Engineering biomimetic tarsal microtissue via structurally zoned hydrogel scaffold for integrated eyelid reconstruction

Xindi Liu; Ke Yao; Meizhu Wang; Xuri Chen; Zhichu Chen; Zheren Sun; Xuanhe Lin; Nianjia Wang; Chunlei Yao; Shenyu Huang; Yijie Wang; Qi Gao; Yong He; Juan Ye

doi:10.1016/j.bioactmat.2026.05.052

. 2026 Jun 17;65:691–711. doi: 10.1016/j.bioactmat.2026.05.052

Engineering biomimetic tarsal microtissue via structurally zoned hydrogel scaffold for integrated eyelid reconstruction

Xindi Liu ^a,¹, Ke Yao ^b,^c,¹, Meizhu Wang ^a,¹, Xuri Chen ^d,¹, Zhichu Chen ^a, Zheren Sun ^b,^c, Xuanhe Lin ^b,^c, Nianjia Wang ^e, Chunlei Yao ^f, Shenyu Huang ^a, Yijie Wang ^a, Qi Gao ^a,^⁎, Yong He ^b,^c,^⁎⁎, Juan Ye ^a,^⁎⁎⁎

PMCID: PMC13310939 PMID: 42376008

Abstract

The tarsus is critical for maintaining eyelid structural integrity, functional stability, and ocular surface health. Tarsal defects result in dual impairments of mechanical support and meibomian gland secretion, leading to severe complications such as corneal exposure and epithelial damage. However, current tarsal substitute materials fail to simultaneously meet the requirements of anatomical biomimicry, mechanical compatibility, and functional tissue integration, thereby constituting a significant bottleneck in clinical repair. To address this challenge, we developed a novel biomimetic tarsal microtissue strategy: based on the natural anatomical structure of the tarsus, a structurally zoned gelatin methacryloyl (GelMA) hydrogel scaffold was fabricated via digital light processing (DLP) 3D printing. Its mechanical properties were further enhanced by mechanical stimulation combined with salting-out, achieving a strength and flexibility comparable to those of native tarsal tissue. Subsequently, the scaffold was seeded with rosiglitazone (Rosi)-induced differentiated human meibomian gland epithelial cells (hMGECs) to construct the biomimetic tarsal microtissue. In vitro and in vivo experiments demonstrated that this microtissue exerted excellent substitution effects. When implanted in situ into rat eyelid defects, it effectively provided mechanical support and promoted tissue regeneration. This integrated strategy offers a promising approach to structural reconstruction and clinical functional amelioration of tarsal defects, with significant potential to improve ocular surface health and clinical outcomes for patients.

Keywords: Tarsus, Eyelid reconstruction, Microtissue, Hydrogel scaffold

Graphical abstract

Highlights

•
Based on native tarsal anatomy, a zonally biomimetic GelMA scaffold was fabricated via DLP 3D printing.
•
Mechanical stimulation plus salting-out enhanced the scaffold's mechanics to match native tarsal strength and flexibility.
•
Rosiglitazone-induced hMGECs were seeded to form microtissue; in vitro/in vivo tests confirmed effective tarsal defect repair.

1. Introduction

As a crucial protective barrier for the eyeball, the structural integrity and functional stability of the eyelids directly dictate the health of the ocular surface and visual function [1,2]. Among the eyelid structures, the tarsus, as the core supportive component of the eyelid's posterior layer [3,4], not only maintains eyelid morphology and motor function via its inherent mechanical properties but also provides critical lipid-based lubrication to the ocular surface through its internally distributed, well-organized meibomian glands and ducts, thereby preventing the development of ocular surface disorders such as dry eye disease [5,6]. However, tarsus defects frequently occur due to factors including trauma, tumor resection, or congenital malformations [7]. In such cases, the eyelid suffers from dual impairment of its supportive and secretory functions. Patients not only experience cosmetic and functional impairments but also face an increased risk of severe complications such as corneal exposure and infection, which may even jeopardize visual acuity. Therefore, the repair of tarsal defects remains a core challenge in oculoplastic surgery [8]. The development of tarsus substitute materials that integrate structural biomimicry, mechanical compatibility, and functional tissue integration has emerged as a key direction to overcome this dilemma.

Currently, the clinically commonly used tarsus repair strategies still have significant limitations [[9], [10], [11]]. Autologous tissue transplantation can restore eyelid support to a certain extent [[12], [13], [14], [15]]; however, it requires additional surgery to harvest donor tissue, which can easily lead to donor site damage and functional impairment. Furthermore, the limited quantity of donor tissue makes it difficult to match the morphological requirements of complex defects. Synthetic polymeric materials exhibit favorable mechanical support performance [[16], [17], [18]], but their poor flexibility prevents them from adapting well to the motor demands of the eyelids. Moreover, these materials exhibit poor biocompatibility and are unable to induce the regeneration of functional tarsus tissue; long-term implantation is prone to complications, such as foreign body reactions and corneal irritation.

Hydrogel materials have emerged as a research focus in tissue engineering due to their excellent biocompatibility, tunable degradability, and physical properties similar to those of soft tissues [[19], [20], [21]]. Among them, gelatin methacryloyl (GelMA) is a biocompatible and photocurable hydrogel composed of gelatin and methacrylate (MA) [22,23]. Constituting over 99% of GelMA's chemical composition, gelatin is produced by collagen hydrolysis and exhibits biocompatibility comparable to that of the extracellular matrix (ECM) [24,25]. It has already demonstrated outstanding potential for repairing soft tissues such as skin and cartilage. However, traditional GelMA hydrogels exhibit weak mechanical strength. The modulus of pure GelMA hydrogels is far lower than that of natural tarsus [2], rendering them incapable of withstanding the mechanical stress generated during eyelid movement and prone to collapse and deformation. In recent years, physical modification strategies, such as mechanical stimulation and the salt particle leaching method, have provided a practical approach to enhancing the mechanical properties of hydrogels [24,26]. Additionally, the rapid advancement of 3D printing technology has opened a new avenue for addressing the challenge of material structural biomimicry [27,28]. By precisely controlling printing parameters, 3D printing technology enables the fabrication of scaffolds with complex microstructures, thereby achieving high-fidelity replication of biological tissues.

Beyond this, tarsus repair not only requires the material's structural and mechanical compatibility but also relies heavily on the involvement of functional cells. As the core functional unit responsible for the secretory function of the tarsus, the differentiation status of meibomian gland epithelial cells (MGECs) directly determines the lipid synthesis capacity of the meibomian glands [29,30]. Thus, the question of how to induce and maintain the adipogenic differentiation phenotype of these cells within the material microenvironment remains another core challenge for achieving functional regeneration of the tarsus. Rosiglitazone, a highly specific peroxisome proliferator-activated receptor γ (PPARγ) agonist, has been shown to effectively induce the differentiation of MGECs, promote the expression of lipid synthesis-related genes, and enhance overall adipogenic capacity [31,32]. Beyond its role as a master transcriptional regulator for meibocyte differentiation and lipid synthesis, PPARγ is deeply involved in local immune microenvironment regulation through a well-established crosstalk with the NF-κB signaling pathway [33]. In the pathophysiological context of tarsal repair, severe surgical trauma and biomaterial implantation inevitably trigger acute postoperative inflammation mediated by NF-κB. If left unchecked, this pro-inflammatory cascade dictates excessive fibrotic scarring and the destruction of fragile de novo glandular architectures. Crucially, compelling literature has established an antagonistic regulatory crosstalk between these two pathways: ligand-activated PPARγ not only drives lipogenesis but also exerts potent anti-inflammatory effects by negatively antagonizing the NF-κB cascade via transcriptional transrepression. Therefore, harnessing this dynamic PPARγ/NF-κB crosstalk provides a dual-action cellular regulatory approach: synergistically ensuring the adipogenic functionalization of the meibomian glands while orchestrating an immune-privileged, pro-regenerative microenvironment.

Based on this, the present study proposes a novel construction strategy for tarsus substitute materials (Fig. 1). The primary hypothesis of this study is that a biomimetic tarsal microtissue, synergistically integrating a mechanically reinforced, structurally zoned hydrogel scaffold with rosiglitazone (Rosi)-induced human MGECs (hMGECs), can simultaneously restore the biomechanical support of the eyelid and promote structural and functional tissue regeneration in tarsal defects. To test this hypothesis, we utilized digital light processing (DLP) 3D printing to fabricate a zoned GelMA scaffold that biomimics the natural meibomian gland architecture, followed by mechanical stimulation and salting-out training to achieve native-like flexibility and strength. Rosi-induced hMGECs were then loaded onto this scaffold. Correspondingly, the primary endpoints of this study are defined as: (1) the in vivo restoration of macroscopic eyelid morphology and biomechanical properties (e.g., eyelid thickness and stress-strain profiles) comparable to native healthy tarsus at 1-month post-implantation; and (2) the morphological reconstruction and functional tissue integration at the defect site, characterized by adipogenic lipid synthesis capacity and a localized anti-inflammatory microenvironment. Furthermore, transcriptomic and single-cell RNA sequencing analyses serve as secondary, exploratory endpoints to elucidate the underlying molecular and cellular interactive mechanisms driving the early repair process. By establishing these clearly defined endpoints, this integrated strategy aims to overcome the limitations of traditional tarsus repair materials and provide robust translatable evidence for the functional reconstruction of severe tarsal defects, thereby improving the ocular surface health and quality of life of patients.

2. Results

2.1. Fabrication and characterization of the hydrogel scaffold

According to the Hofmeister effect, different ions exhibit distinct protein precipitation capacities, which can be leveraged to regulate protein aggregation states by simply adding ions. With the assistance of specific ions, modulus-tunable structures can be constructed using the same protein composition [34]. In addition to increasing molecular concentration, mechanical stimulation significantly promotes the rearrangement of hydrogel molecular chains and greatly accelerates the scaffold's performance enhancement during training. Herein, we propose a combination of molecular and structural engineering approaches for hydrogel fabrication. First, a biomimetic tarsus scaffold was fabricated via DLP 3D printing, followed by mechanical stimulation training in a saline solution (Fig. 2A). During training, the hydrogel was placed in an ammonium sulfate solution for repeated mechanical stimulation. Phase separation occurred via salting-out, in which ammonium sulfate replaced water molecules, facilitating the ordered rearrangement of hydrogel molecular chains [35]. The hydrogel was then transferred to a PBS solution to release pre-stretch stress, with ammonium sulfate being leached out to ensure the biocompatibility of the hydrogel scaffold. After multiple training cycles, robust and tough hydrogel scaffolds were successfully constructed, with structural and functional characteristics highly biomimetic of the native tarsus (Fig. 2B).

Fig. 2 — **Fabrication and Multidimensional Characterization of Biomimetic Tarsal Hydrogel Scaffolds. (A)** Schematic Illustration of DLP 3D Printing Process for Hydrogel Scaffold. **(B)** Schematic of Cyclic Mechanical Stimulation Combined with Salting-Out Assisted Training Process. (C–F) Mechanical Property Characterization of Trained Hydrogel Scaffold (Stress-Strain Curve, Tensile Strength, Modulus, Toughness). **(G)** Diagram of Mechanical Property Data. The reported stress and strain values represent the maximum values recorded immediately before scaffold failure (i.e., ultimate tensile stress and maximum strain). The modulus was calculated from the linear region of the stress-strain curve within the 0-10% strain range. **(H)** Picture of Tensile Test. (I–J) Electron Microscopy Observation and Elemental Analysis of Hydrogel Scaffold Before and After Training. **(K)** DLP-Printed Pore Diameter Gradient Scaffolds for Meibomian Cell Aggregation Screening. **(L)** Morphology of the Final Prepared Biomimetic Tarsal Hydrogel Scaffold. (All quantitative data are presented as mean ± SD, n = 3 independent replicates. ∗∗, P < 0.01).

Furthermore, we evaluated the mechanical properties of the trained scaffold. The scaffold fabricated via cyclic mechanical stimulation and salting-out-assisted training exhibited significantly improved mechanical performance, as shown in its stress-strain curve in Fig. 2C. The trained scaffold's tensile strength reached 953.17 kPa, a 43.8-fold increase over the original scaffold. Its modulus and toughness were also substantially enhanced, with values of 879.01 kPa and 1244.46 kJ/m³, respectively, 29.7- and 85.3-fold higher than those of the initial hydrogel. Additionally, the trained scaffold displayed a fracture elongation of 2.3, a 2.19-fold increase over the initial scaffold, and remarkable resistance to crack propagation (Fig. 2D–H). Furthermore, according to a previous study [2], the tensile modulus of the native rabbit tarsal plate within a 0-15% strain range is 2.554 ± 1.453 MPa. In comparison, the modulus of our trained scaffold within a similar strain range (0-10%) reached 0.88 ± 0.31 MPa. Although slightly lower than that of the native tissue, the modulus of our scaffold remains within the same order of magnitude, demonstrating potential for adequate biomechanical support. These results confirm that the training process significantly enhances the scaffold's mechanical properties, rendering its mechanical behavior more biomimetic of the native tarsus.

We also conducted electron microscopy and elemental analysis to assess the reinforcing effects of training on the scaffold. Initially, the bulk hydrogel network within the scaffold exhibited a porous structure (Fig. 2I), characterized by uniform element distribution and numerous voids. Following mechanical and salting-out-assisted training, the hydrogel's strength and toughness were enhanced through alterations in molecular arrangement and hydrogen-bonding networks, thereby facilitating stress dispersion under external loads. Post-training, the scaffold's molecular network was rearranged, leading to the disappearance of many pores and the formation of a denser structural configuration (Fig. 2J). The remaining pores act as stress-concentration sites, aiding energy absorption and dispersion, thereby improving the hydrogel's impact resistance. Consequently, the trained scaffold's porous architecture enhances mechanical properties, including increased strength, toughness, and ductility [36].

Finally, to determine the optimal size of the printed pore in the scaffold for meibomian cell aggregation, we designed a series of pore diameter gradients to screen for the optimal pore size. As illustrated in Fig. 2K, high-precision DLP printing technology enables the accurate fabrication of the specified pore sizes, facilitating subsequent cell experiments to assess the scaffold's efficacy in facilitating meibomian cell aggregation. The final scaffold after printing and preparation is shown in Fig. 2L.

