Effectiveness of an ocular adhesive polyhedral oligomeric silsesquioxane hybrid thermo-responsive FK506 hydrogel in a murine model of dry eye

2.1. Synthesis of poly(MPOSS-PEG-PPG urethane) (MPEP) copolymers

Poly(MPOSS-PEG-PPG urethane) (MPEP) was derived from macrodiols of MPOSS, PEG, and PPG, with the feed ratio of PEG-PPG (F127) fixed at 2:1, HDI as the chain extender, and DBT as the catalyst. The synthetic method was similar to that used in our previous work [34]. The MPOSS content was set at 0.5, 1, or 2 wt%, and the resulting copolymers were denoted as nMPEP, where n represented the feed weight percentage of MPOSS: MP for MPOSS, E for PEG, and P for PPG. Additional details are provided in the Supporting Information.

2.2. Molecular characterization

Gel permeation chromatography (GPC) was conducted on a Waters GPC System (Shimadzu, Japan) equipped with two Phenogel columns (103 and 105 Å) (size: 300 × 7.80 mm) in series and a refractive index detector. HPLC grade THF was used as an eluent at a flow rate of 1 mL/min at 40 °C. Poly(methyl methacrylate) standards were used to generate a calibration curve.

¹H and ¹³C nuclear magnetic resonance (NMR) spectra were obtained using a 500-MHz NMR spectrometer (JEOL, Japan) at room temperature. The chemical shift was referenced to the solvent peak of CDCl₃ at 7.3 ppm.

Fourier transform infrared (FT-IR) spectra were collected using a Spectrum 2000 FT-IR spectrophotometer (PerkinElmer, USA) at room temperature. The copolymers were mixed with KBr and tableted, undergoing 32 scans in the wavenumber range of 4000–400 cm⁻¹ at a resolution of 4 cm⁻¹.

2.3. Thermal analysis

Thermogravimetric analysis (TGA) was carried out on a Q500 TGA analyzer (TA Instruments, USA) under a nitrogen atmosphere. The temperature was elevated from room temperature to 800 °C at a rate of 20 °C/min.

Differential scanning calorimetry (DSC) was performed on a Q100 photo-DSC analyzer (TA Instruments, USA) using indium for calibration. Two heating−cooling cycles in the range of −80 °C to 200 °C were run at a heating or cooling rate of 20 °C/min, and data from the second cycle were used for analysis.

2.4. Determination of the critical micelle concentration (CMC)

Aqueous copolymer solutions at different concentrations starting from 1 wt% were prepared using the gradient dilution method, and then mixed with DPH methanol solution (0.6 mM) at a volume ratio of 50:1. After reaching equilibrium at 4 °C overnight, UV–vis spectra were obtained on a UV-2501 PC spectrophotometer (Shimadzu, Japan) at room temperature in the range of 320–460 nm.

2.5. Particle size analysis

The particle size and distribution of the aqueous copolymer solutions at different concentrations (0.1, 0.5, and 1 wt%) and different temperatures (25, 37 and 70 °C) were measured by dynamic light scattering (DLS) on a Nano ZetaSizer System (Malvern Instruments, USA) using a 633-nm laser light and a scattering angle of 173°.

2.6. Determination of the lower critical solution temperature (LCST)

The LCST value of the aqueous copolymer solutions (2 wt%) was also estimated by UV–vis spectra generated on a UV-2501 PC spectrophotometer (Shimadzu, Japan) equipped with a temperature control module. The temperature was elevated from 25 °C to 75 °C at an interval of 1 °C and an equilibrium time of 5 min per step.

2.7. Determination of sol-gel transition behaviors

Aqueous copolymer solutions at concentrations varying from 2 wt% to 20 wt% were prepared in 4-mL vials. After being placed at 4 °C for 1 d to ensure full dissolution, the samples were gradually heated in a water bath from 4 °C to 80 °C at an increment of 2 °C and an equilibrium time of 5 min per step. The gelation temperature was defined as the critical temperature at which firm gels were observed in inverted vials.

2.8. Rheological analysis

The rheological measurements were performed on a Discovery DHR-3 hybrid rheometer (TA Instruments, USA) adopting a parallel-plate geometry. Temperature ramps from a low temperature to body temperature and temperature sweeps from 4 °C to 80 °C at a heating rate of 5 °C/min were measured, both with the strain fixed at 1% and the frequency fixed at 1 Hz.

2.9. Determination of the light transmittance of blank hydrogel

Temperature-sensitive copolymers, 1MPEP and 2MPEP, can form hydrogels at ocular surface temperature (34.5 °C) at a concentration of 5%, while the concentration of F127 needs to reach 20% [35]. PBS solutions containing 5% 1MPEP, 5% 2MPEP, and 20% F127 were prepared. A 100-μL aliquot of the samples was placed in a 96-well plate and incubated 34.5 °C to form hydrogels; a commercial liquid Carbomer was used as a control. A multifunctional microplate reader (Multiskan FC, Thermo Fisher Scientific, USA) was used for full-wavelength scanning to determine the absorbance value. The light transmittance of the hydrogel was calculated according to the formula, A = –log T%, where A is absorbance and T% is light transmittance [36]. According to previous studies, hydrogels with a light transmittance greater than 90% in the visible wavelength range of 390–780 nm are considered transparent; those with a light transmittance of 10–90% are considered translucent; and those with a light transmittance below 10% are considered opaque [37].

