A transparent PVA-based polymer sunscreen protects retinal tissue from ultraviolet damage Authors: Siyu Pan, Sa Du, Zeyu Ma, Zhuoyuan Lin, Qiang Zeng, Yen Wei, Bo Wang, Yuwei Wu, Lei Tao Published: 2026 ABSTRACT AbstractUltraviolet (UV) light causes many eye diseases. Wearing sunglasses blocks <50% UV light from entering eyes. Skin sunscreens are banned for eye use due to irritation and safety concerns. Few sunscreens can be used in the eye to block UV light without compromising the eye functions. Herein, we modified poly(vinyl alcohol) (PVA), the active ingredient in commercial artificial tears, using naturally extracted aldehydes through Hantzsch’s reaction and obtained a PVA-syringaldehyde derivative that has a high sun protection factor (SPF 80), excellent UV resistance (>99%) and high transparency (84%) at 1 wt% in water, and a long half-life with little acute irritation and long-term safety when applied to female mouse eyes. Zebrafish protected by this polymer maintained healthy eye tissues and vision-guided behaviors after UV exposure. Polymer treatment on mice eyes effectively protected the retina from UV-induced damage, suggesting its potential as an eye sunscreen.Subject terms: Biomedical materials, Biomedical materials Introduction Apart from the skin, the eye is the organ that is most vulnerable to sunlight-induced damage. Approximately 95% of people are aware of the damage caused by ultraviolet (UV, 280–400 nm) light to the skin, whereas only 7% know that UV light can cause eye disease1. Cumulative UV exposure has been identified as the cause of many ocular diseases, including pterygium, cortical cataracts, and photokeratitis (Fig. 1a, left)2–8. When UV light in sunlight reaches the human eye, more than 95% of the UV energy is absorbed by the cornea, iris, and lens9,10. In adults (≥25 years old), only 1–2% of the UV light reaches the retina, and this number increases to 2–5% in children (8–10 years old), who have larger pupils and clearer lenses (Fig. 1a, right)11,12. Retinal stress caused by UV exposure, which may lead to macular degeneration and vision loss, is more severe in children than in adults13,14. Thus, children are more likely than adults to develop eye problems when they are exposed to UV light. UV damage to eyes and UV protection for eyes. a UV-induced eye diseases and UV reaching the retina in children (8–10 years old) and adults (≥25 years old). b Current UV protection techniques—wearing sunglasses and using traditional UV-blocking agents. c Schematic of a PVA-based sunscreen used to protect eyes from UV damage. The eye is a visual organ that functions only when it is directly exposed to light; thus, conventional UV protection methods do not efficiently protect the eyes from UV light. For example, wearing sunglasses is the most common way to protect the eyes from UV irradiation, albeit at the expense of visual clarity15,16. However, sunglasses that are positioned 6 mm away from the forehead block only ∼45% of ambient UV17. The peripheral, reflected, and scattered UV rays still reach the eye, and even the back-reflected UV rays from the lenses add to the UV burden on the eye (Fig. 1b(i))18. Recently developed UV-resistant contact lenses containing small molecular UV absorbers can adhere to the eye surface, effectively preventing the back reflection of UV light19–21. However, these small molecular UV absorbers may leak during use. The amount of these UV absorbers must be limited to ensure transparency, which can result in inadequate UV-blocking efficacy22. Furthermore, people with dry eyes may feel uncomfortable wearing contact lenses. These drawbacks limit the applicability of UV-resistant contact lenses. Besides, organic UV-blockers (e.g., oxybenzone and avobenzone) and inorganic nanoparticles (e.g., ZnO and TiO2) have been extensively used for the UV protection of skin23–27. However, UV absorbers for eye use should be safe, transparent, and water-soluble. Nanoparticles can block the view, and additives (e.g., ethanol, glycerol) used to increase the water solubility of organic UV-blockers can irritate or damage the eyes. These issues restrict the ocular applicability of conventional UV-blocking agents (Fig. 1b(ii)). Therefore, the UV protection of the eyes is an important and thorny issue that has been neglected; hence, developing a strategy to help people of all ages in blocking UV light from all directions without impairing the visual function of the eyes is an urgent and challenging need. Herein, we developed a tailor-made polymer sunscreen for eye protection against UV light by introducing a natural syringaldehyde (SA) extract into poly(vinyl alcohol) (PVA) via the Hantzsch reaction (Fig. 1c). We used PVA, SA, and the Hantzsch reaction to develop a sunscreen for the eyes owing to the following reasons: (1) PVA has been used as an active ingredient in commercial artificial tears28. It is water-soluble, biocompatible, and easy to modify29–31, making it an ideal framework for developing UV-resistant polymers for use in the eyes. (2) The Hantzsch reaction effectively produces 1,4-dihydropyridine (1,4-DHP), which is characterized by UV resistance, and its UV absorption range can be adjusted by introducing different substituents at the C(4) position of 1,4-DHP32,33. (3) SA is an antioxidant and anti-inflammatory aldehyde that can be derived from natural products and alleviate UV-induced biological damage34,35. The aromatic structure in SA makes it a potential UV absorber36. SA has excellent reactivity as a reaction component for the Hantzsch reaction. In this study, we introduced three aldehydes derived from natural products into PVA via the Hantzsch reaction and selected the SA-containing PVA derivative (PVA-SA) as it exhibits the best UV-blocking and antioxidant properties among these three PVA derivatives. PVA-SA was transparent at the working concentration (1 wt% in water); it had a high sun protection factor (SPF) of ∼80 and shielded the entire UVB (280–320 nm) and UVA (320–400 nm) rays, even at a low concentration (0.5 wt% in water). In our cellular experiments, PVA-SA showed low acute cytotoxicity and long-term cytotoxicity, and effectively protected corneal epithelial cell lines and primary human dermal fibroblasts against UV-induced DNA/RNA damage, inflammation, angiogenesis, and senescence. In the zebrafish experiments, UV light impaired the zebrafish eyes, leading to poor vision and abnormal behavior. PVA-SA effectively protected the eyes from UV damage. There was little difference between the behaviors and ocular histology of the protected and healthy zebrafish. In the mouse experiments, PVA-SA showed very little acute eye irritation or hepatorenal toxicity in both short-term and long-term use, and effectively protected the mouse retina from UV damage. These results demonstrate that the rationally designed PVA derivative is a safe and efficient sunscreen for the UV protection of the eyes and has potential for practical applications. Results Preparation and characterization of PVA derivatives We introduced several aldehydes derived from natural products (benzaldehyde (BA), 4-hydroxybenzaldehyde (HBA), and SA) into the PVA structure through the one-pot Hantzsch reaction and obtained three water-soluble PVA derivatives (PVA-BA, PVA-HBA, and PVA-SA), respectively (Fig. 2a). The characteristic peaks of –NH– groups in 1,4-DHP moieties (8.7–8.9 ppm) and phenyl groups (6.3–7.1 ppm) of all PVA derivatives are clearly visible in the 1H nuclear magnetic resonance (1H NMR) spectra (Fig. 2b). In the Fourier-transform infrared (FT-IR) spectra of PVA and the PVA derivatives (Fig. 2c), the PVA derivatives clearly exhibit peaks associated with the –C–N bond (1642–1630 and ~1170 cm−1) in 1,4-DHP moieties and benzene rings (1530–1480 cm−1). Based on the atomic percentages of N and O obtained from the X-ray photoelectron spectroscopy (XPS) data, the modification degrees of Hantzsch moieties in the PVA derivatives were calculated as 7% in PVA-BA and 5% in PVA-HBA and PVA-SA (Supplementary Table 1). The weight average molecular weight (Mw) of the PVA derivatives obtained via gel permeation chromatography (GPC) (PVA-BA: 33.4 kg mol−1, PVA-HBA: 44.2 kg mol−1, and PVA-SA: 37.1 kg mol−1) is larger than that of PVA (Mw (GPC): 31.6 kg mol−1) (Supplementary Fig. 1). These results indicate that the PVA derivatives are successfully prepared. Preparation and characterization of PVA derivatives. a Illustration of the synthesis of PVA derivatives. b 1H NMR (DMSO-d6, 400 MHz) spectra and c FT-IR spectra of PVA and PVA derivatives. d UV–vis transmittance spectra of aqueous solutions of PVA derivatives (5 mg/mL) and magnified spectra at 280–370 nm. e Scavenging ratios of the RNS and ROS of PVA derivatives. Data are presented as mean ± SD (n = 3 independent samples). f Theoretical energy bandgaps of M-SA calculated using density functional theory (DFT). Source data are provided as a Source Data file. We then tested the UV-shielding performance of the prepared polymers using their aqueous solutions at different concentrations via UV–visible (UV–vis) transmittance analysis (Fig. 2d, Supplementary Fig. 2). When the PVA derivative concentration reaches 5 mg/mL (0.5 wt%), the PVA-SA solution shields almost the entire UV spectrum (UVB: 280–320 nm; UVA: 320–400 nm), whereas some UV light still passes through the PVA-BA and PVA-HBA solutions. Moreover, UV-induced biological damages are closely related to oxidative stress, with PVA-SA showing the highest clearance for reactive nitrogen species (RNS) and reactive oxygen species (ROS) among all PVA derivatives (Fig. 2e). Thus, we selected PVA-SA as a typical PVA derivative for subsequent experiments owing to its excellent UV resistance and antioxidant capacity. We investigated the UV-shielding