Influence of Substrate Stiffness on iPSC-Derived Retinal Pigmented Epithelial Cells

Abstract

Retinal degenerative diseases are a major cause of blindness involving the dysfunction of photoreceptors, retinal pigmented epithelium (RPE), or both. A promising treatment approach involves replacing these cells via surgical transplantation, and previous work has shown that cell delivery scaffolds are vital to ensure sufficient cell survival. Thus, identifying scaffold properties that are conducive to cell viability and maturation (such as suitable material and mechanical properties) is critical to ensuring a successful treatment approach. In this study, we investigated the effect of scaffold stiffness on human RPE attachment, survival, and differentiation, comparing immortalized (ARPE-19) and stem cell-derived RPE (iRPE) cells. Polydimethylsiloxane was used as a model polymer substrate, and varying stiffness (~12 to 800 kPa) was achieved by modulating the cross-link-to-base ratio. Post-attachment changes in gene and protein expression were assessed using qPCR and immunocytochemistry. We found that while ARPE-19 and iRPE exhibited significant differences in morphology and expression of RPE markers, substrate stiffness did not have a substantial impact on cell growth or maturation for either cell type. These results highlight the differences in expression between immortalized and iPSC-derived RPE cells, and also suggest that stiffnesses in this range (~12-800 kPa) may not result in significant differences in RPE growth and maturation, an important consideration in scaffold design.

Graphical abstract

Graphical Abstract.

Significance Statement.

Replacing dysfunctional cells using supportive biomaterials is a promising treatment approach for retinal degenerative diseases. However, the independent effect of scaffold stiffness on stem cell-derived retinal pigmented epithelial cells (RPE) has not yet been characterized. Here, we compare the growth and maturation of stem cell-derived RPE grown on stiffnesses between 12 and 800 kPa. The results suggest that while there are no major differences in gene expression patterns between stiffnesses, RPE cells may exhibit a slight preference toward stiffer substrates, as shown by pigmentation and tight junction expression. These results help to further establish design parameters for retinal cell scaffolds.

Introduction

The retinal pigmented epithelium (RPE) is a vital cell layer of the eye responsible for the phagocytosis of shed photoreceptor outer segments, the control of nutrient and waste transport across the retina, and the reduction of UV and oxidative damage.^1-3 Many retinal degenerative diseases, including retinitis pigmentosa and age-related macular degeneration have pathophysiology related to dysfunction of the RPE. As few treatment options exist, strategies focused on the replacement of the RPE cells via cell transplantation are actively being developed. Early studies in this area showed that injection of RPE cells as a single-cell suspension results in poor cell survival and integration.^4-6 While transplantation of cultured RPE monolayers was shown to enhance cellular survival and integration, RPE cell monolayers are mechanically unstable and difficult to handle surgically.⁷ Thus, biomaterial delivery scaffolds, which support survival and promote polarity of RPE, have become a critical component of RPE cell replacement procedures.^4-6,8-11 However, understanding the ways in which scaffold composition, structure, and stiffness affect procedural parameters and outcomes such as graft handling, tissue integration, and the behavior and function of the RPE cells themselves is critical for ensuring the clinical safety and efficacy of RPE scaffolds.^12-16

Numerous studies in the past 2 decades have highlighted the effect of biomaterials’ stiffness on a multitude of cellular processes including cell viability, migration, proliferation, and differentiation in many systems and tissues.^17-21 Indeed, several previous studies have suggested that substrate stiffness can influence several characteristics of RPE function.^22-25 For example, phagocytotic activity of ARPE-19 cells has been shown to marginally decrease with increasing substrate stiffness (polyacrylamide, 0.5 vs 5 kPa).²² Furthermore, another study showed that although increasing substrate stiffness (polyacrylamide, 5 or 24 kPa vs ≥170 kPa) does not seem to affect ARPE-19 viability, stiffer substrates resulted in significantly higher levels of reactive oxygen species, which may suggest reduced function.²³ Moreover, increasing substrate stiffness (1200 vs 60 kPa) has been shown to increase ARPE-19 cell metabolic activity while also increasing the expression of pro-inflammatory cytokines IL6 and IL8, which are undesirable in RPE transplantation.²⁵ While these studies provide evidence that substrate stiffness is likely an important consideration for RPE cell function, the applicability of the results to RPE transplantation is limited, since ARPE-19 is a spontaneously immortalized cell line. While the use of ARPE-19 cells enables efficient and feasible in vitro experiments, immortalized cell lines may not adequately represent the phenotype and function of native RPE cells. Indeed, comparisons of gene and protein expression between ARPE-19 and native or cultured primary RPE cells suggest that there are major differences between these cell types.^26-30

