Bridging 2D and 3D culture: probing impact of extracellular environment on fibroblast activation in layered hydrogels

Abstract

Many cell behaviors are significantly affected by cell culture geometry, though it remains unclear which geometry from two- to three-dimensional (2D to 3D) culture is appropriate for probing a specific cell function and mimicking native microenvironments. Toward addressing this, we established a 2.5D culture geometry, enabling initial cell spreading while reducing polarization to bridge between 2D and 3D geometries, and examined the responses of wound healing cells, human pulmonary fibroblasts, within it. To achieve this, we used engineered biomimetic hydrogels formed by photopolymerization, creating robust layered hydrogels with spread fibroblasts at the interface. We found that fibroblast responses were similar between 2D and 2.5D culture and different from 3D culture, with some underlying differences in mechanotransduction. These studies established the 2.5D cell culture geometry in conjunction with biomimetic synthetic matrices as a useful tool for investigations of fibroblast activation with relevance to the study of other cell functions and types.

INTRODUCTION

In vitro cell culture models have been useful tools in establishing the importance of extracellular cues and ‘outside-in’ signaling in the function and fate of a variety of cell types, including stem cell differentiation, cancer metastasis, and fibroblast activation.^1,2 Many of the earliest studies of mammalian cell responses to extracellular cues within engineered environments were conducted in two-dimensional (2D) culture, partly owing to limited material cytocompatibility and the difficulty of imaging, amongst other assays, in three-dimensional (3D) culture.³ Over the past decade, a variety of cytocompatible polymerization methods have been established, for example, chain and step growth polymerization of biocompatible polymers, biopolymers, and peptides appended with different reactive handles for the formation of synthetic and hybrid hydrogels, enabling the successful encapsulation and 3D culture of many cell types.⁴ With the accessibility and precise property control afforded by such systems, 3D culture is increasingly becoming the standard for many in vitro biological studies. This paradigm shift also is partly driven by the increased understanding of microenvironments in the body and the need to better mimic them in studies of specific biological processes. For example, in vivo, wound healing cells that reside in interstitial connective tissues, like fibroblasts and stem cells, interact with the extracellular matrix (ECM) on all sides, and generally, 3D cell culture is thought to more faithfully recapitulate this less polarized in vivo environment. However, as 3D cell culture models have become more ubiquitous, the geometry in which cells are cultured, amongst other properties, has been observed to have a significant effect on cell phenotype.⁵ An outstanding question in the field has emerged: are the differences observed between cells cultured in 2D and 3D culture the result of 3D cell culture more accurately reflecting the in vivo environment, or could some of these differences partly be artifacts of specific 3D culture systems? For example, cell confinement (e.g., high matrix crosslink density, small pore size, slow or no degradability) could limit cell spreading, proliferation, migration, or traction force generation, influencing by design or convoluting non-specifically observations of other cellular processes.⁶ Toward addressing this, in this work, we aimed to examine the effects of cell culture geometry on fibroblast activation, a cell response that is mediated by mechanotransduction and relevant in fibrotic disease, and establish materials approaches for multidimensional culture studies that bridge between 2D and 3D culture using the power of photopolymerization.

Fibroblasts are present in many organs throughout the body and are key in the repair of injured tissues.⁷ When an injury occurs, fibroblasts activate, a term which refers to an increase in the synthesis of ECM and adoption of a contractile phenotype that aids in tissue remodeling. Fibroblasts activate in response to a host of extracellular cues, including cytokines released by immune cells and other fibroblasts, changes in protein composition, and changes in matrix mechanics.⁸ Activated fibroblasts, also referred to as myofibroblasts, that persist once the tissue is repaired are a key effector cell in the progression of fibrotic disease.⁹ These cells are often identified by increased alpha smooth muscle actin (αSMA) expression, yes-associated protein (YAP) nuclear localization, and related changes in gene expression.¹⁰ Understanding the nuances of ECM cues and related signaling in fibroblast activation is of great interest to the study of fibrosis and has been identified by the National Heart Lung and Blood Association as a key requirement for the development of better fibrosis treatments, motivating the need for improved model systems to study these complex processes.¹¹

In 2D culture, fibroblasts have been shown to activate in response to increased matrix ‘stiffness’,^12,13 for example, on collagen-functionalized polyacrylamide and RGD-functionalized PEG hydrogels (‘soft’ to ‘stiff’ as measured by Young’s modulus (E): E ~ 1 to 5 kPa and E ~ 7 to 30 kPa, respectively).^12,14 However, recently fibroblast activation has been observed to decrease in 3D culture in similarly stiff cell-degradable PEG-peptide-based matrices (soft to stiff: E ~ 0.24 and 4 kPa to 13 kPa).¹⁵ Fibroblast activation thus is one of several cell behaviors, including cell spreading and proliferation¹⁶ and stem cell differentiation,¹⁷ that have been shown to be responsive to cell culture geometry.¹⁸ Many possible explanations of these observed differences have been explored. In some cases, cell proliferation has been shown to depend on the ability of the cell to degrade or remodel the surrounding matrix in 3D culture.¹⁶ Further, many cell types initially adopt a spread morphology in 2D culture and a rounded morphology in 3D culture with synthetic extracellular matrices, and cell shape has been shown to affect stem cell differentiation.¹⁹ Finally, cell polarization differs between the two geometries, leading to differences in cell morphology and cytoskeletal organization that may influence observed cell functions.²⁰ In vivo, fibroblasts are observed to adopt a spindle-like morphology and activate in response to increased matrix stiffness, which is more similar to the responses observed in 2D culture, but also interact with proteins on all sides, as they would in 3D culture. Approaches are needed for decoupling the effects of cell spreading/shape and cell polarization within culture models and for determining how these two properties affect fibroblast function. Development of such approaches can provide insights into which aspects of the in vivo environment are most important to replicate when testing specific hypotheses and, more broadly, for creating improved in vitro model systems.