mechanism of PVA-SA using a model molecule (M-SA) that has a similar pendant structure to 1,4-DHP in PVA-SA (Supplementary Fig. 3a, b). We calculated the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) of M-SA, which are involved in the UV spectrum37, using density functional theory (DFT) (Fig. 2f). Figure 2f depicts the distribution of HOMO and LUMO, indicating that upon excitation, charge transfer occurs from the SA ring to the 1,4-DHP ring. The EHOMO and ELUMO of M-SA in methanol are calculated to be −5.49 eV and −1.68 eV, respectively, and Egap is 3.81 eV, corresponding to the wavelength of 325 nm. The difference between the theoretical absorption peak and experimental absorption peak (364 nm, Supplementary Fig. 3c) falls within a reasonable deviation range, indicating that M-SA can absorb UV radiation via electron transition from HOMO to LUMO. Cellular UV protection by PVA-SA: cytotoxicity and mechanism analyses We used a human corneal epithelial cell line (HCE-T) to evaluate the cytotoxicity of PVA-SA via Cell Counting Kit-8 (CCK-8) assay. PVA served as a control. After they were co-cultured with different concentrations (1–10 mg/mL) of PVA or PVA-SA for 24 h, the HCE-T cells exhibit a high rate of cell viability (over 90%), indicating that the polymers have low cytotoxicity (Supplementary Fig. 4). We further evaluated the long-term cytotoxicity of PVA-SA through a colony formation assay (Supplementary Fig. 5). The colonies in the groups with single or multiple administration (5 times) of PVA-SA (10 mg/mL) are similar to those in the blank group, demonstrating that PVA-SA has low long-term cytotoxicity. The SPF of PVA-SA (2 mg/cm2) is determined to be 80 according to reported methods38. This value is comparable to commercial sunscreens for the skin (SPF 50+). Moreover, the PVA-SA solution at 10 mg/mL (1 wt%) appears almost colorless and transparent (84%, 400–800 nm) (Supplementary Fig. 6). Thus, PVA-SA at 10 mg/mL was used for cellular UV protection. The HCE-T cells were co-cultured with PVA and PVA-SA (10 mg/mL, in culture medium) and then exposed to UV irradiation (UVA: 365 ± 3 nm, 20 mW/cm2) for 10 min (120,000 J/m2, equivalent to 2.5–12.6 years of average UV dose in sunlight) followed by incubation in a fresh culture medium for 24 h (Fig. 3a). The cells directly exposed to UV radiation were used as a control (UVA group); the cells that were incubated in medium alone were used as a Blank. We used the fluorescein diacetate/propidium iodide (FDA/PI) double staining assay to visually observe cell viability. The number of surviving cells in the PVA-SA group is similar to that of the Blank group, whereas fewer living cells are detected in the UVA and PVA groups than in the Blank and PVA-SA groups (Supplementary Fig. 7a). The quantitative data acquired via the CCK-8 assay agree well with the visual results (Supplementary Fig. 7b). These results indicate that UVA inhibits the proliferation of HCE-T cells, and PVA cannot alleviate the adverse effects of UVA on cells, whereas PVA-SA prevents UV damage to cells. RNA-sequence analysis of HCE-T cells under the protection of PVA-SA after UV exposure. a Scheme of the experimental procedures to evaluate the cellular UV protection of PVA-SA. b Volcano plots display the differentially expressed genes (DEGs) of the UVA + PVA-SA group compared with those of the UVA group; p < 0.05, |log2FC| > 1. Differential expression analysis was performed using DESeq2 (two-sided), and there was no adjustment made for multiple comparisons. c Gene ontology (GO) analysis of DEGs related to the transcription and synthesis of DNA/RNA. The chord represents the relationships between transcriptome functions and genes. The genes are shown on the left, and the heatmap alongside shows the log FC values of the genes; the functions are shown on the right. The chords connect the genes to their corresponding enriched functions. d Results of the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis of DEGs. Differential expression analysis was performed using DESeq2 (two-sided), and there was no adjustment made for multiple comparisons. e Heatmap of DEG expressions in the UVA and UVA + PVA-SA groups. Red: upregulated; blue: downregulated. Control: UVA group. f Protein–protein interactions (PPI) analysis related to DNA/RNA synthesis, cell cycle, inflammation, angiogenesis, and senescence functions; Database: version-11-5.string-db.org/organism/9606. Gene set enrichment analysis (GSEA) shows that DEGs in the gene sets of the RNA degradation pathway (g), the IL-17 signaling pathway (h), and cellular senescence (i) are downregulated in the UVA + PVA-SA group compared with those in the UVA group. We performed whole-transcriptome RNA sequencing on the HCE-T cell lines to investigate the effects of UV and PVA-SA alone on cells. Untreated cells served as the non-treatment control (the NC group). The results demonstrate that both UVA and UVB cause cell damage by inducing DNA/RNA damage, inflammation, angiogenesis, and senescence, while PVA-SA (10 mg/mL) has a negligible effect on cells (Supplementary Figs. 8–10). Then, we comprehensively investigated the cellular protection mechanism following UV irradiation under PVA-SA treatment (Fig. 3a). PVA-SA treatment led to the downregulation of 403 genes and the upregulation of 257 genes relative to the UVA group (Fig. 3b). Gene ontology (GO) analysis shows that differentially expressed genes (DEGs) are mainly enriched in terms related to DNA/RNA transcription and synthesis, such as “the regulation of transcription by RNA polymerase II” and “sequence-specific double-stranded DNA binding” (Fig. 3c). These enriched terms are also closely related to the cell cycle and affect cell growth and proliferation. Furthermore, DEGs are enriched in terms related to angiogenesis and inflammation, such as “the negative regulation of the Notch signaling pathway” and “negative regulation of blood vessel endothelial cell proliferation involved in sprouting angiogenesis” (Supplementary Fig. 11). Meanwhile, the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment of DEGs shows that the remarkably enriched terms of the HCE-T cell line are related to cell growth, adhesion and proliferation (e.g., Pl3K-Akt signaling pathway, ECM-receptor interaction, MAPK signaling pathway, and focal adhesion), and inflammation (e.g., Notch signaling pathway and IL-17 signaling pathway) (Fig. 3d). Moreover, genes related to senescence inhibition (e.g., COL6A3, ITGA2) and angiogenesis inhibition (e.g., THBS1) are upregulated by PVA-SA, whereas genes related to cell cycle inhibition (e.g., CDKN1A, GADD45A), inflammation (e.g., CXCL2, CXCL8, IL1B), and angiogenesis (e.g., VEGFA, ADM2) are downregulated by PVA-SA (Fig. 3e). Figure 3f exhibits the protein–protein interactions (PPI) associated with DEGs and demonstrates the close relationship between the abovementioned biological functionalities. The results of the gene set enrichment analysis (GSEA) demonstrate that the PVA-SA group is negatively correlated with RNA degradation, the IL-17 signaling pathway, and cellular senescence (Fig. 3g–i). These results reveal that PVA-SA effectively protects cells from UVA-induced DNA/RNA damage, inflammation, angiogenesis, and senescence, and provide a genetic mechanism for cellular UV protection by this polymer. UV radiation primarily exerts its toxicity by directly inducing DNA damage, leading to DNA cross-links and secondary double-strand breaks with oxidative stress as a secondary event of the initial damage. To study how PVA-SA protects cells from damage caused by UV, we co-cultured the cells with PVA-SA (10 mg/mL in culture medium) before (the Prevention group) and after (the Rescue group) UVA irradiation (365 ± 3 nm, 120,000 J/m2), respectively. The cells treated with UVA alone and without treatment served as the control and blank groups, respectively. We evaluated the damaged DNA in cells from each group using the γ-H2AX staining assay (Supplementary Fig. 12). Significant green fluorescence could be detected in the UVA and the Rescue groups, while negligible fluorescence was observed in the blank and Prevention groups. The results of the γ-H2AX staining assay agree well with those of the colony formation assay (Supplementary Fig. 13). These results indicate that PVA-SA protects cells by directly blocking UVA light, rather than preventing cells from UV-induced damage by alleviating the oxidative stress, which might be attributed to the fact that the large polymer is difficult to rapidly enter the cells. Behavioral analyses of zebrafish under UV irradiation Zebrafish larvae have recently been used as model organisms for visual study due to their variety of visually directing behaviors, such as their startle response (reduced locomotor activity in sudden light) and phototaxis (swimming from dark areas to light areas)39,40. They exhibit rapid visual development and are similar to mammals in terms of retinal tissue, cell composition, and gene expression41,42. Zebrafish larvae are small and have big eyes; thus, they are easy to observe in high-throughput experiments43,44. In this study, we cultured zebrafish larvae in PVA-SA and PVA solutions (10 mg/mL), 72 h post-fertilization (72 hpf), whose retinal neurogenesis was essentially complete. Subsequently, we placed these larvae under UV irradiation (365 ± 3 nm, 8000 J/m2, equivalent to 2–10 months of average UV dose in sunlight) for approximately 30 s and then transferred them to fresh culture medium. The larvae in the medium only were used as a Blank, and the larvae exposed to UV were used as controls. After 24 h of incubation, the swimming behavior of these larvae (96 hpf) was heavily dependent on visual cues, and we conducted their startle response test using alternating light–dark conditions