p(MPC-co-DMA) was synthesized and characterized as shown in Fig. 1a. The successful synthesis of DMA monomer was confirmed as shown in Figure S1. FTIR spectra of DMA monomer depicted the characteristic peak at 1650 cm− 1, which signifies the stretching vibration of C = O in amide group (Figure S2). The result suggested the amidation reaction and polymerizable unit had been conjugated with catechol motif. In addition, the peak at 3225 cm− 1 was assigned to -OH stretching vibration of catechol unit. The result indicates the availability of DMA for adhesive function. Prior to copolymerization, given that MPC monomer has low solubility in DMSO solvent, the successful p(MPC) synthesis was confirmed using NMR spectra analysis. As shown in Figure S3, the NMR spectrum of p(MPC) showed no proton signals of residual vinylic double bond at δ = 5.5 − 6.5 ppm, indicating that the resultant powder from our synthesis was in polymer configuration.

A series of copolymers with varying compositions were produced by using the feeding molar ratios of MPC and DMA of 1:1, 3:1, and 6:1. Figure 1b and Figure S4 display the 1H NMR spectra of various p(MPC-co-DMA) samples using D2O as the solvent. The signals at 6.79 and 6.93 ppm correspond to 3,4,6-trihydrophenyl groups of DMA. In addition, the peaks at 3.26, 3.71, 4.14 and 4.34 ppm are attributed to methylene groups and the methyl groups in MPC. The functional groups of all polymer compositions were further analyzed using FTIR spectra. Figure S5 showed the characteristic bands at 1244, 1170 and 1082 cm− 1, which are attributed to phosphate groups in MPC moiety. Additionally, the absorbance peak at 967 cm− 1 is assigned to the antisymmetric stretches of the C − N bonds in the N-(CH3)3 of MPC moiety. Notably, in various p(MPC-co-DMA) polymers, a novel peak at 1527 cm− 1 may represent the stretching vibration of C-N bond in amide motif of DMA. To this end, the NMR and FTIR spectra have confirmed the successful synthesis of the p(MPC-co-DMA) via free-radical copolymerization. For all of the polymer samples, the actual ratios of MPC to DMA are determined by dividing the integrated area of the aromatic protons of DMA (δ = 6.79 to 6.93 ppm) by the integrated area of the characteristic peaks for MPC (δ = 3.26 ppm). For feeding ratios of MPC to DMA of 1:1, 3:1, and 6:1, the respective ratios for the resultant samples is 2.77:1, 4.49:1 and 7.09:1 (Table S1). Accordingly, the highest DMA content is achieved at a value of 26.5 mol% for p(MPC1-co-DMA1). In nature, enriched catechol residues have been documented with values of 30 mol% in the interfacial mussel’s foot proteins (mfp-3 and mfp-5) [41]. Therefore, p(MPC1-co-DMA1) closely resembles a mussel’s foot protein in term of catechol content. The respective catechol content of p(MPC3-co-DMA1) and p(MPC6-co-DMA1) is 18.2 and 12.4 mol%. Figure 1c further confirms the higher catechol content in the copolymers at 1 wt% in the order of molar ratio 6:1 < 3:1 < 1:1, which is consistent with the molar ratio recorded using 1H NMR. In addition, the catechol content of copolymers quantified by UV-VIS (Fig. 1c, Figure S6 and Table S1) can support the data of NMR spectra (Fig. 1b and Table S1).

