-8 crystals (Figure S1). The powder X-ray diffraction (XRD) pattern of insulin-embedded ZIF-8 crystals also exhibited all the typical peaks of pure ZIF-8 (Figure 2B). To further ascertain the formation of ZIF-8 crystals and the encapsulation of insulin, Fourier transform infrared spectroscopy (FTIR) and Raman spectroscopy were employed. The FTIR spectrum obtained from pure insulin exhibited absorption peaks at 1640–1670 and 1510–1560 cm⁻¹, corresponding to amide I and amide II bands of insulin, respectively (Figure 2C, black spectrum). Following ZIF-8 encapsulation, the FTIR spectrum (orange spectrum) revealed absorption bands associated with ZIF-8 crystals at 1584 cm⁻¹ corresponding to the C=N stretching of imidazole and at 1400–1500 cm⁻¹ corresponding to the imidazole ring stretching in addition to the amide I and amide II bands of insulin.^[52] Similar results were observed by Raman spectroscopy, which also indicated the presence of insulin in ZIF-8 crystals (Figure S2). To quantify the encapsulation efficiency, residual insulin concentration in the supernatant after crystal formation and removal by centrifugation was determined by sandwich enzyme-linked immunosorbent assay (ELISA). It was found that the encapsulation efficiency was dependent on the concentration of ZIF-8 precursors. Specifically, when the concentrations of zinc acetate and 2-methylimidazole were set to 20 mM and 80 mM, respectively, almost 100% of insulin (0.6 mg/mL) was encapsulated within ZIF-8 crystals. Similar encapsulation efficiency was obtained when increasing the concentrations of zinc acetate and 2-methylimidazole to 30 mM and 120 mM, respectively (Figure S3). The typical rhombic dodecahedral shape of ZIF-8 can be observed at these two conditions, whereas smaller and irregular crystals at the concentrations of 10 mM and 40 mM led to the incomplete encapsulation (Figure S4). In a control experiment, encapsulation efficiency was extremely low (~5%, owing to simple physical adsorption) if we simply mixed the insulin solution with pre-synthesized pure ZIF-8 crystals (not shown). This physical mixing of pre-synthesized ZIF-8 crystals with insulin was in stark contrast with the protein-encapsulating approach (i.e. formation of ZIF-8 crystals in the presence of insulin), which exhibited high encapsulation efficiency (~100%). With nearly complete encapsulation, the loading amount of insulin in the biocomposite was determined using thermogravimetric analysis (TGA). The mass loss profile of insulin-loaded ZIF-8 was significantly different compared to that of the pure ZIF-8. The first weight loss at ~100 °C in both pure and insulin-loaded ZIF-8 crystals (~10% weight loss) corresponds to the removal of guest molecules (mainly H₂O) from the cavities and some unreacted reagents.^[53–54] As opposed to pure ZIF-8, insulin-loaded ZIF-8 crystals exhibited a ~10% weight loss between 200 °C to 400 °C, which can be attributed to the decomposition of insulin (Figure 2D). Taken together, these results indicate that insulin is encapsulated with ZIF-8 crystals with a high encapsulation efficiency.

Figure 1.

Schematic illustration depicting the use of ZIF-8 encapsulation for preserving insulin structure and bioactivity against various environmental stressors. The suspension of insulin-embedded ZIF-8 was subjected to elevated temperatures, agitation and organic solvent. The preserved insulin can be released from ZIF-8 crystals within 1 minute, enabling on-demand usage, thus extending the benefits of advanced protein therapeutics in resource-limited settings.

Figure 2.

(A) Scanning electron microscope image of insulin-embedded ZIF-8. Inset: Transmission electron microscope image of insulin-embedded ZIF-8. (B) Powder X-ray diffraction spectra of insulin-embedded ZIF-8 and pure ZIF-8. (C) Fourier transform infrared spectra of insulin-embedded ZIF-8, pure ZIF-8 and pure insulin. (D) Thermogravimetric analysis of insulin-embedded ZIF-8 and pure ZIF-8.