2.10. Degradation analysis of blank hydrogel

A 500-μL aliquot of 1MPEP, 2MPEP, and F127 solution was added to a microfuge tube. After the hydrogel had formed at 34.5 °C, 500 μL PBS was gently added on top of the hydrogel. The microfuge tubes were placed vertically in a constant-temperature shaker, and degradation was carried out at 34.5 °C and 50 rpm. At a specific timepoints, the upper liquid was removed and the tubes were weighed; the weight of the remaining hydrogel was recorded following subtraction of the weight of the blank tube, and fresh PBS was subsequently added to continue degradation.

2.11. Determination of the concentration of FK506 by high-performance liquid chromatography (HPLC)

HPLC was used to determine the concentration of FK506. Column: ZORBAX SB-C18 (5 μm, 4.6 × 150 mm); mobile phase: acetonitrile: 0.25% phosphoric acid aqueous solution (60:40, v/v); UV detection wavelength: 215 nm; flow rate: 1 mL min⁻¹; column temperature: 50 °C; injection volume: 10 μL.

FK506 powder was accurately weighed and dissolved in acetonitrile at a concentration of 1 mg/mL for use as a stock solution. FK506 was serially diluted in the mobile phase to generate standards of 1, 2, 4, 12, 24, 32, 64, and 128 μg/mL. Each solution was subjected to chromatography, and a concentration–peak area standard curve was created. The fitted curve was y = 19.059x + 13.581, R² = 0.9999.

2.12. Preparation of the FK506 hydrogel

Since FK506 is poorly water-soluble, it was first wrapped in amphiphilic micelles. At a material/drug ratio of 20:1, 1MPEP, 2MPEP, F127, and the same amount of FK506 was dissolved in acetone and the mixture was added dropwise to PBS while stirring. Nitrogen blowing was used to remove the volatile acetone to prepare 1MPEP-FK506(1M-FK506), 2MPEP-FK506(2M-FK506), and F127-FK506 drug-loaded micelle solutions. The precipitate was removed by centrifugation at 2000 rpm, and HPLC determined the FK506 concentration in the supernatant. Its concentration was then calculated and adjusted. A certain weight of material was added to ensure that the final concentration of 1MPEP, 2MPEP, and F127 was approximately 5%, 5%, and 20%, respectively. The final concentration of FK506 was set to 0.1% according to the commonly used clinical concentration and stored for use.

2.13. In vitro drug release experiment

A 500-μL aliquot of 1M-FK506, 2M-FK506, or F127-FK506 solution was added to a microfuge tube and allowed to form hydrogels at 34.5 °C in a constant-temperature water bath. Subsequently, 500 μL 1 × PBS (containing 0.2% Tween) was gently added to the top of the hydrogel as the release medium. The tubes were placed vertically in a constant-temperature shaker at 50 rpm and 34.5 °C. The upper release medium was removed at certain timepoints and replaced with 500 μL fresh PBS until the hydrogel had completely degraded. The release medium was subjected to HPLC to measure the concentration of FK506, which was used to construct a drug release curve.

2.14. Cytotoxicity of the hydrogel on human corneal epithelial (HCE) cells

The drug solutions were prepared using the same method as that for the hydrogel extract. The 1MPEP, 2MPEP, F127, 1M-FK506, 2M-FK506, and F127-FK506 solutions were stored at −80 °C for three days, following which 500 μL of each was added to a cell culture plate and allowed to form a hydrogel at 37 °C. Subsequently, 1 mL of the culture medium was added and incubated at 37 °C for 48 h. The supernatant was collected and serially diluted with medium to obtain 0.5, 0.25, 0.125, 0.0625, and 0.03125 mg/mL; the 0 mg/mL FK506 group was used as a control.

The MTT assay was used to assess cell viability. The immortalized human corneal epithelial (HCE) cell line was purchased from the American Type Culture Collection (ATCC) and cultured in DMEM/F12 medium supplemented with hEGF, insulin, FBS, and penicillin-streptomycin. HCE cells were seeded onto 96-well plates and incubated for 24 h. The diluted hydrogel–drug extracts were added and incubated for 48 h, following which the medium was removed, replaced with MTT solution, and incubated for another 4 h. The supernatant was removed and replaced with DMSO to dissolve the formazan crystals. Finally, the absorbance was measured at 492 nm using a SpectraMax® absorbance reader (Molecular Devices); cell survival curves were subsequently constructed.