To this end, we set out to establish an approach for comparing fibroblast activation across different culture geometries. For ease of comparison to prior work and relevance for widescale use, we selected well-defined step growth PEG-peptide hydrogels as the base matrix, which have been utilized for both 2D and 3D culture of fibroblasts.^13,21–23 Utilizing the capabilities afforded by photopolymerization, we then established an approach for creating a layered hydrogel geometry with these synthetic matrices, where cells are cultured at the interface between layers to create a 2.5D culture for comparison to and as a bridge between 2D and 3D cultures (Figure 1). The term ‘2.5D’ has been used for a variety of cell culture geometries that are not fully 3D cell encapsulations, and these systems been deployed for the culture of variety of cells types including hepatocytes, cancer cells, and fibroblasts.^20,24–28 In 2.5D cell culture, fibroblasts can spread, allowing them to adopt a similar shape as is observed in 2D culture, but also interact with the hydrogel from multiple sides with different degrees of confinement based on the system used, which is more similar to 3D cell culture. For example, one of the more frequently deployed 2.5D approaches for the culture of fibroblasts is ‘sandwiching’ the cells between two non-interacting hydrogel layers, allowing cell spreading, sensing of the matrix on both sides in the z-direction, and unrestricted movement in the x-y direction along the interface.^20,24,25,27 Used to a lesser extent for studies of mechanotransduction, approaches also have been developed to encapsulate fibroblasts and other cell types at the interface between two hydrogels. For example, fibroblasts or cancer cells have been encapsulated within a physically assembled interface between two hydrogel layers utilizing harvested protein-based systems, where cells sense the matrix from all sides and have some restriction to movement^26,28. However, to date, there are few comparisons amongst these geometries. We hypothesized that fibroblast activation and mechanotransduction-related pathways would be influenced by both cell culture geometry and matrix stiffness. To test this, we then cultured human pulmonary fibroblasts in the three cell culture geometries (2D, 2.5D, 3D) at a soft modulus and a stiff modulus. With this approach, we aimed to determine the differential effects of cell shape and cell polarization on cell response by comparing cells that were spread and polarized (2D culture), spread but less polarized (2.5D culture), and cells that were not spread and least polarized (3D culture). These studies both establish a new tool for multidimensional cell culture, which can be translated to the culture of other cell types, and provide insights into key differences and similarities in fibroblast response to matrix stiffness in different culture geometries, giving context to other studies of fibroblast activation in 2D and 3D culture and informing future investigations of fibroblast biology and fibrosis.

Figure 1. Approach for the culture of human pulmonary fibroblasts in 2D, 2.5D, and 3D geometries.

A) Hydrogels were formed from an 8-arm PEG-norbornene, a di-cysteine cell degradable peptide linker, and a mono-cysteine integrin-binding collagen mimetic peptide (POG)₃POGFOGER(POG)₄. These precursors at stoichiometric ratios in aqueous solution were photopolymerized with low cytocompatible doses of long wavelength UV light (10 mW/cm² at 365 nm for 5 minutes with photoinitiator LAP). B) Human pulmonary fibroblasts were cultured on top of hydrogels (2D culture), between two hydrogel layers (2.5D culture) using a new method established in this work, or encapsulated within hydrogels (3D culture).

METHODS

PEG-norbornene functionalization

Functionalization of multi-arm PEG with norbornenes was performed using an established protocol.²⁸ Briefly, N,N-Diisopropylcarbodiimide (DIC) (Sigma) was added to a round bottom flask (5X molar excess relative to the amine functional groups of 8-arm PEG amine). The flask was purged with argon, and then dichloromethane (DCM) was added by syringe while stirring to dissolve the DIC (roughly 30 mL DCM to 2 g DIC). Norbornene carboxylic acid (10X molar excess) (Sigma) was melted anhydrously and transferred by syringe to the flask containing DIC. 8-arm PEG amine (M_n ~ 40 kDa, JenKem) and 4-(dimethylamino)pyridine (DMAP, 0.5X molar excess) (Alfa Aesar) were added to a separate round bottom flask that was then purged with argon, and anhydrous DCM was added to dissolve the PEG and DMAP (roughly 40 mL DCM to 10 g PEG). This polymer solution was transferred to the flask containing the DIC and norbornene solution, which was purged with argon and stirred overnight. The functionalized PEG was purified by precipitation in ice-cold ethyl ether (10X volume) three times with centrifuging between precipitations, dried at room temperature overnight, resuspended in deionized (DI) water, and purified by dialysis against DI water at room temperature for approximately 2 days. The purified polymer was lyophilized, and functionality was determined using ¹H NMR in D₂O on an AV400 NMR spectrometer: 400 mHz δ 6.15 to 6.0 (m, 16H), δ 3.7 to 3.42 (m, 3636H) and disappearance of the peak at δ 3.12 to 3.02 (t, 16H). Typical reactions resulted in, on average, a functionality of ~ 85% norbornene per 8-arm PEG.

Peptide synthesis

Peptides were synthesized by solid phase peptide synthesis using Fmoc chemistry (Protein Tech Inc PS3, Protein Tech Inc Tribute, or CEM Liberty Blue). All amino acids (APPTec) were double coupled. The sequence GCRDVPMSMRGGDRCG was built on Rink Amide MBHA resin (Novabiochem), and CG(POG)₃POGFOGER(POG)₄G was built on H-Rink Amide-Chemmatrix resin (PCAS Biomatrix). Peptides were cleaved from resin using 95% w/v trifluoroacetic acid (Fisher Scientific), 2.5% w/v triisopropylsilane (Fisher Scientific), and 2.5% w/v DI water. The peptides were cleaved up to 3 hours, precipitated in in ice-cold ethyl ether three times, dried at room temperature overnight, resuspended in DI water, and purified using reverse phase high performance liquid chromatography (HPLC, C18 column, Water:ACN 95:5 to 45:55 over 45 min, Waters). The identity of the peptide sequence was confirmed using electrospray ionization (ESI) mass spectroscopy.

Lithium Phenyl-2,4,6-trimethylbenzoulphosphinate (LAP) synthesis

The photoinitiator LAP was synthesized by following a previously published protocol.²⁹ Briefly, dimethyl phenylphosphinate (3 g) (Sigma) and 2,4,6-trimethylbenzoyl (3.2 g) (Sigma) were added to a dry round bottom flask purged with argon and stirred at room temperature overnight. Lithium bromide (6.1 g) (Sigma) was dissolved in 2-butanone (100 mL) (Sigma) and then added to the round bottom flask. The reaction was heated to 50 °C using an oil bath for 10 minutes. The reaction mixture was cooled over several hours and subsequently filtered to recover the powder product, which was desiccated under vacuum. The identity of the product was verified by ¹H NMR in D₂O: 400 mHz δ 7.59 (m, 2H), 7.44 (m, 1H), 7.36 (m, 2H), 6.78 (s, 2H), 2.12 (s, 3H), and 1.90 (s, 6H).

