Nanofiber‐based glaucoma drainage implant improves surgical outcomes by modulating fibroblast behavior

Abstract Biomaterials are implanted in millions of individuals worldwide each year. Both naturally derived and synthetic biomaterials induce a foreign body reaction that often culminates in fibrotic encapsulation and reduced functional lifespan. In ophthalmology, glaucoma drainage implants (GDIs) are implanted in the eye to reduce intraocular pressure (IOP) in order to prevent glaucoma progression and vision loss. Despite recent efforts towards miniaturization and surface chemistry modification, clinically available GDIs are susceptible to high rates of fibrosis and surgical failure. Here, we describe the development of synthetic, nanofiber‐based GDIs with partially degradable inner cores. We evaluated GDIs with nanofiber or smooth surfaces to investigate the effect of surface topography on implant performance. We observed in vitro that nanofiber surfaces supported fibroblast integration and quiescence, even in the presence of pro‐fibrotic signals, compared to smooth surfaces. In rabbit eyes, GDIs with a nanofiber architecture were biocompatible, prevented hypotony, and provided a volumetric aqueous outflow comparable to commercially available GDIs, though with significantly reduced fibrotic encapsulation and expression of key fibrotic markers in the surrounding tissue. We propose that the physical cues provided by the surface of the nanofiber‐based GDIs mimic healthy extracellular matrix structure, mitigating fibroblast activation and potentially extending functional GDI lifespan.

the nanofiber-based GDIs mimic healthy extracellular matrix structure, mitigating fibroblast activation and potentially extending functional GDI lifespan. Pharmacological targeting of biomaterial-associated fibrotic events is particularly challenging due to the presence of physiological transport barriers and natural drug clearance mechanisms, often leading to transient gains in the functional lifespan of biomaterials. Fibroblasts, the effector cells in fibrosis, transdifferentiate into myofibroblasts upon receiving activating stimuli and deposit extracellular matrix (ECM), which encapsulates biomaterials. Physical stimuli arising from healthy ECM, such as mechanical compliance and topography, maintain the quiescence phenotype of fibroblasts, whereas mechanical tension and rigidity activate fibroblasts. 4 The response of fibroblasts to physical cues motivates the development of material-centric approaches to mitigate fibrotic encapsulation of biomaterials.
Glaucoma is a leading cause of irreversible blindness. The only clinically proven approach to prevent glaucomatous vision loss is to reduce intraocular pressure (IOP). 5,6 Glaucoma drainage implants (GDIs), which can be classified into those with and without a reservoir for aqueous drainage, are among the most widely used ocular biomaterials. 7 However, implantation of GDIs for IOP reduction often leads to post-operative complications, including hypotony and fibrosis. [8][9][10][11][12] Conventional subconjunctival GDIs have a modest median functional lifespan of 5 years due to fibrosis, and revision or repeat procedures are burdensome for patients and have an increased risk of failure. [13][14][15] Additionally, the aqueous humor draining into the subconjunctival space contains soluble factors implicated in fibroblast recruitment and activation, such as transforming growth factor-beta (TGF-β), platelet-derived growth factor (PDGF) and vascular endothelial growth factor (VEGF). 16,17 Furthermore, continuous aqueous humor outflow causes chronic mechanical tension in the bleb, triggering tissue resident fibroblast activation. 18 The introduction of the implant material only further exacerbates these cellular processes, leading to recognition of the material as foreign, and subsequent fibrotic encapsulation. Indeed, activated fibroblasts were found in fibrotic tissue surrounding glaucoma implants in preclinical surgical models and in glaucoma patients who have a history of implant failure. 19,20 Thus, there is significant need to engineer GDIs which enable (1) safe and effective IOP lowering, (2) controlled aqueous outflow to avoid hypotony and reduce subconjunctival exposure to aqueous cytokines, and (3) mitigation of fibrosis ( Figure 1). Here, we utilize a versatile nanofiber-based platform to manufacture tube GDIs. We hypothesized that a surface composed of nanofibers would be more structurally and mechanically similar to healthy ECM, and thus, less likely than smooth surfaces to be perceived by fibroblasts as foreign. In cell culture and in rabbit eyes, the nanofiber architecture reduced fibroblast activation and increased cellular integration. Additionally, we demonstrate that tube GDIs with a nanofiber surface reduced subconjunctival fibrosis in the eyes of rabbits compared to GDIs with smooth surfaces. We then evaluated the performance of the nanofiber-coated GDIs in comparison to the clinically available, minimally invasive XEN ® Gel Stent, and the silicone tube portion of the widely used Baerveldt ® GDI (BGI tube). We observed that nanofiber-based GDIs were capable of safe and effective IOP reduction, while minimizing conjunctival fibrosis-related gene expression. We anticipate that the reduced fibrotic response to nanofiber-based tube GDIs could lead to improved functional lifespan and positive long-term outcomes in glaucoma surgery.