2.15. Surface plasmon resonance

Surface plasmon resonance on a BIAcore-T200 biomolecular interaction analyzer was used to analyze the interaction between mucin and polymers. The purchased mucin was dialyzed in ice-cold PBS for 48 h to replace the salt content in the original sample, and the protein was subsequently freeze-dried. Acetate buffers of different pH were configured to screen the appropriate pH for protein coupling. The amino coupling method was used to fix mucin on the CM5 sensor chip, with the final immobilization level of mucin being approximately 10,000 RU. Using PBS as the working buffer, 1MPEP, 2MPEP and F127 were serially diluted to 62.5 μM, 31.25 μM, 15.625 μM, 7.8125 μM, 3.90625 μM, and 0 μM. The polymer solution at various concentrations was passed over the chip with a contact time of 120s, a flow rate of 30 μL/min, and a separation time of 450s. The chip was subsequently washed with pH 2.5 glycine buffer and PBS buffer at a working temperature of 25 °C. Biacore T200 Evaluation software (Version 2.0) was used to analyze the interaction between mucin and polymers, and the curve was fitted with a 1:1 binding model to calculate the kinetic constant KD value.

2.16. Animal dry eye model

Female C57BL/6 mice (weighing 25–35 g; aged 8–12 weeks) were purchased from the Animal Center of Xiamen University (Xiamen, Fujian, China). The animal protocols conformed to the Association for Research in Vision and Ophthalmology (ARVO) Statement for the Use of Animals in Ophthalmic and Vision Research. This research program was approved by the Experimental Animal Ethics Committee of Xiamen University. The mice were kept in a well-ventilated environment at a humidity of 30 ± 3% and a temperature of 23–25 °C. Subcutaneous injections of scopolamine hydrobromide (0.25 g/100 mL) in the flank four times daily for five days established a desiccating stress (DS5) [[38], [39], [40]].

2.17. Topical drug treatment and slit-lamp imaging

Topical drug treatment was performed on the murine dry eye model by applying 5 μL different eye drops formulation (PBS, 1MPEP, 2MPEP, F127, Commercial FK506, F127-FK506, 1M-FK506, 2M-FK506) twice daily for five days without anesthesia. In detail, researchers gently held a mouse and tried their best to comfort it before performing subcutaneous injection with minimal pain and stress. Besides, all the researchers received training, into how to skillfully administer the subcutaneous injection to avoid secondary injury to the mouse.

Following administration of the eye drops and manual blinking of the murine eyes, the corneas were examined and imaged via slit-lamp microscopy to observe the hydrogel on the ocular surface. The process was performed as described previously [41].

2.18. Tear secretion test

The tear secretion test was performed at the end of treatment. Briefly, a phenol red cotton thread was placed into the lower conjunctival fornix and the mouse was forced be stay stationery for 15 s. The secretory rate was measured in millimeters based on the advancement of the red stained tear front.

2.19. Corneal fluorescein staining

Corneal fluorescein staining was conducted by instilling 5% Oregon green dextran (OGD) onto the murine ocular surface for 1 min with manual blinking several times. The mice were subsequently euthanized, and the corneas were rinsed with 1 mL saline five times. The corneas were photographed using a fluorescence dissecting microscope (Leica, Wetzlar, Germany) under 470-nm fluorescence excitation. The fluorescence intensity of corneal OGD staining was quantitatively analyzed using the NIS Elements software (NIS Elements; version 4.1; Nikon, Melville, NY, USA) with a 200,000-pixel circle on the central cornea.

2.20. Histological staining

The murine eyeballs and periocular adnexa were separated and harvested, fixed in 4% formaldehyde for at least 24 h, and infiltrated with paraffin. Paraffinized samples were cut into sagittal sections (5-μm thick), placed on glass slides, and stained using Hematoxylin & Eosin (H&E) and Periodic Acid-Schiff (PAS) staining kits according to the manufacturer's instructions. All stained slides were observed and imaged using a light microscope (Eclipse E400 with a DS-Fi1, Nikon, Melville, NY). For PAS staining, the goblet cells were quantitatively counted using the NIS Elements software including the upper and lower conjunctiva.

2.21. Immunofluorescence staining

After sampling, the tissues were embedded in optimal cutting temperature compound (OCT compound) and stored at −80 °C. Frozen samples were cut into sagittal sections (5-μm thick) and placed on glass slides. For immunofluorescence staining, following fixation in acetone and blocking with 2% bovine serum albumin (BSA), the frozen slices were incubated at 4 °C overnight with anti-matrix metalloproteinase-3 (MMP-3), anti-matrix metalloproteinase-9 (MMP-9), and anti-CD4 antibodies. The following day, the slices were washed in PBS three times and incubated with Alexa Fluor 488-conjugated donkey anti-goat IgG or Alexa Fluor 488-conjugated donkey anti-rabbit IgG for 1 h in the dark at room temperature. Lastly, the slices were mounted with DAPI, observed and then imaged using a fluorescence microscope (LeicaDM2500).

2.22. Western blotting

Murine conjunctival tissues were surgically separated and dissolved in cold RIPA buffer containing protease and phosphatase inhibitor cocktail. The protein solutions were extracted, purified, and subsequently quantitated using a BCA assay kit. Equal amounts of protein were separated on 12% tricine gels and transferred to polyvinylidene difluoride (PVDF) membranes. The membranes were blocked in 5% BSA for 2 h at room temperature, followed by overnight incubation with an anti-CD4 antibody. The next day, the membranes were washed three times with Tris-buffered saline containing 0.05% Tween 20 (TBST), and the membranes were incubated for 1 h with HRP-conjugated goat anti-rabbit IgG. Protein bands were visualized using an Ultra Chemiluminescence Reagent kit and images were captured by the ChemiDoc XRS System (BioRad Laboratories, Inc., Philadelphia, PA, USA). Densitometry was performed using the ImageJ software.