Hydrogel polymerization

Hydrogels were polymerized using thiol-norbornene click chemistry.³⁰ Briefly, precursor solutions were prepared with different final concentrations of functional groups making ‘stiff’ and ‘soft’ hydrogels: i) 15 mM of norbornene groups from norbornene functionalized 8-arm PEG, 13 mM of thiol groups from the cell-degradable peptide linker GCRDVPMS↓MRGGDRCG, and 2 mM of thiol groups from the pendant peptide CG(POG)₃POGFOGER(POG)₄G (stiff composition forE ~ 10 kPa) or ii) 7 mM of norbornene groups from norbornene functionalized 8-arm PEG, 5 mM thiol groups from the cell-degradable peptide linker GCRDVPMS↓MRGGDRCG, and 2 mM of thiol groups from the pendant peptide CG(POG)₃POGFOGER(POG)₄G (soft composition for E ~ 3.5 kPa, which was the lowest modulus that could be achieved with robust and consistent hydrogel formation). The photoinitiator LAP (2.2 mM) was used in all formulations. For cell culture applications, hydrogel precursor solution was added to a sterilized syringe mold and polymerized with 10 mW/cm² at 365 nm light for 5 minutes. Syringe molds were made by removing the tip of a 1-mL syringe; these molds were sterilized before use by immersion in 70% ethanol for at least 15 minutes followed by air-drying in the biosafety cabinet.

For modulus measurements, hydrogels were polymerized in situ on an AR-G2 rheometer with UV-Vis light accessory and 8-mm geometry (10 mW/cm² at 365 nm) (TA Instruments; Omnicure Series 2000, Exfo). During polymerization, the shear modulus was measured over time at a strain of 1% and a frequency of 2 Hz for measurements within the linear viscoelastic regime. The final modulus (e.g., no measurable change in modulus with continued irradiation) was generally reached within two minutes of commencing irradiation. Young’s modulus (E) then was calculated from the measured storage modulus (G2) using rubber elasticity theory, adjusting for theoretical equilibrium swelling.^31,32

Fibroblast culture

Normal Human Lung Fibroblasts were obtained from Lonza (NHLFs from a single donor at passage 2, catalog number CC-2512), who guarantees these cells through 15 population doublings and verifies them negative for von Willebrand Factor Expression/Factor VIII, cytokeratins 18 and 19, alpha smooth muscle actin, mycoplasma, bacteria, yeast, and fungi, HIV-1, hepatitis B, and hepatitis C. These cells were expanded and cryopreserved and, for each experiment, a vial was thawed and subcultured, feeding every other day and passaging at 80% confluency (passage ≤ 8) according to the vendor instructions. For 2D hydrogel culture, cells were seeded on top of polymerized hydrogels (20 μL) at a concentration of 4×10⁴ cells/cm². For 3D hydrogel culture, cells were resuspended in hydrogel precursor solution (20 μL) at a concentration of 1×10⁴ cells/μL to ensure similar cell seeding densities in all geometries, as determined by analysis of the number of cells per z-stack projection of live/dead stained cells after 24 hours, and a similar magnitude to reported seeding densities used for 3D culture of fibroblasts within synthetic matrices.¹⁵ Hydrogel precursor solution (20 μL) was added to a sterilized syringe mold and polymerized with 10 mW/cm² at 365 nm light for 5 minutes. The hydrogel was then immediately immersed in cell culture media and incubated at 37°C and 5% CO₂. Cells were encapsulated within the 2.5D culture system as detailed below. For each experiment, three or more replicates of each type of hydrogel sample were prepared with cells subcultured from the same thawed vial, ensuring that all cells in each hydrogel condition had experienced the same culture history up until the point of seeding or encapsulation in the hydrogel; each experiment was repeated at least once and verified to yield similar results to representative data shown. All cell-gel constructs were cultured within non-tissue culture treated 48 well plate under sterile conditions at 37°C and 5% CO₂, and media was replaced every other day.

Formation of 2.5D hydrogels

To culture cells in a 2.5D geometry, hydrogel precursor solution (10 μL) was pipetted into a sterilized syringe mold and spread with a pipet tip so that it evenly covered the surface of the plunger of the syringe. If even spreading of the solution could not be achieved, an additional 10 μL was added to ‘wet’ the mold, the solution was spread, and then 10 μL was removed from the mold.³³ The 10 μL of precursor solution then was irradiated (10 mW/cm² at 365 nm for 5 minutes) to form the bottom layer of the hydrogel for cell culture. 50 μL of cell solution (8.37×10⁴ cells/mL, chosen so the number of cells per z-stack projection was similar in all three cell culture geometries) subsequently was added to the syringe mold under sterile conditions and incubated at 37°C and 5% CO₂ for 4 hours. Here, four hours of incubation was selected to allow adequate time for fibroblasts to settle, attach, and spread on the bottom layer as single cells before the top layer of hydrogel was added; this approach allowed the experiment to start from single spread cells surrounded by matrix in this 2.5D culture geometry for comparison to single spread cells on top of matrix in 2D culture and to single rounded cells surrounded by matrix in 3D culture. After initial cell seeding, excess media was carefully removed from the hydrogel surface. The second hydrogel layer was then added on top of the attached and spread cells by applying another 10 μL of precursor solution and irradiating (10 mW/cm² at 365 nm light for 5 minutes). The resulting layered hydrogel with cells encapsulated at the interface between the layers was removed from the syringe mold, placed in a non-tissue culture treated 48 well plate with 0.5 mL media, and incubated at 37°C and 5% CO₂.

In a separate experiment, to visualize successful layer formation, maleimide functionalized fluorophores were conjugated into each layer of the 2.5D geometry. For the bottom layer, 0.02 mM of maleimide functionalized DyLight 488 was added to the hydrogel precursor solution and incubated at room temperature for 1 hour for adequate maleimide labeling prior to hydrogel formation. After photopolymerization as described earlier, the bottom layer was then incubated with 50 μL of complete media for 4 hours at 37 °C to mimic the conditions used during cell seeding. A second hydrogel precursor solution containing 0.02 mM maleimide functionalized Alexa Fluor 594 was prepared and incubated for 1 hour. The media subsequently was removed from this first hydrogel layer, and the second precursor solution was added and photopolymerized to form the top layer. The resulting layered construct was imaged with confocal microscopy (LSM 800 Confocal Microscope (Zeiss)). The hydrogel was sliced in the z-direction using a razor blade, and the resulting section was flipped on its side for facile imaging of the interface.