Differentiated RPE cells derived from embryonic or induced pluripotent stem cells are the frontrunner for clinical transplantation and are a compelling research alternative to ARPE-19. There are now several established, publicly available RPE differentiation protocols,^31-34 and compared to ARPE-19, gene and protein expression levels of stem cell-derived RPE cells are more similar to native RPE.^28,29,34-36 Thus, stem cell-derived RPE cells could offer critical insight into how RPE cells respond to various scaffold characteristics, including stiffness. Although numerous studies have investigated the response of stem cell-derived RPE to various biomaterial compositions, far fewer report the independent effect of substrate stiffness on stem cell-derived RPE. In one study, higher stiffness (5.7 vs 2.4 MPa; collagen vitrigels) was reported to lead to decreased pigmentation and reduced expression of key RPE genes in stem cell-derived RPE.³⁷ However, current theory suggests that for a given tissue system, mimicking the native mechanical properties of the matrix may best promote resident cell survival and function. The compressive stiffness of the chorioretinal complex is reported to be approximately 10 kPa,^38-42 while most RPE scaffolds are orders of magnitude stiffer (e.g., 600-6000 kPa).^37,43,44 The effect of substrate stiffness, within the physiological range of native chorioretinal tissue, on stem cell-derived RPE cells has not yet been reported.

In this study, we sought to determine the independent effects of physiologically relevant substrate stiffness (i.e., compressive modulus) on both ARPE-19 and induced pluripotent stem cell-derived RPE (iRPE) cells. We used a non-degradable elastomer, polydimethylsiloxane (PDMS), with varying base-to-curing agent ratios (i.e., cross-linking densities) to achieve a broad range of mechanical properties but maintained relatively consistent substrate composition and surface chemistry. RPE viability and development were assessed for both cell lines using light microscopy, immunostaining, and gene expression analysis across several weeks. Our results help to further define scaffold stiffness as a vital design parameter for the development of RPE replacement scaffolds and continue to lay important groundwork for future studies in tissue engineering for retinal degenerative diseases.

Materials and Methods

PDMS Substrates and Mechanical Properties

PDMS substrates were prepared using commercially available Sylgard 184 (Dow Corning, Midland, MI) with various base-to-curing agent weight ratios (80:1, 60:1, 50:1, 40:1, 30:1, and 10:1). PDMS solutions of each base-to-curing agent weight ratio were mixed in 2 mL microcentrifuge tubes, vortexed, and briefly centrifuged to remove air bubbles. For mechanical property assessment, samples were then fabricated using a mold consisting of a glass slide and a stainless-steel washer (11 mm inner diameter with a height of 2 mm). The washer was placed onto the slide and binder clips were used to clamp the parts together. One hundred and fifty milliliters of the PDMS solution was slowly pipetted into the mold, polymerized at 60 °C for 3 hours, and left to finish curing at room temperature overnight. PDMS samples were then removed from the molds and trimmed to size (8 mm diameter) for mechanical property testing.

Mechanical properties of the PDMS substrates were measured using a Kinexus Ultra + rheometer (Netzsch, Selb, Germany) with an 8-mm-diameter upper plate. Both shear and compressive modulus measurements were conducted to calculate the true compressive modulus, as the sample’s dimensions resulted in the system behaving as a constrained compressive test. For shear measurements, a single frequency strain-controlled oscillatory sequence was performed. Based on the linear viscoelastic region of PDMS (data not shown) a frequency of 1 Hz and strain rate of 1% were chosen. The gap height was set to 2 mm initially, then finely tuned to the exact height of each sample, based on the normal force measurement from the upper geometry. The dynamic strain sequence, which recorded the corresponding force for each strain condition, was run for 1 minute; the shear storage modulus was calculated by the software. Once this sequence was completed, the PDMS samples were compressed (normal to the sample surface) at a rate of 0.01 mm/s to a final compression of 0.2 mm, with normal force and upper geometry position recorded every 0.1 seconds. Compressive stress and strain for each trial were calculated and the data were fit to a linear regression to determine the sample’s apparent compressive modulus. Constrained compression was accounted for using the following set of equations:

$$\begin{array}{l}\frac{E_a}{E}=\frac{1+3v\left(\frac{1-v}{1+v}\right)S^2}{1+3v(1-2v)S^2}\end{array}$$

$$\begin{array}{l}E=2G(1+v)\end{array}$$

Where E_a is the apparent compressive modulus, E is the true compressive modulus, ν is Poisson’s ratio, G is the shear modulus, and S is the ratio of the sample’s radius to its height.⁴⁵