2.2. Cellular Compatibility Evaluation of the hydrogel scaffold

To evaluate the cellular biocompatibility of the high-strength hydrogel scaffold, this study used the scaffold extract as the experimental group and D-KSFM medium (as described in the Methods section) as the control group, and detected the survival status and proliferative activity of the human meibomian gland epithelial cells (hMGECs) through live/dead staining and CCK8 assay, respectively. Specifically, the fluorescence microscopic images of live/dead staining (Fig. 3B) showed that hMGECs in both groups were dominated by live cells emitting green fluorescence, with a very low proportion of dead cells emitting red fluorescence; the quantitative statistical results (Fig. 3C) further confirmed that there was no significant statistical difference in the live cell rate of hMGEC between the experimental group and the control group (P > 0.05), indicating that the scaffold extract did not exert an inhibitory effect on hMGEC survival. Furthermore, the CCK8 assay was used to evaluate the proliferative activity of hMGEC, and the results (Fig. 3D) showed that the absorbance value of hMGEC viability in the experimental group was consistently higher than that in the control group during the culture period, with a statistically significant difference (P < 0.05). This suggests that the high-strength hydrogel scaffold extract not only has no cytotoxicity to hMGEC, but may also exert a certain promoting effect on their proliferation.

Fig. 3 — **Cellular Compatibility Evaluation of Hydrogel Scaffold and Screening of Optimal Pore Diameter for hMGECs Growth. (A)** Schematic diagram of multi-gradient pore diameter design for screening meibomian gland acini-mimetic cell clusters. **(B)** Fluorescence microscopic images of hMGECs after live/dead staining (green: live cells; red: dead cells). **(C)** Quantitative statistical results of live/dead staining. (n = 4; ns, no significant difference) **(D)** Absorbance values of hMGECs viability detected by CCK8 assay during culture. (n = 10; ∗, P < 0.05) (E–F) Growth morphology of hMGECs in scaffolds with different pore diameters after 2 days (E) and 2 weeks (F) of culture. (nuclei labeled with DAPI, cytoskeleton labeled with phalloidin; confocal microscopy imaging; scale bar: 200 μm).

2.3. Screening of pore diameters for cell growth

To screen for the optimal pore diameter that supports the formation of cell clusters in the hMGECs and biomimetically mimics the morphology of meibomian gland acini, this study designed a multi-gradient pore diameter range of 100 μm - 400 μm, referring to the actual acini size observed in human meibomian gland acinar tissue sections. The schematic diagram of this design is shown in Fig. 3A. Experimental results indicated that within the 200 - 300 μm pore diameter range, hMGECs formed obvious cell clusters in the pores after both 2 days and 2 weeks of culture. In contrast, cells barely infiltrated into pores smaller than 200 μm; for pores larger than 300 μm, cells mostly adhered to the pore walls and failed to aggregate into clusters. Fig. 3E shows the growth morphology of hMGECs in different pore diameters after 2 days of culture, and Fig. 3F presents the results after 2 weeks of culture (Dapi for nuclear labeling, phalloidin for cytoskeleton labeling, confocal microscopy imaging).

Based on these morphological observations, the specific spatial parameters were finalized for the structurally zoned tarsal microtissue: a pore diameter of 250 μm was adopted for the biomimetic acinar zone to optimally promote 3D cell clustering, whereas a pore diameter of 100 μm was selected for the ductal zone, utilizing its restricted space to prevent cellular infiltration and maintain a structured conduit.

To explicitly elucidate the biological significance of the scaffold's zoned macro-architecture, we evaluated its capacity to guide spatial-specific morphological assembly using an in vitro co-culture system of hMGECs and L929 fibroblasts (Supplementary Fig. S1). Cytoskeletal F-actin staining demonstrated that the distinct topographical cues strictly dictated cellular morphodynamics. Within the 250 μm pore regions, epithelial cells were physically confined and guided to aggregate into dense 3D spheroids, successfully mimicking the structural morphology of meibomian gland acini. Conversely, the 100 μm pore regions imposed tighter spatial constraints that restricted massive cellular clustering, simulating the narrow physical environment of meibomian ducts.

Importantly, the Stromal region was engineered with a wavy microfiber topology. Based on the mechanobiological principles recently established by our group [25], this wavy topography efficiently provided contact guidance, driving L929 fibroblasts to align along the fibers to mimic the dense connective tissue of the native tarsal plate. Notably, the continuous nature of the co-culture allowed for natural cellular intermingling, resulting in some hMGECs migrating into the stromal zone (simulating the physiological epithelial-stromal boundary). However, the potent structural constraints of the wavy fibers physically prevented these ectopic epithelial cells from forming 3D acinar clusters, forcing them to adopt a spread, aligned morphology instead. These findings powerfully confirm that the primary biological role of the zoned architecture is to serve as a robust topographical template: 3D acinar-like assembly is strictly permitted only within the designated 250 μm physical niches, thereby exerting dominant mechanical control over cellular spatial organization rather than independently driving biochemical differentiation.

2.4. Transcriptome Sequencing Analysis of the Microtissue In Vitro

To explore the repair-oriented transcriptional regulation of hydrogel scaffolds combined with rosiglitazone (Rosi) on hMGECs, samples were divided into four groups (Plate-Ctrl, Gel-Ctrl, Plate-Rosi, Gel-Rosi) based on “whether induced by Rosi” and “whether cultured on hydrogel scaffolds” for transcriptome sequencing.

Principal component analysis (Fig. 4A) showed significant separation of the four groups along PC1 (51.2%) and PC2 (34.1%), indicating distinct transcriptomic differences between groups. The Upset Venn diagram of differentially expressed genes (Fig. 4B) further intuitively presented the number of differentially expressed genes across key comparison pairs, providing basic data support for subsequent differential gene screening. To clarify the scaffold's repair-related regulation, the intersection of differentially expressed genes from “Plate-Ctrl vs Gel-Ctrl” and “Plate-Rosi vs Gel-Rosi” was selected (Figs. 4C and 1058 genes). Analysis based on these genes revealed: up-regulated KEGG pathways (Fig. 4D) enriched in metabolic pathways and cell cycle, supporting cell proliferation and metabolism for repair; up-regulated biological processes (Fig. 4E) focused on mitosis and chromosome segregation; down-regulated biological processes (Fig. 4F) involved skin development and keratinization, suggesting the scaffold can reduce excessive cell keratinization, which is beneficial for maintaining cell phenotypes required for repair.

For the core experimental group “Gel-Rosi”, compared with “Plate-Ctrl”: the volcano plot (Fig. 4G) showed numerous differentially expressed genes; KEGG pathway analysis (Fig. 4H) showed up-regulation of repair-related pathways like p53 and MAPK; up-regulated biological processes (Fig. 4I) included cell migration and epithelial differentiation, core links of tissue repair; down-regulated biological processes (Fig. 4J) still focused on non-repair processes like skin development and keratinization. In conclusion, hydrogel scaffolds combined with Rosi can regulate cell proliferation, migration and reduce excessive keratinization, providing functional support for meibomian gland tissue repair.

2.5. In Vitro Protein Expression validation of the microtissue

To rigorously disentangle the biological modulatory effect of Rosi and justify its necessity prior to in vivo microtissue assembly, we evaluated both adipogenic differentiation and anti-inflammatory response of hMGECs treated with a concentration gradient of Rosi in standard in vitro cultures. As shown in Supplementary Fig. S2, both immunofluorescence staining and RT-qPCR analyses confirmed that Rosi induction markedly upregulated the expression of PPARγ, which in turn promoted the upregulation of adipogenic functional and anti-inflammatory related factors in a dose-dependent manner.

To validate the in vitro protein expression characteristics of the meibomian gland microtissue, this study analyzed key protein expression via Western Blot (WB) and fluorescent staining to distinguish the independent effects of Rosi and hydrogel scaffolds.

WB results showed (Fig. 5A for bands, Fig. 5B–E for quantitative analysis): PPARγ expression in the Plate-Rosi group was significantly higher than in the Plate-Ctrl group, and in the Gel-Rosi group was significantly higher than in the Gel-Ctrl group, with a higher level than the Plate-Rosi group, indicating Rosi upregulated PPARγ and the hydrogel further enhanced this effect. NF-κB and Nox4 showed similar trends: their expression in the Plate-Rosi group was significantly lower than in the Plate-Ctrl group, and in the Gel-Rosi group was significantly lower than in the Gel-Ctrl group, with a lower level than the Plate-Rosi group, demonstrating Rosi downregulated both, and the hydrogel synergistically strengthened this effect. Krt5 showed no significant difference between the Plate-Rosi and Plate-Ctrl groups, but the Gel-Ctrl group had significantly lower expression than the Plate-Ctrl group, with no significant difference between the Gel-Rosi and Gel-Ctrl groups, indicating its downregulation was independently mediated by the hydrogel, unrelated to Rosi. These results confirmed Rosi's known effects (upregulating PPARγ and downregulating NF-κB/Nox4, consistent with the literature) and suggested that the hydrogel scaffold might modulate Krt5 expression and exert synergistic immunomodulatory effects with Rosi.

Fluorescent staining was performed on hydrogels (Fig. 5F–I for staining images, Fig. 5J–M for quantitative analysis): the Gel-Rosi group showed significantly higher fluorescence intensity of PPARγ and ADRP, and significantly lower NF-κB intensity than the Gel-Ctrl group, with no significant difference in Krt5 between the two groups. Consistent with WB trends, these results validated that Rosi's upregulatory effects on PPARγ/ADRP and inhibitory effect on NF-κB remained unimpaired on hydrogel scaffolds, and Krt5 expression was still primarily regulated by the hydrogel.

2.6. Adipogenic lipid synthesis on the microtissue

To evaluate the adipogenic lipid synthesis capacity of hMGECs in hydrogel scaffolds, this study performed Oil Red O staining and Nile red fluorescent staining, with groups divided based on whether induced by rosiglitazone. Oil Red O staining under light microscopy (Fig. 5N) showed that only a small number of red lipid droplets were observed in the Gel-Ctrl group. In contrast, numerous densely distributed red lipid droplets were found in the Gel-Rosi group, directly indicating that rosiglitazone induction significantly promotes lipid synthesis in hMGECs. Nile red fluorescent staining results (Fig. 5O) further verified this trend: the red fluorescent signal (lipids) in the Gel-Ctrl group was weak, with clear blue Dapi nuclear staining; the red fluorescent signal in the Gel-Rosi group was significantly enhanced, consistent with the cell distribution area. Quantitative analysis of Nile red staining (Fig. 5P) showed that the fluorescence intensity in the Gel-Rosi group was significantly higher than that in the Gel-Ctrl group with a statistically significant difference, clearly confirming that rosiglitazone induction can effectively enhance the adipogenic lipid synthesis capacity of hMGECs in hydrogel scaffolds. It is important to note that these assays reflect the robust intracellular lipid synthesis and storage capacity of the seeded hMGECs, which serves as a prerequisite indicator for their secretory potential.

2.7. In Vivo Repair Efficacy of the microtissue

To evaluate the in vivo repair efficacy of the microtissue, a rat tarsal in situ defect repair model was established: Fig. 6A is a schematic diagram of rat tarsal defect in situ model construction, and Fig. 6B is a comprehensive surgical flowchart. Five groups were set up: Healthy (untreated), Control (tarsal defect without repair), Normal Gel (repair with untrained hydrogel), Trained Gel (repair with trained hydrogel), and Microtissue (trained hydrogel loaded with hMGECs).

After one week, gross appearance (Fig. 6C) showed: the Control group exhibited entropion, with hairs irritating the eye and causing corneal damage; other groups showed good repair effects. Histological sections (Fig. 6D, HE and Masson staining) indicated that the model was successfully established with obvious tarsal defects; the Normal Gel group showed significantly more material degradation than the Trained Gel and Microtissue groups. After one month, gross appearance (Fig. 6E) showed: the Control group still had entropion and corneal damage, and the Normal Gel group had more severe damage than the Control group; the Trained Gel and Microtissue groups showed excellent eyelid repair effects. Histological sections (Fig. 6F, HE and Masson staining) showed that the material in the Normal Gel group was completely degraded, losing its supportive function.

For mechanical properties at 1 month post-surgery, eyelid thickness (Fig. 6G) showed: the Control group had significantly lower eyelid thickness than the Healthy group; the Normal Gel group showed a partial increase but still lower than the Healthy group; the Trained Gel and Microtissue groups had no significant difference in eyelid thickness compared to the Healthy group and were significantly higher than the Control group. In the stress-strain curves (Fig. 6H), the Healthy, Trained Gel, and Microtissue groups showed almost identical trends, while the Control group showed a significantly lower increase in stress with strain, and the Normal Gel group was between the two. Stress at 50% strain (Fig. 6I) showed: the Control group had significantly lower stress than the other four groups, with no significant difference among the Healthy, Normal Gel, Trained Gel, and Microtissue groups, suggesting that the mechanical properties of Trained Gel and Microtissue could recover to a level close to healthy.

2.8. In Vivo Ocular Surface Functional Recovery

To directly evaluate the functional recovery of the ocular surface following eyelid reconstruction, critical clinical ophthalmic parameters—including Tear Film Break-Up Time (TBUT), Phenol Red Thread (PRT) test, and corneal fluorescein staining—were longitudinally assessed at 1, 2, 3, and 4 weeks post-surgery (Fig. 7 and Supplementary Fig. S3). The Control and Normal Gel groups exhibited severe tear film instability (significantly shortened TBUT and PRT) and persistent corneal epithelial damage, as evidenced by intense punctate and coalescent corneal fluorescein staining. In striking contrast, the Microtissue group demonstrated TBUT and PRT results that were close to those of the Healthy group and remained stable throughout the 1-month observation period. Furthermore, standardized semi-quantitative evaluation using the NEI scoring system revealed that the Microtissue group showed a reduction in corneal epithelial damage scores, but not a drastic one. Notably, the Microtissue group already exhibited a lower NEI score at the 1-week time point, and the NEI score gradually decreased over the 1-month observation period. These direct functional assessments robustly confirm that the structural and biological restoration provided by the microtissue effectively translates into enhanced tear film stability and improved ocular surface health.