Elevated temperature is the primary detrimental condition during insulin transport and storage, an issue likely to be exacerbated by the increasing epidemic of diabetes in the developing countries, where refrigeration or “cold chain” facilities are not guaranteed.^[55–56] At high temperatures, insulin undergoes both physical (such as unfolding, non-native aggregation and fibrillation) and chemical degradations (such as deamidation, disulfide destruction and reshuffling).^[40] To assess the efficacy of ZIF-8 encapsulation in preserving insulin in solution state under non-refrigerated storage conditions, both ZIF-8 encapsulated insulin and unencapsulated insulin were stored in phosphate-buffered saline (PBS) at 25, 40 or 60°C for 1 week. After 1 week, the ZIF-8 encapsulated insulin was released by adding ethylenediaminetetraacetic acid (EDTA) to dissociate the ZIF-8 crystals by breaking the coordination bonds between zinc and 2-methylimidazole (Figure S5).^[57] The released insulin and unencapsulated insulin were then quantified by ELISA. It was confirmed that both ZIF-8 residues and EDTA did not alter the insulin characteristics and ELISA analysis (Figure S6). The preservation efficacy (insulin recovery %) was calculated by comparing the recovered amount of insulin to the insulin amount prior to incubation. Testing of three different concentrations of ZIF-8 precursors revealed that 20 mM zinc acetate with 80 mM 2-methylimidazole provided the highest preservation efficacy at all three different temperatures (Figure 3A). Specifically, ZIF-8 encapsulated insulin showed more than 95% recovery after 1 week storage at 25 or 40°C, as well as more than 70% at 60°C. Conversely, unencapsulated insulin stored at these temperatures for 1 week exhibited less than 60%, 50% and 30% recovery at 25, 40 and 60°C, respectively. Compared to the optimal precursor concentrations, the low insulin recovery upon using 10 mM zinc acetate with 40 mM 2-methylimidazole can be attributed to the incomplete encapsulation of insulin (Figure S3). In contrast, higher concentrations (30 mM zinc acetate with 120 mM 2-methylimidazole) of the precursors compared to the optimal concentrations, led to incomplete release of insulin (~90%, Figure S7) resulting in slightly lower recovery. Subsequently, using the optimal ZIF-8 precursor concentrations (20 mM zinc acetate with 80 mM 2-methylimidazole), we extended the storage time at different temperatures up to 4 weeks. Different vials of native or ZIF-8 preserved insulin in PBS were sampled at selected time intervals (2, 3 or 4 weeks) to monitor possible changes in the insulin recovery (Figure 3B). With the ZIF-8 encapsulation, 90% insulin was recovered after storage at 25 or 40°C up to 4 weeks (the maximum time tested). Significantly, after 4 weeks at 25 or 40°C, insulin with ZIF-8 encapsulation showed comparable recovery to the unencapsulated insulin in PBS stored at −20 °C (the current “gold standard” as the control, storage temperature required by the manufacturer) (Figure S8). Zinc acetate alone was also tested for preservation efficacy, considering that zinc ions could also increase the thermal stability of insulin by forming insulin hexamer.^[58] However, the preservation efficacy of zinc acetate alone was 30%−40% lower than that of ZIF-8 encapsulation. In addition to solution state, we also compared the insulin recovery between dry powder of ZIF-8 encapsulated insulin and pure insulin powder after 1 week storage at 25, 40 or 60 °C (Figure S9). The results showed that ZIF-8 encapsulation provided excellent stability to insulin in dry state (˃95% recovery at 25 or 40°C and ˃80 % at 60°C), whereas pure insulin powder showed ~70%, ~60 % and ~50% recovery at 25, 40 and 60°C, respectively. These results clearly demonstrate the feasibility and superiority of using ZIF-8 encapsulation for preserving insulin in both solution and dry states at high temperatures.

Figure 3.

(A) Recovery percentage of insulin with ZIF-8 (three different precursor concentrations) or without ZIF-8 encapsulation after 1 week incubation in PBS at 25, 40 or 60 °C. (B) Recovery percentage of insulin with ZIF-8 encapsulation, with addition of zinc ion or without ZIF-8 encapsulation incubated in PBS at 25, 40 or 60°C for different time durations. (C) Recovery percentage of insulin with or without ZIF-8 encapsulation in PBS after subjecting it to agitation and ethyl acetate at room temperature. (D) Circular dichroism spectra of pristine insulin prior to incubation, released insulin from ZIF-8 encapsulation after 1 week incubation at 40°C and insulin without ZIF-8 encapsulation after 1 week incubation at 40°C. Inset: Secondary structure content of the three types of insulin obtained from the CD spectra. Results are the mean and standard deviation from three independent samples.