Based on the results of mechanical testing, PDMS substrates with base-to-curing agent ratios 60:1, 30:1, and 10:1 were chosen for the cell studies, as they corresponded to stiffnesses of experimental interest: the lowest reported compressive modulus of the RPE-choroidal layers is 10-20 kPa (60:1), our previous work has demonstrated that ~100 kPa is a lower limit for a scaffold to be handled effectively during a retinal transplantation surgery (30:1),⁴⁶ and many currently described retinal scaffolds have moduli in or near the MPa range (10:1). To evaluate the stability of PDMS substrates over the length of the experiment, samples were fabricated at the base-to-curing agent ratios of interest, incubated in cell culture media for 4 weeks, and the mechanical properties were measured as described above.

Preparation of PDMS Substrates for Cell Seeding

PDMS substrates were prepared using 10:1, 30:1, and 60:1 base-to-curing agent ratios of Sylgard 184. The polymer base and cross-linking agent were mixed in microcentrifuge tubes and briefly centrifuged to remove air bubbles. Next, 120 µL of each PDMS mixture was pipetted into the wells of a 24-well cell culture plate for a final substrate thickness of approximately 0.5 mm. Samples for immunocytochemistry were made in glass bottom plates, while samples for SEM were made on 20 mm diameter glass coverslips. The PDMS-coated plates were placed in a vacuum chamber for 5 minutes to remove air bubbles, incubated at 60 °C for 3 hours, then left at room temperature overnight for the polymerization reaction to complete. Prior to cell seeding, substrates were disinfected by soaking in 70% ethanol for at least 1 hour, followed by several rinses with sterile water. Before cell culture, PDMS-coated plates were exposed to oxygen plasma (Harrick Plasma PDC-001) to improve protein adsorption and subsequent cell attachment. Briefly, samples were exposed to oxygen plasma at a flowrate of 25 mL/minute at 30 W power for 10 minutes. Substrates and controls were then immediately coated with Cell Basement Membrane (ASC-3035, ATCC) at a concentration of 0.15 mg/mL, left at 4 °C overnight, and incubated at 37 °C for 2 hours prior to cell seeding.

Cell Culture

APRE-19 cells (CRL-2302, ATCC) were plated from frozen and grown on tissue culture plastic until the experiments began. For all experiments, ARPE-19 cells were used between passages 7 and 10. On day 0 of the experiment, cells were plated at 1 × 10⁵ cells/cm² onto the PDMS substrates and grown for 7 or 14 days. Media consisted of DMEM/F12 with phenol red (Life Technologies), supplemented with 10% heat-inactivated FBS (Life Technologies) and 0.2% primocin (Invivogen). Media was changed every 2-3 days.

iRPE cells were differentiated following the established protocol from Foltz and Clegg.³¹ Briefly, human iPSCs derived from a donor with no known ocular disease (derived in-house, previously published)⁴⁷ were grown to 80% confluence in Essential 8 media (Life Technologies), passaged at a surface area ratio of 1:4, and cultured in a variety of growth-factor enriched media over a 14-day period as previously described.³¹ Growth factors included: nicotinamide and CHIR99021 (Sigma); rhNoggin, rhDKK1, IGF1, and basic FGF (R&D Technologies); and Activin A and SU5402 (STEMCELL Technologies). The base RPE differentiation media consisted of DMEM/F12 (Life Technologies) with 1% N2 (Life Technologies), 1% non-essential amino acids (Life Technologies), and 2% B27 (Life Technologies) supplements.