Fig. 7 — **In Vivo Ocular Surface Functional Recovery. (A)** Representative corneal fluorescein staining images of each group at 1, 2, 3, and 4 weeks post-surgery. **(B)** Semi-quantitative analysis of corneal epithelial damage assessed by the National Eye Institute (NEI) grading system. **(C)** Quantitative results of Tear Film Break-Up Time (TBUT) at each time point. **(D)** Quantitative results of the Phenol Red Thread (PRT) test at each time point. (All quantitative data are presented as mean ± SD, n = 3 eyes per group.)

2.9. In vivo protein expression of PPARγ and NF-κB

To investigate the in vivo regulatory effects of the microtissue on PPARγ and NF-κB, four groups (Control, Normal Gel, Trained Gel, Microtissue) were set up, and the detection time points were 1 week and 1 month, analyzed by WB and fluorescent staining respectively.

In WB assay (Fig. 8A for WB images, Fig. 8B–E for quantitative results): There was no significant statistical difference in bulk PPARγ expression among the four groups at 1 week and 1 month. Conversely, the whole-tissue expression of NF-κB at 1 week was significantly reduced in the Microtissue group, suggesting that Microtissue implantation is associated with an attenuated local inflammatory response. This scaffold-mediated immunomodulatory effect is directly supported by our comprehensive spatiotemporal tracking of macrophages (Supplementary Fig. S4–S8), which demonstrates that the microtissue significantly mitigates overall inflammatory macrophage infiltration and promotes an early phenotypic transition toward a pro-resolving state; at 1 month, the inflammation levels of the four groups were all relieved, resulting in no statistical difference.

In fluorescent staining assay (Fig. 8F for fluorescent staining images, Fig. 8G–J for quantitative results): Due to the limitations of the material's own properties and sectioning conditions, it was difficult to obtain fluorescent staining sections with intact materials, so the area immediately adjacent to the material was initially selected for in situ analysis. Interestingly, different from the bulk WB results, the Microtissue group showed significantly higher localized PPARγ levels than other groups at both 1 week and 1 month. The localized expression trend of NF-κB remained highly consistent with that of WB, showing low expression in the Microtissue group and high expression in the Normal Gel group at 1 week, and no significant difference among the four groups at 1 month.

To rigorously resolve the apparent discrepancy between the whole-tissue WB and localized IF results for PPARγ, and to explicitly avoid subjective post hoc assumptions, a double-blinded, region-specific IF quantification was further conducted (Supplementary Fig. S9). Two independent, blinded observers quantified the spatial expression dynamics in two distinct zones: the peri-implant zone (tissue adjacent to the material) and the distant zone (native tissue spatially separated from the implant). This blinded spatial analysis explicitly revealed that the significant upregulation of PPARγ in the Microtissue group was highly restricted to the peri-implant interface, while its expression in the distant zone remained comparable to other groups. This spatially restricted response empirically explains the non-significant WB results, as the highly localized PPARγ signal was heavily diluted by the vast volume of unresponsive distant tissues during whole-eyelid homogenization. In sharp contrast, the blinded quantification of NF-κB demonstrated a robust suppression in both the peri-implant and distant regions. This indicates that the microtissue-mediated anti-inflammatory effect operates as a broader, tissue-wide modulation, perfectly aligning with the significant downregulation globally captured in the bulk WB analysis.

2.10. In Vivo Single-Cell Sequencing Analysis at 1 Week post-implantation

To broadly map the early repair microenvironment, single-cell RNA sequencing (scRNA-seq) was performed at 1 week post-implantation. Because four rat eyelids per group were necessarily pooled to overcome the technical limitations of low viable cell yield from such microscopic tissues, these initial transcriptomic findings are inherently presented as an exploratory and descriptive framework.

Cell clustering results (Fig. 9A) showed that the eyelid tissue could be segregated into multiple cell populations, including Orifice cells, Ductal cells, Acinar cells, Meibocytes, Mast cells, Smooth muscle cells, Fibroblasts, Pericytes, Macrophages, T cells, and Melanocytes. Direct comparison of cell clusters between the two groups (Fig. 9B) demonstrated a basic correspondence between the clusters. In terms of cell number comparison (Fig. 9C), the number of Orifice cells in the Microtissue group was significantly increased compared to the Healthy group, while the number of Meibocytes was significantly decreased, and no significant differences were observed in the counts of other cell clusters.

Based on the GO enrichment analysis of these four key cell types (Fig. 9D), the tarsal defect repair process exhibited a coordinated and differential response across multiple cell populations. The significant increase in Orifice cell numbers is presumably attributed to a microtissue-driven phenotypic differentiation. While we initially hypothesized that the scaffold's porous architecture might provide a physiological opening-mimicking cue, we acknowledge that attributing this solely to porosity lacks the definitive proof of controlled in vivo comparisons with varying pore sizes. However, our independent immunofluorescence validation (Supplementary Fig. S10) provides compelling spatial evidence for this structural dependence: Krt10⁺ orifice cells were found to explicitly localize within the internal porous cavities of the implanted scaffold. This spatial distribution strongly suggests that these specific structural features physically act as a localized niche. Furthermore, our newly conducted pseudotime trajectory analysis (Supplementary Fig. S11) corroborated this phenomenon at the transcriptomic level, demonstrating a distinct developmental shift of the microtissue cell population toward the advanced Orifice cell branch. Collectively, these spatial and bioinformatic findings indicate that the directed cellular transition is likely a synergistic result of the scaffold's specific porous architecture and its comprehensive physicochemical microenvironment, which together support their directed growth and functional homeostasis. Furthermore, functional enrichment analysis revealed that these Orifice cells downregulated lipid and small molecule metabolic processes to maintain intrinsic cellular balance, while upregulating cell migration, differentiation, and anatomical structure development, thereby serving as the core force for the regeneration of meibomian gland orifice structures. In contrast, the significant reduction in Meibocyte numbers is likely due to the insufficient filling of the massive cell loss caused by the tarsal defect model by the supplemented hMGECs and the limited growth time at 1 week postoperatively; these cells preserved their lipogenic phenotypic potential by downregulating extracellular matrix synthesis and developmental processes, while upregulating nutrient responses and immune responses to reserve energy for subsequent functional recovery. Ductal cells activated immune defense mechanisms to purify the repair microenvironment, regulated cell survival and developmental programs to reserve regenerative potential, and temporarily downregulated extracellular matrix synthesis to prioritize the maintenance of their core meibum transport function. Acinar cells exhibited upregulated immune defense and antimicrobial responses to eliminate postoperative foreign bodies, with the significant upregulation of cell adhesion facilitating structural connection and functional coordination among acinar cells, providing critical support for the integrity and functional integration of acinar tissue.

Cell communication analysis (Fig. 9E–F) indicated that the cell-cell communication intensity in the Microtissue group was significantly higher than that in the Healthy group. This is presumably a direct reflection of enhanced intercellular signal crosstalk during tissue reconstruction following tarsal defect replacement surgery, indicating that Microtissue implantation can effectively activate the local cellular interaction network and provide crucial support for tissue repair.

To rigorously validate the descriptive transcriptomic trends with true biological replication, independent flow cytometry and immunofluorescence analyses were conducted on unpooled samples (n = 3 independent eyes per group). Consistent with the scRNA-seq prediction, flow cytometric quantification of viable single cells (Supplementary Fig. S12) revealed a statistically significant increase in the percentage of Krt10⁺ cells (orifice cells) and a concurrent decrease in ADRP⁺ cells (meibocytes) in the Microtissue group. These populational shifts were spatially and quantitatively corroborated by independent immunofluorescence analysis of tissue sections (Supplementary Fig. S10), confirming the robust alterations in Krt10 and ADRP expression. Together, these independent validations firmly substantiate the microtissue-induced cellular dynamics.

3. Discussion

As the core supportive structure of the eyelid and the source of ocular surface lipids, tarsal defect repair has long been confronted with three major clinical challenges: insufficient structural biomimicry, mismatched mechanical properties, and difficulty in functional tissue integration. In traditional repair strategies, autologous tissue transplantation is associated with donor site morbidity and limited availability [37,38], while conventional artificial materials often fail to achieve synchronous structural reconstruction and functional amelioration due to poor biocompatibility, insufficient mechanical strength, or lack of functional activity [18]. Addressing these pain points, this study constructed a biomimetic tarsal microtissue integrating structural biomimicry, mechanical adaptation, and functionality, providing a novel technical approach for tarsal defect repair.

The core innovations of this study are mainly reflected in three aspects: First, a high-strength GelMA hydrogel scaffold was successfully fabricated via DLP 3D printing combined with a synergistic reinforcement strategy of mechanical stimulation and salt leaching. This design overcomes the bottleneck of mechanical fragility and easy collapse of traditional hydrogels. High-precision 3D printing ensures scaffold structure repeatability, while mechanical reinforcement provides the supportive strength required for eyelid movement and the biocompatibility and cell affinity of hydrogels, achieving a balance between mechanical performance and biocompatibility. Second, a zoned structure was designed based on the anatomical characteristics of meibomian gland acini, ducts, and stroma, and pore size parameters suitable for cell growth were screened. The profound influence of distinct pore sizes on the morphodynamic status of hMGECs is fundamentally governed by the mechanobiological principle of spatial confinement. As documented in 3D tissue engineering, the geometric boundaries of microniches strictly dictate the thermodynamic balance between cell-matrix adhesion and cell-cell cohesion. In larger pores (>300 μm), the expansive surface area with low curvature dominantly promotes integrin-mediated cell-matrix adhesion, causing cells to spread as a 2D monolayer. Conversely, overly small pores (<200 μm) impose severe steric hindrance that prevents mass cellular infiltration. The optimally screened 200-300 μm pores provide the exact geometric confinement that matches the physiological scale of native acini. This specific topological restriction physically hinders extensive 2D planar spreading, thereby mechanically shifting the thermodynamic preference towards robust cell-cell cohesion. Consequently, this physical constraint drives the spontaneous in situ self-assembly of robust 3D multicellular spheroids [39,40]. This biomimetic design provides a biomimetic microenvironment for cell adhesion, proliferation, and functional expression by simulating the spatial arrangement of natural tarsus. Third, rosiglitazone-induced hMGECs were loaded onto the biomimetic structure to achieve synergistic biomimicry of structure and function. The scientific justification for selecting PPARγ and NF-κB as our core in vivo regulatory indicators is deeply rooted in the pathophysiological demands of tarsal reconstruction, which requires a delicate balance between active functional regeneration and the mitigation of acute surgical trauma. PPARγ serves as the indispensable master transcription factor driving meibocyte terminal differentiation and de novo lipid synthesis. Conversely, NF-κB operates as the canonical orchestrator of pro-inflammatory cytokine cascades and postoperative inflammatory responses; if left unchecked following scaffold implantation, its activation dictates severe scarring and regenerative failure. Crucially, these two pathways are linked via a well-established regulatory crosstalk: ligand-activated PPARγ exerts profound transrepression on the NF-κB signaling axis. By utilizing rosiglitazone as a local pharmacological trigger, our microtissue achieved a dual therapeutic effect: it explicitly maintained the local upregulation of PPARγ to drive adipogenesis, while this PPARγ activation inherently silenced the NF-κB signaling cascade to suppress local inflammatory factors. As explicitly corroborated by our temporal macrophage polarization data (Supplementary Fig. S4–8), this PPARγ-mediated transrepression of NF-κB effectively fostered a highly pro-regenerative, anti-inflammatory immune microenvironment, uncoupling glandular regeneration from detrimental foreign-body fibrosis and securing the structural integrity of the reconstructed tarsus.

A series of in vitro and in vivo experimental results fully verified the repair potential of the microtissue. In vitro experiments confirmed that the scaffold's biocompatibility and rosiglitazone's inductive effect synergistically promote adipogenic differentiation of hMGECs, with significantly enhanced lipid synthesis evidenced by Oil Red O and Nile red staining. In the in vivo rat in situ tarsal defect model, the microtissue exhibited excellent repair effects at both 1 week and 1 month post-implantation. Gross appearance showed no entropion or corneal damage, which occurred in the Control group, and mechanical tests revealed that eyelid thickness and stress-strain curves were close to those of the Healthy group, indicating that it had achieved synchronous improvement in morphological repair, mechanical support, and ocular surface protection.

Despite the preliminary progress, this study still has several limitations: First, in terms of structural biomimicry, our scaffold design is fundamentally inspired by the macroscopic anatomy of the human tarsal plate to facilitate clinical translation. While the current zoned structure successfully mimics the functional macro-porosity required for cell aggregation, it does not perfectly replicate the microscopic ultrastructural fibrous arrangement of the native extracellular matrix. Furthermore, it has not fully replicated the fine anatomical structure of the natural tarsus, such as the density distribution of acini and the branching morphology of ducts. However, the screened pore size parameters and zoning strategy provide a reference technical framework for subsequent precise biomimicry. Second, regarding the functional activity and in vivo cellular contributions, we acknowledge a critical limitation inherent to our xenograft model. Because we utilized human-derived cells (hMGECs) in immunocompetent rats without specific human-lineage tracing, the long-term survival and direct functional engraftment of these cells in vivo remain unverified and are likely constrained by xenogeneic immune clearance. As accurately noted, the observed long-term macroscopic repair and structural maintenance are therefore largely driven by the excellent biomechanical properties and inherent immunomodulatory capacity of the trained hydrogel scaffold itself. We hypothesize that the implanted hMGECs primarily serve as an “early biological primer”—exerting transient paracrine effects (such as the early downregulation of NF-κB observed at 1 week) to favorably modulate the acute post-implantation microenvironment before eventual clearance. Furthermore, the Rosi-induced hMGECs primarily exhibit adipogenic differentiation rather than full physiological lipid secretion. Consequently, the current microtissue primarily achieves profound histological reconstruction and ocular surface amelioration, rather than the absolute de novo restoration of the tear film lipid layer by the implanted cells. To definitively uncouple the long-term functional contributions of the seeded cells from the scaffold's physical properties, future studies utilizing species-matched homologous models are imperative. Notably, a study published in November 2025 developed, for the first time, mouse and human meibomian gland organoids [41], providing a robust syngeneic cellular model for future functional tarsal repair that can circumvent the xenograft limitations encountered in this proof-of-concept study. Third, regarding the in vitro biomechanical characterization, while uniaxial tensile testing provided essential quantitative baselines—such as elastic modulus and ultimate strength—to verify the scaffold's mechanical comparability with native tarsus, it does not fully replicate the complex dynamic forces the eyelid experiences in vivo. The current study lacks specific biomechanical evaluations for dynamic compliance, bending mechanics against the curvature of the ocular globe, and long-term cyclic fatigue mimicking continuous physiological blinking. These complex dynamic mechanical assessments represent a limitation of our current methodology and will be a crucial focus of our future biomechanical investigations. Fourth, the in vivo observation period in this study was limited to 1 month post-implantation. While the 1-week and 1-month time points effectively capture early-to-mid-term repair dynamics, host immune responses, and preliminary functional stability, this follow-up duration is undeniably insufficient to definitively evaluate the long-term tissue integration, ultimate in vivo degradation profile of the scaffold, chronic late-stage fibrosis, mechanical durability against continuous cyclic fatigue (e.g., blinking), and long-term safety. Future preclinical studies utilizing extended observation periods (e.g., 3 to 6 months) and larger animal models will be imperative to comprehensively validate the ultimate safety and functional durability of the biomimetic microtissue prior to clinical translation. Fifth, while our in vitro and in vivo evaluations clearly demonstrate the scaffold-mediated immunomodulatory phenomena, the precise multi-target binding mechanisms and the direct activation of specific downstream molecular pathways remain to be fully elucidated. Future mechanistic investigations utilizing specific pathway inhibitors, gene knockout models, or advanced multi-omics approaches are required to definitively map these complex molecular interactions.