in a high-throughput manner. The movement of the larvae was monitored using a Noldus behavior recording system (Fig. 4a). Behavior analyses of zebrafish larvae after UV irradiation. a Illustration of experimental procedures for the startle response of larval zebrafish. b The typical tracks of larvae in different groups in one light–dark cycle. c Effect of alternating light–dark periods (2 min of light followed by 2 min of darkness) on the swimming distance in 96-hpf larval zebrafish. The data are presented as mean ± SD (n = 24 independent zebrafish). The results are for a total of 96 larvae. d Differences in swimming distance between the light and dark fields of larvae in different groups. Data are presented as mean ± SD (n = 24 independent zebrafish). This experiment was repeated thrice independently with similar results. Statistical analyses of results were performed by Student’s t-test for independent samples (one-sided), and there was no adjustment made for multiple comparisons. e Illustration of experimental procedures for phototaxis testing of larval zebrafish. f Screenshots of phototaxis testing when the number of larvae in the light area was highest for all groups (n = 30 independent zebrafish). g Statistical data for phototaxis testing. Data are presented as mean ± SD (n = 30 independent zebrafish). This experiment was repeated thrice independently, yielding similar results. Source data are provided as a Source Data file. The startle response refers to the phenomenon in which zebrafish larvae switch from a dark to a light environment and swim much less in light than in the dark45,46. The zebrafish larvae were domesticated in the dark for 10 min and then subjected to three cycles of light and dark cycles (each cycle: 2 min of light followed by 2 min of darkness). Figure 4b shows typical traces of zebrafish larvae in one cycle (Supplementary Video 1), with the larvae in each group swimming farther during the dark period than during the light period. This phenomenon is observed consistently over three alternating periods of light and dark, and the generated quantitative data for the alternating swimming distances (Fig. 4c) indicate that larvae in different groups exhibit various levels of startle response. In particular, compared with the Blank group, the UV group shows a much smaller difference between the swimming distance in the light and dark. The distance difference exhibited by the UV + PVA group is slightly higher than that demonstrated by the UV group. Conversely, the difference in the distance between the UV + PVA-SA group and Blank group is close (Fig. 4d). These results suggest that the larvae in the UV group are less sensitive than those in the Blank group to sudden changes in brightness and darkness, which may have been caused by UV-induced damages to vision, neural regulation, or metabolic function regulation, and that PVA provides poor UV protection to the larvae, whereas PVA-SA effectively protects the larvae from UV damages. We then performed phototaxis testing to evaluate the visual integrity of the larvae after different treatments. We confined the pretreated larvae (96 hpf) to the dark using a baffle. After removal of the baffle, the movement and number of larvae in the light area were recorded (Fig. 4e, Supplementary Video 2). Through direct observation, we observe that zebrafish larvae in the UV and UV + PVA groups are significantly less likely to swim from the dark zone to the light zone than those in the Blank and UV + PVA-SA groups (Fig. 4f). The quantitative data suggest that only 4 and 5 of 30 larvae in the UV and UV + PVA groups, respectively, move from the dark area to the light area, whereas these numbers are 13 and 10 for the Blank and UV + PVA-SA groups, respectively (Fig. 4g). These results indicate that UV irradiation may impair the larvae’s vision, neural regulation, or metabolic function regulation, resulting in decreased phototaxis, and PVA prevents very little damage from UV light, whereas PVA-SA protects the larvae from the damages from UV light; thus, the phototaxis of the protected larvae is comparable to that of the healthy larvae. UV protection of PVA-SA on zebrafish larvae's eyes We further investigated the UV-induced damage to the larvae’s eyes and the protective effect of PVA-SA on the eyes of zebrafish larvae. We stained the aforementioned pretreated larvae (96 hpf) with acridine orange (AO) and observed their ocular areas to detect apoptotic cells. The eyes of the larvae in the UV and UV + PVA groups display bright green spots, which indicate the possible presence of chromatin condensation in apoptotic cells (Fig. 5a, white arrows). In contrast, almost no fluorescent signal is detected in the Blank and PVA-SA groups, indicating negligible apoptosis of their ocular cells (Fig. 5a). The quantitative data for fluorescence intensity agree well with the qualitative results (Supplementary Fig. 14). These results suggest that UV exposure can damage the eye cells of larvae, and PVA barely prevent UV damage, whereas PVA-SA effectively shields against UV exposure, leaving the eye cells almost intact. UV protection of PVA-SA on zebrafish eyes. a Acridine orange (AO) stained images of larvae’s eyes from different groups. The white arrows indicate apoptotic cells. Scale bar: 50 μm. The experiment was repeated thrice independently with similar results. b Hematoxylin and eosin (H&E) stained images of larval eyes from different groups. The arrows indicate the spaces in the ganglion cell layer (GCL), and the circles indicate inflammatory cell infiltration in the UV- and PVA groups. Scale bar: 50 μm. c Numbers of nuclei in the GCL of different groups. Data are presented as mean ± SD (n = 3 independent zebrafish). Statistical analyses of results were performed by Student’s t-test for independent samples (one-sided), and there was no adjustment made for multiple comparisons. d Thickness of the GCL, inner plexiform layer (IPL), photoreceptor layer (PR), retinal pigment epithelium (RPE), and inner nuclear layer (INL) in different groups. Data are presented as mean ± SD (n = 3 independent zebrafish). The thickness of each sample is measured five times at different sites of the sample. Source data are provided as a Source Data file. Figure 5b shows the histological images of the larvae’s eyes. Inflammatory cell infiltration has been observed around the eyes of larvae in the UV and UV + PVA groups (Fig. 5b, red circle), suggesting that necrosis might occur in both groups. However, few inflammatory cells are observed around the eyes of larvae in the UV + PVA-SA group. The apparent gap between the ganglion cell layer (GCL) in the UV and UV + PVA groups can be attributed to nuclear loss and disorganization (Fig. 5b, red arrows, Supplementary Fig. 15), which is also verified by the quantitative data (Fig. 5c). Compared with healthy larvae, UV- and PVA-treated larvae exhibit a significant reduction in the thickness of various eye cell layers, including the GCL, inner plexiform layer (IPL), inner nuclear layer (INL), and photoreceptor layer (PR). However, the cell layer thickness of the UV + PVA-SA group closely resembles that of the Blank group (Figs. 5b, d). Negligible UV-induced DNA breaks were observed in all groups using terminal deoxynucleotide transferase-mediated deoxyuridine triphosphate nick end labeling (TUNEL) staining assay (Supplementary Fig. 16). These results suggest that PVA-SA effectively preserves the larval eye structure under UV irradiation. Acute eye irritation test for PVA-SA We evaluated the local safety of PVA-SA as an eyedrop component by performing an acute eye irritation test (Fig. 6a). We divided the healthy mice into four groups and treated them with different solutions, including saline (the Blank group), sodium dodecyl sulfate (SDS, 2 μL, 100 mg/mL in saline, used as the positive control)47, PVA (2 μL, 100 mg/mL in saline), and PVA-SA (2 μL, 100 mg/mL in saline). After 5 min of administration of the different solutions, the mice’s eyes were washed with saline. We then used a slit-lamp microscope to observe the eye conditions and fluorescein-stained cornea at 0.5, 1.0, 1.5, 24, 48, and 72 h, respectively. After 0.5 h of treatment, mice treated with SDS exhibit swelling of the eyelids, corneal turbidity, corneal defects, and ulcers. The eyelids and corneas of mice treated with saline, PVA, and PVA-SA show few abnormalities (Fig. 6b and Supplementary Fig. 17). After 72 h, the eyes of the saline, PVA, and PVA-SA-treated mice remain healthy, whereas the eyes of the SDS-treated mice do not completely recover (Fig. 6b and Supplementary Fig. 18a). We also measured fluorescein sodium staining scores using a 16-point scale to monitor the corneal conditions48. The scores of the PVA and PVA-SA groups remain at zero for 72 h, similar to those of the saline group (Supplementary Fig. 18b). These results quantitatively verify that the eyes of mice treated with saline, PVA, and PVA-SA remain healthy. In vivo safety of PVA-SA and its UV protection for the mouse retina. a Scheme of experimental procedures of the acute eye irritation test. b Representative slit-lamp images of fluorescein cornea staining of mouse eyes after different treatments, after 0.5 and 24 h. Scale bar: 1 mm. The experiment was repeated thrice independently with similar results. c H&E staining of mouse eyes after different treatments for 72 h. Scale bar: 1 mm. The bottom row shows enlarged images of the cornea. The arrows indicate the ruptured cells of the cornea in the SDS group. Scale bar: 100 μm. The experiment was repeated thrice independently with similar results. d TUNEL staining of the mouse corneal tissues after different treatments for 72 h. Blue: DAPI, green: TUNEL. Scale bar: 200 μm. The experiment was repeated thrice