Following the differentiation, iRPE cells were passaged at 1 × 10⁵ cells/cm² and cultured in RPE supporting media for an enrichment period of 28 days. RPE supporting media consisted of X-VIVO 10 (Lonza) with 0.2% primocin, and Y-27632 Rock inhibitor (ASC-3030, ATCC) added for 48 hours after passaging. Following this enrichment period, cells were then plated at 1 × 10⁵ cells/cm² onto the experimental substrate groups (varying stiffnesses) and cultured in RPE supporting media for 2 or 4 weeks before analysis. For both cell types, bright field images were collected at each time point using an EVOS epifluorescent microscope (Thermo Fisher Scientific).

Immunocytochemistry

At each time point, RPE cells grown on substrates of different stiffnesses were washed with 1× PBS and fixed with 4% paraformaldehyde for 30 minutes. The fixing solution was removed, SuperBlock T20 blocking buffer (Thermo Fisher Scientific) was added, and the samples were incubated at room temperature for at least 1 hour. The primary antibody, anti-human ZO-1 rabbit polyclonal (Novus Biologicals, NBP1-85047), was diluted at 1:100 in SuperBlock blocking buffer (Thermo Fisher Scientific), added to the wells, and incubated overnight at 4 °C. Primary antibody solution was then removed, and cells were washed 3 times with PBS. Secondary antibody, goat anti-rabbit Alexa Fluor 647 (Invitrogen), was diluted at 1:1000 in SuperBlock, and nuclear stain DAPI (Invitrogen) was added at a dilution of 1:2000. The solution of secondary antibody with nuclear stain was added to the wells, incubated at room temperature for 2 hours, and washed 3 more times in PBS. Finally, 2 drops of Aqua-Mount (Epredia) were added, and fluorescent images were captured using an inverted Nikon Ti2 Eclipse A1 confocal microscope.

Gene Expression

Total RNA from the RPE cells cultured on different stiffnesses was extracted using RNeasy kit according to the manufacturer’s instructions (Qiagen, Germantown, MD). Following the validation of purity and yield using a NanoDrop One (Thermo Fisher Scientific), cDNA was transcribed using SuperScript VILO cDNA Synthesis kit (Invitrogen) according to the manufacturer’s instructions and stored at −20 °C until qPCR. Quantitative PCR was performed with PrimeTime Gene Expression Master Mix (IDT, Coralville, IA) on a QuantStudio 3 (Applied Biosystems, Waltham, MA). The amplification procedure was as follows: 50 °C for 2 minutes, 95 °C for 10 minutes, and then 40 cycles of 95 °C for 15 seconds followed by 60 °C for 60 seconds. 18sRNA was used as a housekeeping gene control and fold change was calculated using the ΔΔCt method. Additional information about each gene and primer is available in Supplementary Table S1.

RPE Cell Ultrastructure

Samples were fixed in glutaraldehyde for 30 minutes, then stored in PBS until being stained with osmium tetroxide and dehydrated with increasing concentrations of ethanol following a standard protocol, described elsewhere.⁴⁸ Prior to imaging, a gold-palladium coating was applied using an argon beam K550 sputter coater (Emitech). SEM images were captured using a Hitachi S-4800 SEM (Hitachi High-Technologies) at an accelerating voltage of 1 kV. All sample preparation and SEM imaging were performed at the University of Iowa Central Microscopy Facility.

Statistical Analysis

The compressive modulus of each PDMS base-to-curing agent ratio was measured in triplicate for 2 separately mixed solutions (n = 6 total), while gene expression assays were performed in triplicate (n = 3). For compressive modulus data, one-way ANOVA and Tukey post hoc tests, each at a CI of 95%, were used to determine statistical significance. For gene expression analysis, a 2-way ANOVA was performed for each gene (factors: time and stiffness), followed by Sidak’s multiple comparisons for significant factors, each at a CI of 95%. GraphPad prism was used for all statistical analyses and generating plots. Data are reported as mean ± SD.

Results

Prior to seeding, we characterized the compressive modulus of PDMS substrates with varying base-to-curing agent ratios. As expected, increasing the ratio between the base and curing agent resulted in samples with lower stiffnesses, with an overall compressive moduli range of approximately 10-800 kPa (Supplementary Fig. S1A). The mechanical properties remained constant over time by measuring compressive modulus after a 4-week incubation in cell culture media. While a slight increase in moduli values and greater variability were observed for all samples, the pre- and postexposure values were not statistically different from each other (Supplementary Fig. S1B).