Synthesis routes and characterizations of copolymers. (a) The preparation of DMA monomer and p(MPC-co-DMA) copolymers. (b) 1H NMR spectrum of p(MPC1-co-DMA1) in D2O. (c) UV-VIS spectra of p(MPC-co-DMA) (1 wt%) with different compositions in PBS at pH 7.4

The number-average molecular weight (Mn) of the copolymers determined by GPC is shown in Figure S7 and Table S1. The length of the polymer chain decreases as MPC feeding mass increases, as shown in Experimental section. It is difficult to dissolve DMA homopolymers in organic solvents, and the polymers only demonstrate limited solubility in dimethylformamide [42]. Without sufficient protection, radical scavenging ability by catechol hinders the polymerization reaction because radicals are quenched by semi-quinone radicals, leading to the formation of intra- or inter-crosslinked structure [43, 44]. Therefore, the copolymers with greater MPC feeding content may easily form a long polymer chain. Zhang and colleagues also found that the solubility of copolymer at the molar ratio of 1:3 (ratio between MPC: DMA) was limited at 4 mg/mL [38]. In this study, our attempt to increase DMA content with feeding ratios 1:3 and 1:6 showed a sight of catechol oxidation (a darken color in the polymerization solution) as shown in Figure S8a. Additionally, the vinylic signals of DMA at δ = 5–6 ppm were still visible after the reaction, indicating the low polymerization efficacy of MPC: DMA feeding ratios at 1:3 and 1:6 (Figure S8b). Therefore, the highest DMA portion was only observed at 1:1 for this study.

Lubricating agents can act as an osmoprotectant to ease dry eye discomfort by restoring the physiological osmolarity of tear film and decreasing the effect of hyperosmotic distress to corneal cells [45]. Therefore, the lubrication properties of mucin and synthesized polymers on silicon tubes were investigated. For the pristine silicon tube (Bare), the COF was constant at ∼ 2.5 for all three sliding cycles (Fig. 2). After mucin deposition, the COF value for the surface initially decreases to ∼ 0.57, and then gradually increases to ∼ 0.98 after the 2nd and 3rd cycles. The increase in COF tendency was also observed in other polymer coating substrates. Because a mucin layer was generated on coating surface via non-specific adsorption, the interfacial shear force between the coating and the silicon sheet could remove a portion of adsorbed mucin macromolecules after each sliding cycle. As a result, the stability of the lubrication that the mucin-bound surface provides is compromised. For a polymer-bound surface, the smoothness of the surfaces in which p(MPC) is deposited surfaces increases to give a COF of 0.46 to 0.66 after three cycles but surfaces on which p(MPC1-co-DMA1) is deposited experience only a slight change in COF from 0.36 to 0.37. The difference in COF values for the two polymer compositions probably reflects the functionality of DMA. Catechol has a high adhesive strength on a variety of surfaces [41], which could increase the stability of the coating against shear stress. A significantly lower COF value for p(MPC)- and p(MPC1-co-DMA1)-coated surfaces than for mucin-bound surfaces at the final sliding cycle (p < 0.05) indicates the enhanced lubrication properties of zwitterionic polymers. Considering the high dipole moment of zwitterionic pendants, the strong ionic-driven hydration shell surrounding quaternary amine N+(CH3)3 and phosphonate PO4− is not easily squeeze out under shear or compression, facilitating a greater degree of viscous dissipation than for “non-hydration” water [46, 47]. Therefore, the interfacial friction for surfaces on which MPC is deposited is lower than the friction for mucin-bound surfaces.

Hydration lubrication evaluation of as-prepared polymers. COF of silicon tube (Bare), Mucin-coated, p(MPC)-coated and p(MPC1-co-DMA1)-coated silicon tubes in DI water. The friction tests were performed in each sample at a normal load of 2 N and a sliding speed at 150 mm/mL after three sliding cycles. Values are mean ± SD (n = 3). *p < 0.05 vs. mucin