While the present study establishes a promising in vivo proof-of-concept for microtissue-mediated glandular and eyelid regeneration, several critical translational hurdles must be explicitly addressed before clinical application. First, regarding defect relevance, our study utilized a surgically induced acute defect model to evaluate de novo tissue regeneration. However, real clinical scenarios often involve complex pathophysiology, extensive scarring, and prolonged inflammation, which may influence regenerative outcomes. Second, concerning species differences, inherent anatomical, physiological, and biomechanical disparities exist between rodent and human eyelids (e.g., variations in tarsal plate architecture, gland density, and blink dynamics). Third, scaling up the engineered construct for human-sized defects presents a significant bioengineering challenge. Enlarging the microtissue dimensions requires overcoming critical mass transfer limitations to ensure adequate core nutrient diffusion and oxygenation prior to in vivo host vascularization. Transitioning from rodent models to large animal models will be an essential next step to rigorously evaluate this scalability. Finally, regarding surgical practicality, the current implantation procedure requires an open micro-incision. For realistic clinical translation, handling and precisely localizing soft constructs within the highly dynamic mechanical environment of the human eyelid demands further optimization. Future iterations should explore minimally invasive delivery approaches, such as injectable hydrogel systems or specialized microsurgical instruments, to minimize surgical trauma, prevent excessive post-operative scarring, and enhance clinical feasibility. Recognizing these limitations provides a clear and objective roadmap for the continued clinical translation of engineered microtissue therapies.

In summary, the biomimetic tarsal microtissue constructed in this study provides a new idea and technical support for tarsal defect repair, and subsequent optimization of structural fineness, cell functionalization, and long-term effects will further promote its clinical application value.

4. Materials and methods

4.1. Materials and reagents

Gelatin methacryloyl (EFL-GelMA) and lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP) were purchased from Yongqinquan Intelligent Equipment Co., Ltd. (Suzhou, China). Phosphate-buffered saline (PBS) was acquired from Zhejiang Jinuo Biomedical Technology Co., Ltd. Ammonium sulfate ((NH₄)₂SO₄, AR, 99%) was obtained from Macklin Biochemical Co., Ltd. (Shanghai, China). Tartrazine dye was purchased from Sigma-Aldrich (St. Louis, MO, USA).

For cell culture, immortalized hMGECs (CRL-3472™) and murine L929 fibroblasts were obtained from ATCC. Keratinocyte Serum-Free Medium (K-SFM), Keratinocyte Growth Supplement, F12 medium, Fetal Bovine Serum (FBS), 0.25% trypsin-EDTA, and penicillin-streptomycin were all purchased from Thermo Fisher Scientific. Rosiglitazone (Cat. No. R2408) was purchased from Sigma-Aldrich.

For biochemical assays and staining, the Cell Counting Kit-8 (CCK-8) was acquired from Dojindo. The Live-Dead Cell Staining Kit (C2015S), DAPI (C1006), Modified Oil Red O Staining Kit (C0158S), Nile Red Lipid Droplet Fluorescence Assay Kit (C2051S), RIPA Lysis Buffer (P0013B), BCA Protein Assay Kit (P0010), Fluorescent phalloidin for cytoskeleton labeling (Cat. No. C2207S), and Pre-stained Protein Marker (P0078) were all purchased from Beyotime Biotechnology (Shanghai, China). RNA extraction kits were sourced from Yishan Biotech. Primary and secondary antibodies utilized are explicitly detailed in their respective methodological subsections.

All other general analytical-grade chemical reagents (including 4% paraformaldehyde, Triton X-100, isopropanol, ethanol, xylene), animal anesthetics (sodium pentobarbital, lidocaine), and standard histological dyes (hematoxylin, eosin, Masson's trichrome reagents) were purchased from standard commercial suppliers such as Sigma-Aldrich or Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China) unless otherwise specified.

4.2. Fabrication and characterization of high-strength hydrogel scaffolds

4.2.1. Scaffold fabrication

First, a 0.3% (w/v) LAP photoinitiator solution containing 0.05% tartrazine dye was prepared using PBS. Next, 2 g of dry GelMA macromer was added to a 50 mL centrifuge tube containing 25 mL of the LAP solution (yielding an optimized 8% w/v concentration). The tube was then placed in a 50 °C water bath until the hydrogel was completely dissolved. The solution was subsequently transferred into a 10 cc dispensing syringe and stored in an incubator at 20–30 °C for future printing.

In the first fabrication step, hydrogel scaffolds were printed using a BP8601Pro bioprinter (Yongqinquan Intelligent Equipment Co., Ltd., Suzhou, China). The scaffold was designed with a structurally zoned geometry, incorporating 250 μm pores in the acinar regions for cell aggregation and 100 μm pores in the ductal regions to serve as conduits. To mimic the physiological tissue environment, the temperatures of the printing platform and the resin vat were set to 37 °C. The printing utilized a 405 nm UV light source with an intensity of 20 mW/cm², and the exposure time was set to 12 s per layer. The printed hydrogel scaffolds were temporarily stored at 2–8 °C prior to the mechanical strengthening process.

In the second step, two custom rectangular acrylic boxes (without lids, internal dimensions: 190 × 120 × 45 mm) were fabricated from 5-mm thick acrylic sheets to hold a 50% ammonium sulfate solution and a PBS solution, respectively. For each training cycle, the printed scaffold was completely immersed in the ammonium sulfate solution and subjected to 20 slow uniaxial stretches. Immediately following the stretches, the scaffold was transferred to the PBS solution and immersed for exactly 1 min to release the pre-stretch stress and leach out residual salts. After completing 20 such cycles—a parameter determined by our previous studies to provide the optimal mechanical strengthening effect—the hydrogel scaffold was further crosslinked under 405 nm UV light to finalize the preparation of the biomimetic tarsal scaffold [24,35].

4.2.2. Characterization of the scaffold morphology

To effectively preserve the porous microarchitecture and minimize structural shrinkage, the hydrogel samples were rapidly frozen in liquid nitrogen and subsequently lyophilized (freeze-dried). Following this preparation, the general morphology of the scaffold was observed using an optical microscope (Nikon, Japan) and a scanning electron microscope (SU8010, Hitachi, Japan) after gold sputtering using a sputter coater.

4.2.3. Mechanical test

Prior to and during the mechanical testing, all scaffold samples were kept fully hydrated in PBS at room temperature. A mechanical testing system (electronic universal testing machine (UTM2102, Shenzhen Sun Technology Co., Ltd.)) was used for the tensile tests of different scaffolds. The extension rate was set to 2 mm/s. The elastic modulus was calculated from the initial linear region of the stress-strain curves, specifically within the 0–10% strain range. All the data were processed and analyzed using Origin software.

4.3. Cell experiments

4.3.1. Culture, seeding, and differentiation induction of hMGECs

4.3.1.1. hMGECs culture

Immortalized hMGECs (ATCC® CRL-3472™) were used in this study, and the culture protocol was optimized based on previously established methods with modifications [29]. The complete growth medium was prepared using Gibco™ Keratinocyte Serum-Free Medium (K-SFM, Thermo Fisher Scientific, 17005-042) as the base, supplemented with 0.2% (v/v) Gibco™ Keratinocyte Growth Supplement (Thermo Fisher Scientific, 37000-015) containing recombinant human epidermal growth factor (EGF, 50 ng/mL) and bovine pituitary extract (BPE, 50 μg/mL), as well as 1% (v/v) penicillin-streptomycin (100 U/mL penicillin, 100 μg/mL streptomycin; Thermo Fisher Scientific, 15140-122) to prevent contamination.

Frozen hMGECs were thawed rapidly in a 37 °C water bath, transferred to a centrifuge tube containing 9 mL pre-warmed complete growth medium, and centrifuged at 125 × g for 5 min to remove cryopreservation reagents. The cell pellet was resuspended in 2 mL complete medium, and cells were seeded into T-25 culture flasks at a density of 5 × 10⁴ cells/cm². Cultures were maintained in a humidified incubator at 37 °C with 5% CO₂ and 95% air. The growth medium was refreshed every 2-3 days, and cells were passaged when reaching 70-80% confluency using 0.25% trypsin-EDTA (Thermo Fisher Scientific, 25200-056).

4.3.1.2. hMGECs seeding on high-strength hydrogel scaffolds

Hydrogel scaffolds were pretreated to improve cell adhesion prior to cell seeding: scaffolds were first soaked in 75% (v/v) ethanol for 2 h, rinsed thoroughly with sterile phosphate-buffered saline (PBS, pH 7.4), and then irradiated under an ultraviolet lamp for 2 h for sterilization; subsequently, they were incubated in complete growth medium at 37 °C for 2 h to equilibrate the microenvironment and pre-load nutrients.

hMGECs in the logarithmic growth phase were harvested and resuspended in complete medium to prepare a cell suspension. 100 μL of cell suspension was used per square centimeter of the scaffold surface (the cell density was adjusted according to experimental requirements, and in this study, it was 6 × 10⁶ cells/mL). The cell suspension was uniformly dropped onto the surface of the pretreated scaffolds, and the scaffolds were placed in a culture vessel. They were incubated at 37 °C with 5% CO₂ for 2 h to allow initial cell adhesion. After adhesion, an appropriate amount of complete growth medium was gently added along the wall of the vessel (to cover the material and avoid disturbing cell-scaffold interaction), and culturing was continued for 2 days to form cell-scaffold constructs.

4.3.1.3. In vitro validation of rosi-induced adipogenic differentiation and anti-inflammatory response

To evaluate the dose-dependent modulatory effect of Rosi and rigorously justify its necessity for inducing the functional phenotype of hMGECs, an in vitro concentration-gradient assay was performed prior to microtissue assembly. hMGECs were seeded in standard culture plates and treated with differentiation medium supplemented with varying concentrations of Rosi (0, 10, 30, and 50 μM). Following the designated 4-day induction period, total RNA was extracted from the cells using an RNA extraction kit (RN001-50Rxns, Yishan Biotechnology, China) according to the manufacturer's instructions. Subsequently, 1 μg of total RNA was reverse-transcribed into complementary DNA (cDNA) using the PrimeScript™ RT Master Mix (Takara, Japan).

Quantitative real-time PCR (qRT-PCR) was then performed using SYBR Green PCR Master Mix. The thermal cycling conditions consisted of an initial denaturation at 95 °C for 30 s, followed by 40 cycles of 95 °C for 5 s and 60 °C for 30 s. The relative mRNA expression levels of key adipogenesis-related and anti-inflammatory genes were calculated using the 2^(-ΔΔCt) method, utilizing GAPDH as an internal reference. All primer sequences used in this assay are detailed in Supplementary Table S1. Concurrently, immunofluorescence staining was conducted to visualize and quantitatively analyze the expression of PPARγ, a core regulator mediating both adipogenic differentiation and anti-inflammatory effects, across the different Rosi concentration groups.

4.3.1.4. Induction of hMGECs differentiation on hydrogel scaffolds

Differentiation induction was initiated when hMGECs on the scaffolds reached approximately 80% confluency (2 days after seeding), which was determined by observation under an inverted phase-contrast microscope. The complete growth medium was aspirated and replaced with differentiation medium, which consisted of F12 medium (Thermo Fisher Scientific, 11765-054) supplemented with 10 ng/mL EGF, 50 μg/mL BPE, 1% penicillin-streptomycin, and 30 μM rosiglitazone (Rosi, Sigma-Aldrich, R2408) — the concentration of Rosi was selected based on dose-response studies [31], which showed that 30 μM achieves maximum lipid synthesis induction without significant cytotoxicity. The cell-scaffold constructs were cultured in differentiation medium for a standardized duration of 4 days, an optimized time point determined by preliminary morphological screening and literature support [31]; the medium was refreshed every 2 days to maintain stable concentrations of growth factors and drugs.

4.3.1.5. In vitro sequential Co-culture on topographically zoned scaffolds

To explicitly evaluate the topography-guided spatial cell arrangement and morphological assembly within the structurally zoned scaffolds, an in vitro co-culture system comprising epithelial cells (hMGECs) and stromal cells (L929 murine fibroblasts) was established to physically simulate the physiological epithelial-stromal boundary.

A sequential cell-seeding strategy was purposefully employed to facilitate initial glandular assembly and prevent the highly migratory fibroblasts from preemptively occupying the designated epithelial niches. First, hMGECs in the logarithmic growth phase were seeded onto the pre-treated biomimetic scaffolds and cultured in the standard complete hMGEC growth medium (K-SFM based) at 37 °C with 5% CO₂ for 2 days. This critical pre-culture period allowed for initial epithelial focal adhesion and the onset of topography-guided 3D aggregation, particularly within the 250 μm pore regions.

Following the 2-day pre-establishment phase, L929 fibroblasts were harvested and sequentially seeded onto the cell-scaffold constructs. To optimally support the metabolic demands of both distinct cell populations while maintaining phenotypic stability, the culture medium was immediately transitioned to a specialized co-culture medium. This customized medium consisted of the original complete hMGEC growth medium supplemented with 1% FBS (Fetal Bovine Serum) (Gibco 10099141C).