To lay the foundation for understanding how substrate stiffness influences RPE cell fate, we began our study using ARPE-19 cells, which are well-characterized and ubiquitous in retinal research. Cells were plated on PDMS substrates with increasing stiffnesses (12, 100, and 800 kPa) and evaluated by light microscopy at 1- and 2-week time points. Regardless of stiffness, all substrates supported the formation of a complete monolayer of tightly packed cells (Fig. 1). Across all stiffnesses, there was little evidence of cells exhibiting a classic cobblestone morphology, although cells did seem to become more tightly packed between weeks 1 and 2. However, substrate stiffness did not appear to impact cell packing, morphology, or degree of pigmentation of ARPE-19 cells (Fig. 1).

Figure 1.

Representative light microscopy images of ARPE-19 cells after 1 and 2 weeks on increasing substrate stiffnesses. Scale bars = 400 μm, inset scale bars = 100 μm. Abbreviation: TCP, tissue culture plastic.

We determined whether subtle, molecular changes in ARPE-19 occur on increasing substrate stiffness by assessing the expression of an array of RPE-specific genes using qPCR. Relative to the day 0 control, only slight differences in expression were observed at the week 1 time point (Fig. 2A). At this time point, expression of pigmented epithelial-derived factor (PEDF) seemed to be upregulated for cells on PDMS substrates compared to tissue culture plastic. By week 2, this differential expression was much more pronounced, and expression of the proliferative marker, Ki67, was also dramatically downregulated on TCP compared to PDMS substrates (Fig. 2B). These observations were corroborated by statistical analysis of each independent gene (Supplementary Table S2): stiffness was a significant factor for PEDF expression (P < .05), and both stiffness and time were significant factors for expression of Ki67 (P < .01 and P < .001, respectively). Furthermore, expression of RLBP1 was significantly upregulated at week 2 compared to week 1 (P < .001; Fig. 2B; Supplementary Table S2). RPE65 and BEST1 were not expressed at detectable levels in any group over the course of the experiment, including the day 0 reference group (which is characteristic of this cell line).

Figure 2.

Relative fold change of gene expression in ARPE-19 cells cultured for 1 or 2 weeks, compared to day 0 ARPE-19 with 18sRNA as a reference gene. N = 3; data is shown as mean ± SD. RPE65 and BEST1 genes had no measurable expression levels in all groups, including day 0. Abbreviation: TCP, tissue culture plastic.

The tight junction protein ZO-1 is a classic marker of epithelial cell identity and thus, we evaluated ARPE-19 changes in response to substrate stiffness at the protein level by immunocytochemistry of ZO-1 (Fig. 3). ZO-1 was expressed and localized to the membranes of ARPE-19 cells on all substrates, including TCP. Qualitatively, differences in localization across samples were apparent, with TCP and 12 kPa samples (Fig. 3A, 3D) exhibiting denser cell packing (i.e., number of nuclei per field) and greater membrane localization than 800 or 100 kPa samples (Fig. 3B, 3C). However, these observations are based on incomplete information, as our number of replicates for this assessment was low, and thus we did not quantify ZO-1 expression or cell packing density.

Figure 3.

Representative immunocytochemistry images of ARPE-19 cells on increasing substrate stiffness showing ZO-1 expression (red) with nuclear counterstain (DAPI, blue). (A) Tissue culture plastic, (B) 800 kPa, (C) 100 kPa, (D) 12 kPa. Scale bar = 100 μm.

Given the known limitations of ARPE-19 cells (e.g., lack of expression of RPE cell markers including BEST1 and RPE65), we focused the next phase of our study on a cell type that is more representative of native RPE cells: human iPSC-derived RPE (iRPE). Briefly, we seeded iRPE cells on PDMS substrates with different stiffnesses (12, 100, and 800 kPa) as well as a TCP control group, as in the ARPE-19 experiments. Regardless of substrate stiffness, iRPE cells demonstrated tight packing with regions of cobblestone morphology by week 2 (Fig. 4). At week 4, pigmentation was noted in all samples (Fig. 4, arrows), and seemed to be more frequent for higher stiffness substrates (TCP and 800 kPa) compared to softer substrates (100 and 12 kPa).

Figure 4.

Light microscopy images of iRPE cells after culturing for 2 and 4 weeks show little morphological differences between substrate stiffnesses, arrows show areas of observed pigmentation. Scale bars = 400 μm, inset scale bars = 100 μm. Abbreviation: TCP, tissue culture plastic.