| Nanofiber-based stents retain architecture and promote cell integration in vivo
We previously reported the development of nanofiber-based glaucoma GDIs with a degradable PGA core to enable controlled outflow of aqueous humor. 21 These pressure control shunts (PCS) prevent acute post-operative hypotony and allow for greater volumetric outflow at later post-operative stages for sustained IOP reduction. 21 In order to evaluate the effect of device stiffness and dimensions on GDI efficacy, we developed and evaluated two PET/PGA nanofiberbased designs in a long-term rabbit model: (1) a shorter GDI with a thicker, more rigid wall (PCS1: length 6 mm, OD 450 μm) and (2) a longer, more flexible GDI (PCS2: length 9 mm, OD 350 μm) ( Figure 2a). PCS2 showed a greater maximal ΔIOP of À5 ± 1.3 mmHg ($30% lower than the non-operated contralateral eye at Day 56) compared to À3 ± 1.7 mmHg for PCS1 (Figure 2b). Similarly, PCS2 provided $1.5-fold increased cumulative ΔIOP (area under the curve, AUC) compared to PCS1 over the 91-day study period (Figure 2c). given that even commercially used GDIs have been shown to fail due to fibrotic encapsulation after only 1-2 months in rabbits. [22][23][24]

| Characterization of fibroblast interactions with nanofiber and smooth scaffolds in vitro
The results of PCS2 evaluation led us to hypothesize that designing flexible GDIs with nanofiber surfaces may improve the functional device lifespan. However, the fully nanofiber-based PCS GDIs had mechanical limitations observed by the surgeon during handling and implantation, including irreversible deformation under compressive stress. Further, PET was too mechanically rigid when heated to evaluate a smoothsurfaced GDI in vivo (not shown). Thus, we incorporated polyurethane (PU) in later designs, due to its increased mechanical resiliency and flexibility when annealed into smooth surfaces. Further, smooth PU surfaces were amenable to coating with electrospun nanofibers post-annealing, which allowed us to directly compare smooth and nanofiber surface architecture. To characterize the effect of surface topography on cell activation in vitro, we electrospun PET and PU nanofiber scaffolds (Figure S1a) that were annealed to either maintain the nanofiber network under hydration or heated to the melting point and gradually cooled to room temperature to create smooth surfaces ( Figure S1b). Nanofiber diameters, analyzed using SEM micrographs, ranged from 400 to 1000 nm (Figure S1c spindle-shaped appearance. Further, fibroblasts on nanofibers had a more branched appearance with significantly smaller cell bodies F I G U R E 1 Nanofiber-based, partially degradable glaucoma drainage implant. Post-operative hypotony and fibrotic encapsulation lead to complications and failure of glaucoma surgery. Here, we engineered partially degradable, nanofiber-based GDIs and compared functional outcomes to implants with a smooth surface. We aimed to prevent hypotony and limit aqueous outflow in the acute post-operative phase using a degradable core design. We further hypothesized that imparting nanofiber architecture to glaucoma GDIs to mimic healthy ECM would support fibroblast quiescence to preclude the fibrotic processes that lead to device failure. (area 1004 ± 431 μm 2 [ Figure S2c], mean ferret diameter 154 ± 28 μm    Figure 4c). It is notable that as a collagen-binding integrin, ITGA2 is also an important regulator of integrin signaling, and ITGA2 knockout mice display an enhanced fibrotic response. 25 Additionally, MMP1 is a protein known to degrade ECM. 26 We also observed decreased expression of the matrix stiffness sensor ITGA6  Figure S3a. We then profiled the fibroblast response to the scaffolds using a cassette of genes identified previously as a signature of fibrotic glaucoma surgeries. 27 In the absence of