Live/Dead staining

At time points of interest, hydrogels were washed 2X with warmed PBS for 5 minutes, incubating at 37°C and 5% CO₂. During incubation, calcein AM and ethidium homodimer-1 from a LIVE/DEAD viability/cytotoxicity kit (Thermo Fisher) were diluted in PBS, 0.5 μL/mL and 2 μL/mL, respectively. The reagents were incubated with the hydrogels for 20–30 minutes at 37°C, and then the hydrogels were washed 2X with warmed PBS for 5 minutes before imaging on an LSM 800 Confocal Microscope (Zeiss). Three hydrogels were imaged per condition and three images were taken per hydrogel with >100 cells counted per condition, and error was calculated from n = 3 hydrogels.

Immunofluorescent staining

At 48 hours, hydrogels were washed with PBS and then fixed with 4% PFA (Fisher Scientific) for 20 minutes at room temperature. Fixed samples were washed with PBS (3X for 20 minutes each) and then blocked with 5% BSA in PBS for one hour while rocking. Hydrogel samples were permeabilized with 0.2% v/v triton-X (Fisher Scientific) for 20 minutes and then washed in 0.2% v/v Tween-20 (Sigma) (3X for 20 minutes each) while rocking. Primary antibody with 1.5% BSA in PBS (rabbit anti-YAP (Abcam ab52771, 1:200), mouse anti-alpha smooth muscle actin (Abcam ab7817, 1:50) was added to blocked samples and incubated at 4°C overnight. The samples were washed with 0.2% v/v Tween-20 (3X for 20 minutes each) while rocking. The secondary antibody (1:200) (goat anti-rabbit AlexaFluor 594 (Invitrogen A11012), goat anti-mouse AlexaFluor 488 (Invitrogen A1101) or goat anti-rabbit AlexaFluor 647 (Invitrogen A21244)) and rhodamine-labeled phalloidin to label F-actin (5 μg/mL) (Sigma) were added in a 1.5% BSA in PBS solution and incubated overnight at 4 °C. Hydrogels were washed with PBS (4X for 5 minutes each) while rocking and incubated with DAPI (Invitrogen) for 1 hour. Samples were stored in PBS at 4 °C protected from light until imaging. Three hydrogels were imaged per condition using an LSM 800 confocal (Zeiss) and three images (100 μm z stacks with 4 μm between each slice) were taken per hydrogel with >100 cells counted per condition, and error was calculated from n = 3 hydrogels.

Image analysis

For quantitative analysis of live/dead staining, the number of objects in the 488 nm channel (live) and the number of objects in the 594 nm channel (dead) were counted using the “Find Objects” function in Volocity (PerkinElmer). Touching bodies were separated in 2D and 2.5D culture where cell clustering was observed using the “separate touching objects” function with the object size guide set to 25,000 μm³. The % live cells was calculated by dividing the number of live cells by the total number of cells in the image.

For αSMA staining, channels were separated in Fiji, and nuclei were counted manually. The 488 nm channel (αSMA) and the 405 nm channel (nuclei) were then overlaid and the number of cells expressing αSMA was counted. The % αSMA positive cells was calculated by dividing the number of cells expressing αSMA by the total number of cells in the image.

YAP nuclear localization was assessed following a modified version of published protocols.^10,35 Briefly, the “Find Objects” function in Volocity (PerkinElmer) was used to find nuclei (denoted as Population 1) and cell bodies (f-actin) (denoted as Population 2) in the DAPI (405 nm) and red (594 nm) channels, respectively. The “Analyze” function then was used to determine the total fluorescence intensity for YAP in the far-red channel (647 nm) within each population. To determine the fluorescence intensity of cytosolic YAP, the total fluorescence intensity of Population 1 was subtracted from the total fluorescence intensity of Population 2 using the “Subtract” function. This value was used to calculate the ratio of nuclear YAP/ cytosolic YAP.

To quantify the interface of the maleimide labeled 2.5D gels, images from confocal microscopy were analyzed using MATLAB. Briefly, fluorescence intensity of each channel was measured across the z-dimension, and the total distance in which fluorescence intensity was detected from both fluorophores above the background levels was considered the thickness of the interface.

Quantitative PCR analysis

To assess gene expression, RNA was isolated from 2D, 2.5D, and 3D cultures 48 hours after seeding using a modified Trizol-based protocol.³¹ Two hydrogels per RNA sample were incubated in 100 μL of 1000 U/mL collagenase (Thermo Fisher) for 30 minutes at 37°C to degrade the synthetic matrix. 500 μL Trizol (Invitrogen) subsequently was added to each tube, in a fume hood, and then incubated at room temperature for 5 minutes. 100 μL of chloroform (Sigma) was added, and tubes were briefly vortexed and then incubated at room temperature for 3 minutes. Tubes were centrifuged (12000 × g) for 15 minutes, then the aqueous phase, which contains RNA, was removed and added to a new tube. 250 μL of isopropanol (Sigma) and 2.5 μL of glycoblue (Thermo Fisher) were added to the aqueous phase and incubated at room temperature for 10 minutes to precipitate the RNA. The RNA was pelleted by centrifugation (12000 × g) for 10 minutes and resuspended in 20 μL RNAse free water. Subsets of resuspended RNA solutions then were pooled to make three 60-μL samples per condition, providing sufficient RNA per sample for downstream processing and analysis. The pooled samples were treated with a DNA-free DNA removal kit (Invitrogen) according to manufacturer instructions. RNA was further purified by adding 400 μL phenol-chloroform-isoamyl alcohol, centrifuging for 5 minutes to separate out the aqueous layer, which was removed to a new tube, treated with 200 μL chloroform, and then centrifuged and transferred to a new tube again; this step was repeated. RNA was re-precipitated using glycoblue, and the pellet was washed in 500 μL ice cold ethanol, and then resuspended in RNAse free water.

RNA purity was assessed via UV-vis absorbance using a Nanodrop, and then cDNA was synthesized from RNA using an iScript kit (Bio Rad). Quantitative PCR was performed using SYBR green Master Mix (Thermo Fisher). Specifically, 50 ng of cDNA was included in each reaction. Primers for CDH11, Serpine1, TGFB1, and ITGB1 were purchased from Bio Rad. 18S was used as a housekeeping gene with the following primer sequence: forward primer – GATCAAAACCAACCGGTCA, reverse primer – GATCGGCCCGAGGTTATCTA. Primers reported in Liu et al.¹² were used to detect COL1A1 gene expression: forward primer – AAGAGGAAGGCCAACTCGAG, reverse primer – CACACGTCTCGGTCATGGTA.