We evaluated stiffness-induced iRPE molecular changes by quantifying gene expression via qPCR on a panel of RPE-specific target genes (Fig. 5). At week 2, most genes were expressed at a similar level to the day 0 control, except for BEST1, which seemed to be slightly upregulated for all substrates (Fig. 5A). Although expression of BEST1 seems to be upregulated in the 100 kPa samples compared to the other substrates, high sample variability at this time point complicates the interpretation of this observation. At week 4, the upregulation of BEST1 compared to the day 0 control is more pronounced than at week 2 for all substrates (Fig. 5B). Compared to week 2, Ki67 was also significantly downregulated (P < .05) and RLBP1 was significantly upregulated (P < .05) for all substrates (Fig. 5B, Supplementary Table S3). Stiffness was not a significant factor for the expression of any of the genes we evaluated (Supplementary Table S3).

Figure 5.

Relative fold change of iRPE cells cultured for 2 or 4 weeks compared to day 0 iRPE, using 18sRNA as a reference gene. N = 3; data is shown as mean ± SD. Abbreviation: TCP, tissue culture plastic.

To assess tight junction integrity and ultrastructural characteristics of iRPE on different stiffnesses, we performed immunocytochemistry and scanning electron microscopy (Fig. 6). On all substrates, iRPE cells formed a tightly packed monolayer with hexagonal appearance, as shown by immunocytochemistry. Cells cultured on the softest substrate (12 kPa) had less ZO-1 expression and membrane localization compared to other substrates, indicating a potentially less mature RPE phenotype. Ultrastructural analysis revealed that iRPE exhibited similar morphological features on all PDMS substrates, including tightly packed monolayers and microvilli, presumably indicating some level of polarity. Cracks between groups of cells were also observed in these images; these are typical, and likely result from necessary sample dehydration during SEM preprocessing. In this case, differences between the substrate and cells in terms of water content and mechanical properties likely contribute to this phenomenon. For example, the observed larger cracks (i.e., more exposed substrate) in the higher stiffness substrates (100 and 800 kPa) corresponds to a larger mismatch in properties, with a stiffer PDMS substrate being less compliant than the cell monolayer.

Figure 6.

Representative confocal and SEM images of iRPE cells cultured for 4 weeks on increasing substrate stiffness. In confocal images: blue represents DAPI, red represents ZO-1, and scale bar = 50 μm. In SEM images, scale bar represents 10 μm. Abbreviation: TCP, tissue culture plastic.

Discussion

Cell support scaffolds are necessary to facilitate retinal transplantations. However, tuning scaffold parameters to ensure efficient and effective cellular growth and function remains an ongoing process. In this work, we investigated the independent effect of scaffold stiffness on ARPE-19 and iRPE cell growth and maturation. We used a nonbiodegradable synthetic polymer, PDMS, to ensure chemical functionality remained similar between groups. While we do not recommend PDMS for clinical transplantations, its use as a model polymer system enables us to isolate the effects of substrate stiffness on RPE cells. Testing across a range of stiffnesses (12-800 kPa) without substantially altering the chemistry of the substrate enables a comprehensive view of substrate mechanical properties while also focusing on the less studied, lower end of reported stiffnesses for the RPE-choroidal complex. Moreover, PDMS’ constant mechanical stiffness over time is an advantage compared to other material systems (e.g., collagen) that are often used to probe the impact of stiffness on cell fate.