2.23. Optical coherence tomography (OCT) examination, fundus imaging, and fluorescence fundus angiography

OCT, fundus imaging, and fluorescence fundus angiography were performed using an OPTOPROBE system (China). Briefly, different FK506-containing eye drops (Commercial FK506, F127-FK506, 1M-FK506, 2M-FK506) were administered twice daily for 7days, following which the mice were anesthetized by abdominal cavity injection of 1% pentobarbital sodium (0.0075 mL/g), and tropicamide phenylephrine eye drops were used to dilate the pupils. Retinal OCT and fundus images of each eye were captured, centered on the optic nerve head. Corneal OCT images were conducted in the same manner as the retinal OCT process following administration of the drugs and manual blinking. Prior to fluorescence fundus angiography, the mice were injected with 1% fluorescein sodium (0.0075 mL/g) in the abdominal cavity. Fluorescence fundus angiography images were captured in the same manner as the fundus imaging process.

2.24. Statistical analysis

All data are expressed as the mean ± standard error of mean (S.E.M.). Based on the normality of the data distribution, statistical significance was evaluated by one-way ANOVA followed by Tukey's post-hoc test using the GraphPad Prism 8.0 software (GraphPad Software; San Diego, CA, USA). P < 0.05 was considered statistically significant.

3.1. Molecular characteristics of poly(MPOSS-PEG-PPG urethane) (MPEP) copolymers

Random multiblock copolymers, poly(MPOSS-PEG-PPG urethane) (MPEP), with various amounts of MPOSS were synthesized from the polyaddition of MPOSS, PEG, and PPG, which were linked by diisocyanate HDI under a DBT catalyst (Scheme 1a). According to the GPC results in Table 1 and Fig. S3, the MPEP copolymers share a similar molecular weight and dispersity (Đ_M). After a 24 h reaction, there were 15–18 blocks on average in each polymer chain.

The chemical structures of the MPEP copolymers were verified by both the ¹H NMR and ¹³C NMR spectrum (Fig. 1a, S1 and S2). Typically, PEG blocks have characteristic peaks at 3.6 ppm generated by CH₂. PPG blocks present peaks at 1.1 ppm, 3.4 ppm, and 3.5 ppm belonging to CH₃, CH₂, and CH, respectively. HDI linkages show specific peaks at 1.3 ppm, 1.5 ppm, and 3.1 ppm ascribed to CH₂ groups [42]. Referring to the ¹H NMR spectrum of MPOSS, the presence of MPOSS blocks in MPEP copolymers was confirmed by peaks at 0.9 ppm and 0.6 ppm that were attributed to seven CH₃ and CH₂ on the isobutyl groups. The integral ratio of all of these peaks is consistent with the theoretical value. A detailed description of the ¹³C NMR spectrum is provided in the Supporting Information. Moreover, the composition of the MPEP copolymers was calculated from the integral ratio and it was close to the feed ratio (Table 1).

The similarity between the FT-IR spectra of the MPEP copolymer and their macrodiols yielded further evidence that the synthesis protocol was successful (Fig. 1b). New absorption bands appeared at around 1705 cm⁻¹ and 1535 cm⁻¹, representing C O and N–H in the urethane linkage generated from the hydroxyl/isocyanate reaction [43]. An overlapping absorption band for C–O–C in PEG and PPG blocks, as well as Si–O–Si in the MPOSS block, was observed at around 1106 cm⁻¹. A small amount of MPOSS also produces a moderate absorption band at around 1253 cm⁻¹ with Si–C [44]. In addition, absorption bands in the range of 3000–2800 cm⁻¹ corresponded to saturated C–H existing in all three blocks.

The thermal properties of MPEP copolymers were studied by TGA and DSC, and the relative parameters are shown in Table 1. The TGA results demonstrate good thermal stability of all three copolymers, whose degradation temperature for 5% weight loss was above 300 °C. Meanwhile, MPOSS underwent two steps of degradation after 266 °C and after 392 °C, attributed to the decomposition of vertex groups and transformation of the Si–O–Si cage structure into ceramics, respectively (Fig. 2a) [45]. Such changes were reflected in the TGA curves of MPEP copolymers, where the first step overlaps with the degradation of PEG and PPG blocks and the second step takes place after roughly 570 °C.

Regarding the DSC results, a single T_g for MPEP copolymers is presented in the heating process (Fig. 2d), indicating good miscibility of MPOSS with polymers and no microphase separation. The heating curve of MPOSS exhibits two endothermic peaks at 0 °C and 162 °C, likely corresponding to two crystal types (Fig. 2c) [46]. When incorporated into the polymer system, bulk aggregation of MPOSS is restricted due to its low content. Thus, for MPEP copolymers, only one endothermic peak caused by PEG blocks was observed at around 38 °C, lower than the T_m of pure PEG as a consequence of copolymerization. In reverse, the cooling curves of MPEP copolymers show an exothermic peak at around −19 °C with a weak shoulder on the left (Fig. 2b), which may be due to the wide distribution of molecular weight [47].