Apart from elevated temperatures, therapeutic proteins are often subjected to mechanical agitation during transport and formulation. In the case of insulin solution, it is known that mechanical agitation can cause partial unfolding and irreversible aggregation that contains high levels of non-native, intermolecular β-sheet structures.^[59–60] Moreover, therapeutic proteins can also be exposed to an aqueous-organic interface during diverse formulation processes such as emulsion or coacervation, which can also be detrimental to the protein structure.^{[44, 61]} Hence, we investigated the efficacy of ZIF-8 encapsulation in preserving insulin against mechanical agitation or organic solvent that would normally lead to protein denaturation. To mimic the scenario during transport or formulation, insulin in PBS with or without ZIF-8 encapsulation was vortexed at 200 rpm for 48 h. As shown in Figure 3C, ZIF-8 encapsulated insulin was recovered over 90%, in contrast to less than 50% recovery from unencapsulated insulin in PBS. Further increasing the intensity of vortex to 400 and 600 rpm did not compromise the insulin recovery when encapsulated with MOF, whereas the un-encapsulated insulin recovery further decreased (Figure S10). In another case, 1 mL of insulin in PBS or ZIF-8 encapsulated insulin suspension in PBS was first mixed with ethyl acetate (1 mL), and instantly stirred at 200 rpm for 6 h. The ethyl acetate on the top of aqueous solution was removed by vacuum oven before performing ELISA to determine the recovered insulin concentration. The unencapsulated insulin in PBS exhibited less than 60% recovery, whereas ZIF-8 encapsulated insulin was recovered over 80%. The excellent recovery after organic solvent exposure and mechanical agitation can be attributed to the tight confinement of the biomacromolecules within ZIF-8 framework, which significantly lowers the free volume available for chain mobility. SEM images also showed intact insulin-embedded ZIF-8 shape after each stress test (Figure S4).

To further confirm that ZIF-8 encapsulation preserves the insulin structure, circular dichroism (CD) spectroscopy was employed to characterize the secondary structure of insulin with and without ZIF-8 encapsulation after 1 week incubation at 40 °C (Figure 3D). As expected, elevated temperature caused a significant change (an increase in the β-sheet content along with the decrease in the α-helical content) of the secondary structure of unencapsulated insulin in PBS, as shown in the CD spectrum. In contrast, the secondary structure of ZIF-8 encapsulated insulin was found to be very similar to that of the pristine insulin, indicating that ZIF-8 encapsulation is able to maintain the structure of insulin. In addition to CD spectroscopy, high-performance liquid chromatography (HPLC) also demonstrated insulin preservation by ZIF-8 encapsulation, against elevated temperatures, either in suspension or as dry powder (Figure S11). In contrast, unpreserved insulin stored in solution at 40°C showed evidence of degradation. Similar results were observed upon subjecting the encapsulated and unencapsulated insulin to mechanical agitation and organic solvent exposure (Figure S12 and S13). Notably, mechanical agitation converted the unencapsulated insulin from an overall rich helical to dominant β-sheet structure, which indicates severe aggregation of the protein.^[62–63] Overall, CD spectroscopy and HPLC provide direct evidence for the preservation of the insulin structure with ZIF-8 encapsulation.

After establishing the chemical stability and structural integrity of ZIF-8 encapsulated insulin, we assessed the biological activity of ZIF-8 encapsulated insulin by measuring the effectiveness of ZIF-8 encapsulated insulin for treating hyperglycemia in streptozotocin-induced type 1 diabetic mice. The mice were randomly divided to four groups and intravenously injected with PBS solution, insulin stored at −20 °C, ZIF-8 encapsulated insulin or unencapsulated insulin stored at 40 °C for 1 week, respectively. The ZIF-8 encapsulated insulin was released by adding EDTA before injection. The blood glucose concentrations of mice in each group were then monitored for 12 h. As shown in Figure 4A, for mice treated with insulin stored at −20°C and with ZIF-8 encapsulated insulin stored at 40 °C, the blood glucose levels comparably and rapidly decreased to normoglycemic (70–200 mg/dL) within 1 h, were maintained in the normoglycemic range for 4 h, and then increased to hyperglycemic range (~550 mg/dL) within 12 h. This unequivocally indicated the comparable bioactivity of ZIF-8 encapsulated insulin to the refrigerated equivalence. In contrast, insulin without ZIF-8 encapsulation and storage at 40°C only moderately decreased glucose concentrations. The partial loss of insulin bioactivity here could be attributed to the loss of structure integrity of insulin at elevated temperatures as confirmed by the aforementioned ELISA, CD and HPLC experiments. Overall, the in vivo experiments clearly demonstrate the excellently preserved bioactivity of insulin through ZIF-8 encapsulation.

Figure 4.