Commercial porcine stomach mucin was used to study the mucoadhesive properties of p(MPC-co-DMA) copolymers. The interaction between catechol-containing copolymer and mucin was first measured using UV-VIS spectroscopy. 100 µL of copolymer solutions (10 mg/mL) were added to 1 mL of mucin solutions (1 mg/mL) to achieve a mass ratio of 1:1 in PBS buffer at pH 7.4. As shown in Figure S9a, the mucin solution exhibits a strong absorbance signal at 260 nm. There is no significant difference in the position of the absorbance peak for mucin and mucin-p(MPC). In contrast, UV-VIS demonstrated a slight red shift if mucin solution is mixed with catechol-functionalized polymer. The absorbance maxima for mucin shifts toward a lower wavelength region and there is an increase in the catechol content, which demonstrates that a mucin-catechol complex forms [34, 48, 49]. This behavior is also noted by other study, suggesting a change in peptide strands of mucin molecules and their hydrophobicity [48]. As a zwitterionic moiety, MPC segments of the copolymer resist non-specific interaction with biomacromolecules [50]. Hence, mucin-copolymer complex formation is solely contributed by DMA segments. As the catechol degree increases, more reactive sites are available for conjugation. Furthermore, copolymers separately recorded show no sign of catechol oxidation [51, 52], displaying only one absorbance peak centered at 280 nm in PBS buffer (Figure S9a). Hence, bidentate catecholic hydroxyls can induce strong hydrogen bonds [30, 49], allowing initial contact for mucin complexation.

Mucins play a vital role in the innate protection of the eyes [53]. Hence, the mucoadhesive properties of p(MPC-co-DMA) were further studied using a mucin-deposited substrate prior to in vitro and in vivo studies. Mucins were coated onto a plasma-treated silicon wafer by passive adsorption for 1 h. To verify the stability of the coating, WCA measurement was used to qualify the relative hydrophilicity after mucin deposition. Figure S9b shows that there is increased wetting of the surface of the silicon wafer so there is a significant reduction in the value of the WCA from 68° to 26.4°. Mucins exhibit a high binding affinity with water molecules [54]. This renders coated silicon surface with improved hydrophilicity. After deposition with mucin, copolymers were allowed to deposit on the substrates for 1 h. Interestingly, p(MPC1-co-DMA1) further induced wetting of mucin film (WCA at 23.2°), while a similar result was not observed in samples coated with other polymer compositions. Because zwitterions can exhibit superior wetting via ionic solvation [15], the decrease in the WCA implies the abundance of MPC segments on the mucin layer of p(MPC1-co-DMA1).

The surface elemental composition of the modified surface was investigated using XPS (Fig. 3). The characteristic phosphorus P2p signal at 133.8 eV is only attributed to the phosphate group of MPC [55]. Figure 3a shows that the signal is absent for the mucin layer. For different polymer compositions, only catechol-containing copolymers are immobilized on mucin-coated surfaces. The high-resolution spectra for the phosphorus to carbon ratio (P/C) also shows that there is an increase from 0.07 to 0.14 (Table S2), with increasing degree of catechol: p(MPC) < p(MPC6-co-DMA1) < p(MPC3-co-DMA1) < p(MPC1-co-DMA1). The N1s core-level spectra features a peak for mucin at 400.3 eV, which is associated with the primary and secondary amines at the N-terminus and peptide bonds in mucins (Fig. 3b) [53, 54]. After the copolymer deposition, a new N1s peak at 402.7 eV appears, which demonstrates the existence of the quaternary amine N+(CH3)3 in MPC. The presence of phosphorus and charged N+ are clear evidence that zwitterionic moieties are immobilized [58]. Moreover, introducing acrylamide CH-NH-CH section into DMA increases N elemental signals on mucin-deposited surface. As a result, the atomic ratio of nitrogen to carbon (N/C) is greater for samples on which a copolymer is deposited than for the mucin layer. These results demonstrate that DMA features an anchoring functionality.

Mucoadhesion evaluation of as-prepared polymers. XPS spectra of the modified substrates: Core-level binding energy of elements: P2p (a), N1s (b), C1s (c), and O1s (d). Changes in frequency Δf with time were recorded with eQCM for mucin deposition (e) and the interaction between polymers on mucin layer (f)

.

