Digital light processing-bioprinted poly-NAGA-GelMA-based hydrogel lenticule for precise refractive errors correction

Refractive disorder is the most prevalent cause of visual impairment worldwide. While treatment of refractive errors can bring improvement to quality of life and socio-economic benefits, there is a need for individualization, precision, convenience, and safety with the chosen method. Herein, we propose using pre-designed refractive lenticules based on poly-NAGA-GelMA (PNG) bio-inks photo-initiated by digital light processing (DLP)-bioprinting for correcting refractive errors. DLP-bioprinting allows PNG lenticules to have individualized physical dimensions with precision achievable to 10 µm (μm). Material characteristics of PNG lenticules in tests included optical and biomechanical stability, biomimetical swelling and hydrophilic capability, nutritional and visual functionality, supporting its suitability as stromal implants. Cytocompatibility distinguished by morphology and function of corneal epithelial, stromal, and endothelial cells on PNG lenticules suggested firm adhesion, over 90% viability, phenotypic maintenance instead of excessive keratocyte-myofibroblast transformation. In-vitro immune response analyzed by illumina RNA sequencing in human peripheral blood mononuclear cells indicated that PNG lenticules activated type-2 immunity, facilitating tissue regeneration and suppressing inflammation. In-vivo performance assessed using intrastromal keratoplasty models in New Zealand white rabbits illustrated that implantation of PNG lenticules maintained stable optical pathway, induced controlled stromal bio-integration and regeneration, avoided complications such as stromal melt, interface scarring, etc, but exerted no adverse effects on the host. Postoperative follow-up examination on intraocular pressure, corneal sensitivity, and tear production remained unaffected by surgery up to 1-month post-implantation of PNG lenticules. DLP-bioprinted PNG lenticule is a bio-safe and functionally effective stromal implants with customizable physical dimensions, providing potential therapeutic strategies in correction of refractive errors.