independently with similar results. e Retention time of PVA-SA in mouse eyes. The data are presented as mean ± SD (n = 3 independent eyeballs). f H&E staining images of the mouse retinas after different treatments. Scale bar: 50 μm. g Numbers of nuclei in the ONL of different groups. Data are presented as mean ± SD (n = 5 independent eyeballs). Statistical analyses of results were performed by Student’s t-test for independent samples (one-sided), and there was no adjustment made for multiple comparisons. h Scheme of PVA-SA administration via intravenous injection. Different serum biomarker levels (i urea; j crea; k aspartate aminotransferase (AST); and l albumin (ALB)) in healthy mice and mice intravenously injected with PVA (30 mg) and PVA-SA (30 mg) after 72 h. p-values were obtained by comparing the data with those of the healthy group. Data are presented as mean ± SD (n = 6 independent mice). Statistical analyses of results were performed by Student’s t-test for independent samples (one-sided), and there was no adjustment made for multiple comparisons. Source data are provided as a Source Data file. After 72 h, we euthanized the mice and removed their eyes to histologically evaluate the structural and morphological changes in the eyes after different treatments. H&E staining shows that the corneal epithelial cells of mice treated with SDS are apparently peeled and broken (black arrow), whereas the corneal epithelial cells of mice treated with normal saline, PVA, and PVA-SA are arranged in a dense, orderly, and integrated manner (Fig. 6c), and the difference in corneal thickness between the PVA, PVA-SA, and normal saline groups is negligible (Supplementary Fig. 19a). Furthermore, the iris and retina are histologically normal in all groups, indicating that the different solutions mainly affect the cornea (Supplementary Fig. 19b). We also used the TUNEL assay to observe the apoptosis of different corneal cells. The damage caused by SDS to the corneal tissue is severe, showing bright fluorescence, whereas the damage caused by PVA-SA is hardly detectable (Fig. 6d). Meanwhile, the expressions of tumor necrosis factor-α (TNF-α) and interleukin-6 (IL-6) in the PVA and PVA-SA groups are also nearly negligible (Supplementary Fig. 20). These results indicate that PVA-SA causes almost no acute eye irritation. Ocular retention ability of PVA-SA We then investigated the ocular retention ability of PVA-SA through fluorescein testing. Briefly, we treated mice with PVA-SA (2 μL, 100 mg/mL in saline), then collected their eyes at different time intervals after the administration of PVA-SA, and soaked them in saline. The 1,4-DHP moieties in PVA-SA have intrinsic fluorescence; thus, the resident PVA-SA in the eye was measured by detecting the fluorescence intensity of the eye-containing solutions according to the standard curve for PVA-SA (Supplementary Fig. 21). As shown in Fig. 6e, the residual amount of PVA-SA in the eyes decreases to 30.3% within 30 min, which may have been due to the rapid loss of PVA-SA through the nasolacrimal duct and from outside the eyes by normal physiological behaviors, such as blinking and free movement of the mice. After 60 and 90 min of administration, the amounts of PVA-SA retained on the eyes are 13.0% and 1.9%, respectively. The half-life of PVA-SA in the eye is estimated to be approximately 15 min. UV protection of PVA-SA on the mouse retina We used a mouse model to evaluate the protective effect of PVA-SA. We dropped a solution of PVA-SA (2 μL, 100 mg/mL, in saline) into the eyes of BALB/C mice, and then exposed the eyes to UV light (365 ± 3 nm, 60,000 J/m2) for 120 s (equivalent to 1.2 − 6 years of average UV dose in sunlight). After 24 h, we euthanized the mice and removed their eyes for histological assessment of the structural and morphological changes in the eyes. Healthy mice and mice exposed to UV light without any treatment were used as the blank group and the control group (the UV group), respectively. H&E staining results showed that compared with the blank group, the thickness of the retinal cell layers (including outer plexiform layer (OPL), outer nuclear layer (ONL), and inner segment/outer segment (IS/OS)) in the UV group has been significantly reduced (Fig. 6f, Supplementary Fig. 22). Meanwhile, the ONL becomes disordered and has obvious nuclear loss, the boundary between IS/OS becomes blurred, and the IS also becomes disordered with obvious gaps. On the contrary, the retinal cell layers in the UV + PVA-SA group are very similar to those in the blank group, with orderly cell layer arrangement, normal morphology, a clear IS/OS boundary, and unreduced nuclei number in ONL (Figs. 6f, g, Supplementary Fig. 22). Additionally, immunofluorescence staining results clearly showed the green signals of IL-6 in the retina in the UV group, indicating that UV light induces cell damage, while the IL-6 signals were hardly observed in the blank group and the UV + PVA-SA group (Supplementary Fig. 23). These results verified that UV light causes damages to the retina of mice and the anti-UV PVA-SA effectively protects the retina of the mice exposed to UV light. In vivo safety assessment of PVA-SA We tested the toxicity of PVA-SA (in saline) in mice via intravenous injection (Fig. 6h). PVA is a well-known biosafe polymer that has been approved by the US Food and Drug Administration for human use49; thus, PVA was used as a control. Healthy mice were used as a blank. All mice have survived high doses of PVA (30 mg) and PVA-SA (30 mg, 150 times the dosage for eye administration) without evident weight loss (Supplementary Fig. 24) or exhibiting abnormal behavior during experiments. After 72 h of injection, half of the mice were sacrificed, and their hearts, livers, spleens, lungs, and kidneys were isolated for histological analysis by H&E staining. Histological analysis of the organs reveals minimal damage and few lesions (Supplementary Fig. 25). Blood was collected to measure the serum biomarkers of liver and renal function. No significant changes are observed in the urea, creatinine, aspartate aminotransferase (AST), or albumin (ALB) levels compared with healthy mice (Fig. 6i–l). The remaining mice were sacrificed and tested using the same methods 7 days after injection. Negligible differences are observed between the polymer-injected and healthy mice in the histological analysis of major organs and serum biomarker levels (Supplementary Fig. 26). These results indicate that PVA-SA has minimal adverse effects on the liver and kidneys in mice and has a biosafety similar to PVA. In vivo long-term toxicity of PVA-SA We investigated the in vivo long-term toxicity of PVA-SA by repeatedly applying PVA-SA (2 μL, 100 mg/mL in saline) to the BALB/C mouse’s eyes twice a day (once at 8:00 am and once at 8:00 pm) for 14 days. Saline (2 μL) served as the blank. We used a slit-lamp microscope to observe the eye conditions and fluorescein-stained cornea on the 3rd, 7th, and 14th day, respectively (Supplementary Fig. 27). During the 14-day observation period, the eyelids and corneas of mice treated with saline and PVA-SA showed few abnormalities. No significant green fluorescence has been detected in both saline and PVA-SA-treated groups, indicating negligible corneal defects and ulcers. After 14 days, we euthanized the mice and removed their eyes to histologically evaluate the structural and morphological changes in the eyes after PVA-SA treatments. Negligible bright fluorescence was observed in the saline and PVA-SA groups in TUNEL staining (Supplementary Fig. 28a). H&E staining showed that corneal epithelial cells of mice treated with PVA-SA are arranged in a dense, orderly, and integrated manner, similar to those of mice treated with saline. Meanwhile, the iris and retina in both groups were histologically normal (Supplementary Fig. 28b, c). The expression levels of TNF-α and IL-6 in the PVA-SA group showed no significant difference compared with those in the blank (Supplementary Fig. 28d, e). These results demonstrate that the long-term use of PVA-SA causes nearly no irritation to the eyes. Moreover, the hearts, livers, spleens, lungs, and kidneys of these sacrificed mice were isolated for histological analysis by H&E staining. Histological analyses of the organs showed minimal damage and few lesions (Supplementary Fig. 29a). Blood was collected to measure the serum biomarkers of liver and renal function. No significant changes were observed in the urea, creatinine, AST, or ALB levels compared with those of healthy mice (Supplementary Fig. 29b−e). These results indicate that PVA-SA is slightly toxic when used in long-term in vivo experiments. Discussion In this study, we selected three natural extract aldehydes with aromatic structures for their high reactivity in the Hantzsch reaction and broad application as spices in the food industry. These aldehydes derived from natural products are safe and also exhibit biological activities (e.g., antioxidant and anti-inflammatory) that alleviate UV-induced biological damage. However, SA alone has almost no protection on UVA-treated cells, and weakly protects cells against UVB (Supplementary Fig. 30). It needs DMSO to increase the solubility in water, and is unsuitable for eye use. We introduce SA into the PVA structure via the Hantzsch reaction. The water solubility of SA is dramatically improved owing to the water-soluble PVA backbone, and the UV-shielding ability of SA is also enhanced due to the combination with the inherently UV-resistant 1,4-DHP structure formed by the Hantzsch reaction. Therefore, the covalent linkage of SA on PVA via the Hantzsch reaction amplifies the effect of SA to achieve the target water-soluble, anti-UV polymer for eye use. In future research, other natural/synthetic