3.2. Self-assembly properties of MPEP copolymers

In aqueous solution, amphiphilic MPEP copolymers tend to self-assemble from unimers into micelles, driven by hydrophobic interaction. With a hydrophobic core of MPOSS and PPG blocks and a hydrophilic corona of PEG blocks, such micelles can encapsulate hydrophobic molecules in aqueous solution, thus yielding stabilization and solubilization effects [48]. Based on this, CMC values for MPEP copolymers were determined according to the UV–vis absorption spectrum using hydrophobic DPH as the fluorescent probe. Increasing the concentration from 0.0005 to 1 wt% caused the absorption peaks at around 344, 358, and 378 nm to become increasingly larger (Fig. 3a). By plotting graphs of the difference in absorbance at 378 nm and 400 nm (A₃₇₈–A₄₀₀) versus logarithmic concentration, CMC values of 0.5MPEP, 1MPEP, and 2MPEP were estimated to be 0.082 wt%, 0.075 wt%, and 0.071 wt%, respectively, decreasing slightly with increasing MPOSS content (Fig. 3b and S4).

Since PPG is inherent to a hydrophilicity/hydrophobicity transition depending on temperature, the self-assembly behaviors would change in order to reach the delicate balance between hydrophilicity and hydrophobicity. As illustrated in Scheme 1b, the micelles formed by MPEP were further reversibly aggregated at higher temperatures or concentrations. On the macro level, the aqueous MPEP solution turns cloudy, and the temperature at which a 50% reduction in the transmittance at 500 nm occurs is typically defined as the LCST for copolymers at that given concentration [49]. From Fig. 3c, it can be seen that the LCST values for MPEP copolymers adapted according to MPOSS content, which were estimated to be 53.6 °C for 0.5MPEP, 60.0 °C for 1MPEP, and 66.0 °C for 2MPEP at a concentration of 2 wt%.

The assumption of such micelle transformation at low concentrations was also confirmed by the change in particle size. The mean diameters of the particles in aqueous MPEP solution at different temperatures and concentrations as measured by DLS are shown in Table 2 (Supporting Information). By comparison, the samples mostly demonstrated a minor decrease in particle size when the temperature was increased from 25 °C to 37 °C. Different levels of particle size increase were observed after heating the samples to 70 °C, which is more evident at relatively higher concentrations due to the larger portion of micelle aggregates (Fig. 3d, 3e). Moreover, the MPOSS content of copolymers also affected the particle size, especially at relatively high temperatures. As shown in Fig. 3f, the particle size at 70 °C is positively correlated with MPOSS content, and the variation among different concentrations also increased because of the strong hydrophobicity of MPOSS which can facilitate the self-assembly and aggregation of micelles [50]. A detailed description of the MPEP copolymer morphology is provided in Supporting Information.

3.3. Sol-gel transition behavior of MPEP copolymers

Furthermore, as the concentration of aqueous MPEP solution continued to increase, the micelles they first packed into a network and then gradually became a non-flowable gel (Scheme 1b). Our group also reported that block copolymers from methacrylate POSS monomers and PEG macroinitiators self-assemble into micelles and as their concentration increases, they next become packed into macrogels. Moreover, more robust hydrogels were obtained by adding octa-vinyl POSS and carrying out UV curing [50,51]. The sol-gel transition behaviors of MPEP copolymers were investigated by the tube-inverting method. From the temperature- and concentration-dependent phase diagrams (Fig. 4a, S5a and S5b), aqueous MPEP solution may undergo four periods, including clear sol, clear gel, turbid gel, and dehydrated gel, as the temperature increases from 4 °C to 80 °C. Among these, the boundary between clear and turbid gels was hard to identify and they were classified together as being in the gel phase. In contrast, commercially available Pluronic F127 (PEG-PPG) underwent three transitions, including clear sol, clear gel, and clear sol (Fig. 4b). The opaqueness of MPEP thermogels probably results from the special mesh size of their micelle networks close to the visible-light wavelength. A different final state may be due to excessive hydrophobic interactions among multiple MPOSS and PPG blocks, which collapses micelle networks and repels water at high temperatures [52]. Moreover, it is obvious that the critical gelation temperature (CGT) for MPEP copolymers decreases with increasing concentrations and is controlled by MPOSS content. At the same concentration, a lower CGT is presented for MPEP copolymers with a higher MPOSS content. Meanwhile, the CGC for 0.5MPEP, 1MPEP, and 2MPEP was estimated to be 7 wt%, 5 wt%, and 5 wt%, respectively, which is much lower than that of Pluronic F127 and advantageous for biomedical applications.