(A) In vivo studies of insulin preservation efficacy of ZIF-8 encapsulation. Blood glucose concentrations in streptozotocin-induced diabetic mice after administration of PBS solution, insulin stored at −20°C for 1 week, ZIF-8 encapsulated insulin and unencapsulated insulin stored at 40°C for 1 week. The ZIF-8 encapsulated insulin was released by adding EDTA before injection. Results are the mean ± standard deviation (n=3). (B) Histology study to assess toxicity. Histological evaluation of the major organs of the mice at 5 days after intravenous injection of PBS or stabilized insulin. No symptoms of inflammation and/or lesion were observed in the hematoxylin and eosin stained images.

Finally, the biocompatibility of insulin-embedded ZIF-8 (after fully dissociated by EDTA) was assessed by determining its cytotoxicity using MTT assay (Figure S14). The mouse embryonic fibroblast 3T3 cells were used as the model cell line. After 24 h incubation with relatively high concentration (1000 μg/mL) of dissociated crystals, the cell viability was found to be higher than 80%, indicating the low cytotoxicity of the dissociated products. To further evaluate the biocompatibility of insulin-embedded ZIF-8, the mice treated with ZIF-8 encapsulated insulin and PBS were sacrificed 5 days after insulin administration for histological analysis. The haematoxylin and eosin (H&E) stained images of various organs from the two groups showed similar structure (Figure 4B). There were no apparent histopathological abnormalities or lesions observed in the heart, liver, spleen, lung and kidney. In addition, there was no weight loss in either group after 5 days administration (Table S1). Overall, the insulin-embedded ZIF-8 after dissociation shows excellent biocompatibility. We also tested the feasibility of removing dissolved ZIF-8 residues before insulin administration. The ZIF-8 encapsulated insulin was first released by adding EDTA and then filtered to remove any ZIF-8 byproduct by centrifugation through a 3 kDa filter. After washing three times, HPLC-mass spectrometry analysis showed that more than 99% of 2-methylimidazole can be removed (Figure S15). This purification step mitigates the toxicity concern from ZIF-8 residue, especially for repeated drug administration as is the case with insulin.

3. Conclusion

In conclusion, we have demonstrated that ZIF-8 encapsulation preserves the structure and bioactivity of insulin under various environmental stressors including elevated temperature, organic solvent and mechanical agitation. Apart from standard bioanalytical tool (ELISA), CD measurements provide direct evidence for the preserved secondary structure of insulin upon ZIF-8 encapsulation. For the first time, the preserved protein bioactivity (insulin in this work) is evaluated in vivo using a diabetic mouse model. ZIF-8 encapsulated insulin at an elevated temperature (40°C) shows comparable bioactivity to insulin stored at −20°C. The ZIF-8 residues exhibit low cytotoxicity and do not cause any side effects to animals, and can be completely removed by a simple purification step before insulin administration. Overall, we believe this facile approach can be generalized to various protein therapeutics, thus extending the benefits of advanced protein therapeutics to resource-limited settings and under-served populations/regions. While the results presented here provide a proof-of-concept for preserving the bioactivity of insulin with ZIF-8 encapsulation, the technique can be easily extended to other protein therapeutics and MOFs^[47].

4. Experimental section

Materials:

2-methylimidazole, zinc acetate dihydrate, ethylenediaminetetraacetic acid (EDTA), ethyl acetate, (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide) (MTT), dimethyl sulfoxide (DMSO) and phosphate-buffered saline (PBS) were purchased from Sigma-Aldrich. Human recombinant insulin and insulin sandwich ELISA kit (detection range, 15.60–1,000 pmol/L) were purchased from R&D systems. The Pierce bicinchoninic acid (BCA) protein assay kit and trypsin were obtained from Thermo Fisher Scientific. Dulbecco’s modified eagle medium and Dulbecco’s phosphate-buffered saline (DPBS) were purchased from Gibco. All experiments were performed using nanopure water with a resistivity of 18.2 MΩ•cm.

Synthesis of insulin-embedded ZIF-8:

To form insulin-embedded ZIF-8, 2-methylimidazole solution (0.5 mL, in nanopure water) and zinc acetate dihydrate solution (0.5 mL, in nanopore water) were added into 5 mL of insulin solution (0.6 mg/mL in PBS). The final concentrations of 2-methylimidazole after mixing were 40 mM, 80 mM, and 120 mM. The final concentrations of zinc acetate dihydrate after mixing were 10 mM, 20 mM and 30 mM. The molar ratio of 2-methylimidazole and zinc acetate was controlled to be 4:1. The resultant mixture was incubated at room temperature for 12 h to form insulin-embedded ZIF-8 crystals. For subsequent characterization, the insulin-embedded ZIF-8 crystals were collected by centrifugation (at 13.4 k rpm for 20 min), washed twice by nanopure water and vacuum dried at room temperature. Pure ZIF-8 was synthesized via the similar approach without adding insulin.