The ΔΔC_t method was used to evaluate gene expression. Gene expression for all samples was normalized to the 2D soft condition to facilitate pairwise comparisons. cDNA synthesis and qRT-PCR were run on a CFX96 detection system (Bio-Rad). Statistical significance was determined from a two-tailed t-test of ΔC_t values. Error was calculated from the Ct values for each RNA pool (collected from 6 hydrogel samples) and propagated to ΔΔC_t values, which were used to calculate fold change error according to the following formula:

$$\mathit{\text{Positive Error}}={\mathit{2}}^{-(\Delta \Delta \text{Ct}+\Delta \Delta {\text{Ct}}_{\text{error}})}$$

$$\mathit{\text{Negative Error}}={\mathit{2}}^{-(\Delta \Delta \text{Ct }-\Delta \Delta {\text{Ct}}_{\text{error}})}$$

Statistical analysis

Results are reported as mean ± standard error with three or more independent samples for all conditions, where exact numbers of replicates are noted above for specific experiments. To determine statistical significance (p < 0.05), a two-tailed t-test was performed.

RESULTS

A new approach was established for 2.5D culture using layered PEG-peptide hydrogels

Multi-layer engineered hydrogels have been used to enable the co-culture of multiple cell types^34,35 or the creation of regions with different matrix compositions in a single hydrogel.^36,37 These multi-layer engineered hydrogels are often fabricated by polymerizing (bio)polymers decorated with reactive handles to create covalently-crosslinked hydrogel layers on top of each other, resulting in a single stratified hydrogel.^34,36 We were inspired by these designs to develop a straightforward method to culture fibroblasts between two well-defined, biomimetic hydrogel layers in an effort to decouple the effects of cell spreading and polarization on fibroblast behavior, which often are confounded when comparing between 2D and 3D culture. While 3T3 fibroblasts previously have been shown to sense and respond to 2.5D culture by changing morphology, comparisons to 3D culture within engineered hydrogels are not possible in these systems. For example, fibroblasts ‘sandwiched’ between poly(acrylamide) hydrogels or physically encapsulated within collagen at the interface with a poly(acrylamide) hydrogel are not strictly confined on all sides as they are when encapsulated within a synthetic covalently-crosslinked matrix for 3D culture, and precursor cytotoxicity precludes the use of poly(acrylamide) for cell encapsulation.^20,28 Approaches are needed that allow the use of a synthetic matrix across all culture geometries to facilitate comparisons between 2D, 2.5D, and 3D culture. Deploying the capabilities of photopolymerizable synthetic matrices, here, we specifically sought to understand how 2.5D culture, in comparison to and as a bridge between 2D and 3D culture, affects fibroblast protein and gene expression, and more broadly, the role cell culture geometry plays in directing fibroblast activation.

We began by establishing a method for 2.5D culture within PEG-peptide hydrogels formed with photoinitiated thiol-ene click chemistry that previously had been utilized for 2D and 3D culture of fibroblasts, facilitating comparison of cell response amongst different geometries.^{13,21–23,38} These hydrogels were formed with 8-arm norbornene-functionalized PEG to allow tuning of modulus over a wide and relevant range for mimicking healthy to fibrotic tissues (E ~ 1–5 kPa to 5–10+ kPa for lung tissue),³⁹ a thiol-functionalized matrix metalloproteinase (MMP)-cleavable peptide GCRDVPMS↓MRGGDRCG that degrades in response to a variety of MMPs secreted by fibroblasts,³⁸ 2mM of thiol-functionalized collagen-mimetic GFOGER peptide to promote cell adhesions, and a photoinitiator (Figure 1A). As these studies were conducted with pulmonary fibroblasts, the integrin-binding GFOGER peptide was chosen to mimic the collagen-rich interstitial tissue of the lung and was added at a concentration that has been previously shown to promote cell adhesion.^13,40,41 These hydrogels rapidly form upon the application of cytocompatible doses of long wavelength UV light, within less than 1 minute with 10 mW/cm² at 365 nm (Figure S1). To create the 2.5D culture environment (Figure 2A), precursor solution for each hydrogel composition was added (10 μL) to a syringe mold and polymerized with 10 mW/cm² at 365 nm light for 5 minutes, a time well beyond that required for hydrogel formation to ensure complete polymerization of the first layer.³⁷ After formation of the bottom hydrogel layer, human pulmonary fibroblasts resuspended in cell culture media were added to the syringe mold and incubated to allow initial attachment and spreading as single cells (Figure S2) before removing excess media and adding a second layer of hydrogel (10 μL). This approach produced robust layered constructs with spread cells encapsulated at the interface between the well-defined hydrogel-based synthetic matrix (Figure 2B and 2C).

Figure 2. Layered hydrogels for 2.5D cultures.

A) Method of fabrication. B) Confocal z-stack side view (x-z orthogonal projection) of a hydrogel formed with fluorophores in each layer for visualization. C) Confocal z-stack side view (x-z orthogonal projection) and top-down view (x-y orthogonal projection) of cells in 2.5D culture (here, 6 h after encapsulation at the interface between the hydrogel layers).

To ensure that the processing method for formation of the 2.5D cultures did not significantly impact cell viability, we measured the viability of fibroblasts in 2.5D culture in ‘stiff’ hydrogels, as well as traditional 2D and 3D cell culture methods, using a live/dead cytotoxicity assay (Figure 3). Viability was measured at an early timepoint (24 hours) and a later timepoint (72 hours). Confocal microscopy images demonstrated that the fibroblasts adopted a spread and slightly clustered morphology in 2D and 2.5D cell culture at both timepoints and a more rounded morphology in 3D cell culture. In all cases viability was observed to be high (≥ 80%), supporting that the process of 2.5D encapsulation did not impact cell viability or inhibit cell function within the 2.5D cell culture geometry and enabling side by side comparisons of cell behavior in all three cell culture geometries.

Figure 3. Viability of human pulmonary fibroblasts in 2D, 2.5D, and 3D cell culture geometry.

A) Human pulmonary fibroblasts were cultured in three cell culture geometries, and at 24 hours and 72 hours, a cytotoxicity kit was used to stain live (green) and dead (red) cells. Samples were imaged with confocal microscopy (representative confocal z-stack projections shown) and B) quantitatively analyzed to determine viability.