Introduction
The refractive function of our visual system supports reliable perception of our surroundings through precise focusing of ambient light onto the light-sensitive cells (photoreceptors) of the retina [1]. Of the eye's total refractive power, 70% is attributed to the cornea, where the stroma layer normally accounts for 90% of its overall volume, and is considered a significant determinant of visual acuity [2]. Refractive errors, including myopia, astigmatism and hyperopia, are the top cause of distant or near visual impairment worldwide and of great epidemiologic and socioeconomic relevance according to the World Health Organization [3]. Conventional refractive correction by using external devices such as glasses and contact lens are usually effective but may be associated with unacceptable or burdensome inconvenience for the individual's quality of life and activities of daily living [4]. While corneal laser correction including laser-assisted in-situ keratomileusis (LASIK) and small incision lenticule extraction (SMILE) can often achieve the 'spectacle-free' goal, they also involve permanent and irreversible ablation of stromal tissue in the central optical region of the cornea [5]. This means that a LASIK/SMILE-thinned cornea are at risk of biomechanical instability and a reduced ability to withstand an unfavorable relationship between intraocular pressure (IOP) and external pressure according to the micro-anatomy of corneal stroma (which consists of approximately 200 lamellae of proteoglycan-coated collagen fibrils running parallel to each other but orientating along orthogonal directions), and possibly leading to postoperative corneal ectasia [6]. To address this, implantable collamer lens has been recently used as a minimally invasive and additive device, but it requires an intraocular procedure, with potentially significant complications such as premature cataract development, endophthalmitis, and endothelial decompensation [7]. Hence, an individualized, convenient, reversible, and safer means to correct refractive errors is desirable to address concerns of currently available treatments.
Offering a high degree of convenience, safety and reversibility, the concept of adding volume to instead of removing from stromal layer to treat moderate or high hyperopia and presbyopia as well as keratectasia and perforations, was first established in the 1990s [8]. Additive keratoplasty is gaining momentum as a new perspective for refractive errors management, where intrastromal corneal ring segments implantation is one of these successful applications but is limited to myopia of less than 6.0 diopters (D) [9]. The first human donor lenticule implantation in a human subject was reported by Pradhan et al, where they implanted a −10D lenticule into an femtosecond laser (FSL)-created pocket (this new surgical procedure was termed lenticule intrastromal keratoplasty (LIKE) by Seiler et al) of a patient with +11.25D hyperopia, resulting in a decreased spherical equivalent and improved corneal topography [10,11]. As an alternative to LASIK/SMILE, the minimally invasive and tissue-preserving nature of this procedure allows for potential reversibility and recovery of preoperative refractive status, if desired [8]. However, considering the magnitude of suitable corneal tissues required and the unsteady supply of donor corneas, as well as potential complications associated with donor corneas including transmission of infectious agents (HIV, Hepatitis, etc), there is an unmet need for synthetic corneal lenticules [12]. Bioengineered corneal lenticule could work for refractive (hyperopia or myopia) and tectonic purposes, where it can improve corneal tissue strength and integrity (such as in ectasia), and require less demand on storage, transportation, biobanking system condition than humansourced tissues [13]. Compared with corneal donor graft, adoption of non-toxic and non-allogeneic biomaterials makes decellularization procedures unnecessary, eliminating the use of immunosuppressant agents with their associated risks and complications. Moreover, the risk of rejection for thin lenticule implantation is theoretically lower than full-thickness corneal graft owing to a less antigenic load to elicit an immunological reactions and no triggering factors for immune responses due to indirect contact with tear and aqueous humor [13]. What's more, bioengineered lenticule could provide additional functions such as delivering drugs in a slow, controlled, and localized manner, enhancing the ocular bioavailability and prolongs the physiological effects [14].
Changing the paradigm of tissue engineering, 3D bioprinting has enabled deposition of biological materials in a prescribed pattern corresponding with the anatomy of an organotypic model with the potential to regenerate functional organs, including corneas, in a customized manner [15]. Based on this, digital light processing (DLP)-bioprinting was introduced in 2006, utilizing projection technology that allows the polymerization of biomimicry inks in a layer-by-layer fashion to precisely produce predesigned structures with high efficiency, resolution, and fidelity [16]. In terms of corneal constructs, DLPbioprinting owns extreme superiority in displaying smooth surface without strip-like patterns compared with those from bioprinting technologies adopting dot-by-dot or line-by-line approaches [17]. In addition, this micron-grade technology brings greater accuracy to customization of the refractive lenticule's spatial volume with its effect on achieving the desired refractive outcome by depositing the selected bio-ink in a predicted 3D space [18]. What's more, owing to its configurability and good dimensional accuracy, individually DLP-bioprinting on demand offers an economical choice regarding time, cost, and labor than mass manufacturing in advance, since DLPbioprinting allows for a tailor-made design according to individual needs [19]. Therefore, DLP-bioprinting is considered a means for mass-producing made-toorder stromal refractive lenticules which could be customizable in size, thickness and swelling, personalized to the patient's requirements, thereby achieving their best-corrected visual acuity especially when combined with FSL-assisted LIKE techniques. However, there is currently limited evidence on DLP-bioprinted refractive lenticules attaining sufficient level of clinical functionality and practicality for correction of refractive errors.
To achieve personalized refractive function and tectonic stability against deformity, the selected biomaterials should possess sufficient tensile strength [20]. Although having good biocompatibility, biomaterials such as collagen, gelatin, and GelMA are mechanically weak and susceptible to enzymatic degradation [21,22]. Inversely, excessive hardness of the refractive lenticules may cause outward epithelial or inward endothelial erosion, which in turn may compromise globe integrity and expose the eye to potentially blinding complications such as endophthalmitis (fails to keep bacteria or toxins outside) or hypotony (fails to keep aqueous humor inside) [23,24]. In addition, suboptimal optical properties such as poor transparency or unsuitable refractive index (RI) may limit visual function as well as appears aesthetically less pleasing [1]. Furthermore, deficiency in oxygen and nutrient permeability may lead to complications such as stromal melt, tissue necrosis, long-term calcification, or retroprosthetic membrane formation [25,26]. Apart from these, prominent anti-swelling, anti-fouling characteristics are also critically required for refractive lenticules biomaterials. For example, poly (ethylene Glycol)/ poly (ethylene glycol) diacrylate, although its mechanical strength is with a broad range of modulus tunability, the deformation upon swelling may result in implant failure [27]. Also, the occurrence of biofouling of polymer hydrogel materials such as poly (2-hydroxyethyl methacrylate) and poly (acrylamide) is recognized to elicit unwanted fibrosis (and subsequent corneal scarring and loss of transparency) [28]. Ongoing studies on non-toxic, mechanically stable, optically bionic, nutritionally functional, anti-swelling and anti-fouling stromal implants are underway, as an ideal one which is non-inferior to donated corneal tissues in meeting all these needs simultaneously has yet to be found.
Poly-NAGA-GelMA (PNG) hybrid hydrogel is a three-dimensional dual-network hydrophilic polymers of N-Acryloyl glycinamide (NAGA) and gelatin methacrylate (GelMA). This supramolecular binary copolymer owns hybridization of permanent covalent crosslinkage of PNAGA and GelMA chains (with existence of three main bonding forms: PNAGA's double-bond crosslinking, GelMA's double-bond crosslinking, and NAGA-GelMA double-bond crosslinking), together with the triple hydrogen bonding formed by PNAGA's dual amide motif in side chain. PNAGA features outstanding anti-swelling ability owing to shielding of hydrogen (H)-bonds microdomains from water molecule attack, superior anti-fouling capability due to strong H-bonding interactions with water, tissue-mimicry optical and mechanical behaviors resulting from high H-bonded crosslinking density [29,30]. What's more, it is bioinert and has less immune-reactive and lower foreign body reaction [30]. Moreover, NAGA and GelMA bioink is photo-crosslinkable, thus avoiding the use of chemical crosslinkers and their potential toxicity, and does not require post-processing steps to gain sufficient stability, making it highly advantageous for simplifying and upscaling the manufacturing process of refractive lenticules. In addition, PNG gelled quickly and bound tightly inter-layers to self-support against delamination or fragmentation during DLPbioprinting. While the bio-ink of NAGA offers sufficient stability, GelMA renders bio-affinitive cell adhesion and proliferation capability [31,32]. The double network enhances its mechanical properties, and balances between the tectonic stability to preserve the optimal refractive power and the bio-integration capability to sustain the normal stromal function. Meanwhile, the competitive hydrophilicity and sufficient water-retention capability ensures the water like optical clarity and nutrients permeabilization of PNG lenticules.
Importantly this study conceives a corneal refractive errors correction alternative-intrastromal implantation of the DLP-bioprinted PNG lenticule and assesses its physicochemical, biological, ophthalmic, histopathological properties in sequence. Our results demonstrate that a mechanically, optically and biologically balanced bio-ink and deposition using layer-by-layer additive manufacturing can jointly contribute to an uncomplicated, safer, reversible, and customizable refractive errors solution.