Well-characterized immortalized cell lines such as ARPE-19 offer researchers a cost-effective and easy-to-use system for in vitro experiments. As such, in the first part of our study, we used ARPE-19 cells as a model system to begin to understand how substrate stiffness influences RPE cell growth and function. We did not observe major differences in cell morphology, as ascertained using light microscopy (Fig. 1): ARPE-19 cells grew well on all substrates, forming a monolayer by week 1, yet lacked typical RPE cobblestone organization. However, this morphology, as well as the lack of pigmentation, aligns with several other studies of ARPE-19 cells in standard media at these time scales.^29,49-51 Immunostaining of ARPE-19 cells revealed some potential, slight differences in cell packing and tight junction maturation (ie, ZO-1 membrane localization) between different stiffnesses (Fig. 3). However, drawing conclusions from this data would be conjectural due to a low number of replicates and lack of quantification. On the molecular level, expression of the proliferative marker Ki67 decreased and expression of RLBP1, which encodes for an essential visual cycle protein, increased over time (weeks 1-2) for all groups (Fig. 2). These changes were likely due to contact inhibition, which involves tight junctions and their interactions with the cytoskeleton, and ARPE-19 maturation, respectively. Despite the lack of apparent morphological and immunocytochemical differences between groups, stiffness did have a significant impact on the molecular level in some cases. Namely, the expression of Ki67 and PEDF, which encodes for the secreted protein PEDF, were upregulated in the PDMS groups compared to tissue culture plastic (Fig. 2). These differences could have resulted from a difference in stiffness, as polystyrene TCP is reported to have a stiffness on the order of GPa.⁵² However, it is important to note that substrate chemistry may be a confounding factor in this case, which complicates the interpretation of the results.

Although the results of our ARPE-19 experiments disagree with the current general theory regarding cells and substrate stiffness, they align with some previous studies involving this specific cell type. For example, Chen et al²³ reported that while substrate stiffness (5-170 kPa) impacts levels of reactive oxygen species in ARPE-19 cells, it does not substantially affect cell viability or morphology. Additionally, White et al²⁵ showed that the genes encoding IL8 and IL6 are differentially expressed in ARPE-19 cultured on 60 versus 1200 kPa substrates, but expression of RLBP1 is not different between groups. Taken together, these results suggest that substrate stiffness may influence some aspects of ARPE-19 cell behavior and function, but that the differences with respect to characteristic RPE markers are either not significant or not detectable. Drawing overarching conclusions about the effect of substrate stiffness on ARPE-19 cell fate is further convoluted by the propensity for ARPE-19 cells to “de-specialize” as they undergo epithelial-to-mesenchymal transition during standard culture and repeated passaging. In fact, in our current study, neither RPE65 nor BEST1 were expressed at detectable levels by ARPE-19 cells in any experimental group; these and other key gene expression differences between ARPE-19 and native RPE cells have been reported previously.^27,29 Thus, our study serves to further highlight that ARPE-19 cells may not be the most suitable choice for understanding RPE cell interactions with engineered substrates.

In the second phase of our study, we sought to determine the influence of substrate stiffness on iRPE, which, compared to ARPE-19, are reported to be more similar to native RPE with respect to the expression of characteristic RPE genes, cellular morphology, and pigmentation. Briefly, we cultured differentiating iRPE cells on PDMS substrates at 12, 100, and 800 kPa compressive modulus for up to 1 month, and characterized morphological and molecular changes using light microscopy, SEM, immunocytochemistry, and qPCR. After 2 and 4 weeks, iRPE cells formed a complete monolayer with distinct cobblestone morphology, indicative of RPE maturation, on all substrates. We did not observe any meaningful differences in morphology between groups at week 2. However, at week 4, there was a noticeable increase in pigmentation for iRPE cultured on stiff (TCP and 800 kPa) compared to soft (12 and 100 kPa) substrates (Fig. 4). In one previous study, iRPE differentiated on stiff (5700 kPa) substrates had less pigmentation than soft (2400 kPa) substrates.³⁷ Given that the stiffnesses used in that study are analogous to or greater than TCP, our collective results suggest that substrate stiffnesses between 800 and 2400 kPa may encourage stem cell-derived RPE to mature to greater extent than stiffer or softer substrates. Immunostaining of iRPE highlighted some potential, subtle differences across substrate stiffnesses (Fig. 6). Namely, the softest substrate (12 kPa) appeared to have more disorganized, less membrane-localized ZO-1 expression compared to all other groups, perhaps indicating a lower limit of suitable stiffness for iRPE maturation. Ultrastructrual analysis via SEM indicated that iRPE cells comprised normal cell bodies with microvilli on all PDMS groups, similar to the morphology of native RPE.^37,43,53 While cracking of the cell monolayer was more prominent on stiffer (100 and 800 kPa) substrates, this is likely a result of the fixing and desiccation process and a mismatch of properties between the PDMS and iRPE cells, as opposed to differences in the iRPE between samples.⁴³ At the molecular level, iRPE expression of Ki67 significantly decreased between weeks 2 and 4, while expression of RLBP1 significantly increased. In addition, the expression of BEST1, which encodes for the RPE-characteristic ion transport protein bestrophin-1, appeared to be upregulated at week 2 and to further increase at week 4, although these differences across time were not statistically significant. Decreased proliferation and increased expression of RLBP1 and BEST1 were expected, as they are associated with typical iRPE differentiation and maturation.^27,54