The co-culture constructs were continuously maintained in a humidified incubator at 37 °C with 5% CO₂ for an additional 2 days. At the designated endpoint, the constructs were harvested and processed for comprehensive morphological evaluation via cytoskeletal F-actin (phalloidin) and nuclear (DAPI) staining, as detailed in the “Cytoskeleton Staining” section, to analyze the topographical dominance over cell fate.

4.3.2. Biocompatibility verification of high-strength hydrogel scaffolds

4.3.2.1. Preparation of material extract

According to the sample preparation principle of ISO 10993-12:2021 [42], high-strength hydrogel scaffolds were first soaked in 75% (v/v) ethanol for 2 h, rinsed thoroughly with sterile phosphate-buffered saline (PBS, pH 7.4), and then irradiated under an ultraviolet lamp for 2 h for sterilization. After sterilization, the scaffolds were added to the complete growth medium for hMGECs culture (same composition as Section 2.1.1) at a ratio of 0.1 g/mL, incubated in an incubator at 37 °C, 5% CO₂, and 100% relative humidity for 24 h. The resulting solution was stored at 4 °C for later use as the material extract.

4.3.2.2. CCK-8 cell viability assay

hMGECs in the logarithmic growth phase were seeded into 96-well plates at a density of 5 × 10³ cells/well and divided into two groups: normal medium group (containing complete growth medium) and extract group (containing material extract), with 3 replicate wells in each group. After culturing for 24 h, 10 μL of CCK-8 reagent (Dojindo, CK04) was added to each well, followed by incubation for 2 h. The absorbance (OD) was measured at 450 nm using a microplate reader to assess cell proliferation.

4.3.2.3. Live-dead cell staining

hMGECs were seeded into 24-well plates at a density of 2 × 10⁴ cells/well, also divided into normal medium group and extract group, with 3 replicate wells in each group. After 24 h of culture, the staining was performed according to the instructions of the Live-Dead Staining Kit (Beyotime, C2015S): the medium was aspirated, an appropriate amount of staining working solution (calcein-AM for live cells, PI for dead cells) was added, incubated for 15 min in the dark, rinsed with PBS, and then observed and photographed under a fluorescence microscope. The ratio of live to dead cells was counted to assess cytotoxicity.

4.3.2.4. Cytoskeleton and nuclear staining

To evaluate the growth morphology and cluster formation of hMGECs within the hydrogel scaffolds (as shown in Fig. 3E and F), cytoskeleton and nuclear staining were performed. After 2 days and 2 weeks of culture, the cell-scaffold constructs were fixed with 4% paraformaldehyde for 15 min at room temperature and washed three times with PBS. The cells were then permeabilized with 0.1% Triton X-100 for 10 min. To visualize the cytoskeleton, the samples were incubated with fluorophore-conjugated phalloidin (Actin-Tracker Red-Rhodamine, Beyotime, C2207S) at room temperature for 45 min in the dark. After three washes with PBS, the nuclei were counterstained with DAPI (Beyotime, C1006) for 5 min. Finally, the morphological distribution of the cells within the porous scaffolds was observed and captured using a confocal laser scanning microscope.

4.3.3. Transcriptome analysis

hMGECs were divided into four groups (Plate-Ctrl, Gel-Ctrl, Plate-Rosi, Gel-Rosi) with 3 biological replicates in each group. Total RNA was extracted from each group using the RNA extraction kit (RN001-50Rxns, Yishan Biotech) following the manufacturer's instructions. RNA purity (OD₂₆₀/₂₈₀ ratio 1.8–2.1) and integrity were verified by Nanodrop spectrophotometer and agarose gel electrophoresis, respectively. Qualified RNA samples were sent to GeneDenovo for transcriptome sequencing on the Illumina platform (paired-end sequencing, PE150). Raw sequencing data were processed on GeneDenovo's online platform: To ensure analytical reproducibility and high data quality, raw sequencing reads were subjected to rigorous Quality Control (QC) utilizing the fastp software to remove adapter sequences, poly-N runs, and low-quality reads. The high-quality clean reads were subsequently aligned to the reference genome. Differential expression analysis was robustly performed utilizing the DESeq2 software package in R. To strictly control the false positive rate in multiple hypothesis testing, differentially expressed genes (DEGs) were identified using stringent thresholds of False Discovery Rate (FDR) < 0.05 and |log₂FC| ≥ 1. Subsequent functional enrichment analyses, including Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways and Gene Ontology (GO) biological processes, were evaluated with statistical significance strictly defined by a Q-value <0.05.

4.3.4. Western Blot

hMGECs were divided into four groups (Plate-Ctrl, Gel-Ctrl, Plate-Rosi, Gel-Rosi) with 3 biological replicates in each group. Total proteins were extracted using RIPA Lysis Buffer (Beyotime, P0013B) supplemented with 50 × Protease and Phosphatase Inhibitor Cocktail (Beyotime, P1045) on ice. The cell lysates were centrifuged at 12,000 × g for 15 min at 4 °C to collect the supernatant. Protein concentration was measured using the BCA Protein Assay Kit (Beyotime, P0010) according to the manufacturer's instructions. Equal amounts of protein (15 μg per sample) were mixed with 5× SDS-PAGE Loading Buffer (Beyotime, P0286-15 ml) and denatured by boiling at 95 °C for 10 min. Electrophoresis was conducted using FuturePAGE™ 10% 15-well precast gels (ACE Biotechnology, ET15010LGel) with MOPS-SDS Running Buffer (ACE Biotechnology, F00004Gel) prepared following the product manual. The BeyoColor™ Pre-stained Protein Marker (15-120 kDa, Beyotime, P0078) was loaded as the molecular weight standard. Electrophoresis was performed at 160 V (maximum 180 V) until the bromophenol blue indicator migrated to the bottom of the gel. Proteins were then transferred from the gel to polyvinylidene fluoride (PVDF) membranes via a wet transfer system. The membranes were blocked with 5% non-fat milk in Tris-buffered saline with Tween-20 (TBST) for 1 h at room temperature, followed by overnight incubation at 4 °C with primary antibodies: PPARγ (Proteintech, 16643-1-AP), Nox4 (Proteintech, 67681-1-Ig), Krt5 (abcam, ab64081), NF-κB (CST, 8242T), and GAPDH (Proteintech, 60004-1-Ig) as the internal control. After three washes with TBST, the membranes were incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies for 1 h at room temperature: HRP-conjugated Mouse secondary antibody (Earthox, E030110-01) and HRP-conjugated Rabbit secondary antibody (Earthox, E030120-01). Protein bands were visualized using enhanced chemiluminescent (ECL) substrate (Millipore, WBKLS0100) and imaged with a gel imaging system. The relative gray intensity of each protein band was quantified using ImageJ software and normalized to GAPDH expression.

4.3.5. Immunofluorescence staining

To further investigate whether the induction of hMGECs by Rosi is affected by hydrogel scaffolds, cells were divided into two groups: Gel-Ctrl and Gel-Rosi, with 3 biological replicates in each group. After corresponding treatment, cells were fixed with 4% paraformaldehyde at room temperature for 15 min and rinsed with PBS for 3 times. Then, cells were permeabilized with 0.1% Triton X-100 at room temperature for 10 min and rinsed with PBS for 3 times. After blocking with 5% BSA at room temperature for 1 h, primary antibodies were added: PPAR gamma Polyclonal Antibody (Invitrogen, PA5-120040), NF-κB p65 (D14E12) XP Rabbit mAb (CST, 8242T), Anti-Cytokeratin 5 antibody (Abcam, ab64081), and ADRP antibody (Proteintech, 15294-1-AP), followed by incubation at 4 °C overnight. The next day, after rinsing with PBS for 3 times, fluorescent secondary antibodies were added: Goat anti-Rabbit IgG H&L (Alexa Fluor® 488, Abcam, ab150077) and incubated at room temperature in the dark for 1 h. After rinsing with PBS for 3 times, nuclei were stained with DAPI (Beyotime, C1006) for 5 min and rinsed with PBS for 3 times. Z-stack imaging was performed using a confocal microscope, and images were stacked along the Z-axis. For quantitative analysis, due to the uneven three-dimensional distribution of cells, the “single-cell fluorescence intensity calculation method” was adopted: the total fluorescence intensity of the entire field of view was measured and divided by the number of cells in that field to evaluate the protein fluorescence expression level.

4.3.6. Evaluation of adipogenic lipid synthesis

To evaluate the adipogenic lipid synthesis capacity of hMGECs on hydrogel scaffolds, Oil Red O and Nile Red staining were performed. For Oil Red O staining, after the Rosi-induction period, the cell-scaffold constructs were washed with PBS and fixed with 4% paraformaldehyde for 15 min. The samples were then incubated with the freshly prepared Oil Red O working solution (Beyotime, C0158S) for 15-20 min at room temperature. The scaffolds were thoroughly washed with distilled water to remove background dye and observed under a light microscope.

For Nile Red fluorescent staining, the fixed constructs were incubated with the Nile Red staining solution (Beyotime, C2051S) for 10 min at room temperature in the dark, which specifically binds to intracellular neutral lipids and emits intense red fluorescence. After washing with PBS, cell nuclei were counterstained with DAPI for 5 min. Z-stack images were acquired using a confocal microscope, and the relative fluorescence intensity of the lipid droplets was quantitatively analyzed using ImageJ software.

4.4. Animal experiments

4.4.1. Construction of rat eyelid defect in situ model and microtissue implantation

To evaluate the in vivo repair efficacy of the biomimetic tarsal microtissue, a rat in situ repair model of inferior eyelid tarsal defect was established, with the schematic diagram of the model construction shown in Fig. 6A and the actual surgical flowchart presented in Fig. 6B.

SPF-grade SD rats with a body weight of 250g were used in this study. The experiments were divided into five groups: Healthy group (untreated healthy rats), Control group (tarsal defect without repair), Normal Gel group (repair with untrained hydrogel scaffold), Trained Gel group (repair with trained hydrogel scaffold), and Microtissue group (repair with trained hydrogel scaffold loaded with hMGECs). Two sampling time points were set: 1 week and 1 month post-surgery. For each detection item, ≥3 biological replicates were allocated per group to ensure the reliability of the experimental results. All rats were housed in an SPF-grade animal facility in accordance with standard rat feeding specifications. Rats had free access to standard rodent feed and sterile drinking water, and were adaptively fed for 1 week before the experiment.

Rats were anesthetized intraperitoneally with sodium pentobarbital (dose: 50 mg/kg, adjusted appropriately according to body weight). After the rats lost consciousness and their pain reflexes, the hair around the inferior eyelid was cut with ophthalmic scissors, and the skin and conjunctival surface around the eyelid were disinfected with povidone-iodine from the inside out. Given the sensitivity of eyelid tissue, 0.5% lidocaine solution was used for local infiltration anesthesia on the conjunctival surface of the inferior eyelid; this procedure also reduced adhesion between the conjunctiva and the tarsus, facilitating subsequent tissue separation. Under a surgical microscope, a 3 mm-long arc incision was made on the conjunctival surface of the rat's inferior eyelid near the eyelid margin. The conjunctival tissue was bluntly separated to fully expose the tarsus. The full-thickness tarsal tissue in the central area of the inferior eyelid tarsus was precisely excised with microscissors to form a 3mm × 2 mm defect area. The pre-prepared corresponding repair material (Normal Gel, Trained Gel, or Microtissue)—whose macroscopic dimensions and thickness were specifically customized to match the native rat tarsal plate—was implanted into the defect area to ensure a tight fit between the material and the defect edge. Subsequently, the underlying conjunctival tissue was separated and transposed to cover the graft surface. The wound was intermittently sutured with 1-2 stitches using 8-0 absorbable sutures, and the suture knots were fixed on the lateral side of the eyelid, as far away from the eyelid margin as possible to avoid corneal irritation. Immediately after the operation, antibiotic eye ointment was instilled into the rat's conjunctival sac to prevent infection. Within 3 days post-surgery, antibiotic eye drops were instilled 1-2 times daily. During this period, the rat's mental state, eyelid healing, and corneal integrity were closely observed.

Gross appearance photos of the rat's eyelid were taken on postoperative day 1, 1 week, 2 weeks, 3 weeks, and 1 month to record the repair status. At 1 week and 1 month post-surgery, rats were euthanized by an overdose of sodium pentobarbital via intraperitoneal injection. The entire inferior eyelid tissue was excised, quickly rinsed in precooled PBS buffer to remove residual blood, and then processed according to the experimental purpose. All experiments and animal maintenance were performed in strict accordance with the Association for Research in Vision and Ophthalmology (ARVO) statements for the use of animals in ophthalmic and vision research. They were approved by the Ethics Committee of the Second Affiliated Hospital, School of Medicine, Zhejiang University (approval number: 2024–396).

4.4.2. Preparation of tissue sections and multimodal staining analysis

After harvesting the inferior eyelid tissues at 1 week and 1 month post-surgery, the rat tissues were immediately fixed in 4% paraformaldehyde for 24 h, followed by gradient dehydration in ethanol (70%, 80%, 90%, 95%, 100%), transparency in xylene, and paraffin infiltration, after which 5 μm-thick serial paraffin sections were prepared. Subsequently, HE staining, Masson trichrome staining, and immunofluorescence staining were performed sequentially.

For HE staining, the sections were dewaxed and rehydrated, then stained with hematoxylin for 5 min, differentiated with hydrochloric acid-ethanol for several seconds, and stained with eosin for 2 min, followed by dehydration, clearing, and mounting, with histological structures observed under an optical microscope; for Masson staining, the dewaxed and rehydrated sections were stained with hematoxylin for 5 min, acid fuchsin for 5 min, phosphomolybdic acid solution for 5 min, and aniline blue for 5 min, rinsed with acetic acid solution, then dehydrated, cleared, and mounted to observe collagen fiber distribution; for immunofluorescence staining, after dewaxing and rehydration, antigen retrieval was conducted in citrate buffer at 95 °C for 20 min, followed by cooling at room temperature and rinsing with PBS, then blocked with 5% BSA for 1 h at room temperature, incubated with primary antibodies PPARγ Polyclonal Antibody (Invitrogen, PA5-120040) and NF-κB p65 (D14E12) XP Rabbit mAb (CST, 8242T) overnight at 4 °C, rinsed with PBS, incubated with fluorescent secondary antibodies (e.g., Alexa Fluor®-conjugated goat anti-rabbit/mouse IgG) for 1 h at room temperature in the dark, stained with DAPI for 5 min, mounted with fluorescent mounting medium, and images were acquired using a confocal microscope.