| Effect of nanofibers on fibroblast cell cycle in vitro
The broad transcriptomic changes in fibroblasts in response to the nanofiber and smooth scaffolds suggested globally regulated phenomena. In other cell types, including pluripotent stem cells, tissue-specific progenitor cells, and adaptive immune cells, resistance to activating stimuli is associated with quiescence. 30 We hypothesized that physical cues from nanofibers induce quiescence in fibroblasts. To explore this hypothesis, we analyzed the cell cycle of fibroblasts exposed to nanofibers or smooth surfaces using flow cytometry. When fibroblasts were cultured on smooth PET surfaces for 24 h, approximately 60% of the cell population were present in the G0/G1 phase and $36% of cells in S/G2 phase (Figure 4f) indicating normal cell cycle progression.
When transferred to nanofiber scaffolds for a period of 24 h, the percentage of fibroblasts in the G0/G1 phase increased to 81% and the percentage of cells in the S/G2 phase reduced to $9% (Figure 4f).
When the cells were then transferred back to smooth scaffolds, within 24 h the percentage of cells in the G0/G1 phase was then reduced to $64% (Figure 4f), indicating that nanofibers induced cell cycle arrest.
Moreover, after culturing cells for 48 h on the nanofibers, $31% of cell population exited the cell cycle, indicating induction of a quiescence phenotype ( Figure S4). Taken together, these data suggest that exposure of cells to nanofibers promoted a quiescent, non-fibrotic phenotype.

| Comparative evaluation of nanofiber-based and commercially available GDIs
Next, we sought to evaluate the performance of the leading 9 mm Nano GDI in comparison to clinically available GDIs. We chose two of the most commonly used subconjunctival GDIs: the silicone tube portion of Baerveldt Glaucoma Implant (BGI tube) and the XEN Gel Stent.
Bleb morphology in the XEN and Nano groups was largely unchanged through POD 28 (Figure 6c and Figure S6a Table S4. These data suggest that the nanofiber surface also attenuated fibroblast activation in response to the GDI in the conjunctival tissue in vivo.