aldehydes with unsaturated structures, which are also potential UV absorbers, can be similarly investigated to expand the molecule library and develop different sunscreens for the eyes. UV light is significantly attenuated by water. We took this factor into account in this research. We put the UV detector under the zebrafish-containing dish or cell-containing plate during these experiments, and we subtracted the dose that was absorbed by the empty dish or plate. Meanwhile, the zebrafish and cells were at the bottom of the containers. Thus, the doses mentioned in the manuscript may not be exactly the same as the actual dose received by the eyes of the zebrafish embryos or the cells in culture, but the difference should be relatively small. In this study, we used UVA as the main light source for the cellular and zebrafish experiments because it constitutes approximately 95% of UV light in the Earth’s atmosphere50,51. However, UVB (∼5% in the atmosphere, and more abundant at high-altitude areas or near-Earth space than at the Earth's surface) has a higher energy than UVA for the same dose and may cause more serious eye damage. For example, UVB is an important environmental factor that results in cataracts, which cause blindness in over 15 million adults (≥50 years old)52. Thus, we also evaluated the protection of cells against UVB using PVA-SA. After exposure to UVB (312 nm, 125 μW/cm2) for 40 min (3000 J/m2, equivalent to 1.2–6 years of average UV dose in sunlight), the FDA/PI and CCK-8 assays indicate that UVB probably inhibits the proliferation of HCE-T cells, and PVA fails to prevent the adverse effects of UVB on the cells, whereas PVA-SA prevents UVB-induced damage to the cells (Supplementary Fig. 31). Moreover, the RNA-sequencing results suggest that, compared with UVA, the protective effect of PVA-SA against UVB is more focused on inhibiting DNA/RNA damage, inflammation, and cell adhesion (Supplementary Fig. 32). The γ-H2AX staining and colony formation assays further indicate that PVA-SA protects cells by directly blocking UVB light, rather than prevents cells from UV-induced damage by alleviating the oxidative stress (Supplementary Figs. 33 and 34). These results indicate that PVA-SA is also an excellent protective agent against UVB. The liquid layer thickness of the PVA-SA solution will affect its UV protective effect. At the same polymer concentration, a thicker liquid layer will absorb more UV light. In this study, the liquid layer of PVA-SA solution is about 2 mm thick in the experiments of cells and zebrafish due to the requirement of enough culture media, and is less than 1 mm thick when used in the mice eyes. These thicknesses are much higher than the thickness of the tear film between the cornea and the eyelid (3–40 μm)53. However, the doses of UVA and UVB irradiation for the cell experiments are 120,000 J/m2 and 3000 J/m2, respectively. The dose of UVA irradiation for the zebrafish experiment is 8000 J/m2. The doses of these UV rays are much stronger than the actual level. Therefore, the results obtained in this research (using solutions thicker than the actual situation to shield the UV rays, much stronger than the actual level) only demonstrate the potential of this polymer for UV protection. The actual UV protection effect of this polymer still needs to be evaluated in practical applications. Besides immortalized HCE-T cell lines, we also investigated the cytotoxicity and protective effect of PVA-SA on primary human dermal fibroblasts, which are non-immortalized and can undergo UV-induced senescence. The results of the CCK-8 assay and colony formation assay indicate that PVA-SA has low acute cytotoxicity and long-term cytotoxicity, respectively (Supplementary Figs. 35 and 36). Meanwhile, the CCK-8 results demonstrate that PVA-SA can effectively protect primary cells against both UVA (365 ± 3 nm, 20 mW/cm2 for 10 min, 120,000 J/m2 in total) and UVB (312 nm, 125 μW/cm2 for 40 min, 3000 J/m2 in total) irradiation (Supplementary Fig. 37). The RNA-sequencing data reveal that PVA-SA prevents UV-induced senescence by protecting against DNA damage and promoting cell growth and proliferation (Supplementary Figs. 38 and 39). The γ-H2AX staining assay and colony formation assay further show that PVA-SA directly blocks UV and protects primary cells against the initial insults as DNA damage, rather than preventing cells from UV damage by alleviating the oxidative stress (Supplementary Figs. 40−43). These results indicate the low cytotoxicity and excellent UV protection of PVA-SA for both cell lines and primary cells. In the zebrafish experiments, the co-culture time of zebrafish and PVA-SA was within one minute. The polymer added to the water may still be absorbed via the gills and induce systemic effects in such a short time. We will conduct detailed research on the absorption of polymer by zebrafish and the resulting systemic effect in the future. Meanwhile, the zebrafish larvae are too small to have only their eyes exposed to UV light in this study; thus, the vision-related abnormal behavior of larvae could be caused by the UV-induced damage to eyes, neural regulation, or metabolic function regulation. Therefore, in future research, we will use more complex mammalian models to evaluate the UV protection effect of polymers on the eye and draw reliable conclusions. Most eyedrops are poorly available due to precorneal loss and nasolacrimal drainage54,55. Approximately 60% of the active ingredients in eyedrops are eliminated within 2 min after eyedrop administration, and they are almost eliminated after 15–25 min56. In this study, the half-life of PVA-SA in mice’s eyes is approximately 15 min, which is much longer than the values reported in the literature. We measured the viscosity of PVA-SA solution (10 wt%) using a rheometer. When the shear rate ranges from 10 to 2000 s-1, the PVA-SA solution performs as a Newtonian fluid with a steady viscosity of approximately 67 mPa·s. When the shear rate exceeds 2000 s-1, it exhibits a shear thinning behavior as a non-Newtonian fluid (Supplementary Fig. 44). Meanwhile, the commercial product Eyestil PF that contains water and hyaluronic acid (0.2%, Molecular weight: 500–2000 kg/mol) as the main components, shows the viscosity of ~20 mPa·s at the shear rate of 10 s-1 and the significant shear thinning at the range from 10 to 10,000 s-153. The molecular weight of PVA-SA is 37.1 kg/mol, which is quite low compared with the hyaluronic acid in Eyestil PF (500–2000 kg/mol). Thus, the long retention time of PVA-SA is probably attributable to the high viscosity of the solution due to the high concentration of PVA-SA (100 mg/mL, i.e., 10 wt%) and the hydroxyl groups in PVA-SA that can interact with the polar corneal surface through hydrogen bonding. However, in practical applications, an ideal eye sunscreen should have a relatively low viscosity but a long retention time on the eye surface. A possible strategy is to use PVA with high molecular weights to develop UV-resistant polymers. The PVA backbone contains a large number of hydroxyl groups that can enhance the interaction between the polymer and the eye surface, thereby enabling the resulting UV-resistant polymer to remain attached to the eye surface for a long time, even at a low concentration and a low viscosity. Many ocular diseases are closely related to oxidative stress, such as dry eye syndrome57. Future research may explore the use of PVA-SA, which possesses excellent antioxidant capacity for treating these ocular diseases. In summary, we developed a sunscreen that can be used directly on the eyes to block UV from all directions without compromising eye function. We prepared three PVA derivatives (PVA-BA, PVA-HBA, and PVA-SA) by introducing, via the Hantzsch reaction, BA, HBA, and SA derived from natural products into PVA, respectively, which is the component of artificial tears. After a series of screening, we obtained PVA-SA that has a high SPF value (80), and its aqueous solution (1.0 wt%) has high transparency (84%) and excellent UV-shielding rate (>99%). PVA-SA exhibits low cytotoxicity in short-term and long-term use and effectively safeguards corneal epithelial cells and primary human dermal fibroblasts against UV-induced DNA/RNA damage, inflammation, angiogenesis, and senescence, thereby conferring protected cells with cellular viability comparable to those not exposed to UV light. In the in vivo experiments, PVA-SA effectively protects zebrafish larvae from UV damage; thus, the zebrafish protected by PVA-SA maintains almost intact vision-guided behaviors (startle response and phototropism) and healthy eye tissue after UV irradiation. Moreover, PVA-SA also exhibits excellent long-term biosafety with minimal acute eye irritation and a long retention time in the eyes of mice, and effectively protects the mouse retina from UV-induced damage, demonstrating that PVA-SA is a safe and effective anti-UV material for eye use. This goal-oriented study successfully developed an eye sunscreen with excellent safety and anti-UV properties using natural products and multicomponent reactions, thereby demonstrating an effective strategy for developing high-value-added biomaterials. Methods Preparation of PVA derivatives (PVA-BA, PVA-HBA, and PVA-SA) The PVA derivatives were synthesized according to the previous literature58. Take the preparation of PVA-BA as an example. Briefly, PVA-0588 was purified by precipitation and dialysis to remove antioxidant/anti-UV additives prior to use. The purified PVA (1.1 g, n(OH) ~ 22 mmol) was dissolved in anhydrous 1-methyl-2-pyrrolidinone (NMP, 10 mL), then diketene (184.9 mg, 2.2 mmol) was added, and the mixture was stirred at room temperature for 1 h. Then, benzaldehyde (256.8 mg, 2.42 mmol), 