To assess the thickness of the resulting interface, we employed a technique developed by Bryant and coworkers with similar layered PEG-peptide hydrogels.³⁷ A minimal amount of free thiols in each precursor solution were covalently labeled with fluorophores (0.02 mM DyLight488 maleimide and AF594 maleimide) prior to formation of the bottom layer and top layer, respectively. This labeling allowed visualization of each layer (green and red, respectively) and the interface between them (yellow where fluorophores overlap) with confocal microscopy. With this approach, we observed the interface between the two layers to be on the order of ~ 50-μm thick (Figures S3). Indeed, Bryant and coworkers have shown that these interfaces of small but measurable thickness between layers arise from some transport of monomer from the top layer precursor solution into the bottom layer hydrogel prior to photopolymerization of the top layer, promoting the formation of covalent bonds between the layers and, to a lesser extent, the formation of a semi-interpenetrating network. Importantly, in their studies, nanoindentation measurements also demonstrated that the interface exhibits moduli reflective of the top and bottom layers.³⁷ Based on these observations, we speculate that the two layers are integrated at the interface through a combination of i) dangling end groups at the surface of the first layer that react with fresh monomer from the top layer precursor solution and ii) entangled chains between the top and bottom layer that are ‘locked’ into place upon photopolymerization of the top layer. To assess cell sampling of the interface, we analyzed confocal z-stack cross-sections of immunostained cells in non-labeled hydrogels. We observed the spreading of cells at the interface in this 2.D geometry (Figure 2C), with cells sampling ~ 100-μm thick region around the interface after 2 days in these cultures (Figure S4). Starting from single cells at the interface, the fibroblasts form growing clusters of cells that spread into the top and bottom layers. In this context, cells interact with both the top and the bottom layer of the hydrogel centered around the interface between the layers, with increasing cell-cell contact from cell clustering over time.

Differences in the activation of pulmonary fibroblasts cultured in different hydrogel geometries indicated by αSMA expression

Having determined that fibroblasts remained viable in all three cell culture geometries for a time frame relevant for observing changes in gene and protein expression, we investigated the effects of hydrogel geometry on fibroblast activation. As fibroblast response to stiffness has been shown to be affected by hydrogel geometry, we chose to culture human pulmonary fibroblasts in 2D, 2.5D, and 3D geometries with hydrogels of different stiffnesses. We chose a soft condition (E ~ 3.5 kPa) aimed at mimicking the modulus of healthy lung tissue and a stiff condition (E ~ 10 kPa) aimed at mimicking the modulus of fibrotic lung tissue (Figure S5).³⁹

To assess fibroblast activation, we cultured human pulmonary fibroblasts in all three cell culture geometries for 48 hours and then stained the cells for αSMA. αSMA is a commonly recognized marker for fibroblast activation⁴² and is expressed by fibroblasts to enable a contractile phenotype during matrix remodeling. We expected αSMA expression to be highest for cells cultured on stiff substrates in 2D culture and lowest in 3D culture based on prior reports¹⁵ and wanted to establish if 2.5D culture correlated more or less with either of them (Figure 4).

Figure 4. Fibroblast activation in 2D, 2.5D, and 3D cell culture.

A) Human pulmonary fibroblasts were cultured in three cell culture geometries for 48 hours and then stained for αSMA and with DAPI nuclear dye (representative confocal z-stack projections shown). B) The percentage of cells expressing αSMA was quantified: high levels of activation were observed with both ‘soft’ and ‘stiff’ hydrogels in 2D and 2.5D culture; in ‘stiff’ hydrogels, a significant percentage of activated fibroblasts were observed in 2D and 2.5D culture relative to 3D culture (**p<0.01).

We observed most cells to be αSMA positive in both the 2D and 2.5D cell culture conditions and that, in comparison, fewer cells were αSMA positive in the 3D cell culture condition, with the difference being statistically different in the 3D stiff case. Our observations correlate with others observations that fewer fibroblasts cultured in stiff 3D substrates are αSMA positive compared to fibroblasts cultured in soft 3D substrates, a trend that is reversed in 2D cell culture.¹⁵ Unfortunately, the high levels of αSMA positive cells we observed in the 2D and 2.5D geometry in both low and high modulus conditions make it difficult to use this measure of fibroblast activation to determine what effect stiffness may be having in these geometries. Despite this, response of fibroblasts in 2.5D resembled more that of 2D culture, both of which exhibited a spread morphology and high levels of αSMA expression at the protein level like that observed in wound healing or fibrosis.

We hypothesize that these high levels of αSMA positive cells in both the 2D and 2.5D cultures may be due to proto-myofibroblast formation promoted by the collagen mimetic peptide GFOGER,⁴³ which was used to promote cell adhesion and mimic the collagen-rich environment of human lung tissue, and the high levels of cell-cell contact its binding promotes. Indeed, in our prior studies with the GFOGER peptide, we observed its binding to promote significant activation of lung fibroblasts in 2D culture, including high levels of αSMA expression and the formation of large cell clusters reminiscent of the activated foci that form during lung fibrosis.⁴⁶ These observations are consistent with those more broadly that both cell-cell contact and increased collagen content promote fibroblast activation in fibrosis.⁹ It is possible that, at a Young’s modulus of 3.5 kPa, the soft condition tested was not ‘soft’ enough to reduce αSMA expression in the fibroblast population studied here, which had been propagated on tissue culture plastic over the time scale of weeks which is known to promote some level of fibroblast activation.⁴⁴ Attempts to form softer hydrogels unfortunately resulted in inconsistent polymerization. However, the 2.5D culture approach established here provides many opportunities for future studies with different matrix designs to further probe and identify a soft 2.5D condition in which fibroblasts are not activated, including the use of different integrin-binding peptides, lower cell seeding densities, or the application of relevant growth factors. For example, while the total amount of integrin binding peptide needs to be held constant (~ 2 mM in the precursor solution for parody between cell seeding density applied to and attached to the substrate), ratios of different integrin binding peptides can be explored such as fibronectin/vitronectin mimic RGDS with GFOGER.³¹ To further investigate the ability of fibroblasts to sense and respond to stiffness in 2D, 2.5D, and 3D culture within the collagen mimetic hydrogels studied here, we examined YAP nuclear localization.

YAP nuclear localization indicated differences in mechanotransduction in pulmonary fibroblasts cultured in different hydrogel geometries

YAP nuclear localization has been previously used as an early marker of fibroblast activation and is associated with mechanotransduction.⁴⁵ In particular, nuclear localization of YAP, as well as its transcriptional coactivator with PDZ-binding motif (TAZ), has been shown to be essential for mesenchymal cell response to the stiffness of the ECM. In 2011, Dupont et al. demonstrated that YAP/TAZ signaling was required for stem cell differentiation in response to stiffness.² In later studies, YAP/TAZ nuclear localization was identified as necessary for fibroblast activation and an early marker of fibroblast activation, where YAP and TAZ were observed to accumulate in the nucleus of fibroblasts in fibrotic but not healthy lung in vivo and during culture on pathologically stiff substrates in vitro.^15,45 We hypothesized that differences in early activation and mechanosensing of fibroblasts may be observed in YAP nuclear localization even if differences in fibroblast activation were not detectable by αSMA expression. To test this, we stained human pulmonary fibroblasts for YAP, F-actin to visualize the cell body, and DAPI to identify the nucleus. We then calculated the ratio of YAP localized in the cell body compared it to YAP localized to the nucleus using Volocity (Figure 5).