DLP-bioprinting process
The DLP printing device (EFL-BP-8600, Company, Suzhou, China) was used for 3D bioprinting of PNG lenticules. The entire printing system was placed in biological safety cabinet and was sterilized using 75% alcohol after use and UV light before next use. To generate pre-designed 3D volumes, computer-aideddesigned models (Materialize 3-matic 11.0.0.241) in STL format was loaded into the bioprinting software. Right before printing, the hybrid bio-inks were heated to 37 • C for full dissolution, vortexed for several minutes for complete mix, and filtered by a 0.22 µm filter for proper disinfection. After completing DLPbioprinting, PNG lenticules was transferred to sterile DPBS in tissue culture plates under standard cell culture conditions.

Mechanical tests
PNG bio-inks were printed into 10 × 50 × 0.5 mm strips for tensile test, and columns of φ5 × 5 mm for compression test. All mechanical tests were performed at room temperature after achieving a swelling equilibrium in DPBS using Instron tensile and compression tester (INSTRON LEGEND 2344) (n = 3). The extension rate was controlled at 50 mm min −1 for tensile test and the crosshead speed at 10 mm min −1 for compression test. Young's modulus or compression modulus were calculated as the gradient of the linear region in the stress-strain curve of the tensile test or compression test, respectively. Ultimate tensile strength or compression strength was calculated as the highest measured stress before failure in the tensile or compression test [33,34].

Dynamic swelling behaviors
PNG bio-inks were printed into φ10 × 1 mm tablets for swelling testings.
PNG tablets were separately immersed in sterile DPBS at 37 • C for six weeks and were taken out at different time intervals (0, 2, 4, 6, 24, 48, 72, 168, 504, 1008 h), to measure the mass (n = 6). 1 ml solution was taken out and refreshed with another 1 ml solution daily for the first 4 d and subsequently every alternate date until 6 weeks [27]. The percentage residual mass of PNG tablets were calculated according to the following equation [35]: where W 0 is the initial weight of the hydrogel and W t is the weight of the hydrogel at each time point. Furthermore, fully hydrated PNG tablets were dried superficially with filter paper and weighed on a microbalance to record the wet weight (m wet ). Then the weighed tablets were lyophilized to constant weight (m dry ) [34].
EWC is defined as: Before swelling tests, PNG tablets were lyophilized and weighed as the initial dry weight. After immersing in DPBS at 37 • C for 24 h, the wet weight were recorded.
SR is defined as: where m dry , i is the initial dry weight of the hydrogel, and m wet , t is the weight of the hydrogel at 24 h.

Optical characterization
PNG bio-inks were printed into 10 × 40 × 0.5 mm strips for measuring light transmittance and 15 × 35 × 0.5 mm strips for measuring RI, using UV-vis spectrophotometer and Abbe refractometer, respectively (n = 3). Before both optical characterizations, the PNG strips were immersed in DPBS (pH = 7.4) at 37 • C till fully hydrated and DPBS served as the blank control. Light transmittance was measured using monochromatic light at wavelength of 630 nm. For the refractive test, PNG strips were placed between the sample prisms for reading the RI values at room temperature.

Water contact angle analysis
PNG bio-inks were printed into 20 × 10 × 0.1 mm films for water contact angle analysis. The instant water contact angles at 60 s were measured by an optical contact angle meter (n = 5).