As a previous study indicated that POSS incorporation into the polymer chain markedly altered the rheological behavior of copolymers [53]. Rheological measurements were used to characterize the temperature responsiveness of MPEP copolymers. The concentrations of the tested samples were chosen to be 8 wt% for 0.5MPEP, 6 wt% for 1MPEP, and 5 wt% for 2MPEP, whose corresponding CGT values in the phase diagrams were approximately 32 °C, 24 °C, and 26 °C. According to the temperature sweep results over a the range from 4 to 80 °C (Fig. 4c, S5c, S5d), the storage modulus G′ and loss modulus G″ both increased with rising temperature at different rates and exhibited a crossover representing the sol-gel transition. The CGT determined from this method was approximately 32.0 °C, 16.3 °C, and 17.8 °C for 0.5MPEP, 1MPEP, and 2MPEP, respectively. While the CGT of 0.5MPEP obtained by the two methods matches well, the other two data points obtained from rheological measurements were lower than those obtained from the phase diagrams, probably because the temperature at which G′ is equal to G″ is actually a semi-solid state and not firm enough to be regarded as a gel in the tube-inverting method. From a temperature of approximately 70 °C, G′ and G″ began to decrease slightly and then became close again, which suggests the collapse of gels and coincides with the results of the tube-inverting method. In addition, the temperature ramp from a lower temperature such as 15 °C to body temperature was conducted to evaluate the response rate of the materials to the temperature change. As can be seen from Fig. 4d, S5e, and S5f, a fast sol-gel transition occurred that was almost in sync with the temperature change. This agreement is promising for utilizing injectable materials for in situ formation of hydrogels in biomedical applications. The temperature of the healthy corneal surface is around 34.5 °C, administering the drug formulation onto a surface at such a temperature could induce the sol-to-gel transition [54]. As such, a transparent surface adherent film will be formed which prolongs the drug retention on the surface and further leads to an increased therapeutic efficacy.

3.4. Light transmittance of the blank hydrogel

According to previous studies, hydrogels are considered to be transparent if their light transmittance is greater than 90% over the visible wavelength range of 390–780 nm [37]. Since eye hydrogels require high transparency, we tested the absorbance of 1MPEP, 2MPEP, and F127 after gelling, and calculated the light transmittance according to the formula A = –log T% (Fig. 5a). We found that the light transmittance of the hydrogel was greater than 90% within the visible light wavelength range, indicating that the transparency of 1MPEP, 2MPEP, and F127 is excellent and will not affect vision. In addition, we also calculated the light transmittance of the commercial liquid carbomer and found that it was not significantly different.

3.5. Degradation of the hydrogel and in vitro release of FK506

Previous studies have shown that it is beneficial to formulate copolymers into the hydrogels for ocular applications, because the hydrogel delivery system used on the ocular surface can not only offer a slow-release rate of encapsulated payloads for stable drug concentration, but also effectively resist the ocular surface clearance mechanism for improving the bioavailability [55,56]. We prepared 1MPEP, 2MPEP, and F127 hydrogels and conducted in vitro degradation experiments to compare the degradation rates. As shown in Fig. 5b, the degradation rate of F127 was very fast and almost completely degraded on day 1; however, degradation of the 1MPEP hydrogel was slower and was completely degraded only on day 6, whereas that of the 2MPEP hydrogel was even slower, approaching complete degradation on day 14. The slow degradation of the hydrogel on the ocular surface could reduce the rate at which the drug delivery system turns into an aqueous solution. Such an effect will retard flushing of the drug delivery system from the ocular surface. Besides, the degradation rate of the hydrogel affects the release rate of the drug from the hydrogel. The sustained release of the drug can maintain the drug concentration at a relatively high level on the ocular surface for a longer time. Previous studies compared the pharmacokinetics of topical hydrogels and eye drops, the results indicate that administering a drug in a hydrogel sustains its release and significantly improves its bioavailability. These characteristics also explain the reason for our preference in using a hydrogel formulation as drug delivery system [11,57,58].

Moreover, we prepared 1MPEP, 2MPEP, and F127 hydrogels containing FK506. FK506-loaded polymer micelles were prepared first; the FK506 encapsulation efficiency of 1MPEP and 2MPEP reached 93.4% and 92.8%, respectively, and the FK506 loading capacity was 4.46% and 4.43%, respectively. Subsequently, the blank copolymers were added to reach the correct hydrogel concentration, and the temperature was raised to transform the solution into a gel state. As shown in Fig. S6, in comparison with Commercial FK506, the transparency of the FK506 solution encased in F127, 1MPEP, and 2MPEP polymers was significantly improved, which indicates that the polymer micelles significantly improved the solubility of FK506 in aqueous solution. To explore the FK506 release rate from the hydrogel, an in vitro drug release experiment was carried out (Fig. 5c). The results show that FK506 was released fastest from the F127 hydrogel, while it was released slower from the 1MPEP hydrogel, and even more so from the 2MPEP hydrogel, which indicates that the hydrogel containing POSS groups had a higher strength and better slow-release effects.

3.6. In vitro cytotoxicity of the hydrogel

To test the cell compatibility of the hydrogel, we used the MTT assay to evaluate the toxicity of the hydrogel extract towards HCE cells (Fig. 5d). The results showed that irrespective of the presence of FK506, the F127 hydrogel extract displayed the highest cytotoxicity, the 1MPEP extract was much less cytotoxic, and the 2MPEP extract was the least cytotoxic. This may be because the degradation rate of F127 is the fastest, resulting in more material entering the extract whereas the 1MPEP and 2MPEP degradation rates are slower, which accounts for them being less cytotoxic. Since the POSS structure has been proven to have very good biocompatibility [59,60], we used the hydrogel extract method to detect the cytotoxicity of the hydrogel. The degree of degradation of the hydrogel was inconsistent at 24h or 48h. It can be seen from the hydrogel degradation experiment that the degradation rate of 2MPEP is slower than 1MPEP, so the concentration of copolymer leached into the medium is lower, so it appears to be less cytotoxic.