4.4.3. In situ mechanical testing

For in situ mechanical testing, the inferior eyelid tissues harvested at 1 month post-surgery were used. The tissues were immediately placed in pre-cooled PBS buffer to keep moist and then tested using the biomechanical tester EFL-MT5600 (Suzhou Yongqinquan Intelligent Equipment Co., Ltd.). Due to the small volume of rat eyelid tissue, which made effective clamping for tensile testing difficult, the in situ compression testing was performed instead. To ensure experimental standardization and reproducibility, the eyelid tissues containing the implanted material were preprocessed and trimmed to a uniform loaded area of 0.5 cm². Specifically, the processed eyelid tissue was placed on the instrument's testing platform, and vertical pressure was applied after setting the instrument parameters, with dynamic changes in pressure and deformation recorded simultaneously to generate stress-strain curves. Meanwhile, the thickness of the eyelid tissue was quantitatively measured using the instrument's supporting function. All tests were completed within 1 h after tissue harvesting to ensure the stability of tissue mechanical properties.

4.4.4. In vivo ocular surface clinical assessments

To dynamically and longitudinally evaluate the functional recovery of the ocular surface following eyelid reconstruction, comprehensive clinical ophthalmic examinations—including the Phenol Red Thread (PRT) test, Tear Film Break-Up Time (TBUT), and corneal fluorescein staining—were systematically conducted at 1, 2, 3, and 4 weeks post-surgery.

Phenol Red Thread (PRT) Test: Tear secretion volume was measured using the PRT test. To accurately reflect physiological basal tear production and strictly eliminate the well-documented suppressive interference of anesthetics on lacrimal gland function, this procedure was performed on unanesthetized rats. The folded end of a standard phenol red cotton thread (Type I, Tianjin Jingming New Technology Development Co., Ltd., China) was gently placed into the lateral third of the lower conjunctival fornix. After exactly 15 s, the thread was removed, and the length of the color-changed wetted portion (turning from yellow to red due to the alkaline nature of tears) was measured in millimeters (mm) to quantify tear volume.

Tear Film Break-Up Time (TBUT): To minimize animal distress and prevent reflex blinking from interfering with stable microscopic observation, subsequent ocular surface evaluations were performed under light inhalation anesthesia using isoflurane. A standardized volume of 2 μL 1% sodium fluorescein solution (Sigma-Aldrich, Cat. No. F6377) was instilled into the conjunctival sac. The eyelids were manually blinked three times to ensure a uniform distribution of the dye across the ocular surface. Evaluated under a slit-lamp biomicroscope equipped with a cobalt blue filter, the TBUT was recorded as the time (in seconds) elapsed between the last complete manual blink and the appearance of the first dark, unstained spot in the fluorescent tear film. For each eye, three consecutive measurements were performed, and the average value was calculated.

Corneal Fluorescein Staining and Standardized Clinical Grading: Immediately following the TBUT measurement, high-resolution digital photographs of the fluorescein-stained corneas were captured. To ensure objective evaluation, the severity of corneal epithelial damage was semi-quantitatively assessed utilizing the standardized National Eye Institute (NEI) grading system. Briefly, the cornea was conceptually divided into five regions: central, superior, inferior, nasal, and temporal. Each region was independently graded on a scale of 0 to 3 based on the density of punctate staining (0 = no staining; 1 = 1–15 punctate spots; 2 = 16–30 punctate spots; 3 = >30 punctate spots or coalescent staining/ulcers). The total corneal staining score was calculated by summing the scores of the five regions, yielding a maximum possible score of 15 per eye. To eliminate subjective bias, all clinical scorings were independently performed by two observers who were blinded to the experimental groupings, and their scores were averaged for statistical analysis.

4.4.5. Western Blot

To analyze the in vivo protein expression of PPARγ and NF-κB, Western Blot (WB) was performed on rat inferior eyelid tissues harvested at 1 week and 1 month post-surgery. The tissues were divided into four groups: Control, Normal Gel, Trained Gel, and Microtissue, with ≥3 biological replicates per group. The detailed procedure was as follows: The harvested intact eyelid tissues were cut into small pieces and placed in pre-cooled RIPA Lysis Buffer (supplemented with 50× Protease and Phosphatase Inhibitor Cocktail, Beyotime, P1045). After sufficient grinding on ice, ultrasonic disruption was performed to achieve complete lysis. The lysates were then centrifuged at 12,000×g for 15 min at 4 °C to collect the supernatant. For subsequent procedures including protein quantification, denaturation, electrophoresis, membrane transfer, blocking, antibody incubation, and development quantification, refer to Section 2.4.

4.4.6. Double-blinded region-specific immunofluorescence quantification

To objectively resolve spatial expression dynamics and explicitly eliminate subjective bias, the quantitative evaluation of in vivo immunofluorescence was conducted using a stringent double-blinded, region-specific protocol. Digital images of tissue sections were randomized and encoded. Two independent, blinded observers were tasked with defining and quantifying Regions of Interest (ROIs) in two distinct spatial zones per section: (1) the Peri-implant zone (tissue directly adjacent to the microtissue construct), and (2) the Distant zone (native tissue spatially separated from the implant interface). Using ImageJ software, the positive fluorescence area fractions within these specific spatial ROIs were independently quantified, enabling the precise differentiation between localized contact-mediated responses and broader tissue-wide paracrine effects.

4.4.7. Single-cell sequencing

To explore the early repair mechanism of microtissue, single-cell sequencing was performed to analyze the cell composition and functional characteristics of rat eyelid tissues at 1 week post-surgery. Two groups were established: the Healthy group (untreated healthy rat eyelids) and the Microtissue group (rat eyelids implanted with microtissue). Due to the small volume of a single rat eyelid tissue, the number of cells obtained from dissociating one eyelid was insufficient to meet the sample size requirement for single-cell sequencing. Therefore, 4 inferior eyelid tissues from 4 rats were mixed in each group to ensure sufficient cell quantity. The detailed procedure was as follows: The eyelid tissues harvested at 1 week post-surgery were quickly rinsed with pre-cooled PBS buffer to remove residual blood and impurities. After rinsing, the tissues were immediately placed in tissue preservation solution provided by Gene Denovo Biotechnology Co., Ltd. (to maintain cell viability before transportation and subsequent processing) and then sent to the company.

Gene Denovo Biotechnology Co., Ltd. performed the single-cell suspension preparation and quality control: gentle digestion using a tissue dissociation kit (containing a composite enzyme system such as collagenase and hyaluronidase) combined with mechanical pipetting to prepare single-cell suspensions, followed by filtration through a 40 μm cell strainer to remove tissue debris, and trypan blue staining to detect cell viability (ensuring a viability rate of ≥90%). After passing quality control, the company continued to perform the core single-cell sequencing procedures, including suspension concentration calibration, single-cell transcriptome library construction based on the 10x Genomics Chromium platform, and high-throughput sequencing after library quality inspection.

After retrieving the raw sequencing data, our team performed the subsequent bioinformatics analysis using the R language (version 4.0) and relevant professional bioinformatics packages. First, to guarantee computational reproducibility, raw scRNA-seq data were processed utilizing the standard Seurat package in the R programming environment. Stringent Quality Control (QC) metrics were implemented at the single-cell level: cells expressing fewer than 200 or greater than 5000 detected genes (to meticulously exclude empty droplets and potential cell doublets, respectively), as well as cells with a mitochondrial gene expression ratio exceeding 10% (to exclude apoptotic or highly stressed cells), were systematically filtered out. Following QC, the data were subjected to log-normalization, scaling, and batch-effect correction. Then, principal component analysis (PCA) was used for dimensionality reduction, followed by clustering to obtain cell clusters. Cell types were annotated using known cell marker genes [43], covering 11 cell types, including meibomian gland-related cells (Orifice cells, Ductal cells, Acinar cells, Meibocytes), as well as immune cells and stromal cells. Subsequently, the distribution of cells per cluster in the two groups was counted and compared for differences. For key functional cell populations of the tarsus (Orifice cells, Meibocytes, Ductal cells, Acinar cells), differentially expressed genes were extracted for Gene Ontology (GO) functional enrichment analysis to explore changes in cell functional characteristics. Finally, a cell communication analysis tool was used to quantify the intensity of intercellular signaling between the two groups and to analyze differences in the activation of the local cell interaction network after microtissue implantation.

4.4.8. Independent validation via flow cytometry and immunofluorescence

To independently validate the descriptive scRNA-seq findings, flow cytometry and immunofluorescence analyses were performed on unpooled, biologically independent rat eyelid samples (n = 3 individual eyes per group).

Flow Cytometry Analysis: Freshly harvested eyelid tissues were finely minced and enzymatically dissociated in DMEM/F12 supplemented with 0.4 mg/mL Collagenase IV at 37 °C for 90 min. Following filtration through a 70 μm cell strainer, the single-cell suspensions were incubated with the LIVE/DEAD™ Fixable Blue Dead Cell Stain Kit (Invitrogen, L34961) in protein-free PBS for 30 min in the dark to explicitly exclude dead cells from downstream quantitative analysis. Subsequently, cells were fixed with 4% paraformaldehyde, permeabilized with 0.1% Triton X-100 containing 1% BSA, and intracellularly stained with primary antibodies against Krt10 (Proteintech, 18343-1-AP) and ADRP (Proteintech, 80362-2-RR). After incubation with corresponding fluorochrome-conjugated secondary antibodies, data were acquired via flow cytometry to quantify the viable positive cell populations.

Quantitative Immunofluorescence: Tissue sections from the unpooled independent biological replicates (n = 3) were stained for Krt10 (Proteintech, 18343-1-AP) and ADRP (Proteintech, 15294-1-AP) to spatially validate the cellular phenotypes. Digital images were acquired, and the positive fluorescence area fraction (%) was quantitatively analyzed using ImageJ software. To ensure representative sampling and quantitative robustness across the tissues, multiple measurement regions of interest (ROIs) were selected, yielding a total of 6 evaluated data points (n = 6 ROIs derived from 3 biologically independent eyelids) per group for statistical comparison.

4.5. Statistical analysis

All quantitative data are expressed as the mean ± standard deviation (SD). To ensure maximum data transparency, individual biological replicates are displayed as distinct scatter points superimposed on the bar charts. Statistical analyses and graphing were performed using GraphPad Prism 10 software (GraphPad Software, Boston, MA, USA). The exact sample size (n) for each experiment is explicitly stated in the respective figure legends. Prior to parametric testing, the normality of data distribution was assessed using the Shapiro-Wilk test, and the homogeneity of variance was verified utilizing the Brown-Forsythe test. For comparisons between two independent groups, a two-tailed unpaired Student's t-test was employed. For analyses involving three or more groups, a one-way analysis of variance (ANOVA) was conducted, followed strictly by Tukey's post-hoc test to correct for multiple comparisons. Statistical significance was predefined at a threshold of P < 0.05. The significance levels in the figures are denoted as follows: ns indicates no significant difference (P ≥ 0.05), ∗ for P < 0.05, ∗∗ for P < 0.01, ∗∗∗ for P < 0.001, and ∗∗∗∗ for P < 0.0001.

Data availability statement

The raw bulk RNA sequencing and single-cell RNA sequencing datasets generated and analyzed during the current study have been deposited in the NCBI Sequence Read Archive (SRA) repository. The data will be made publicly available under the BioProject accession number PRJNA1437397 upon publication of this article.

Ethics approval and consent to participate

All experiments and animal maintenance were performed in strict accordance with the Association for Research in Vision and Ophthalmology (ARVO) statements for the use of animals in ophthalmic and vision research. They were approved by the Ethics Committee of the Second Affiliated Hospital, School of Medicine, Zhejiang University (approval number: 2024–396).

CRediT authorship contribution statement

Xindi Liu: Conceptualization, Formal analysis, Investigation, Methodology, Validation, Writing – original draft. Ke Yao: Conceptualization, Formal analysis, Investigation, Methodology, Validation. Meizhu Wang: Formal analysis, Investigation, Methodology, Validation. Xuri Chen: Conceptualization, Formal analysis, Investigation, Methodology, Writing – review & editing. Zhichu Chen: Formal analysis, Methodology. Zheren Sun: Formal analysis, Methodology. Xuanhe Lin: Formal analysis, Methodology. Nianjia Wang: Investigation, Validation. Chunlei Yao: Methodology, Validation. Shenyu Huang: Methodology. Yijie Wang: Methodology. Qi Gao: Conceptualization, Methodology, Writing – review & editing. Yong He: Conceptualization, Methodology, Writing – review & editing. Juan Ye: Conceptualization, Methodology, Writing – review & editing.

Declaration of competing interest

The authors declare no conflict of interests.

Acknowledgments

This work was financially supported by Key Program of the National Natural Science Foundation of China (82330032), Key Research and Development Program of Zhejiang Province (2024C03204), National Natural Science Foundation Regional Innovation and Development Joint Fund (U20A20386), Key Program of the Natural Science Foundation of Zhejiang Province (LKLY25H180001).

Footnotes

Peer review under the responsibility of editorial board of Bioactive Materials.

^{Appendix A}

Supplementary data to this article can be found online at https://doi.org/10.1016/j.bioactmat.2026.05.052.

Contributor Information

Qi Gao, Email: gaoqi8977@zju.edu.cn.

Yong He, Email: yongqin@zju.edu.cn.

Juan Ye, Email: yejuan@zju.edu.cn.