Glucose permeability
The glucose permeability of the PNG films were determined using a self-modified two-chamber device filled with glucose solution and MQ water, respectively and measured with glucometer. Calculation of the diffusion coefficient of PNG films followed previous protocols [33].

In-vitro studies 2.8.1. Cell culture
The study was conducted according to the tenets of the Declaration of Helsinki. All cultures were cultivated at 37 • C under an atmosphere with 5% CO2, and the medium was refreshed every two days. Human corneal epithelial cells transfected with SV40 (hCE-Ts, donated by Dr Kaoru Araki-Sasaki, Japan and Dr Joseph CHAN, HK) were cultured in the complete medium that constituted Dulbecco's modified Eagle's medium/Ham's F12 media supplemented with 10% v/v FBS and 1% P/S [36]. Fibroblast Medium (ScienCell® #2301) supplemented with 2% FBS and 1% P/S were used for culturing human keratocytes (hKs, ScienCell® #6520, donated by Dr Kendrick SHIH) of passage 3-5 in vitro [37]. Human umbilical cord mesenchymal stem cells (hUCMSCs, PCS-500-010, ATCC®) of passage 1-5 were grown in mesenchymal stem cell basal media supplemented with mesenchymal stem cell growth kit (2% FBS, 5 ng ml −1 rh FGF basic, 5 ng ml −1 rh FGF acidic, 5 ng ml −1 rh EGF) and used according to the manufacturer's protocol [38]. A protocol published by a Japanese research team in 2018 were followed and modified to induce hUCMSCs transdifferentiated corneal endothelial cells (hUTCECs) [39]. Primary rabbit corneal endothelial cells (prCECs) were isolated from rabbit eyes based on a previously published protocol [36]. prCECs of passage 1-2 were cultured in F99 (1:1 mixture of Medium 199 and Ham's F-12 Nutrient Mix) supplemented with 10% newborn calf serum and 1% P/S. Human periphery blood mononuclear cells (PBMCs) were isolated from the buffy coat (Blood Storage & Issue Section, Hong Kong Red Cross) by a density gradient centrifugation method  [40,41]. Microscopic images were taken with a fluorescence microscope using SPOT Advanced software.

SEM visualization
(1) Ultrastructural adhesion morphology of the confluent hCE-Ts or hKs seeded on PNG-25%/5% tablets or glass at an initial density of 2000 cells ml −1 , (2) microstructure of the rabbit stroma or PNG lenticules were assessed using SEM. Samples were criticalpoint-dried and visualized under LEO 1530 FEG or Hitachi S4800 FEG scanning electron microscope.

2.8.5.
In-vitro scratch-healing assay PNG bio-inks were printed into φ10 × 1 mm tablets for 6-well-transwell plates co-culture. PNG-25%/5% tablets were transferred onto the transwell insert (0.4 µm Polyester Membrane), and then the insert was assembled with six-well plates pre-seeded with hKs. The scratch-healing assay followed our previously published protocol [42].
2.8.6. Immunofluorescence staining hCE-Ts, hKs, hUCMSCs, prCEns, and hUTCECs laden on glass coverslip or PNG-25%/5% tablets were stained with its relevant primary antibody (table 2), then subsequently with the corresponding secondary antibody ( To examine the safety and efficacy of the DLP-bioprinted PNG lenticules, the relevant rabbits underwent controlled surgical procedures of intrastromal keratoplasty (ISK) under general anesthesia-sedation with intramuscular acepromazine (1 mg kg −1 ); analgesia with subcutaneous buprenorphine (0.02 mg kg −1 ); anesthesia with a combination of intramuscular ketamine (50 mg kg −1 ) and xylazine (5 mg kg −1 ). Lidocaine eye drops was instilled 10 min before surgery as local (topical) anesthesia. Betadine solution was applied for disinfection of the ocular surface, followed by rinsing with normal saline. Postoperatively, buprenorphine (0.02 mg kg −1 ) subcutaneous injections were used for analgesia every 12 h for seven days. Rabbits were randomly assigned to three different groups (n = 6): Wound, Sham-ISK, and PNG-ISK groups. Untreated eyes were used as healthy controls.
In the Sham-ISK and PNG-ISK groups, the central corneal surface was lightly marked with φ6 mm trephine, then a partial-thickness (100 µm depth) intrastromal pocket was created with a crescent knife to the same diameter through a tiny cut (1 mm). Next, in PNG-ISK group, PNG lenticules (φ3.5 × 0.05 mm) were gently folded and inserted into the corneal pocket through the tiny cut with 0.12 fine forceps to examine the host's immune reaction and tissue integration. In the wound-control group, the central superficial corneal stroma was removed to about 100 µm depth with φ6.0 mm trephine and crescent knife ( figure 6(a)). TRICIN eye ointment was applied two times a day postoperatively. Animals were monitored daily for signs of discomfort and status of wound or implant for the first week after surgery and then twice weekly for subsequent weeks.