Despite apparent changes in the pigmentation of iRPE cells and possible differences in ZO-1 localization on varying substrate stiffnesses, we did not detect any significant differences in gene expression that were attributable to substrate stiffness (Fig. 5). One potential explanation for this discrepancy is related to the heterogeneity of cells on the substrate in our study and localized differences in pigmentation. When compared to the entire image area, total areas of pigmentation were still relatively small across all groups. However, RNA was collected from a large surface area of cells, potentially causing a dilution of any relevant upregulated genes related to maturation. Although the heterogeneity we observed is common for iRPE, it does highlight the limitations of using traditional qPCR in this context. Single-cell RNA sequencing or RNA in situ hybridization would likely shed additional light on localized differences in maturation as a result of scaffold stiffness. Additionally, it is possible that the time points of RNA collection in our study did not align with the transcriptional changes that led to enhanced iRPE maturation, or that additional relevant regulatory mechanisms (e.g., RNA silencing) were not captured by the assays we used here.

In general, the results of our iRPE study do not support our original hypothesis, contradicting with the results from other tissue systems that suggest that differences in substrate stiffness in this range influence cell morphology and molecular identify. However, the results do suggest there is a lower limit of advantageous substrate stiffness around 12-100 kPa, as evidenced by less pigmentation and less ZO-1 membrane localization. Furthermore, while the range of substrate stiffnesses included in our study encompassed several orders of magnitude, it may not have included the upper limit of advantageous substrate stiffness in a way that facilitates separation of this variable from substrate chemistry. Another limitation of our study is that prior to seeding PDMS substrates with iRPE cells, we followed the standard practice of a 28-day enrichment period, in which iPSCs are differentiated toward RPE on tissue culture plastic. It may be that cells strongly “commit” to RPE fate during this period, causing underlying bulk substrate stiffness to become less relevant than cell-cell connections and cell-mediated extracellular matrix properties. Future studies designed to understand the effects of substrate stiffness on initial RPE differentiation would lend clarity to this area, as would examination of any differences between the composition of new ECM deposited by iRPE. Additionally, functional assays such as transepithelial resistance or outer segment phagocytosis would further elucidate the influence of substrate stiffness on iRPE growth, maturation, and function. Furthermore, while this study investigates the influence of substrate stiffness on cellular response, the mechanical properties of the scaffold could impact the subretinal space post-transplantation. For example, compared to a soft scaffold, a stiff scaffold may result in more mechanical irritation because of its mismatch with the relatively soft retina, potentially leading to an exacerbated immune response. Determining the impact of scaffold stiffness on retinal tissue response is a complex problem that warrants further investigation.

Conclusion

This study provides further evidence that ARPE-19 and iRPE cells have distinct phenotypes, and that their differences are important in the context of understanding RPE cell interactions with engineered biomaterials. Furthermore, our results show that while substrate chemistry may play a role, substrate stiffnesses within the test range (12-800 kPa) do not drastically impact ARPE-19 growth and maturation. However, iRPE cells seemed to preferentially mature on substrates with higher stiffnesses (100 kPa or greater). While these results provide evidence that RPE are relatively insensitive to substrate stiffness in the sub-MPa range, further investigation with a wider range of stiffnesses, as well as RPE functional assays, are needed to fully understand the influence of substrate stiffness on iRPE growth, maturation, and function.

Supplementary Material

Supplementary material is available at Stem Cells Translational Medicine online.

Conflict of Interest

K.W. declared research funding from Photopolymerization Industry University Cooperative Research Center (IUCRC). The other authors declare no potential conflicts of interest.

Author Contributions

R.W.: conception and design, collection and/or assembly of data, data analysis and interpretation, manuscript writing, final approval of manuscript; B.T.: conception and design, financial support, provision of study material, data analysis and interpretation, final approval of manuscript; K.W.: conception and design, financial support, data analysis and interpretation, manuscript writing, final approval of manuscript.

Data Availability

The data underlying this article will be shared on reasonable request to the corresponding author.