5,5-dimethyl-1,3-cyclohexanedione (339.2 mg, 2.42 mmol), ammonium acetate (279.8 mg, 3.63 mmol), and glycine (18.2 mg, 0.24 mmol) were added to the reaction system, and the mixture was kept at 80 °C for 3.5 h to carry out the Hantzsch reaction directly. The mixture was precipitated in acetone three times, and dried under vacuum to obtain the targeted PVA-BA (1.50 g, yield ~89%). PVA-HBA and PVA-SA were similarly prepared using 4-hydroxybenzaldehyde (295.5 mg, 2.42 mmol) and 3,5-dimethoxy-4-hydroxybenzaldehyde (440.9 mg, 2.42 mmol) instead of benzaldehyde, respectively to obtain the targeted polymers (PVA-HBA: 1.55 g, yield ~91%; PVA-SA: 1.68 g, yield ~92%). UV-blocking performance and transparency of polymer solutions The UV-shielding performance and transparency of polymer solutions were investigated by UV–vis analysis. The transmittance data were used to calculate the blocking percentage of UVA (320–400 nm) and UVB (280–320 nm), and the visible light transmittance (400–800 nm). The blocking percentage of UVA was calculated by equation 1, where T(λ) is the transmittance of the samples at the wavelength of λ, dλ is the bandwidth, and λ is the wavelength. The blocking percentage of UVB was calculated by equation 2, The transmittance of visible light was calculated by equation 3, Sun protection factor (SPF) value determination We measured the SPF value of PVA-SA using the method reported in the literature37. A piece of quartz glass (thickness: 2 mm) was coated with a 3 M medical tape (4 cm2), and then a PVA-SA aqueous solution (100 µL, 80 mg/mL in H2O) was dropped onto the tape. The sample was dried for 20 min in a dark room prior to measuring transmittance spectra at 290–400 nm (spot interval: 1 nm). A quartz glass with a 3 M medical tape was used as a background for baseline calibration. The SPF value was calculated using the following equation 4: where Eλ = CIE erythemal spectral effectiveness, Sλ = solar spectral irradiance, and Tλ = spectral transmittance of the sample. The antioxidant ability of PVA derivatives The antioxidant ability of PVA derivatives was evaluated by radical-scavenging tests (ABTS+• as a typical example of RNS and •OH as a typical example of ROS). The ABTS+• solution was prepared according to the previous literature59. Briefly, ABTS+• was prepared from an aqueous solution (20 mL) containing 2, 2′-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) diammonium salt (ABTS, 4.0 mM) and potassium persulfate (1.41 mM). This ABTS+• solution was kept in darkness for 14 h. The ABTS+• solution was diluted by deionized water to the absorbance at 734 nm of ~0.8, then added into a 96-well plate (190 μL), followed by the addition of PVA-BA solution (10 μL, 2 mg/mL). The absorbance of ABTS+• solution at different time intervals was measured by a microplate reader at 734 nm. PVA-HBA and PVA-SA solutions were simultaneously tested. Deionized water was used as a blank. The absorbance at 15 min was used for the calculation of the scavenging ratio of RNS. The results were presented as mean ± SD, n = 3. The scavenging ratio of ROS was measured using a previous method60. Briefly, aqueous solutions of salicylic acid (20 μL, 10 mM), FeSO4 (20 μL, 10 mM), H2O2 (20 μL, 10 mM), and PVA-BA (100 μL, 10 mg/mL) were mixed in a 96-well plate. The mixtures were incubated at 37 °C for 15 min; then, the characteristic absorption of 2, 3-dihydroxybenzoic acid, which is converted from salicylic acid by •OH, was recorded at 510 nm. PVA-HBA and PVA-SA were similarly tested. Deionized water (100 μL) was used as a blank. The inhibition efficiency of 2, 3-dihydroxybenzoic acid by PVA derivatives reflects the anti-ROS ability of PVA derivatives. The results were presented as mean ± SD, n = 3. The scavenging ratio of RNS or ROS was calculated based on the change in absorbance using following equation 5: where Ab indicates the absorbance of the blank group, As indicates the absorbance of the experimental group. Preparation of the model molecule (M-SA) The model molecule was synthesized via the Hantzsch reaction. Ethyl acetoacetate (130.1 mg, 1.0 mmol), 5,5-dimethyl-1,3-cyclohexanedione (140.2 mg, 1.0 mmol), 3,5-dimethoxy-4-hydroxybenzaldehyde (182.2 mg, 1.0 mmol), ammonium acetate (115.6 mg, 1.5 mmol), and glycine (7.51 mg, 0.1 mmol) were dissolved in acetonitrile (1.0 mL). The mixture was kept at 70 °C for 3 h. Then, the model molecule was acquired by precipitation in deionized water and diethyl ester three times as a gray powder (394.8 mg, yield ~95%). 1H NMR (400 MHz, DMSO-d6, δ/ppm) 9.08 (s, 1H, CHNC), 8.14 (s, 1H, OH), 6.39 (s, 2H, ph), 4.79 (s, 1H, CCHC), 4.04 (d, J = 7.1 Hz, 2H, CH2CH3), 3.67 (s, 6H, OCH3), 2.46–2.27 (m, 2H, CH2C = O), 2.28 (s, 3H, NHCCH3), 2.21–1.97 (m, 2H, CH2CNH), 1.21 (t, J = 7.1 Hz, 3H, CH2CH3), 1.04 (s, 3H, CH3CCH3), 0.94 (s, 3H, CH3CCH3). 13C NMR (100 MHz, DMSO-d6, δ/ppm) 195.07, 167.57, 150.13, 147.88, 144.88, 138.57, 134.29, 110.38, 105.38, 104.57, 59.54, 56.24, 50.71, 35.81, 32.56, 29.83, 26.64, 18.69, 14.75. IR (υ/cm-1): 3547, 3277, 3199, 3075, 2959, 1691, 1604, 1484, 1424, 1378, 1312, 1280, 1211, 1170, 1150, 1116, 1069, 1030, 979, 918, 855, 778, 707. ESI-MS: observed (expected): 438.1890 (438.1887) [M+Na+]. Theoretical calculations of M-SA All calculations were carried out with the Gaussian 16 software package61. Geometry optimizations were conducted under the B3LYP62-D3(BJ)63/def2-SVP64 level. Frequency calculations were carried out at the same level of theory as the geometry optimization to identify the stationary points as minima (no imaginary frequency). The solvation effects of methanol were taken into consideration by using the solvation model density (SMD)65 implicit solvation model. Wave function analysis was performed by Multiwfn66, and orbitals were visualized by virtual molecular dynamics (VMD)67. Cell culture HCE-T cells (Cat. No.: CL-0743, purchased from Procell Life Science & Technology) were cultured in a Dulbecco’s Modified Eagle Medium/Nutrient Mixture F12 (DMEM/F12) supplemented with 15% fetal bovine serum, 5 μg/mL insulin, 10 ng/mL human EGF, and 1% penicillin and streptomycin (HCE-T complete culture medium). Primary human dermal fibroblasts were cultured in fibroblast medium (FM) supplemented with 5% fetal bovine serum, 1% fibroblast culture additives, and 1% penicillin and streptomycin. The cells were cultured at 37 °C in 5% CO2, and the culture medium was changed every day to maintain the exponential growth of cells. Short-term and long-term cytotoxicity of PVA-SA The short-term cytotoxicity of PVA-SA was evaluated by cell count kit-8 (CCK-8) assay. Briefly, HCE-T cells (∼1 × 105 cells/mL) were seeded in a 96-well plate in HCE-T complete culture medium (100 μL). After attachment, cells were washed with PBS and incubated in 100 μL of medium containing 1, 2, 4, 8, 10 mg/mL of PVA-SA, respectively. After 24 h, cells were washed with PBS and cultured in 100 μL of culture medium with 10% CCK-8 for 2 h before observation. The plate was put into a microplate reader (VICTORTM X3 PerkinElmer 2030 Multilabel Plate Reader) to record the absorbance at 450 nm. The results were compared to a positive control (cells cultured in culture medium only; 100%) and a negative control (without cells; 0%) to calculate the cell viability. PVA solutions were similarly prepared and tested for HCE-T cell tests. The long-term cytotoxicity of PVA-SA was evaluated by a colony formation assay. HCE-T cells (~100 cells/well in 12-well plates) were incubated with PVA-SA (10 mg/mL in culture medium) for 24 h at 37 °C in 5% CO2. The cells were continually cultured for another 10 days with normal culture medium (the single administration group) or culture medium containing PVA-SA (10 mg/mL) (the multiple administration group, 5 times). The colonies were fixed with paraformaldehyde (4% in H2O) for 15 min (25 °C) and stained with crystal violet aqueous solution (0.1%) for 15 min (25 °C), followed by washing with PBS twice. The colonies containing more than 50 cells were recorded as survivors; data were presented as mean ± SD (n = 3), and the colony number of untreated cells was used as the blank group and defined as 100%. The short-term cytotoxicity of PVA-SA for primary human dermal fibroblasts was similarly tested. The long-term cytotoxicity of PVA-SA for primary human dermal fibroblasts was also similarly tested for another 14 days instead of 10 days. The crystal violet fraction was calculated by ImageJ (1.51), and the crystal violet fraction of untreated cells was used as the blank and defined as 100%. Cellular UV protection by PVA-SA: viability, γ-H2AX staining assay, and colony formation assay The cellular UV protection by PVA-SA was evaluated by fluorescein diacetate/propidium iodide (FDA/PI) double staining assay. Briefly, HCE-T cells (∼1 × 105 cells/mL) were seeded in a 96-well plate in HCE-T complete culture medium (100 μL). After attachment, cells were washed with PBS and incubated in 100 μL of medium containing 10 mg/mL of PVA-SA. Then, the cells were irradiated by UVA (365 ± 3 nm, 20 mW/cm2, 10 min) and UVB (312 nm, 125 μW/cm2, 40 min), respectively. The doses of UVA and UVB irradiation that reached the cells were 120,000 J/m2 and 3000 J/m2, respectively. After irradiation, the cells were washed with PBS and incubated with fresh culture medium for 24 h. Then, PBS-FDA-PI mixed solution (FDA: 3 μg/mL; PI: 3 μg/mL) was added to simultaneously observe the live cells (λex = 488 nm, λem = 493–560 nm) and dead cells (λex = 543 nm, λem = 565–700 nm) via a Laser Scanning Confocal Microscopy (ZEN black 2.3). The viability of cells was also analyzed through a CCK-8 assay as mentioned above. The viability of cells in pure culture medium without UV irradiation was defined as 100%. The PVA solution was similarly prepared and tested. The γ-H2AX staining