Figure 5. YAP nuclear localization in human pulmonary fibroblasts in 2D, 2.5D, and 3D culture.

A) Human pulmonary fibroblasts were cultured in three cell culture geometries for 48 hours then stained for YAP, F-actin, and nuclei (representative confocal z-stack projections shown). B) YAP nuclear localization was determined by quantitative image analysis. In 2D, YAP nuclear localization was greater on stiff matrices than on soft matrices (***p < 0.001), which correlates with previous studies of fibroblast activation in 2D substrates. The opposite trend was observed in 3D (*p < 0.05), correlating with observations of αSMA expression. Interestingly, no difference was observed between soft and stiff matrices in 2.5D culture. However, in both soft and stiff matrices, YAP nuclear localization was greater in 2.5D culture than in 2D culture (*p < 0.05).

In 2D and 3D cell culture geometries, we found that YAP nuclear localization was greater in stiff matrices compared to soft matrices, consistent with previous reports of observations of fibroblast mechanosensing in these geometries.^15,45 In 2.5D, we did not observe a significant change in YAP nuclear localization between the stiff and soft conditions, suggesting that fibroblasts may be more activated in this geometry potentially driven by different cell-cell interactions and thus less responsive to stiffness. Interestingly, YAP nuclear localization was statistically different between the 2D and 2.5D cell culture geometry, with the 2.5D geometry having higher levels of YAP nuclear localization. These results indicated that, while fibroblast morphology and αSMA expression appear to be similar in 2D and 2.5D culture, differences exist in the way fibroblasts sense and respond to matrix cues in the 2D and 2.5D geometries and that more fibroblasts may be at an early stage of activation in 2.5D relative to 2D culture. Further, no statistical difference was observed in YAP nuclear localization between the 2.5D and 3D cell culture geometries, suggesting some potential similarities in mechanotransduction between 2.5D and 3D culture despite clear differences in fibroblast activation within these geometries based on αSMA expression.

Cell culture geometry affected the expression of genes associated with cell-cell and cell-matrix interactions

To determine whether the differences in cell response observed amongst the different geometries at the protein level translated to the gene level, we conducted quantitative reverse transcription PCR (RT-qPCR) analysis of different genes related to cell-cell signaling, cell-matrix signaling, and fibroblast activation, outlined in Table 1. These genes are associated with key cell-cell and cell-matrix interactions involved in mechanotransduction and fibroblast activation and were investigated to provide a better understanding of the role mechanosensing plays in directing fibroblast behavior in these three cell culture geometries. Specifically, collagen I and TGFβ1 are both secreted by, and are promoters of, activated fibroblasts and can be used as measures of fibroblast activation.^8,12 Cadherin 11 is up-regulated in fibroblast activation and plays an important role in mediating cell-cell contact, which we hypothesized would be much higher in the 2D and 2.5D geometries where cells were observed to form activated clusters.^46–48 Integrin β1 is associated with binding to collagen I and is targeted by the GFOGER peptide, which was included in the hydrogel-based matrices to promote cell adhesion.⁴⁹ Serpine1 encodes the serine protease inhibitor plasminogen activator inhibitor-1, which has been implicated in fibrosis, and is regulated by the YAP/TAZ pathway, serving as a measure of mechanosensing.¹⁰ All gene expression data were normalized to the 2D soft condition to facilitate pairwise comparisons.

Table 1.

Genes associated with cell-cell interactions, cell-matrix interactions, and fibroblast activation.

Gene	Protein	Function	Ref
COL1A1	collagen I	secreted by activated fibroblasts during matrix remodeling	12
TGFB1	transforming growth factor β1	cytokine secreted by activated fibroblasts	8
CDH11	cadherin 11	mediates cell-cell contact	46,47
ITGB1	integrin β1	mediates cell-protein contact	49
Serpine 1	plasminogen activator inhibitor 1	serine proteinase inhibitor involved in mechanotransduction	10

Significant differences were observed in CDH11, ITGB1, and Serpine 1 expression (Figure 6, Tables S1 and S2). Expression of these three genes associated with cell-cell, cell-matrix, and mechanosensing, respectively, was significantly lower in 3D cell culture compared to 2D and 2.5D cell culture. Differences in gene expression were less stark between the 2D and 2.5D cell culture geometries, with the only statistical differences being increased CDH11 expression in the 2.5D condition compared to the soft 2D condition. Although not statistical, CDH11 expression trended up with increased stiffness in both 2D and 3D culture while remaining similarly high in both soft and stiff conditions in 2.5D culture. COL1A1 expression levels, within error, were relatively flat and not statistically different among the three cell culture geometries or the two stiffnesses, as has been observed with other fibroblast types with these type of PEG-peptide hydrogels at early times in culture (Figure S6, Tables S1 and S2); more significant differences in COL1A1 may be observed at later times with persistent activation.¹⁵ While not statistical, TGFβ1 gene expression trends correlated with observations of increased cell activation (Figure S6, Tables S1 and S2). Specifically, increased TGFβ1 gene expression was observed to correlate with increased YAP nuclear localization, CDH11 gene expression, and αSMA positive cells for 2D and 2.5D cultures and with increased matrix stiffness in 2D and 3D cultures. Interestingly, in 3D culture, TGFβ1 gene expression trended up with increased stiffness while the percentage αSMA positive cells was statistically down relative to the soft condition; this observation may suggest that fibroblasts were at earlier stages of activation in response to stiffness in the 3D culture geometry. Note, while specific trends were elucidated from these gene expression data, some differences may be masked by high variance due to population heterogeneity and pooling of multiple samples for adequate RNA extraction. Despite this, these observations at the gene level were generally consistent with those made at the protein level, with additional insights into the potential importance of cell-cell contact in the observed cell responses.

Figure 6. Expression levels of genes associated with cell-cell and cell-matrix interactions for cells in 2D, 2.5D, and 3D culture.