Follow-up examination of the animals
Clinical examinations were performed by two independent observers on 1 d preoperatively and 1, 7, 14, and 28 d postoperatively. Corneal nerve function (touch sensitivity) and IOP were assessed in the (host's) peripheral cornea 2 mm anterior to the limbus in conscious animals without local anesthesia or systematic sedation. To elicit a response, the nylon filament of a Roger-Bonnet esthesiometer probing the ocular surface was progressively reduced from 60 mm (maximum) to 0 mm (minimum). Each sensitivity test was repeated three times, advancing in 5 mm steps. Two positive responses in the three attempts at each filament length were regarded as a positive result. The most extended filament length causing a positive result was recorded as the threshold value of sensitivity for that cornea. IOP measurements were performed 5-10 consecutive times till the eventual value is available on the screen and repeated for three times in total by rebound tonometer. Normal IOP for rabbits range from 10 to 20 mmHg [43]. Tear production was evaluated using Schirmer's test with local anesthesia, where sterile test strips are placed in the lower fornix for 5 min, and then the length of the hydrated strip is recorded in mm (with normal defined as >5 mm) [44].
At each follow-up visit, corneal status was examined for any signs of infection or inflammation and documented by slit-lamp photography. Corneal staining for any epithelial defects was captured 90 s after applying fluorescein and evaluated in a masked fashion.

Tissue processing and histological examination
At postoperative 4 weeks, the relevant rabbits were sacrificed painlessly, according to protocol approved from CULATR of the University of Hong Kong. In histological evaluation, the harvested corneas were stained with hematoxylin and eosin (H&E), toluidine blue, and Masson's trichrome. Then immunofluorescence staining of LUM and COLI was conducted to examine any stromal remodeling. Next, immunohistochemical staining, including IL6 and IL10, were performed to examine the in-vivo immune response after implantation of PNG lenticules. Images were digitalized using light microscopy. In the morphological examination, tissue samples were cut into 2 × 2 mm cubes and stored under fixative for TEM processing. Samples were imaged using a Philips CM100 transmission electron microscope (Queen Mary Hospital, HK) with Olympus SIS Tengra CCD camera (2.3k × 2.3k pixels). Evaluation of systemic toxicity was performed by histopathological examination of H&E-stained sections of the heart, liver, kidney, and spleen.

Statistical analysis
All the experimental procedures were performed in the same order, by the same observers, and in the same period to minimize any statistical fluctuation or bias. All instruments were calibrated prior to each session of measurements. All experiments were conducted in triplicates as minimum. Data analysis was performed using statistical software (Graph-Pad Prism Version 8.0.0). Values were expressed as mean ± standard error (SEM). The above data with two groups were analyzed with two-tailed Student's t-test, and more than two groups with one-way analysis of variance. P < .05 was considered statistically significant for all experiments.

DLP-bioprinting of PNG lenticules
Prior to DLP-bioprinting, a polymer-strengthened, photocurable hydrogel bio-ink was prepared by copolymerizing NAGA with GelMA. After DLPbioprinting optimization, the sol-gel transition of PNG-25%/5% bio-inks were settled at around a layer thickness of 10 µm and a layer exposure time of 8 s ( figure 1(a)). At current stage, we DLP-bioprinted rabbits' equal-scaling corneal model obtained by micro-CT scanning together with KeraKlear-alike model obtained by online sources [45] for printability demonstration, but planar refractive lens for the in-vitro and in-vivo experiments ( figure 1(c)). Future in an actual application, PNG hydrogel is expected to be printed as a biconvex refractive lens with minimum thickness at the peripheral edge and maximum thickness at the center (which is determined by the refractive power to be corrected, i.e. approximately 15 µm in the center for 1D spherical equivalent) so as to achieve personalized refractive power [13].

Tunable biomechanics
As sketched in figure 1(b), PNG is a photo-crosslinked hydrogel network that combines covalent GelMA-PNAGA crosslinking with multiple reinforced Hbonding from PNAGA which the higher mechanical strengths are primarily attributed to. However, the excessively high crosslinking density of GelMA may interfere with the formation of PNAGA H-bonds, resulting in a decline in mechanical strength. In our preliminary testing by two experienced eye surgeons, it was noted that increasing concentrations of GelMA beyond 5% resulted in greater brittleness and friability. Therefore, 5% was chosen as the optimal concentration ratio of GelMA and was fixed for further experiments. To validate the mechanical performance of the hybrid hydrogel, Instron tests on elasticity and compressibility of PNG were performed (figures 2(a) and (b)). The results showed mechanical strength increased with increments in monomer contents of NAGA from 15% to 30%, evidenced by the increasing Young's modulus, compressive modulus, tensile strength, break strain, compressive strength, and compressive failure strain (table 3). Importantly, PNG-25%/5% showed comparable structural strength to human donor cornea (table 3), whose effective Young's modulus is 0.281 ± 0.214 MPa for the anterior stroma [46]. In mechanical compression test, PNG lenticule withstood high compression forces and recovered perfectly after removal of the compressive load, manifesting PNG lenticule's reliability in bearing external load and retaining the integrity of ocular globe. These in together elucidated that PNG lenticules displayed mechanical stability to human donor graft, allowing for optimal surgical handling [47].