Catechols can react with thiol and amine via Schiff base and Michael addition reactions [59]. Thiol-containing residues of mucins have been reported by other studies [9, 60]. However, the cysteine components of mucin vary between purification batches due to the variations in animals or purification efficiency [61]. Furthermore, cysteine is also not as abundant as amine residues [34]. Thus, the mucoadhesive properties of polymers are assumed to be most obvious between catechol and amine portions in mucins. To investigate this possibility, the core-level spectra of C1s and O1s were further analyzed in the mucin layer with and without deposition of p(MPC1-co-DMA1) (Fig. 3c and d). The C1s spectral signal for mucin layer shows three respective peak components for C-C/C-H, C-O/C-N, and CONH/COOH at 284.8, 286.4, and 288.2 eV (Fig. 3c) [56, 62]. In the p(MPC1-co-DMA1) spectrum, the peak intensities for C-C/C-H and C-O/C-N increase, which indicates the presence of the copolymer backbone and the tethered residues (acrylamide in DMA and acrylate in MPC) on mucin. In the O1s spectra, there is an increase in O = C for a binding energy of 531.2 eV in p(MPC1-co-DMA1) [63, 64], which shows that catechol moieties were oxidized (Fig. 3d). Interestingly, catechol moieties were in a reduced state as shown in UV-VIS spectra, which differs from the result observed in XPS. The oxygen-induced conversion of catechol to quinone can occur above pH 5; however, the process is rather dull, requiring days of monitoring [57]. Hence, it is likely that catechol-to-quinone conversion did not occur after hours under a physiological condition during UV-VIS measurement. Furthermore, amine-rich mucins may accelerate the oxidation of catechol residues by promoting Michael addition and Schiff base reactions, similar to the co-deposition between DA and polyethyleneimine in the work of Xu and colleagues [62].

Real-time measurement of p(MPC1-co-DMA1) adsorption on the mucin adlayer used eQCM to analyze adhesion behavior. The reduction in frequency shift (Δf) is associated with adsorbed mass on the sensor. Herein, we first examined the adlayer formation of mucin used in this study. Mucin (0.025 mg/mL) was injected and allowed to adsorb in the flow cell for 35 min, forming a mucin adlayer with a stable mass density, as shown in Fig. 3e. The adlayer was then rinsed with PBS for 30 min to remove loosely-bound mucin. The polymer compositions were then allowed to flow for 30 min before rinsing with PBS as shown in Fig. 3f. For p(MPC) homopolymer, there was a gradual decrease in Δf after injection, then a small increase and halt at a mass density of 21.6 ng/cm2. The main driving force for non-specific adsorption of p(MPC) is Van der Waal forces, which are common but relatively weak secondary bonds in close proximity [65]. After p(MPC1-co-DMA1), there was an abrupt decrease (phase I) followed by a steady decrease (phase II) in Δf. In phase I, the copolymer passively diffuses and is absorbs onto mucin adlayer, and the interaction between catechol and mucin provides anchoring sites to allow stable adhesion of the copolymer. In Phase II, the catechol groups that remain on the adsorbed copolymers also allow inter-chain interactions with free-floating chains, leading to an increase in the adsorbed mass density. As a result, the final mass load for the copolymer is 213.2 ng/cm2, which is approximately 10 times the value for p(MPC). To sum up, the adhesive mechanisms for catechol-functionalized polymers have three aspects: (1) catecholic functionalization promotes the immobilization of zwitterions on the mucin surface, as shown by the XPS spectra; (2) bidentate catecholic residues of the copolymer allow strong and instantaneous hydrogen bonding with the mucin layer; and (3) the catechol-amine adducts that gradually form between the mucin and the copolymer increase the mucoadhesive properties of the copolymers.