3.7. The interaction between mucin and POSS-polymers

Generally, mucus can interact with additives through electrostatic interactions, van der Waals forces, hydrophobic forces, hydrogen bonds, etc. [61] The mucosal surface of the eyeball is covered by a layer of mucus, whose main component is mucin. Various systems have been used to extend the adhesion of drugs by enhancing their interaction with the mucin on the surface of the eyeball [62,63]. There are reports that POSS-containing polymers can enhance adhesion to mucin since the POSS group facilitates protein anchoring [33,64]. With a view to exploring the mechanism underlying the enhancement of mucosal adhesion by POSS-containing polymers, we used surface plasmon resonance technology (Biacore T200 system) to verify the binding affinity between the polymers and mucin. Binding affinity is generally measured and reported as the equilibrium dissociation constant (KD); the smaller the KD value, the greater the binding affinity of the ligand for its target. As shown in Fig. 6, the binding force between polymer F127 and mucin was relatively weak, with a KD value of approximately 11.380 μM. Interestingly, for the polymers 1MPEP and 2MPEP containing POSS groups, the affinity to mucin was significantly improved, with KD values of 0.508 μM and 0.409 μM, respectively. The results demonstrate that the addition of POSS groups to the polymer chain helps improve its binding affinity to mucin. It is speculated that the strong hydrophobic POSS groups can interact with the exposed hydrophobic domains around mucin [61,65]. Therefore, POSS-containing polymers may be used to improve adhesion of the drug-carrying system to the ocular surface and prolong the drug retention time.

3.8. Hydrogel gelation and retention on the ocular surface and ocular pharmacokinetics studies

In comparison with commonly used eye drops, hydrogels may remain on the ocular surface for a longer period of time, providing more sustained effects [66]. Blinking and tear flushing cause drug loss [67], whereas the adhesive properties of hydrogels contribute to resistance of their removal from the ocular surface. To evaluate the in situ thermo-reversible crosslinking of the hydrogels on the ocular surface, slit-lamp and OCT were conducted. The slit-lamp images in Fig. 7a show that no residue remained on the cornea following topical administration of Commercial FK506. In contrast, following topical administration of F127-FK506 (PEG-PPG-FK506), 1M-FK506 (1MPEP-FK506), and 2M-FK506 (2MPEP-FK506), the formulations gelled on the cornea; however, after blinking, the F127-FK506 and 1M-FK506 hydrogels underwent detachment (Fig. 7a, the red arrow indicates the hydrogel position on the cornea and the green arrow indicates the hydrogel detached from the cornea). The 2M-FK506 hydrogel attached well to the cornea and its integrity was not damaged by blinking, suggesting that the 2M-FK506 hydrogel provided stronger adhesion. Subsequently, we attempted to visualize via optical coherence tomography (OCT), its attachment to the cornea surface. The gelation of the F127-FK506 hydrogel resulted in a fragmented coating on the cornea following blinking (red arrow in Fig. 7b, F127-FK506). The 1M-FK506 hydrogel behaved as a nonuniform gel on the cornea after blinking (red arrow in Fig. 7b, 1M-FK506). In contrast, the 2M-FK506 hydrogel formed a smooth gel on the cornea and blinking did not cause loss or damage to the hydrogel (red arrow in Fig. 7b, 2M-FK506), indicating that the 2M-FK506 hydrogel effectively enhanced drug duration on the ocular surface.

In the ocular pharmacokinetics studies, it was observed that 2M-FK506 provided a significantly increased drug concentrations in rabbit corneas at 1, 2, 4 and 8h after administration compared with F127-FK506 and Commercial FK506 (Fig. 7c). Besides, it could be found that 2M-FK506 provided a significantly increased FK506 concentrations at 1 and 2h after administration compared with 1M-FK506, F127-FK506 and Commercial FK506 in the rabbits’ aqueous humor (Fig. S10a and S10b). When compared with F127-FK506, MPEP-FK506 hydrogel effectively resisted the ocular surface clearance by adhering to the ocular surface, which significantly increased the drug concentration of FK506 in the cornea and aqueous humor. Similarly, the thermo-responsive hydrogel system effectively increased the concentration of the FK506 in the cornea and aqueous humor compared to Commercial FK506. Its delivery in the hydrogel system enabled a slow and sustained release of FK506. Therefore, MPEP-FK506 hydrogel with enhanced ocular surface adhesive and thermo-responsive properties was found to increase the efficiency of the ocular surface drug delivery.