Swelling and optical properties
As determined by the crosslinking density of PNG polymers, the tightness of the hydrogel network not only impacts on the mechanical but also the hydration properties of the PNG lenticules. It may be that increasing NAGA concentration introduces more Hbonds of PNAGA and photocrosslinks between the PNAGA and GelMA chains, thus restricting the diffusion of water molecules into the network. Therefore, any incrimination in either direction may affect the balance between the stability and biocompatibility that they represent [48]. Figure 2(c) presents the equilibrium water contents and swelling ratios of PNG. In general, with increasing NAGA concentration, the equilibrium water content of PNG decreased. Notably, the equilibrium water content declined and plateaued when NAGA increased to 25%, at around 78% (similar to natural cornea), meeting the requirement of oxygen and other-soluble metabolite exchange [49]. In swelling ratio testing, it was noted that although the wet weight increased after swelling, the total volume of PNG remained almost unchanged. The phenomenon that no serious solubilization deformation occurred has important guiding significance for material implantation in vivo. To mimic in-vivo physiological environment, dynamic swelling behaviors of PNG were evaluated in sterile DPBS (mimicking physiological aqueous condition that covers the ocular surface) by monitoring the percentage of remaining weight over time. The results showed that PNG remained sufficiently stiff and stable, showing a negative degradation rate till 6 weeks of immersion in DPBS ( figure 2(d)). Also, the dynamic swelling curve of PNG-25%/5% was closest to 100% residual mass ratio compared with other concentrations.
Both visible wavelength (400-700 nm) transparency and refractive property of the cornea are closely associated with water retention capacity [50]. Our DLP-bioprinted PNG lenticules exhibited good gross transparency ( figure 1(c)). The light transmission at 630 nm remained >85% over a range of NAGA concentrations (figure 2(e)), comparable to normal human cornea, as was the RI of around 1.37 similar to the reported average value of 1.3765 ± 0.0005 for human cornea (figure 2(f)) [50,51]. According to Lensmaker's formula P = 1 f = (µ − 1) × 1 R , where P = optical power (diopters, D), f = focal length (meters), µ = RI, R = radius of curvature (meters), the refractive power of PNG was calculated to be around 46D (approximately 43D in human cornea)   [52], given that the average R of anterior corneal surface is 7.80 mm. As the PNAGA network of PNG is robust and non-degradable, excessive water accumulation and edema with resulting loss of corneal clarity and/or refractive errors could be prevented [53]. Also, the shape retention by the undergraded outer shell contributes to maintaining individualized, longterm refractive characteristic of the cornea so that with the appropriate RI possessed by PNG hydrogel, and the appropriate corneal curvature from the DLP-bioprinting, the desired refractive power can be predetermined and produced according to clinical needs [54]. Figure 1(g) shows the water contact angle analysis, where all concentration ratios of PNG lenticules displayed a hydrophilic nature (below 90 • ). Interestingly, hydrophilicity peaked (smallest contact angle of 47.4 • ) when concentration is of 25%/5%. Complimentary to the comparable water contents, the high hydrophilicity of PNG lenticules to normal cornea facilitates permeability to nutrients, supporting interaction with the host cells, thereby reducing long-term failure from calcification and/or stromal melt [55]. Permeability assessment of PNG lenticules showed glucose diffusion coefficient of PNG-25%/5% was 2.11 × 10 −6 cm 2 s −1 , close to the human cornea value of 2.5 × 10 −6 cm 2 s −1 (figure 2(h)), and negative correlation of glucose permeability with increasing concentrations of NAGA [56], which makes PNG lenticules more resistant to nonspecific protein adsorption that undesirably triggers retroprosthetic membrane formation [57].