assay was used to test damaged DNA in cells after UV irradiation. Briefly, we treated the cells with PVA-SA (10 mg/mL in culture medium) before or after UV irradiation (UVA: 365 ± 3 nm, 120,000 J/m2; UVB: 312 nm, 3000 J/m2). The cells treated with UV alone and the untreated cells served as the control and blank groups, respectively. Then, we evaluated the DNA damage of each group using the γ-H2AX staining assay according to the manufacturer’s protocol (Phospho-Histone H2AX (Ser139) Rabbit Monoclonal Antibody (Beyotime): Category: Rabbit Monoclonal Antibody (RabMAb): Primary antibody; Product lD: AF1201; isotype: lgG). The green fluorescence of DNA-damaged cells (λex = 495 nm, λem = 515–520 nm) was observed via a Laser Scanning Confocal Microscopy. The colony formation assay was used to test radiation-induced damage and damage repair. Briefly, HCE-T cells (~100 cells/well in 12-well plates) were exposed to UV (UVA: 365 ± 3 nm, 120,000 J/m2 or UVB: 312 nm, 3000 J/m2), and the culture medium was refreshed after irradiation. The cells were continually cultured for another 10 days at 37 °C in 5% CO2. The colonies were fixed with paraformaldehyde (4% in H2O) for 15 min (25 °C) and stained with crystal violet aqueous solution (0.1%) for 15 min (25 °C) followed by washing with PBS twice. In the prevention group, the cells were co-cultured with PVA-SA (10 mg/mL), then exposed to UV, followed by incubation in fresh culture medium after irradiation; in the rescue group, the cells were irradiated, then cultured with PVA-SA (10 mg/mL) for 24 h, followed by incubation in fresh culture medium. The colonies containing more than 50 cells were recorded as survivors; data were presented as mean ± SD (n = 3), and the colony number of untreated cells was used as the blank group and defined as 100%. The UV protection for primary human dermal fibroblasts by PVA-SA was similarly tested using the colony formation assay for another 14 days instead of 10 days. The crystal violet fraction was calculated by ImageJ (1.51), and the crystal violet fraction of untreated cells was used as the blank and defined as 100%. RNA isolation and library preparation HCE-T cells (∼1 × 105 cells/mL) were seeded in a 96-well plate in HCE-T complete culture medium (100 μL). After attachment, cells were washed with PBS and incubated in 100 μL of medium containing 10 mg/mL PVA-SA. Then, the cells were subjected to the following two treatments, respectively. (1) PVA-SA alone: the cells were incubated in culture medium containing PVA-SA (10 mg/mL) for 10 min and 40 min, respectively. (2) PVA-SA protection: The cells were irradiated by UVA (365 ± 3 nm, 20 mW/cm2, 10 min) and UVB (312 nm, 125 μW/cm2, 40 min), respectively. After irradiation, the cells were washed with PBS and incubated with fresh culture medium for 24 h. Then, total RNA was extracted using the TRIzol reagent (Invitrogen, CA, USA) according to the manufacturer’s protocol. RNA purity and quantification were evaluated using the NanoDrop 2000 spectrophotometer (Thermo Scientific, USA). RNA integrity was assessed using the Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). Then the libraries were constructed using the VAHTS Universal V6 RNA-seq Library Prep Kit according to the manufacturer’s instructions. The transcriptome sequencing and analysis were conducted by OE Biotech Co., Ltd. (Shanghai, China). The primary human dermal fibroblasts were similarly tested. RNA sequencing and differentially expressed gene analysis The libraries were sequenced on an Illumina Novaseq 6000 platform, and 150 bp paired-end reads were generated. Raw reads of fastq format were first processed using fastp, and the low-quality reads were removed to obtain the clean reads. The clean reads were mapped to the reference genome using HISAT2 software. Differential expression analysis was performed using DESeq2. P-value < 0.05 and fold change >1.5 were set as the threshold for significantly differential expression gene (DEGs). Hierarchical cluster analysis of DEGs was performed using R (v 3.2.0) to demonstrate the expression pattern of genes in different groups and samples. Based on the hypergeometric distribution, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis of DEGs were performed to screen the significantly enriched terms using R (v 3.2.0), respectively. Gene Set Enrichment Analysis (GSEA) was performed using GSEA software. The analysis was based on a predefined gene set, and the genes were ranked according to the degree of differential expression in the two types of samples. Then, it is tested whether the predefined gene set was enriched at the top or bottom of the ranking list. Animal experiments All animal procedures in this study were performed in conformity with the Guidelines for Care and Use of Laboratory Animals of Tsinghua University and approved by the Animal Ethics Committee of Tsinghua University (approval number: 20-TL1, 20-TL1.G25-1). All experimental animal procedures were performed under anesthesia, and all efforts were made to minimize suffering. All zebrafish embryos and larvae used in this study were from the Tuebingen strain. Fertilized eggs were bought from China Zebrafish Resource Center and raised in Holtfreter’s solution (59 mM NaCl, 0.67 mM KCl, 0.76 mM CaCl2, and 2.4 mM NaHCO3). The larvae (72–96-h post-fertilization, hpf) were used for the experiments. All embryos and larvae were kept in a 28.5 ± 0.5 °C incubator with a 14 h-light/10 h-dark cycle (with lights on at 8:30). No exogenous feeding was provided, as all fish were used before 168 hpf. Welfare monitoring was performed daily by visual inspection of each larva for general appearance, swimming behavior, body position, and responsiveness to gentle mechanical stimulation. Any individual showing signs of distress (e.g., loss of equilibrium, abnormal posture, or lack of response) was immediately euthanized. Euthanasia was achieved by overdose of 3-aminobenzoic acid ethyl ester methanesulfonate (MS-222). BALB/C mice (7 weeks old, 16–20 g, female) were raised in a 10 h-dark/14 h-light housing conditions (20−26 °C, and 40−70% humidity) according to the GB 14925-2023 standard. The mice were fed a standard autoclaved diet (equivalent to LabDiet 5K52 formulation, 6% fat content) and pure water. Welfare monitoring was performed daily by visual inspection of each animal for general appearance (ruffled fur, hunched posture), body weight, and responsiveness to gentle handling. Any individual showing signs of distress (e.g., persistent weight loss or lack of response) was immediately euthanized. Euthanasia was achieved by carbon dioxide (CO₂) inhalation. UV irradiation on Zebrafish larvae The larvae were reared and handled exactly as described above. Firstly, 60 larval zebrafish (72-hpf) were placed in PVA-SA solution (10 mg/mL, in Holtfreter’s solution). Then, larvae were exposed to UVA exposure (365 ± 3 nm) for about 30 s. The doses of UVA irradiation were about 8000 J/m2. After irradiation, larvae were transferred to fresh Holtfreter’s solution for further study. The larvae in other media (pure Holtfreter’s solution; Holtfreter’s solution containing 10 mg/mL PVA; 60 larvae/group) were similarly tested. The larvae in pure Holtfreter’s solution without UV irradiation were used as a blank. Startle response testing After 24 h of UV irradiation, larvae (96-hpf) were randomly selected from the abovementioned groups (24 larvae per group) and transferred to a 96-well plate and housed individually for ease of tracking and statistical analysis. After 10 min of acclimation in the dark, the larvae were exposed to three cycles of alternating 2-min light and 2-min dark periods (for a total of 12 min). Movement of each fish was monitored using a Noldus behavior recording system (Noldus Information Technology). The temperature in the testing room was kept at 25 ± 0.5 °C. Tracks were analyzed from videos using EthoVision XT16 software (Noldus Information Technology). Phototaxis testing After 24 h of UV irradiation, larvae (96-hpf) were randomly selected from the abovementioned groups (30 larvae each group) and transferred to different glass dishes. The glass dishes were placed on a half-black-half-white round circle to mimic dark and light conditions. The larvae in each group were kept in the dark side of the glass dish by a baffle. After the baffle was removed, the behavior of the larvae was recorded for 1 min by a camera, and the number of larvae on both dark and light sides was also recorded and analyzed. Apoptosis evaluation of the ocular cells of larval zebrafish The acridine orange (AO) staining method was applied to detect the sites of apoptosis in larval zebrafish. After 24 h of UV irradiation, larvae (96-hpf) were randomly selected from each group (3 larvae per group) and transferred to the AO solution (5 μg/mL, Holtfreter’s solution). After staining in the dark at 28.5 °C for 30 min, the larvae were washed with fresh Holtfreter’s solution three times to remove extra AO. Then, the larvae were anesthetized with 0.03% MS-222 for 2 min and fixed in 3% methylcellulose. The head position of larvae was adjusted to keep the two eyes in the same vertical plane and overlapping. The apoptotic ocular cells were observed via a Laser Scanning Confocal Microscopy (λex = 488 nm, λem = 500–530 nm). Quantitative data were analyzed using ImageJ software. Histological analysis of larval zebrafish eyes After 24 h of UV irradiation, the 96-hpf larvae were fixed in a 4% paraformaldehyde solution, then embedded in paraffin and stained with hematoxylin and eosin (H&E). The histological morphology of larvae’s eyes was observed by a digital pathology section scanning system. Quantitative data were analyzed using ImageJ software. Acute eye irritation test Eight BALB/C mice (7 weeks old, 16–20 g, female) were randomly separated into four groups (4 eyes for each group). Then, PVA aqueous solution (2 μL, 10% in saline) and PVA-SA aqueous solution (2 μL, 10% in saline) were added dropwise to the mice’s eyes. The group treated with saline was used as a blank. The group treated with sodium dodecyl sulfate (SDS, 2 μL, 10% in saline) was used as a control. After a 5-min incubation, the treated eyes were washed three times with PBS. Photos and corneal fluorescein sodium staining test were taken at different time intervals (0.5 h, 1 h, 1.5 h, 24 h, 48 h, and 72 h). Then, mice were sacrificed 72 h after administration, and their eyes were collected for H&E staining and immunofluorescence staining by tumor necrosis factor-α (TNF-α Monoclonal antibody (Proteintech): Category: TNF-alpha Monoclonal Antibody for WB, IHC, IF/ICC, IF-P, ELISA; Cat No. 60291-1-lg; CloneNo. 