3.9. Evaluation of the hydrogels on the dry eye murine model

Different eye drops were used to assess the efficacy of FK506-carrying hydrogels on the murine dry eye model. The cornea was stained using OGD dye and it appeared green in damaged deepithelialized regions of the outer layer. As can be seen from Fig. 8a, the OGD staining was severe in the model, DS5, PBS-treated, F127-treated, 1MPEP-treated, and 2MPEP-treated groups; however, the Commercial FK506, F127-FK506, 1M-FK506, and 2M-FK506 groups displayed decreased OGD staining, indicating that these four formulations had good reparative effects on corneal epithelial defects caused by desiccating stress. It is noteworthy that statistical analysis of corneal OGD staining showed that topical application of 2M-FK506 significantly decreased its intensity (Fig. 8c). Owing to stronger adhesion to the cornea, continued slow release of FK506 from this hydrogel protected the corneal epithelium from exposure to a tear deficient environment for a long period of time. Tear production is a very important clinical parameter for the evaluation of the development of dry eye since tears protect the corneal epithelial cells from being lost due to inadequate hydration of the ocular surface [68]. Topical application of 2M-FK506 significantly increased tear production as compared with the other groups, with the exception of F127-FK506, for which the difference was not significant (Fig. 8b). According to these results, 2M-FK506 is an efficient formulation for reversing insufficient tear production due to dry eye.

MMP-3 and MMP-9 expression levels are key indicators of corneal barrier function injuries caused by dry eye; therefore, immunofluorescence staining was employed to determine their levels of MMP-3 and MMP-9 in the corneal epithelium (Fig. 8d, S9). Topical application of Commercial FK506, F127-FK506, 1M-FK506, and 2M-FK506 decreased the expression and staining of MMP-3 in the corneal epithelium. Among these hydrogels, topical application of 2M-FK506 decreased the levels of MMP-3 the most, suggesting improvement of the corneal barrier function under desiccating stress (Fig. 8e, 8f).

Dry eye not only causes damage to the corneal barrier function but also roughens the smooth corneal epithelial surface. According to corneal H&E staining, 2M-FK506 hastened more the recovery of corneal epithelial morphology than that caused by other groups (Fig. S8a). Moreover, goblet cells are simple columnar epithelial cells that secrete mucin to form the tear film, which resists changes in the environment of the ocular surface [69], but inflammation caused by dry eye results in goblet cell loss. The 2M-FK506 hydrogel significantly rescued goblet cell loss under desiccating stress (Fig. 9a, 9c, S8b), indicating that 2M-FK506 suppressed inflammation over a long period of time and played an active role in promoting the recovery of ocular surface injury due to long-term retention of the hydrogel. Dry eye is a chronic inflammatory immune disease characterized by infiltration of CD4⁺ T cells into the conjunctiva [70]. FK506 (Tacrolimus) binds the immunophilins cyclophilin and FKBP12, which inhibit calcineurin activity and direct the metabolic reprogramming of CD4⁺ T cells to boost their immunosuppressive effects [71]. The results of the present study indicate that the topical application of 2M-FK506 effectively suppressed the DS-induced infiltration of CD4⁺ T cells in the conjunctiva (Fig. 9b). Considering that CD4⁺ T cells play a key role in FK506 treatment of dry eye, the expression of CD4 was further verified. In accordance, 2M-FK506 effectively inhibited the expression of CD4 (Fig. 9d), indicating that 2MPEP allowed greater efficacy of FK506.

3.10. Safety of the hydrogel in mice

Zhu et al. speculated that the induction of a chronic immune response to nanoparticles might injure the layers of retinal vessels, degenerate retinal cells, and induce neovascularization [72]. Depending on the particulate makeup, it may even elicit a systemic inflammatory response. The therapeutic effects of the MPEP-FK506 formulation have previously been explored on the dry eye murine model; however, its safety still requires verification. Therefore, fundus imaging, fluorescence fundus angiography, and OCT examination were performed following topical administration of different formulations for 7 days. Fundus imaging is a common examination technique to observe the morphology of the retina, optic disc, macular area, and retinal blood vessels, as well as whether there is hemorrhage, exudation, hemangioma, retinal degeneration, retinal hiatus, new blood vessels, atrophy spots, pigment disorder, and other changes to the retina [73]. Fundus angiography is mainly used for further observation of tiny structural changes in fundus blood vessels. From the fundus images and fluorescence fundus angiography, no obvious leakage of fundus vessels, vascular enlargement, or abnormal routes were observed after treatment (Fig. 10a, 10b). OCT is a diagnostic technique in ophthalmology that possesses the advantages of being non-invasive, having a high resolution at the cellular level and a fast-imaging speed, and solving the problem of in vivo study of the retina and cornea [74]. OCT examination of the retina on day 7 showed that the structure of each retinal layer was arranged in an orderly fashion with no obvious retinal detachment or degeneration, and no difference was observed in retinal thickness among the groups (Fig. 10c). Besides, the appearance of the cornea remained normal after treatment among the groups (Fig. 10d). Variation in corneal or retinal thickness indicates the presence of pathological lesions. However, the present data shows that hydrogel administration to the ocular surface did not change the thickness of the cornea and retina compared to the control group irrespective of the time of exposure (Fig. 10e, 10f). Therefore, hydrogel administration does not pose any risk to ocular surface health.