In-vitro evaluation of cytocompatibility
As cell affinity is a prerequisite for any bioactive scaffold intended to facilitate tissue regeneration, we evaluated a range of bio-inks (table 2) with PrestoBlue viability analysis using 2D seeding of hCE-Ts on PNG [58]. OD values representing cell viability increased with incubation time and peaked at 25% of NAGA among various concentrations of GelMA (5%, 10%, 15% and 20%) ( figure 3(a)). To further evaluate the effect of PNG lenticules on different corneal cells, calcein AM and ethidium homodimer staining was performed, and after 7 d of cultivation, corneal epithelial, stromal, and endothelial cells grew well on PNG lenticules with very strong green signal (over 90%), suggesting no significant toxicity on corneal cell population (figures 3(b) and (c)). Although the viability of hKs slightly decreased when seeding on PNG tablets compared to blank well plates, the viability of seeded hCE-Ts or hUTCECs was not significantly affected. This demonstrated PNG lenticules' potential in discriminatingly supporting stromal cell deposition on the anterior instead of the posterior surface, which may prevent the formation of retroprosthetic membranes (that can impair vision) while supporting epithelialization or endothelialization [59]. As vinculin/F-actin represent focal adhesion and cell stretchability, respectively, the vinculin/Factin/DAPI staining revealed that hCE-Ts adhered well to, and spreaded readily on PNG lenticules, indicating its suitability to support epithelial growth [60]. In addition, F-actin staining of hKs' cytoskeleton on PNG presented stellate fibroblast-like morphology instead of elongated myofibroblast-like shape. Meanwhile, the ultrastructural morphology observed by SEM confirmed no significant changes in epithelial or stromal cell morphology when adhered to PNG (figures 4(a) and (b)). Then, we assessed keratocyte cytofunctions using KERA, CD34, or ALDH1A1 as phenotypic markers ( figure 4(b)). The positive expression of KERA suggested controlled stromal regeneration with uniform-length fibril production capability when cultured on PNG lenticules [61]. The presence of surface protein CD34 in keratocytes indicated a less-differentiated class of fibroblast-like dendritic cells when seeding hKs on PNG lenticules [62]. The existence of ALDH1A1 displayed the crystalline contents of hKs for the maintenance of corneal transparency when planted on PNG lenticules [63]. Next, in-vitro scratch assays were performed to determine the influence of PNG lenticules on hKs' migration. Scratch-healing results showed delayed migration of keratocytes when co-cultured with PNG, indicating a suppressed state of these keratocytes instead of excessively activated (myo)fibroblastic state ( figure 4(c)). All these results demonstrated that PNG lenticules are biocompatible scaffolds for maintaining the phenotype and cytofunctions of corneal cells.
Cornea contains a significant resident macrophage population, whereby M1-type promotes positive immune response, and contrarily M2-type contributes to immunosuppressive function playing an important role in wound healing and tissue repair [64]. To understand the immunomodulatory role of PNG lenticules to immune cells during corneal regeneration, hPBMCs derived macrophages were seeded on the surface of PNG lenticules for 48 h. As shown in figure 5(a), these macrophages exhibited distinctive cell morphologies and presented settlement distribution on PNG lenticules. To further evaluate the effects of PNG lenticules on macrophages, a transcriptomic profile of hPBMCs derived macrophages exposed to PNG lenticules or control was conducted. As shown in figure 5(c), the volcano plots showed 1494 up-regulated and 2697 down-regulated genes (PBMC-Ctrl versus PBMC-PNG), indicating a wide range of gene expression differences. Interestingly, type-2 immunity markers such as IL10, CCL2, STAT3, IL4R, CD163 were significantly increased in PNG groups; on the other hand, type-1 immunity markers such as NOS2 and TNF were not, suggesting PNG promotes macrophage-initiated repair immunity [65,66]. All differentially expressed genes were then collected to perform GO database analysis and were divided into three classes, including biological process, cellular component, and molecular  [67]. Then, the KEGG and Reactome pathway analysis were performed to analyze the underlying signaling pathways [68]. The significantly enriched Reactome and KEGG enriched pathways were displayed in figure 5(g). Interestingly, cytokinecytokine receptor interaction, ECM-receptor interaction pathway, interleukin-4, interleukin-13, and interleukin-10 signaling shown are demonstrated to be related to type-2 immunity activation, implying PNG promotes M2 polarization in macrophages [69]. Therefore, differences in PNG-induced gene expressions, including interleukin-4, interleukin-13, interleukin-10 signaling, and collagen formation were next focused on. The heatmap analysis showed that MMP family like MMP1, MMP2, MMP3 and MMP7 were significantly increased in PNG groups, which has been widely proven associated with tissue repair and ECM regeneration (figures 5(h)-(j)) [70]. Moreover, cell adhesion CCL2, immune regulation IL10, IL18, CXCL8 and collagen formation LAMA3, LAMB3, LAMC2 were also upregulated in PBMC-PNG group [71][72][73]. Taken together the transcriptomic profile of PBMC-PNG versus PBMCcontrol implied that PNG lenticules activated type-2 immunity in macrophages, facilitating tissue regeneration and suppressing inflammation.