7B8A11; Mouse/lgG2b.), interleukin-6 (IL-6 Polyclonal antibody (Proteintech): Category: IL-6 Polyclonal Antibody for WB, ELISA; Cat No. 26404-1-AP; Rabbit/lgG.) and terminal deoxynucleotidyl transferase-mediated deoxyuridine triphosphate nick end labeling (TUNEL), respectively. Fluorescein cornea staining score After staining with 1% fluorescein sodium for 90 s, the corneal epithelium in each group (4 eyes/group) was evaluated by a slit lamp with the cobalt blue light at each time interval (0.5 h, 1 h, 1.5 h, 24 h, 48 h, and 72 h). The cornea was divided into four quadrants, which were scored separately. The final score of an eye was obtained by adding all the scores. Scoring criteria: 0 points, no staining; 1 point, less than 30 stained dots; 2 points, more than 30 non-diffuse stained dots; 3 points, with remarkable diffuse staining but without plaque staining; 4 points, with a patchy stain. Ocular retention time of PVA-SA Eighteen BALB/C mice (7 weeks old, 16–20 g, female) were randomly separated into six groups (6 eyes for each group) and treated with PVA-SA (2 μL, 100 mg/mL in saline). The mice treated with saline (2 μL) were used as a blank. At different time intervals (10 min, 20 min, 30 min, 60 min, and 90 min) after administration of PVA-SA, mice were euthanized, and their eyes were collected and soaked in saline (2 mL) for 1 h. Then, the fluorescence intensities of the eye-containing solutions were measured (λex = 365 nm, λem = 442 nm) by a fluorescent photometer. The fluorescence intensities at different time points were normalized by subtracting the solution fluorescence of the blank group to monitor the amounts of PVA-SA in the eyes at different time points. UV protection of PVA-SA on the mouse retina We randomly separated nine BALB/C mice (7 weeks old, 16–20 g, female) into three groups (6 eyes for each group). Then, we dropped a solution of PVA-SA (100 mg/mL, in saline) into the eyes of BALB/C mice, and then exposed the eyes to UV light (365 ± 3 nm, 60,000 J/m2) for 120 s (equivalent to 1.2−6 years of average UV dose in sunlight). Healthy mice and mice exposed to UV light without any treatment were used as the blank group and the control group (the UV group), respectively. After 24 h of UV irradiation, the mice were euthanized, and their eyes were fixed in a 4% paraformaldehyde solution, then embedded in paraffin and stained with H&E. Their eyes were also collected for immunofluorescence staining by IL-6. Quantitative data were analyzed using ImageJ software. In vivo acute toxicity of PVA-SA The in vivo toxicity of PVA-SA to BALB/C mice (7 weeks old, 16–20 g, female) was evaluated via intravenous injection of PVA-SA (30 mg, in saline) to mice (12 mice/group). Mice (6 mice/group) were euthanized 3 days after administration of PVA-SA by carbon dioxide asphyxia. Blood was collected, and serum was separated by centrifugation (3000 rpm, 30 min, 20 °C) to measure the AST, ALB, urea, and creatinine values by an automatic Biochemistry Analyzer (Sinnowa, Nanjing, China) at 37 °C; major organs (hearts, lungs, livers, kidneys, and spleens) were isolated for histological studies by H&E staining. Healthy mice served as a blank. After 7 days of injection, the rest of the mice (6 mice/group) were sacrificed and tested by the same methods. The mice injected with PVA (30 mg, in saline) were similarly tested. In vivo long-term toxicity of PVA-SA Ten BALB/C mice (7 weeks old, 16–20 g, female) were randomly separated into two groups (five mice, 10 eyes for each group). Saline (2 μL) and a PVA-SA solution (2 μL, 10% in saline) were added dropwise to the mice’s eyes twice a day (once at 8:00 am and once at 8:00 pm) for a continuous 14 days, respectively. Photos and corneal fluorescein sodium staining test were taken on the 3rd, 7th, and 14th day. Then, mice were euthanized 14 days after administration of PVA-SA or saline by carbon dioxide asphyxia. Their eyes were collected for H&E staining and immunofluorescence staining by TNF-α, IL-6, and TUNEL, respectively. Their blood was collected, and serum was separated by centrifugation (3000 rpm, 30 min, 20 °C) to measure the AST, ALB, urea, and creatinine values by an automatic Biochemistry Analyzer (Sinnowa, Nanjing, China) at 37 °C; major organs (hearts, lungs, livers, kidneys, and spleens) were isolated for histological studies by H&E staining. Cytotoxicity of SA The cytotoxicity of SA was evaluated by CCK-8 assay. Briefly, HCE-T cells (∼1 × 105 cells/mL) were seeded in a 96-well plate in HCE-T complete culture medium (100 μL). After attachment, cells were washed with PBS and incubated in the presence of SA with different concentrations (0.1−2 mg/mL, in 0.2% v/v DMSO/culture medium). After 24 h, cells were washed with PBS and cultured in 100 μL of culture medium with 10% CCK-8 for 2 h before observation. The plate was put into a microplate reader (VICTORTM X3 PerkinElmer 2030 Multilabel Plate Reader) to record the absorbance at 450 nm. The results were compared to a positive control (cells cultured in culture medium only; 100%) and a negative control (without cells; 0%) to calculate the cell viability. Cellular UV protection by SA The cellular UV protection by SA was evaluated by CCK-8 assay. Briefly, HCE-T cells (∼1 × 105 cells/mL) were seeded in a 96-well plate in HCE-T complete culture medium (100 μL). After attachment, cells were washed with PBS and incubated in 100 μL of medium containing 0.2 mg/mL of SA. Then, the cells were irradiated by UVA (365 ± 3 nm, 20 mW/cm2, 10 min) and UVB (312 nm, 125 μW/cm2, 40 min), respectively. The doses of UVA and UVB irradiation that reached the cells were 120,000 J/m2 and 3000 J/m2, respectively. After irradiation, the cells were washed with PBS and incubated in fresh culture medium for 24 h. Then, the cells were cultured in 100 μL of culture medium with 10% CCK-8 for 2 h before observation. The plate was put into a microplate reader (VICTORTM X3 PerkinElmer 2030 Multilabel Plate Reader) to record the absorbance at 450 nm. The results were compared to a positive control (cells cultured in culture medium only; 100%) and a negative control (without cells; 0%) to calculate the cell viability. Statistical analysis The results were analyzed using SPSS Statistics v.25.0; values were presented as mean ± SD. For the two-group comparison, p-values were derived from the one-way Student t-test to determine the difference. Statistical significance was set at p < 0.05. For the animal experiment, all samples were randomly selected and analyzed; no sample was optionally excluded from the analysis. Blinding was done in the analyses and quantifications. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Supplementary information Supplementary Information Description of Additional Supplementary Information Supplementary Video 1 Supplementary Video 2 Reporting Summary Transparent Peer Review file Source data Source Data Acknowledgements This research was supported by the National Key Research and Development Program for Government-to-Government International Scientific and Technological Cooperation (2024YFE0107100 (Y.Wu)), the National Natural Science Foundation of China (22471140 (L.T.), 52473286 (Y.Wu)), and the Key Research and Development Plan (2022YFB3804901 (L.T.)). Author contributions L.T. initiated and supervised the research. S.P., Z.M., B.W. and Z.L. performed the experiments. S.P., S.D., Q.Z. and Y.Wu analyzed the gene data. Y.Wei provided resource support. L.T., B.W. and Y.Wu provided financial and resource support. L.T. and S.P. wrote the manuscript. All authors discussed and revised the manuscript. Peer review Peer review information Nature Communications thanks Hao Sun and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available. Data availability The data generated in this study are provided in the article and the Supplementary Information file. Source data are provided with this paper. The RNA-sequencing dataset (GSE329195) is available via the link (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE329195). Source data are provided with this paper. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Bo Wang, Email: youxihema@gmail.com. Yuwei Wu, Email: yuweiwu@bjmu.edu.cn. Lei Tao, Email: leitao@mail.tsinghua.edu.cn. Supplementary information The online version contains supplementary material available at 10.1038/s41467-026-73448-8. References Associated Data Supplementary Materials Supplementary Information Description of Additional Supplementary Information Supplementary Video 1 Supplementary Video 2 Reporting Summary Transparent Peer Review file Source Data Data Availability Statement The data generated in this study are provided in the article and the Supplementary Information file. Source data are provided with this paper. The RNA-sequencing dataset (GSE329195) is available via the link (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE329195). Source data are provided with this paper.