Prior to the antioxidant and anti-inflammatory evaluation, the cytotoxicity of p(MPC) and p(MPC-co-DMA) is determined. The biocompatibility of all synthetic polymers is determined using a MTS assay and live/dead staining. For this work, SIRC cells from the cornea of normal rabbit was used. The mitochondrial dehydrogenase activity (MTS activity) for the Ctrl group was defined as 100%. The metabolic activities and viability levels of SIRC cells after exposure to polymer samples were depicted in Figure S10 and S11. After two days of incubation, all the polymer samples did not alter the cell morphology regardless of their concentration (ranging from 0.1 to 1 mg/mL) (Figure S10a and b). Quantitative data for MTS activity and live/dead assay also supported data from the microscopic images in that the values for metabolic activity and cell viability are 95% of the values for the Ctrl group (Figure S10 and S11). In addition to cell cytotoxicity, an alkaline comet assay was conducted to investigate the potential genotoxicity of the polymers. Figure S12 indicated intact nuclei with smooth edges in all samples. Moreover, the tail lengths of comet in all compositions showed no apparent difference with the Ctrl (0.9 ∼ 1.2 μm). The result shows that no DNA breakage was caused by the polymers [66, 67]. Overall, the synthetic polymers used in this study exhibited no harmful effect on SIRC cells. Therefore, the in vitro and in vivo studies were conducted and evaluated using the highest polymer concentration of 1 mg/mL.

An increase in the amount of ROS is a major factor of the DED etiology [68]. The over-expression of intracellular ROS on the ocular surface has various adverse outcomes, including DNA, lipids, protein damage and cell apoptosis [69]. To determine the antioxidant activity of the polymers, this study uses DPPH to determine the free radical scavenging ability for various p(MPC-co-DMA) groups. The results are shown in Fig. 4a and b. DPPH antioxidant solution has a maximum absorbance at 517 nm [70]. As the radical scavenging capability of antioxidant agent increases, the color of DPPH solution gradually changed from dark purple to light yellow [71]. Quantitatively, the respective radical inhibition percentages of p(MPC1-co-DMA1), p(MPC3-co-DMA1), and p(MPC6-co-DMA1) was 87.52 ± 5.36%, 60.14 ± 4.89% and 35.63 ± 4.94%. In contrast, the value of p(MPC) was similar to blank sample. These results highlighted that the antioxidant capability was closely associated with the degree of catechol groups within copolymers, which has obtained wide recognition in other literature [38, 44]. To support this assumption, we continued to conduct Folin-Ciocalteu assay to determine the polyphenol content of p(MPC-co-DMA). The color of a p(MPC1-co-DMA1) solution turned darker blue as shown in Fig. 4c, implying a higher level of polyphenols [72]. The polyphenol content at 85.7 ± 4.7 mg, 59.0 ± 5.0 mg, and 39.9 ± 3.9 mg for gallic acid equivalent (GAE)/g polymer decreases in the order (Fig. 4d): p(MPC1-co-DMA1) > p(MPC3-co-DMA1) > p(MPC6-co-DMA1).

Anti-oxidation evaluation of as-prepared polymers. Photographs of the reaction of DPPH reagent (a) and Folin-Ciocalteu reagent (c) with the test samples including p(MPC), p(MPC1-co-DMA1), p(MPC3-co-DMA1) and p(MPC6-co-DMA1). DPPH scavenging activity (b) and total phenolics content (d) of the various samples were analyzed by UV-VIS spectrophotometry. Values are mean ± SD (n = 5). *p < 0.05 vs. all groups. The blank group: without p(MPC-co-DMA) sample

Cellular models of oxidative stress were adopted to assess the antioxidant capabilities of p(MPC-co-DMA) materials in vitro. DCFH-DA, which is a cell-permeable non-fluorescent probe, can be hydrolyzed to become non-permeable DCFH within cells. DCFH can be further oxidized by increased intracellular ROS (superoxide O2−), forming visible fluorescent signals as shown for HP group (Fig. 5a). The fluorescent signals from cells decrease significantly after two days of incubation with catechol-containing polymers. The greater the concentration of catechols in p(MPC1-co-DMA1), the more the ROS level reduces to the value for the Ctrl group (Fig. 5b). Catechol can quench free radicals by supplying hydrogen atoms from the phenolic hydroxyl residues, but catechol can also reduce other reactive agents through electron transfer, and the resulting phenoxyl radical can further interact with the second radical to form stable quinone structures [73]. The regulation of calcium has a crucial role in chemical signaling, thus the over-expression of intracellular calcium is considered a contributing factor leading to cell death [74, 75]. Therefore, further studies on intracellular calcium levels were conducted to verify cell survivability (Fig. 5c and d). The intracellular calcium level after H2O2 treatment decreased significantly when cells were co-incubated with p(MPC-co-DMA) copolymers in the order: Ctrl < p(MPC1-co-DMA1) < p(MPC3-co-DMA1) < p(MPC6-co-DMA1) < p(MPC) = HP. These results confirm that catechol-functionalized polymers inhibit ROS damage.