A) CDH11 expression varied significantly among all three cell culture geometries (*p < 0.05). CDH11 expression also differed between 2D soft and 2.5D stiff and soft hydrogels (*p < 0.05). B) ITGB1 expression was significantly lower in soft 3D matrices compared to stiff or soft 2D (**p < 0.01) and 2.5D (*p < 0.05) matrices. C) Serpine1 expression was significantly lower in soft 3D matrices compared to stiff or soft 2D (*p < 0.05) and soft or stiff 2.5D (**p < 0.01) matrices. Expression was also significantly lower in 3D stiff matrices compared to 2.5D soft matrices (*p < 0.05).

Observations at the cell, protein, and gene levels support the relevance of 2.5D culture for the study of fibroblast activation and fibrosis

Taken together, our observations at the cell, protein, and gene level suggest that fibroblast response to 3D cell culture geometry is significantly different than fibroblast response to 2D and 2.5D cell culture geometry at early times in culture, within the range of conditions probed, and this response is likely mediated by differences in cell-cell and cell-matrix signaling. Fibroblasts adopted a more rounded morphology in a 3D culture geometry: starting from encapsulation, cells were surrounded on all sides by matrix and had a low surface area to volume ratio with limited to no cell-cell contact, qualitative observations that were reinforced by related quantitative differences in the expression of genes associated with cell-matrix and cell-cell interactions, respectively. In contrast, and by design, fibroblasts adopted a more spread morphology in 2D and 2.5D geometries over the same timeframe: starting from seeding, cells were allowed to spread whether unencumbered or before the addition of the synthetic matrix on top, respectively. Furthermore, the percentage of αSMA positive cells was higher in 2D and 2.5D culture compared to 3D culture, where higher levels of expression are consistent with in vivo observations of increased αSMA during fibrosis progression.⁴² While differences in αSMA expression were not observed between 2D and 2.5D culture, increased YAP nuclear localization was observed in 2.5D culture, aligning more closely with the 3D culture condition and observations in vivo during fibrosis progression.^15,45 At the gene level, expression levels for activation and outside-in signaling were similar between the 2.5D and 2D culture. A notable difference was CDH11, which was upregulated in 2.5D culture with both soft and stiff matrices.

CDH11 expression yields stronger intercellular connections that improve force development in myofibroblasts, aiding in contraction.^50,51 Increased expression of CDH11 is observed in vitro upon TGFβ1-stimulated activation of lung fibroblasts into myofibroblasts and in vivo during lung fibrosis.⁵² Consequently, increased CDH11 expression is both an indicator of myofibroblastic activation and thought to be a part of the positive feedback loop that fuels activation and persistence of myofibroblasts in fibrosis.^52–54 Given our observation of trends in increased TGFβ1 expression with statistical increases in YAP nuclear localization and CDH11 expression in 2.5D culture on both soft and stiff matrices relative to 2D culture, we speculate that having matrix on all sides surrounding highly-activated cells (i.e., spread, clustered, αSMA positive cells in 2.5D culture) may provide a positive feedback loop that further promotes cell activation. For example, in 2.5D culture, being surrounded by matrix may influence the contraction of activated cells and their force generation,⁵⁵ where cells can pull in all directions on the matrix, or influence local concentrations of cell-secreted TGFβ1 and other proteins, where diffusion would be hindered by having a thick synthetic matrix on both sides of the cell rather than media on one side and matrix on the other. With the 2.5D culture platform established here, future mechanistic studies could be performed to parse out the relative influences of cell-cell and cell-matrix interactions on myofibroblastic activation and persistence in these multidimensional cultures, including local force generation, the secretome, and matrix mechanical properties over time. Further, future studies could employ the unique handles afforded by this photopolymerizable system to generate gradients at the interface and examine related cell responses: for example, purposefully mismatching the mechanical properties of the top and bottom layers or allowing a ‘soak time’ for diffusion of the precursor solution of the top layer far into the bottom layer to generate thick interfacial layers with gradients in properties as pioneered by Bryant and coworkers.^37,56

As fibroblast activation is mediated by both cell-cell and cell-matrix interactions,¹⁰ our observations together suggest the importance of both mechanosensing and cell-cell interactions in the differences between 2.5D and 2D culture at this early time point in culture. More fundamentally, our results suggest that decreasing cell polarization alone, through the addition of a layer of matrix on top of cells in 2.5D culture, does not have a significant effect on many key fibroblast responses. Rather, cell morphology, shape, and confinement, which initially are significantly different between 2D and 2.5D culture and 3D culture, may play a larger role in directing fibroblast activation than cell polarization. These studies clarify that differences observed by others in 2D and 3D cell culture geometry are not due to polarization alone. Indeed, cell shape has been observed to play a significant role in directing cell behavior,¹⁹ and our results support this observation, where fibroblasts were spread in 2D and 2.5D culture and rounded in 3D culture.

CONCLUSIONS

In this study, we established a method for 2.5D culture of human pulmonary fibroblasts to decouple cell shape and cell polarization, providing insight into the fundamental differences observed between cell responses in 2D and 3D culture models. We observed differences in αSMA expression between fibroblasts cultured in the 3D geometry, where fibroblasts were rounded, and the 2D and 2.5D geometries, where fibroblasts were highly activated and spread, reminiscent of activated fibroblasts observed during fibrosis. Statistical differences in activation in response to matrix stiffness were not observed, potentially owing to the activating tendencies of GFOGER peptide and resulting high levels of cell-cell contact from cell clustering. With this consideration, there exists an opportunity for future investigations into the effects of matrix composition. Interestingly, we observed differences in YAP nuclear localization and CDH11 expression, related to mechanosensing and cell-cell signaling, respectively, between fibroblasts cultured in 2D and 2.5D cell culture geometry. These results indicate that, while decreased cell polarization from 2D to 2.5D culture may lead to differences in mechanotransduction and cell-cell contact, other factors (e.g., cell spreading and shape) likely also contribute to previously observed differences in fibroblast phenotype in 2D and 3D cell culture geometries. While trends were observed in the expression of investigated genes, fewer yielded statistical differences, potentially owing to cell population heterogeneity that contributes to large variances in gene expression data and highlighting a need for continued innovation in single cell and real-time approaches for assessing cell response to matrix cues and culture geometries. Due to the fundamental differences observed in cell shape between 2.5D and 3D approaches, each model provides benefits for addressing specific hypotheses. In particular, 2.5D cell culture offers unique opportunities for investigations at early timepoints, when 3D culture approaches would restrict cell spreading until later times when cells have secreted enough proteinases to sufficiently degrade the matrix. These studies highlight not only the utility of this 2.5D culture approach as a bridge between 2D and 3D culture for studies of microenvironment interactions, but also areas for future investigations with this approach, including the effects of matrix degradation and receptor binding in addition to the responses of other cell types.