Characterization of insulin-embedded ZIF-8:

Scanning electron microscopy (SEM) images were obtained using a FEI Nova 2300 field-emission SEM at an acceleration voltage of 10 kV. Transmission electron microscopy (TEM) micrographs were collected by a JEM-2100F (JEOL) field emission instrument. Thermogravimetric analysis (TGA) was performed using TA Instruments Q5000 IR Thermogravimetric Analyzer in air (at rate of 5 °C min⁻¹). Fourier transform infrared spectroscopy (FTIR) measurements were performed using a Nicolette Nexus 470 spectrometer. The Raman spectra were obtained using a Renishaw inVia confocal Raman spectrometer mounted on a Leica microscope with a 50× objective and a 514 nm wavelength diode laser as an illumination source. The X-ray diffraction (XRD) measurements of the samples were recorded on a Bruker D8-Advance X-ray powder diffractometer using Cu Kα radiation (λ = 1.5406 Å) with scattering angles (2θ) of 5–35°.

Evaluation of preservation efficacy under environmental stressors using ELISA:

To evaluate the preservation efficacy under non-refrigerated temperatures, the vials with suspension of insulin embedded ZIF-8, pure insulin or insulin with adding zinc acetate solutions were sealed and stored at 25, 40 and 60 °C for different time intervals (including 1, 2, 3 and 4 weeks). Insulin sandwich ELSIA was used to quantify the insulin recovery after storage. Before ELISA measurement, the ZIF-8 encapsulated insulin was released by adding EDTA (with same molar amount as zinc acetate). The preservation efficacy was calculated by comparing the recovered insulin amount to day 0 insulin amount prior to heating. In addition to solution state, dry powder of ZIF-8 encapsulated insulin and pure insulin powder were sealed and stored at 25, 40 and 60 °C for 1 week for comparison. To examine the preservation efficacy under agitation, insulin in PBS (0.5 mg/mL, 1 mL) with or without ZIF-8 encapsulation was vortexed at 200, 400 or 600 rpm for 48 h. To assess the preservation efficacy under organic solvent, 1 mL of insulin in PBS or ZIF-8 encapsulated insulin suspension in PBS was first mixed with ethyl acetate (1 mL), and instantly stirred at 200 rpm for 6 h. The ethyl acetate on the top of aqueous solution was removed by vacuum oven before ELISA measurement.

Circular dichroism (CD) spectroscopy:

The CD measurements were performed using a spectropolarimeter JASCO J-810. The spectrum was collected at the rate of 20 nm per minute at a response time of 16 seconds. Before CD measurement, the ZIF-8 encapsulated insulin was first released by adding EDTA and then filtered to remove any ZIF-8 byproduct using centrifuge tube with 3 kDa filter. The insulin recovered from various treatments was quantified by BCA assay and diluted to 100 μg/mL in PBS. The secondary structures of insulin (α-helical content, β-sheet content) were analyzed using CDPro software from CD spectra.

High-performance liquid chromatography (HPLC) analysis of insulin:

Insulin was analyzed using a method based on the United States Pharmacopeia Insulin Monograph (http://www.pharmacopeia.cn/v29240/usp29nf24s0_m40520.html) and Waters Application Note Final Transferred UPLC Method (http://www.waters.com/webassets/cms/library/docs/720001396en.pdf) with minor modification. Before HPLC measurement, the ZIF-8 encapsulated insulin was first released by adding EDTA and then filtered to remove any ZIF-8 byproduct using centrifuge tube with 3 kDa filter. HPLC-UV was conducted on an Agilent (Santa Clara, CA) HPLC 1100 series system composed of a binary pump with a micro vacuum degasser, thermostatted column oven, high performance micro well plate autosampler, and variable wavelength detector. ChemStation B.04.03 software was utilized for instrument control, data acquisition, peak integration and data analysis. Chromatographic separation was achieved utilizing a Kinetex Core-Shell analytical column (100 × 2.1 mm, 2.6μm, Phenomenex, Torrance, CA). A 0.25 μM inline filter was additionally added prior to the sample entering the column. The injection volume was 5 μL. Insulin retention time was approximately 26 min.