Appendix A. Supplementary data

The following is the Supplementary data to this article:

Multimedia component 1

mmc1.docx^{(28.5MB, docx)}

References

1.Coban I., Sirinturk S., Unat F., Pinar Y., Govsa F. Anatomical description of the upper tarsal plate for reconstruction. Surg. Radiol. Anat. 2018;40(10):1105–1110. doi: 10.1007/s00276-018-2064-7. [DOI] [PubMed] [Google Scholar]
2.Gao Q., Xu P., Hu S., Ye J. The micro-structure and biomechanics of eyelid tarsus. J. Biomech. 2022;133 doi: 10.1016/j.jbiomech.2021.110911. [DOI] [PubMed] [Google Scholar]
3.Yan Y., Ji Q., Yang J., Yin X., Liu S., Karalkin P.A., et al. Bioengineering autologous cartilage grafts for functional posterior lamellar eyelid reconstruction: a preliminary study in rabbits. Acta Biomater. 2024;179:106–120. doi: 10.1016/j.actbio.2024.03.025. [DOI] [PubMed] [Google Scholar]
4.Kim Y.S., Hwang K. Shape and height of tarsal plates. J. Craniofac. Surg. 2016;27(2):496–497. doi: 10.1097/SCS.0000000000002369. [DOI] [PubMed] [Google Scholar]
5.Gao S., Lu B., Zhou R., Gao W. Anatomical and histological study of the upper tarsus in Asian. J. Craniofac. Surg. 2023;34(7):2195–2198. doi: 10.1097/SCS.0000000000009397. [DOI] [PubMed] [Google Scholar]
6.Lv Z., Li S., Zeng G., Yao K., Han H. Recent progress of nanomedicine in managing dry eye disease. Adv. Ophthalmol. Pract. Res. 2024;4(1):23–31. doi: 10.1016/j.aopr.2024.01.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Jennings E., Krakauer M., Nunery W.R., Aakalu V.K. Advancements in the repair of large upper eyelid defects: a 10-year review. Orbit. 2021;40(6):470–480. doi: 10.1080/01676830.2020.1820045. [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Liu J., Zhang M., Zhou M., Wang Q., Jiang X., Huang Q. Exploring biomaterial scaffolds for eyelid reconstruction: a synthesis of experimental findings. Tissue Eng. Part B Rev. 2025 doi: 10.1089/ten.teb.2024.0364. [DOI] [PubMed] [Google Scholar]
9.Fin A., De Biasio F., Lanzetta P., Mura S., Tarantini A., Parodi P.C. Posterior lamellar reconstruction: a comprehensive review of the literature. Orbit. 2019;38(1):51–66. doi: 10.1080/01676830.2018.1474236. [DOI] [PubMed] [Google Scholar]
10.Bertsch C., Marechal H., Gribova V., Levy B., Debry C., Lavalle P., et al. Biomimetic bilayered scaffolds for tissue engineering: from current design strategies to medical applications. Adv. Healthcare Mater. 2023;12(17) doi: 10.1002/adhm.202203115. [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Chisci E., Chisci D., Chisci E., Chisci V., Chisci G. Biomaterials for posterior eyelid reconstruction: our experience and suggestions. Br. J. Oral Maxillofac. Surg. 2023;61(8):577–578. doi: 10.1016/j.bjoms.2023.08.212. [DOI] [PubMed] [Google Scholar]
12.Mandal S.K., Fleming J.C., Reddy S.G., Fowler B.T. Total upper eyelid reconstruction with modified cutler-beard procedure using autogenous auricular cartilage. J. Clin. Diagn. Res. 2016;10(8):NC01–NC04. doi: 10.7860/JCDR/2016/20303.8239. [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Eser C., Kesiktas E., Gencel E., Tabakan I., Yavuz M. Total or near-total lower eyelid defect reconstruction using malar myocutaneous bridge and nasojugal flaps and septal chondromucosal graft. Ophthalmic Plast. Reconstr. Surg. 2016;32(3):225–229. doi: 10.1097/IOP.0000000000000481. [DOI] [PubMed] [Google Scholar]
14.Shi Y., Zhou X., Yu J., Liu H. Reconstruction of full-thickness eyelid defects following malignant tumor excision: the retroauricular flap and palatal mucosal graft. J. Craniofac. Surg. 2016;27(3):612–614. doi: 10.1097/SCS.0000000000002543. [DOI] [PubMed] [Google Scholar]
15.Jin M.J., Gao Y. Using buccal mucosa and auricular cartilage with a local flap for full-thickness defect of lower eyelid. J. Craniofac. Surg. 2021;32(7):e660–e661. doi: 10.1097/SCS.0000000000007770. [DOI] [PubMed] [Google Scholar]
16.Chen L., Yan D., Wu N., Zhang W., Yan C., Yao Q., et al. 3D-Printed poly-caprolactone scaffolds modified with biomimetic extracellular matrices for tarsal plate tissue engineering. Front. Bioeng. Biotechnol. 2020;8:219. doi: 10.3389/fbioe.2020.00219. [DOI] [PMC free article] [PubMed] [Google Scholar]
17.Gao Q., Hu B., Ning Q., Ye C., Xie J., Ye J., et al. A primary study of poly(propylene fumarate)-2-hydroxyethyl methacrylate copolymer scaffolds for tarsal plate repair and reconstruction in rabbit eyelids. J. Mater. Chem. B. 2015;3(19):4052–4062. doi: 10.1039/c5tb00285k. [DOI] [PubMed] [Google Scholar]
18.Wu K.Y., Fujioka J.K., Goodyear E., Tran S.D. Polymers and biomaterials for posterior lamella of the eyelid and the lacrimal system. Polymers (Basel) 2024;16(3) doi: 10.3390/polym16030352. [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Wei F.L., Zhai Y., Wang T.F., Zhao J.W., Wang C.L., Tang Z., et al. Stem cell-homing biomimetic hydrogel promotes the repair of osteoporotic bone defects through osteogenic and angiogenic coupling. Sci. Adv. 2024;10(44):eadq6700. doi: 10.1126/sciadv.adq6700. [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Nie K., Fan Z., Sun W., Wu B., Zhou S., Zhao H., et al. Ultrafast crosslinking, strongly adhesive de novo protein hydrogels promote cartilage regeneration. Bioact. Mater. 2026;56:368–385. doi: 10.1016/j.bioactmat.2025.10.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
21.He S., Liang W., Tang Y., Zhang J., Wang R., Quan L., et al. Robust super-structured porous hydrogel enables bioadaptive repair of dynamic soft tissue. Nat. Commun. 2025;16(1):3198. doi: 10.1038/s41467-025-58062-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
22.He J., Sun Y., Gao Q., He C., Yao K., Wang T., et al. Gelatin methacryloyl hydrogel, from standardization, performance, to biomedical application. Adv. Healthcare Mater. 2023;12(23) doi: 10.1002/adhm.202300395. [DOI] [PubMed] [Google Scholar]
23.Yao K., Xia P., Kong W., Liu N., Lv S., Zhang Y., et al. 3D printing of gradient biomimetic scaffold via electrochemical molecular lock for tissue regeneration. Adv. Mater. 2025 doi: 10.1002/adma.202513484. [DOI] [PubMed] [Google Scholar]
24.Yao K., Hong G., Yuan X., Kong W., Xia P., Li Y., et al. 3D printing of tough hydrogel scaffolds with functional surface structures for tissue regeneration. Nano-Micro Lett. 2024;17(1):27. doi: 10.1007/s40820-024-01524-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
25.Yao K., Lv S., Zhang X., Shen K., Chen Y., Ma Z., et al. 3D printing of multiscale biomimetic scaffold for tendon regeneration. Adv. Funct. Mater. 2025;35(4) [Google Scholar]
26.Sun X., Mao Y., Yu Z., Yang P., Jiang F. A biomimetic "Salting Out-Alignment-Locking" tactic to design strong and tough hydrogel. Adv. Mater. 2024;36(25) doi: 10.1002/adma.202400084. [DOI] [PubMed] [Google Scholar]
27.Huo H., Shen J., Wan J., Shi H., Yang H., Duan X., et al. A tough and robust hydrogel constructed through carbon dots induced crystallization domains integrated orientation regulation. Nat. Commun. 2025;16(1):6221. doi: 10.1038/s41467-025-61535-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Zheng L., Wang L., Zhang J., Wang Y., Ma X., Chen Z., et al. Rapid high-resolution visible-light 3D printing of hydrogels via nanocluster-triggered dual-pathway photoinitiation. Adv. Mater. 2025 doi: 10.1002/adma.202517325. [DOI] [PubMed] [Google Scholar]
29.Phan M.A.T., Madigan M.C., Stapleton F., Willcox M., Golebiowski B. Human meibomian gland epithelial cell culture models: current progress, challenges, and future directions. Ocul. Surf. 2022;23:96–113. doi: 10.1016/j.jtos.2021.11.012. [DOI] [PubMed] [Google Scholar]
30.Duong H.T., Phan M.A.T., Madigan M.C., Stapleton F., Wilcsek G., Willcox M., et al. Culture of primary human meibomian gland cells from surgically excised eyelid tissue. Exp. Eye Res. 2023;235 doi: 10.1016/j.exer.2023.109636. [DOI] [PubMed] [Google Scholar]
31.Kim S.W., Xie Y., Nguyen P.Q., Bui V.T., Huynh K., Kang J.S., et al. PPARgamma regulates meibocyte differentiation and lipid synthesis of cultured human meibomian gland epithelial cells (hMGEC) Ocul. Surf. 2018;16(4):463–469. doi: 10.1016/j.jtos.2018.07.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Chen L., Yan D., Wu N., Yao Q., Sun H., Pang Y., et al. Injectable bio-responsive hydrogel for therapy of inflammation related eyelid diseases. Bioact. Mater. 2021;6(10):3062–3073. doi: 10.1016/j.bioactmat.2021.02.040. [DOI] [PMC free article] [PubMed] [Google Scholar]
33.Straus D.S., Glass C.K. Anti-inflammatory actions of PPAR ligands: new insights on cellular and molecular mechanisms. Trends Immunol. 2007;28(12):551–558. doi: 10.1016/j.it.2007.09.003. [DOI] [PubMed] [Google Scholar]
34.He Q., Huang Y., Wang S. Hofmeister effect-assisted one step fabrication of ductile and strong gelatin hydrogels. Adv. Funct. Mater. 2018;28(5) [Google Scholar]
35.Yuan X., Zhu Z., Xia P., Wang Z., Zhao X., Jiang X., et al. Tough gelatin hydrogel for tissue engineering. Adv. Sci. (Weinh.) 2023;10(24) doi: 10.1002/advs.202301665. [DOI] [PMC free article] [PubMed] [Google Scholar]
36.Kong W., Lu Y., Yuan X., He Y. Restructuring of hydrogel polymer networks via ion storming. Small. 2025;21(24) doi: 10.1002/smll.202502436. [DOI] [PubMed] [Google Scholar]
37.Glahn J.Z., Huelsboemer L., Pomahac B. Face transplantation, social death , and bias in health care resource allocation. Ann. Surg. 2024;279(6):920–922. doi: 10.1097/SLA.0000000000006195. [DOI] [PubMed] [Google Scholar]
38.Zhao Y., Zan T., Li Q. Donor-driven advancements in reconstructive surgery: 20 years of facial transplantation reconstruction. Chin. J. Plastic Surg. 2025;41(5):441–446. [Google Scholar]
39.Ozdemir T., Fowler E.W., Liu S., Harrington D.A., Witt R.L., Farach-Carson M.C., et al. Tuning hydrogel properties to promote the assembly of salivary gland spheroids in 3D. ACS Biomater. Sci. Eng. 2016;2(12):2217–2230. doi: 10.1021/acsbiomaterials.6b00419. [DOI] [PMC free article] [PubMed] [Google Scholar]
40.Bao M., Xie J., Piruska A., Huck W.T.S. 3D microniches reveal the importance of cell size and shape. Nat. Commun. 2017;8(1):1962. doi: 10.1038/s41467-017-02163-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
41.Yu C., Wan X., Wei J., Xu Z., Wu X., Wang X., et al. Human Meibomian gland organoids to study epithelial homeostasis and dysfunction. Protein Cell. 2025 doi: 10.1093/procel/pwaf095. [DOI] [PMC free article] [PubMed] [Google Scholar]
42.British Standards I . BSI; London: 2024. BS EN ISO 10993-12:2021/Amd 1 Biological Evaluation of Medical Devices. Part 12: Sample Preparation and Reference Materials. [Google Scholar]
43.Zhu X., Xu M., Portal C., Lin Y., Ferdinand A., Peng T., et al. Identification of Meibomian gland stem cell populations and mechanisms of aging. Nat. Commun. 2025;16(1):1663. doi: 10.1038/s41467-025-56907-6. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Multimedia component 1

mmc1.docx^{(28.5MB, docx)}

Data Availability Statement

[bib1] 1.Coban I., Sirinturk S., Unat F., Pinar Y., Govsa F. Anatomical description of the upper tarsal plate for reconstruction. Surg. Radiol. Anat. 2018;40(10):1105–1110. doi: 10.1007/s00276-018-2064-7. [DOI] [PubMed] [Google Scholar]

[bib2] 2.Gao Q., Xu P., Hu S., Ye J. The micro-structure and biomechanics of eyelid tarsus. J. Biomech. 2022;133 doi: 10.1016/j.jbiomech.2021.110911. [DOI] [PubMed] [Google Scholar]

[bib3] 3.Yan Y., Ji Q., Yang J., Yin X., Liu S., Karalkin P.A., et al. Bioengineering autologous cartilage grafts for functional posterior lamellar eyelid reconstruction: a preliminary study in rabbits. Acta Biomater. 2024;179:106–120. doi: 10.1016/j.actbio.2024.03.025. [DOI] [PubMed] [Google Scholar]

[bib4] 4.Kim Y.S., Hwang K. Shape and height of tarsal plates. J. Craniofac. Surg. 2016;27(2):496–497. doi: 10.1097/SCS.0000000000002369. [DOI] [PubMed] [Google Scholar]

[bib5] 5.Gao S., Lu B., Zhou R., Gao W. Anatomical and histological study of the upper tarsus in Asian. J. Craniofac. Surg. 2023;34(7):2195–2198. doi: 10.1097/SCS.0000000000009397. [DOI] [PubMed] [Google Scholar]