| DISCUSSION
A wide range of biomaterials from bioinert metals to synthetic polymers have been applied to GDIs that have been studied in both preclinical models and clinical trials. Post-operative outcomes across the range of different types of GDI materials [34][35][36] highlight the need for alternative material-centric strategies to prevent biomaterialsassociated fibrosis. Scleral, conjunctival, and tenon's fibroblast responses to fibrotic stimuli have been characterized previously. 4,27,37,38 Specifically, VEGF production in response to fibrotic stimuli as well as in vitro proliferation responses to various small molecule drugs have been well characterized in tenon's fibroblasts. 39 Additionally, conjunctival and scleral fibroblast sub-populations have been shown to be immunomodulatory and involved in tissue remodeling and collagen synthesis. 38,[40][41][42] Notably, scleral fibroblasts produce a robust pro-fibrotic response to activating stimuli such as TGF-β. 43 Fibroblast migration from the scleral and subconjunctival tenon's fas- TGF-β is a potent fibroblast activator that is implicated in fibrosis following placement of conventional GDIs. 34,35,54,55 By binding to cell surface receptors on fibroblasts, TGF-β triggers transdifferentiation of dormant fibroblasts into myofibroblasts, which are characterized by increased cell contractility and size, ECM remodeling, and expression of pro-fibrotic biomarkers such as αSMA. 56,57 TGF-β overexpression in the aqueous humor leads to fibrosis of the trabecular meshwork, the outflow pathway of the eye. 54,58,59 Earlier studies in normotensive rabbits found that blocking the lumen of GDIs to prevent aqueous outflow into the subconjunctival space reduces fibrosis. 23 Thus, the Nano GDIs described in our studies may prevent fibrosis through both the nano-architecture and degradable inner core that restricts outflow in the early post-operative period. We found that nanofibers successfully induce resistance to TGF-β-and LPA-driven activation of human scleral fibroblasts in vitro, suggesting the response was regulated by a central signaling mechanism rather than receptor-driven. Further, transcriptomic differences across actin-regulated signaling, rho-kinase signaling, and cytoskeletal regulation in fibroblasts cultured on nanofibers further suggested cell quiescence. Interestingly, ITGA2, a known collagen-binding integrin highly expressed in tissue-resident dendritic cells and macrophages, 18,25,60 was significantly upregulated in fibroblasts cultured on nanofibers. The absence of ITGA2 not only results in loss of the quiescence phenotype, but also activates cellular programs associated with cell migration in animal models of prostate, gastric, colorectal, and breast cancer. [61][62][63] Indeed, in ITGA2 knockout mice, ECM deposition was exacerbated in preclinical models of kidney fibrosis. 60 Previously, Yu-Wai-Man et al. validated a cassette of transcripts that correlate with post-operative fibrosis in the conjunctiva after glaucoma surgeries. 27 In this genome-wide RNA sequencing study analyzing patient-derived fibroblasts, increased MMP-10, CD34, and IL-33 expression and decreased MYOCD expression were most predictive of successful surgical outcomes. 27 We observed that fibroblasts cultured on nanofibers potentiated a non-fibrotic transcriptomic signature. Further, we observed that fibroblasts interacting with nanofibers displayed a reversible G0/G1 cell cycle arrest, and that extended nanofiber exposure resulted in fibroblasts exiting the cell cycle.
Together, these findings suggest that biomaterials that mimic ECM could reduce fibroblast activation and promote more successful and durable surgical outcomes.
The in vivo performance of the Nano GDI in rabbits was favorable compared to existing surgical options. Due to the aggressive fibrotic response in rabbit eyes, normotensive rabbit models of glaucoma surgery are primarily used to evaluate biocompatibility, rate of fibrosis, and GDI functionality. 22

| Scaffold manufacturing for in vitro experiments
PET was dissolved in HFIP at 10% (w/v) by stirring overnight at 50 C. PU was dissolved in HFIP at 6% (w/v). Following dissolution, the polymer solution was loaded into a 3 ml syringe and dispensed at a flow rate of 850 μl/h through a 20-gauge stainless steel nozzle using a syringe pump.
A voltage differential of 12.5 kV was maintained between the nozzle and a static aluminum collector. Nanofibers were cut into circular disks with a diameter of 3.5 cm ( Figure S1a). Next, PET nanofibers were annealed at 120 C for 16 h and sterilized using isopropyl alcohol followed by UV exposure for 2 h. Smooth PET films were cut to 3.5 cm disks and sterilized using identical methods. Scaffolds were affixed to wells in a six-well tissue culture plate prior to seeding. PU scaffolds were either annealed at 80 C in the nanofiber group or 150 C in smooth surface group. PU scaffolds underwent identical sterilization procedures to PET scaffolds prior to cell seeding. For protein adsorption, scaffolds were incubated with 1640 RPMI medium containing 10% FBS for 16 h. Following this, scaffolds were washed twice in PBS and adsorbed protein was extracted using 2% SDS. Protein concentrations were measured using BCA assay kit following the manufacturer's instructions.