In-vivo assessment of biocompatibility and safety
To evaluate the safety and efficacy of PNG lenticules, studies of ISK models of New Zealand rabbits in vivo were conducted. Figure 6(a) depicted the surgical procedures of manual dissection of a stromal pocket for intrastromal implantation of PNG lenticules, which is a modified and more economic model compared with FSL-assisted LIKE. While in the actual translation to bedside, PNG lenticule could be applied through FSL-assisted LIKE where the laser removes diseased or scarred corneal stromal tissue prior to insertion and creates the pocket for PNG lenticule replacement with higher accuracy. Corneal sensitivity to mechanical touch showed an average value of 60 mm, comparable to the response of healthy control eyes (figure 6(b)) [74]. In addition, IOP in all operated eyes remained within the preoperative range of 10-20 mm Hg at all postoperative time points within one month (figure 6(c)) [43]. Tear secretion in the operated eyes remained normal and over 5 mm in 5 min (figure 6(d)) [44]. Slit-lamp examination was conducted at predetermined postoperative time points, with no eyes having any wound leak, while the anterior chamber remained well formed, with good corneal transparency (figure 6(e)). Overall, all implanted PNG lenticules remained free of major complications (such as stromal edema, infection, erosion, or toxicity of host tissue) or signs of rejection (like neovascularization or prolonged inflammation) during the 1 month follow up period.
After 1 month implantation, the rabbits were sacrificed, and their corneas and critical organs were collected and analyzed by histopathological staining. H&E staining showed that the host cornea remained uninfluenced in PNG-ISK group ( figure 7(a)). In addition, critical organs, including the heart, kidney, liver, and spleen, did not reveal any deviation from normal by histopathological analysis, postoperatively ( figure 7(b)). This suggested that intrastromal implantation of PNG lenticules has no apparent adverse effects on endothelium or other native structures compared with the respective control groups and can generally be considered safe.
To reveal the effect on extracellular matrix and highlight the amount and distribution of fibrosis, Masson staining was carried out (collagen stains blue while muscle stains red), which showed that fibrosis occurred in the wound group instead of the ISK-PNG group [75]. Next, lumican (LUM) and collogen type-1 (COL1), proteins involved in collagen fibril organization, were assessed for signs of stromal regeneration and integration [76]. LUM was uniformly positive in both sham-ISK and PNG-ISK groups, but heterogeneously distributed in the wound group. Additionally, the expression of LUM was significantly greater in the PNG-ISK group, indicating controlled stromal regeneration with uniform and orthogonally aligned ECM production in this group [77]. COL1 appeared strongly positive near the core area for the wound group but stained uniformly weak in the sham ISK and PNG-ISK groups, suggesting uncontrolled regeneration with excessive scarring in the former [78,79]. Taken together, DLP-bioprinted PNG lenticules induced stromal regeneration and restoration with organized ECM expression instead of the irreversible myofibroblastic scarring. In addition, evidenced by signs of controlled stromal regeneration, our in-vivo results confirmed the occurrence of gradual degradation of GelMA and semi-degradability of PNG hydrogel.
We then performed immunohistochemical staining for both M1 (inflammatory stage) and M2 (anti-inflammatory stage) markers to validate the inflammatory and wound healing activities, which showed the presence of IL6 and  IL10 in both wound and ISK-PNG groups, but being more substantial in the former, consistent with the transcriptomic results [80,81]. To conclude, implanted PNG lenticules displayed transcriptomic match with the regenerated host cornea.
To further investigate the effects of PNG lenticules on host cornea, ultrastructural morphology of corneal tissues was examined by TEM and SEM scanning (figures 7(c)-(l)). The representative TEM images revealed no significant deviation in PNG implanted corneas compared to naive corneas at 28 d post-implantation [82]. Close to that of natural rabbit cornea, the ultrastructure of PNG lenticules displayed irregular pore sizes and a three-dimensional network (figure 7(c)), indicating its potential in recruiting cells and cytokines for further regeneration [83]. This indicated that NAGA's stable network provides structural support for the diffusive exchange of metabolites, cytokines, and cells between implant and recipient host while the gradual degradation of GelMA takes place for tissue integration.

Conclusions
This study introduces our design of a 3D printed double network hydrogel construct for individualized refractive errors correction. Collectively, DLPbioprinted PNG lenticule is mechanically and optically stable, nutritionally and refractively functional while potentially avoiding long-term complications of stromal melt, calcification, formation of retroprosthetic membrane. Therefore, intrastromal implantation of DLP-bioprinted PNG lenticule offers a safer and more convenient, reversible and recoverable, individualizable and customizable solution compared with the current available refractive spectacles or surgeries.

Data availability statement
All data that support the findings of this study are included within the article (and any supplementary files).