[bib6] 6.Lv Z., Li S., Zeng G., Yao K., Han H. Recent progress of nanomedicine in managing dry eye disease. Adv. Ophthalmol. Pract. Res. 2024;4(1):23–31. doi: 10.1016/j.aopr.2024.01.008. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib7] 7.Jennings E., Krakauer M., Nunery W.R., Aakalu V.K. Advancements in the repair of large upper eyelid defects: a 10-year review. Orbit. 2021;40(6):470–480. doi: 10.1080/01676830.2020.1820045. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib8] 8.Liu J., Zhang M., Zhou M., Wang Q., Jiang X., Huang Q. Exploring biomaterial scaffolds for eyelid reconstruction: a synthesis of experimental findings. Tissue Eng. Part B Rev. 2025 doi: 10.1089/ten.teb.2024.0364. [DOI] [PubMed] [Google Scholar]

[bib9] 9.Fin A., De Biasio F., Lanzetta P., Mura S., Tarantini A., Parodi P.C. Posterior lamellar reconstruction: a comprehensive review of the literature. Orbit. 2019;38(1):51–66. doi: 10.1080/01676830.2018.1474236. [DOI] [PubMed] [Google Scholar]

[bib10] 10.Bertsch C., Marechal H., Gribova V., Levy B., Debry C., Lavalle P., et al. Biomimetic bilayered scaffolds for tissue engineering: from current design strategies to medical applications. Adv. Healthcare Mater. 2023;12(17) doi: 10.1002/adhm.202203115. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib11] 11.Chisci E., Chisci D., Chisci E., Chisci V., Chisci G. Biomaterials for posterior eyelid reconstruction: our experience and suggestions. Br. J. Oral Maxillofac. Surg. 2023;61(8):577–578. doi: 10.1016/j.bjoms.2023.08.212. [DOI] [PubMed] [Google Scholar]

[bib12] 12.Mandal S.K., Fleming J.C., Reddy S.G., Fowler B.T. Total upper eyelid reconstruction with modified cutler-beard procedure using autogenous auricular cartilage. J. Clin. Diagn. Res. 2016;10(8):NC01–NC04. doi: 10.7860/JCDR/2016/20303.8239. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib13] 13.Eser C., Kesiktas E., Gencel E., Tabakan I., Yavuz M. Total or near-total lower eyelid defect reconstruction using malar myocutaneous bridge and nasojugal flaps and septal chondromucosal graft. Ophthalmic Plast. Reconstr. Surg. 2016;32(3):225–229. doi: 10.1097/IOP.0000000000000481. [DOI] [PubMed] [Google Scholar]

[bib14] 14.Shi Y., Zhou X., Yu J., Liu H. Reconstruction of full-thickness eyelid defects following malignant tumor excision: the retroauricular flap and palatal mucosal graft. J. Craniofac. Surg. 2016;27(3):612–614. doi: 10.1097/SCS.0000000000002543. [DOI] [PubMed] [Google Scholar]

[bib15] 15.Jin M.J., Gao Y. Using buccal mucosa and auricular cartilage with a local flap for full-thickness defect of lower eyelid. J. Craniofac. Surg. 2021;32(7):e660–e661. doi: 10.1097/SCS.0000000000007770. [DOI] [PubMed] [Google Scholar]

[bib16] 16.Chen L., Yan D., Wu N., Zhang W., Yan C., Yao Q., et al. 3D-Printed poly-caprolactone scaffolds modified with biomimetic extracellular matrices for tarsal plate tissue engineering. Front. Bioeng. Biotechnol. 2020;8:219. doi: 10.3389/fbioe.2020.00219. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib17] 17.Gao Q., Hu B., Ning Q., Ye C., Xie J., Ye J., et al. A primary study of poly(propylene fumarate)-2-hydroxyethyl methacrylate copolymer scaffolds for tarsal plate repair and reconstruction in rabbit eyelids. J. Mater. Chem. B. 2015;3(19):4052–4062. doi: 10.1039/c5tb00285k. [DOI] [PubMed] [Google Scholar]

[bib18] 18.Wu K.Y., Fujioka J.K., Goodyear E., Tran S.D. Polymers and biomaterials for posterior lamella of the eyelid and the lacrimal system. Polymers (Basel) 2024;16(3) doi: 10.3390/polym16030352. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib19] 19.Wei F.L., Zhai Y., Wang T.F., Zhao J.W., Wang C.L., Tang Z., et al. Stem cell-homing biomimetic hydrogel promotes the repair of osteoporotic bone defects through osteogenic and angiogenic coupling. Sci. Adv. 2024;10(44):eadq6700. doi: 10.1126/sciadv.adq6700. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib20] 20.Nie K., Fan Z., Sun W., Wu B., Zhou S., Zhao H., et al. Ultrafast crosslinking, strongly adhesive de novo protein hydrogels promote cartilage regeneration. Bioact. Mater. 2026;56:368–385. doi: 10.1016/j.bioactmat.2025.10.009. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib21] 21.He S., Liang W., Tang Y., Zhang J., Wang R., Quan L., et al. Robust super-structured porous hydrogel enables bioadaptive repair of dynamic soft tissue. Nat. Commun. 2025;16(1):3198. doi: 10.1038/s41467-025-58062-4. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib22] 22.He J., Sun Y., Gao Q., He C., Yao K., Wang T., et al. Gelatin methacryloyl hydrogel, from standardization, performance, to biomedical application. Adv. Healthcare Mater. 2023;12(23) doi: 10.1002/adhm.202300395. [DOI] [PubMed] [Google Scholar]

[bib23] 23.Yao K., Xia P., Kong W., Liu N., Lv S., Zhang Y., et al. 3D printing of gradient biomimetic scaffold via electrochemical molecular lock for tissue regeneration. Adv. Mater. 2025 doi: 10.1002/adma.202513484. [DOI] [PubMed] [Google Scholar]

[bib24] 24.Yao K., Hong G., Yuan X., Kong W., Xia P., Li Y., et al. 3D printing of tough hydrogel scaffolds with functional surface structures for tissue regeneration. Nano-Micro Lett. 2024;17(1):27. doi: 10.1007/s40820-024-01524-z. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib25] 25.Yao K., Lv S., Zhang X., Shen K., Chen Y., Ma Z., et al. 3D printing of multiscale biomimetic scaffold for tendon regeneration. Adv. Funct. Mater. 2025;35(4) [Google Scholar]

[bib26] 26.Sun X., Mao Y., Yu Z., Yang P., Jiang F. A biomimetic "Salting Out-Alignment-Locking" tactic to design strong and tough hydrogel. Adv. Mater. 2024;36(25) doi: 10.1002/adma.202400084. [DOI] [PubMed] [Google Scholar]

[bib27] 27.Huo H., Shen J., Wan J., Shi H., Yang H., Duan X., et al. A tough and robust hydrogel constructed through carbon dots induced crystallization domains integrated orientation regulation. Nat. Commun. 2025;16(1):6221. doi: 10.1038/s41467-025-61535-1. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib28] 28.Zheng L., Wang L., Zhang J., Wang Y., Ma X., Chen Z., et al. Rapid high-resolution visible-light 3D printing of hydrogels via nanocluster-triggered dual-pathway photoinitiation. Adv. Mater. 2025 doi: 10.1002/adma.202517325. [DOI] [PubMed] [Google Scholar]

[bib29] 29.Phan M.A.T., Madigan M.C., Stapleton F., Willcox M., Golebiowski B. Human meibomian gland epithelial cell culture models: current progress, challenges, and future directions. Ocul. Surf. 2022;23:96–113. doi: 10.1016/j.jtos.2021.11.012. [DOI] [PubMed] [Google Scholar]

[bib30] 30.Duong H.T., Phan M.A.T., Madigan M.C., Stapleton F., Wilcsek G., Willcox M., et al. Culture of primary human meibomian gland cells from surgically excised eyelid tissue. Exp. Eye Res. 2023;235 doi: 10.1016/j.exer.2023.109636. [DOI] [PubMed] [Google Scholar]

[bib31] 31.Kim S.W., Xie Y., Nguyen P.Q., Bui V.T., Huynh K., Kang J.S., et al. PPARgamma regulates meibocyte differentiation and lipid synthesis of cultured human meibomian gland epithelial cells (hMGEC) Ocul. Surf. 2018;16(4):463–469. doi: 10.1016/j.jtos.2018.07.004. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib32] 32.Chen L., Yan D., Wu N., Yao Q., Sun H., Pang Y., et al. Injectable bio-responsive hydrogel for therapy of inflammation related eyelid diseases. Bioact. Mater. 2021;6(10):3062–3073. doi: 10.1016/j.bioactmat.2021.02.040. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib33] 33.Straus D.S., Glass C.K. Anti-inflammatory actions of PPAR ligands: new insights on cellular and molecular mechanisms. Trends Immunol. 2007;28(12):551–558. doi: 10.1016/j.it.2007.09.003. [DOI] [PubMed] [Google Scholar]

[bib34] 34.He Q., Huang Y., Wang S. Hofmeister effect-assisted one step fabrication of ductile and strong gelatin hydrogels. Adv. Funct. Mater. 2018;28(5) [Google Scholar]

[bib35] 35.Yuan X., Zhu Z., Xia P., Wang Z., Zhao X., Jiang X., et al. Tough gelatin hydrogel for tissue engineering. Adv. Sci. (Weinh.) 2023;10(24) doi: 10.1002/advs.202301665. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib36] 36.Kong W., Lu Y., Yuan X., He Y. Restructuring of hydrogel polymer networks via ion storming. Small. 2025;21(24) doi: 10.1002/smll.202502436. [DOI] [PubMed] [Google Scholar]

[bib37] 37.Glahn J.Z., Huelsboemer L., Pomahac B. Face transplantation, social death , and bias in health care resource allocation. Ann. Surg. 2024;279(6):920–922. doi: 10.1097/SLA.0000000000006195. [DOI] [PubMed] [Google Scholar]

[bib38] 38.Zhao Y., Zan T., Li Q. Donor-driven advancements in reconstructive surgery: 20 years of facial transplantation reconstruction. Chin. J. Plastic Surg. 2025;41(5):441–446. [Google Scholar]

[bib39] 39.Ozdemir T., Fowler E.W., Liu S., Harrington D.A., Witt R.L., Farach-Carson M.C., et al. Tuning hydrogel properties to promote the assembly of salivary gland spheroids in 3D. ACS Biomater. Sci. Eng. 2016;2(12):2217–2230. doi: 10.1021/acsbiomaterials.6b00419. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib40] 40.Bao M., Xie J., Piruska A., Huck W.T.S. 3D microniches reveal the importance of cell size and shape. Nat. Commun. 2017;8(1):1962. doi: 10.1038/s41467-017-02163-2. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib41] 41.Yu C., Wan X., Wei J., Xu Z., Wu X., Wang X., et al. Human Meibomian gland organoids to study epithelial homeostasis and dysfunction. Protein Cell. 2025 doi: 10.1093/procel/pwaf095. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib42] 42.British Standards I . BSI; London: 2024. BS EN ISO 10993-12:2021/Amd 1 Biological Evaluation of Medical Devices. Part 12: Sample Preparation and Reference Materials. [Google Scholar]

[bib43] 43.Zhu X., Xu M., Portal C., Lin Y., Ferdinand A., Peng T., et al. Identification of Meibomian gland stem cell populations and mechanisms of aging. Nat. Commun. 2025;16(1):1663. doi: 10.1038/s41467-025-56907-6. [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Engineering biomimetic tarsal microtissue via structurally zoned hydrogel scaffold for integrated eyelid reconstruction

Xindi Liu

Ke Yao

Meizhu Wang

Xuri Chen

Zhichu Chen

Zheren Sun

Xuanhe Lin

Nianjia Wang

Chunlei Yao

Shenyu Huang

Yijie Wang

Qi Gao

Yong He

Juan Ye

Abstract

Graphical abstract

Highlights

1. Introduction

Fig. 1.

2. Results

2.1. Fabrication and characterization of the hydrogel scaffold

Fig. 2.

2.2. Cellular Compatibility Evaluation of the hydrogel scaffold

Fig. 3.

2.3. Screening of pore diameters for cell growth

2.4. Transcriptome Sequencing Analysis of the Microtissue In Vitro

Fig. 4.

2.5. In Vitro Protein Expression validation of the microtissue

Fig. 5.

2.6. Adipogenic lipid synthesis on the microtissue

2.7. In Vivo Repair Efficacy of the microtissue

Fig. 6.

2.8. In Vivo Ocular Surface Functional Recovery

Fig. 7.

2.9. In vivo protein expression of PPARγ and NF-κB

Fig. 8.

2.10. In Vivo Single-Cell Sequencing Analysis at 1 Week post-implantation

Fig. 9.

3. Discussion

4. Materials and methods

4.1. Materials and reagents

4.2. Fabrication and characterization of high-strength hydrogel scaffolds

4.2.1. Scaffold fabrication

4.2.2. Characterization of the scaffold morphology

4.2.3. Mechanical test

4.3. Cell experiments

4.3.1. Culture, seeding, and differentiation induction of hMGECs

4.3.1.1. hMGECs culture

4.3.1.2. hMGECs seeding on high-strength hydrogel scaffolds

4.3.1.3. In vitro validation of rosi-induced adipogenic differentiation and anti-inflammatory response

4.3.1.4. Induction of hMGECs differentiation on hydrogel scaffolds

4.3.1.5. In vitro sequential Co-culture on topographically zoned scaffolds

4.3.2. Biocompatibility verification of high-strength hydrogel scaffolds

4.3.2.1. Preparation of material extract

4.3.2.2. CCK-8 cell viability assay

4.3.2.3. Live-dead cell staining

4.3.2.4. Cytoskeleton and nuclear staining

4.3.3. Transcriptome analysis

4.3.4. Western Blot

4.3.5. Immunofluorescence staining

4.3.6. Evaluation of adipogenic lipid synthesis

4.4. Animal experiments

4.4.1. Construction of rat eyelid defect in situ model and microtissue implantation

4.4.2. Preparation of tissue sections and multimodal staining analysis

4.4.3. In situ mechanical testing

4.4.4. In vivo ocular surface clinical assessments

4.4.5. Western Blot

4.4.6. Double-blinded region-specific immunofluorescence quantification

4.4.7. Single-cell sequencing

4.4.8. Independent validation via flow cytometry and immunofluorescence

4.5. Statistical analysis

Data availability statement

Ethics approval and consent to participate

CRediT authorship contribution statement

Declaration of competing interest

Acknowledgments

Footnotes

Contributor Information

2.10. In Vivo Single-Cell Sequencing Analysis at 1 Week post-implantation