In vitro study using mouse embryonic fibroblast (cell line: 3T3):

3T3 cell line was harvested with trypsin and resuspended in Dulbecco’s modified eagle medium at a concentration of 5 × 10⁴ cells per mL. 100 μL per well of the cell suspension was transferred into 96-well plates to preculture for 24 hours. The medium was replaced by a fresh medium that contained different concentrations of dissociated insulin-embedded ZIF-8 crystals. After 24 hours incubation, the medium was removed and cells were washed by DPBS. 100 μL of 1.2 mM MTT medium solution was then added to each well. After 4 hours, the MTT medium was removed and 200 μL DMSO was added to each well. After incubation for 10 min, the absorbance at 570 nm was determined with a plate reader.

In vivo study using streptozotocin (STZ)-induced type 1 diabetic mice:

STZ-induced male C57BL/6 (6–10 weeks) type 1 diabetic mice were purchased from Jackson Laboratory (USA). The blood glucose levels of mice were tested 1 day before administration by collecting blood (~3 μL) from the tail vein and measuring using the Clarity GL2Plus glucose monitor (VWR, USA). The mice were randomly divided into four groups (3 mice each group) and intravenously injected via lateral tail vein with PBS solution, insulin stored at −20 °C, ZIF-8 preserved insulin (dissociated by EDTA) and insulin alone stored at 40 °C for 1 week, respectively. The insulin dose for each mouse was 1 mg/kg (125 μg/mL, ~200 μL). The blood glucose level was measured from tail vein blood samples (~3 μL) of mice at different time points (at 10, 20, 40, and 60 min, and once per hour afterward for the first 12 h in the day of administration). Mice were anesthetized with 2% isoflurane. The mice treated with ZIF-8 preserved insulin and PBS were sacrificed after 5 days administration, and major organs were collected and sliced for haematoxylin and eosin (H&E) staining.

HPLC-mass spectrometry analysis of 2-methylimidazole residues before and after purification:

HPLC-mass spectrometry analysis was performed on an ultra-fast liquid chromatography system (Shimadzu Scientific Instruments, Columbia, MD) with a CMB-20A system controller, two LC-20AD XR pumps, DGU-20A3 degasser, SIL-20AC XR autosampler, FCV-11AL solvent selection module, and CTO-20A column oven, and an external Valco divert valve installed between the LC and mass spectrometer. The LC system was coupled to an API 4000 linear ion trap triple quadrupole (QTRAP) tandem mass spectrometer operated with Analyst 1.5.2. Multiquant 3.0.1 (AB Sciex) was utilized for peak integration, generation of calibration curves, and data analysis. Chromatographic separation was achieved with a Sunfire C18 (150 × 2.1 mm, 3.5 μM, Waters, Milford, MA) analytical column equipped with a C18 VanGuard cartridge (2.1 mm × 5 mm, 3.5 μM, Waters, Milford, MA). A 0.25 μM inline filter was additionally added prior to the sample entering the column. The flow rate was 0.4 mL/min with a mobile phase consisting of 20mM ammonium formate aqueous (A) and 20 mM ammonium formate in methanol (B). The column was equilibrated with 0% B, maintained after injection for 1.0 min, then a linear gradient to 80% B applied over 1.25 min and held for 2.75 min, then reverted back to 0% B over 0.1 min and re-equilibrated for 2.9 min. Total run time was 8 min. The injection volume was 10 μl. The column oven was at 40°C. Under these conditions, approximate retention time for 2-methylimidazole was 1.77 minutes and for 4-methylimidazole (internal reference) was 3.18 minutes. The mass spectrometer electrospray ion source was operated in positive ion multiple reaction monitoring mode. The [M+H]⁺ transitions were optimized for 2-methylimidazole 83.0→42.2 and 4-methylimidazole 83.0→56.2. Mass spectrometer settings for the declustering potential (66, 56 V), collision energy (29, 25 V), entrance potential (10 V), and collision cell exit potential (4, 8 V) were optimized. Optimized global parameters were: source temperature 550°C, ionspray voltage 5000 V, nitrogen (psig) curtain gas 30, gas 1 50, gas 2 50, collision gas medium. Similar as CD experiments, the ZIF-8 encapsulated insulin was first released by adding EDTA and then filtered to remove any ZIF-8 byproduct using centrifuge tube with 3 kDa filter (three times washing with water and using 3000 rpm for 5 min at each time). HPLC-mass spectrometry was used to quantify the 2-methylimidazole before and after washing.