| Primary cell culture and maintenance
Primary human scleral fibroblasts were isolated and cultured as previously described. 4 Briefly, eyes from non-glaucomatous donors were received from the National Disease Research Interchange, dissected, and 1 Â 1 mm scleral segments were placed in complete 1640 RPMI media supplemented with 10% FBS, non-essential amino acids, 1% pen-strep, and sodium pyruvate, inside collagen-coated 35 mm Petri dishes for 14 days. Following this, cells were passaged and maintained in DMEM supplemented with 1% FBS, 1% pen-strep, and sodium pyruvate. 1.5 Â 10 5 cells were seeded onto each scaffold and allowed to acclimatize for a period of 24 h prior to stimulation. Cells were used between passage 3 and 8 for all experiments. Z-stacks were 3D reconstructed using IMARIS (Oxford Instruments, UK) to generate false volume images.

| Atomic force microscopy
For roughness characterization, samples were manufactured as  Kit was used to obtain RNA from tissue using manufacturer's instructions. Fifty ng of cDNA was analyzed per sample along with primer sequences listed in Table S3. To perform all qRT-PCR analysis, the ΔΔCt method was used. Expression was normalized to levels detected in untreated fibroblasts on smooth surfaces for in vitro studies and healthy, non-operated conjunctiva, and sclera tissue in rabbit studies.

| PET GDI manufacturing
PET GDIs were manufactured as described previously. 21  Black light illumination was used to photograph whether the fluorescein flowed out of the anterior chamber through a patent GDI.

| Histology and immunostaining tissue
At the endpoint of the animal studies, rabbits were euthanized and eyes were enucleated. Conjunctival and scleral issue surrounding the implant was dissected and stored in formalin until further processing.
Tissues were embedded in paraffin and allowed to cool. Five μm sections of tissues in paraffin blocks were mounted on microscope slides. xylene. Subsequently, tissue sections were exposed to an ethanol gradient (100%, 90%, 80%, 70%, 50% ethanol in water) followed by rinsing gently in deionized water. Following this, antigen retrieval was performed using a trypsin-based antigen retrieval kit (Abcam, Cambridge, UK) following the manufacturer's protocol. Following antigen retrieval, tissue was stained overnight with a mouse monoclonal alpha-smooth muscle actin (αSMA) (1A4 [asm-1]) primary antibody at 4 C. After washing three times for 5 min each in PBST, tissue was incubated at room temperature with a secondary antibody conjugated to Alexa Fluor 555 for 2 h. Finally, samples were counterstained with SYTOX red and mounted on glass slides using an antifade mounting solution.

| Image analysis
H&E and MT images were obtained using a NIKON (Eclipse Ni) light microscope and NIS elements imaging software (Nikon, IL, USA, version 5.11.0). RGB histograms were generated using ImageJ (version 1.53k14) and collagen intensity was recorded from the blue channel.
Immunofluorescence images were obtained using a confocal laser scanning microscope (Ziess, LSM710), processed using the Zen imaging software (Blue, version 3.4) and quantified using ImageJ. Image analysis and quantification were conducted in a masked manner using at least three images per animal and averaged.

| Statistical methods
All statistical analyses were performed using GraphPad Prism (ver. 9.0). To compare data sets with two groups, a two-tailed student's t test was used, and to compare data sets with multiple groups, analysis of variance (ANOVA) was used. Unless otherwise mentioned, data are represented as mean ± SD. All experiments were performed with at least three replicates. For qRT-PCR analysis, three independent experiments were performed and averaged. For cell cycle analysis, histograms are from six independent samples that were pooled and analyzed. All collected data were included in the analyses.

DATA AVAILABILITY STATEMENT
All data, code, and materials used in this study and analyses are available within the paper or supplementary materials.