In vitro anti-oxidation and anti-inflammation of as-prepared polymers. (a) Representative fluorescent images and (b) intracellular levels of ROS measured by the fluorescence intensity of DCFH-DA of the SIRC cells after incubation with different samples for 24 h and further exposure to H2O2 for 24 h. (c) Representative fluorescent images of the SIRC cells and (d) intracellular levels of calcium measured by the fluorescence intensity of Fura-2, AM after incubation with different samples for 24 h and further exposure to H2O2 for 24 h. Levels of (e) IL-6 and (f) TNF-α from the p(MPC-co-DMA) samples (p(MPC), p(MPC1-co-DMA1), p(MPC3-co-DMA1) and p(MPC6-co-DMA1)). Scale bars: 100 μm. Values are mean ± SD (n = 5). *p < 0.05 vs. all groups; #p < 0.05 vs. LPS, p(MPC), p(MPC3-co-DMA1) and p(MPC6-co-DMA1) groups; ^p < 0.05 vs. Ctrl, p(MPC1-co-DMA1), p(MPC3-co-DMA1) and p(MPC6-co-DMA1) groups

Inflammation is an underlying factor for DED [69]. Hence, the pro-inflammatory factors (IL-6 and TNF-α) were further examined. Lipopolysaccharide-induced cells were also used as a disease control group (LPS) to mimic DED symptoms. As depicted in Fig. 5e and f, the expression of IL-6 and TNF-α decreases significantly (p < 0.05) for LPS groups that are subject to p(MPC-co-DMA) treatment. Moreover, p(MPC1-co-DMA1)-treated group exhibits the highest anti-inflammatory capability of the catechol-containing groups, which demonstrates their antioxidant capacity.

In this study, the normal rabbit corneas were monitored under a slit lamp for h12 and 4 days post-instillation to validate any corneal injury caused by polymers. The rabbit eyes receiving topically instilled polymeric drops showed no signs of corneal redness, edema, or inflammation (Figure S13). Quantitatively, all observations of the modified Draize test resulted in scores of zero as shown in Table S3. Furthermore, intraocular pressure (IOP) value after polymer instillment remained comparable with Ctrl group, which confirmed the safety of the synthetic polymers (Figure S14).

The bio-adhesive properties of p(MPC-co-DMA) copolymers were further evaluated in vivo. In this investigation, the content of p(MPC-co-DMA) retained on the cornea was examined by SEM-EDS after h12 and 4 days of topical administration of nanotherapeutic formulations (Fig. 6a). No elemental phosphorous (red dot) was detected in the Ctrl group. The meibomian lipid layer is constituted by the structural organization of polar lipids (of which phospholipids are the most important) [76]. However, the precorneal tear film is not a part from corneal tissue. This result may explain the absence of phosphorus signal on ocular surface. Hence, the measurement for elemental phosphorous can only be attributed to phosphorylcholine pendants from MPC-containing polymers. For p(MPC) group, the phosphorus portion was ∼ 31% after h12 of treatment. In contrast, at least 50% of the total phosphorus signal, which corresponds to p(MPC-co-DMA), remained visible on the ocular surface (75%, 62%, and 53% for the feeding ratios of 1:1, 3:1 and 6:1, respectively). The results suggested the concentration of catechol significantly affect retention efficiency. Remarkably, phosphorus coverage of p(MPC1-co-DMA1) re