Evaluation of focal adhesion mediated subcellular curvature sensing in response to engineered extracellular matrix

Abstract

Fibril curvature is bioinstructive to attached cells. Similar to natural healthy tissues, an engineered extracellular matrix can be designed to stimulate cells to adopt desired phenotypes. To take full advantage of the curvature control in biomaterial fabrication methodologies, an understanding of the response to fibril subcellular curvature is required. In this work, we examined morphology, signaling, and function of human cells attached to electrospun nanofibers. We controlled curvature across an order of magnitude using nondegradable poly(methyl methacrylate) (PMMA) attached to a stiff substrate with flat PMMA as a control. Focal adhesion length and the distance of maximum intensity from the geographic center of the vinculin positive focal adhesion both peaked at a fiber curvature of 2.5 μm^-1 (both ∼2× the flat surface control). Vinculin experienced slightly less tension when attached to nanofiber substrates. Vinculin expression was also more affected by a subcellular curvature than structural proteins α-tubulin or α-actinin. Among the phosphorylation sites we examined (FAK397, 576/577, 925, and Src416), FAK925 exhibited the most dependance on the nanofiber curvature. A RhoA/ROCK dependance of migration velocity across curvatures combined with an observation of cell membrane wrapping around nanofibers suggested a hybrid of migration modes for cells attached to fibers as has been observed in 3D matrices. Careful selection of nanofiber curvature for regenerative engineering scaffolds and substrates used to study cell biology is required to maximize the potential of these techniques for scientific exploration and ultimately improvement of human health.

I. INTRODUCTION

Cell biology seeks to understand cellular functions, including the integration of information found in an extracellular environment. The importance of the physical extracellular environment, not just other nearby cells, as an input affecting what a cell does has been seen with increasing importance, especially in phenotypic functions such as attachment and migration.¹ Many studies have been conducted, logically so, in well-controlled in vitro environments, predominantly the Petri dish² providing reproducible data. However, discrepancies in the function of a cell in vitro versus in vivo are well recognized. Biomaterial fabrication techniques have been developed that allow complex in vitro extracellular matrix mimetic³ structures that improve the accuracy of in vitro studies.

The extracellular matrix is a hydrated fibrillar cell adhesive structure found throughout the body. Hydrogels mimic many parts of this structure; however, the fibrillar nature can be difficult to control. Indeed, fibers have been combined with hydrogels and 3D-printed scaffolds⁴ to take advantage of the composite material properties. Electrospun nanofibers are fibrillar structures that can be controlled using electrospinning process inputs and parameters,^5–9 making nanofibers a potentially useful substrate from a few perspectives. For cell biologists, nanofiber substrates provide an easy to construct and long shelf-life in vitro substrate that may mimic the in vivo environment better than flat polystyrene dishes providing the ability to control specific factors in the cellular environment. In addition, for regenerative engineers, a nanofiber substrate can be a core component of tissue engineering scaffolds.¹⁰ Thus, the study of cell behaviors on nanofiber substrates benefits investigations across fields.

One of the fiber characteristics that can be controlled by adjusting electrospinning parameters is the fiber curvature (1/diameter, convex subcellular curvature). Curvature can also be studied in the context of nanoparticles,¹¹ but here we consider the cylindrical out-of-plane curvature. Assaying the effect of fiber curvature on cell behavior not only demonstrates its effect as a design parameter for regenerative engineering scaffolds, but also can provide insight into the contribution of physical curvature to the effect of extracellular matrix fibrils of different protein composition. Our constructs are single layer for in vitro observation; however, processes are in development to enable full thickness seeding of cells in electrospun scaffolds.¹² We utilized two overlapping sets of fibers that cover the range of extracellular matrix (ECM) fibrils by controlling solution concentration of high molecular weight poly(methyl methacrylate) (PMMA).^13,14 We then focused on the contribution of curvature to cell functions and characteristics.

The focal adhesion (FA) connects the cytoskeleton to the extracellular environment. It is comprised of hundreds of proteins and is a dynamic center of mechanotransduction that translates spatial information into a diverse set of phenotypes.¹⁵ The FA protein vinculin is a canonical member of this mechanobiology center discovered in 1979 by Geiger¹⁶ and remains an essential subject of investigation particularly in understanding 3D environments. Vinculin is known to promote traction force generation, cell elongation, and directional migratory persistence.¹⁷ An example of differing responses to physical environmental changes is the formation of focal adhesions. Vinculin differed on 3D fibronectin,¹⁸ suggesting that the mode of migration may change with the dimensionality of the environment.

The long standing paradigm of mesenchymal migration is that cdc42 activation at the leading edge of the cell is followed by Rac-1 lamellipodia formation pushing the membrane of the cell in the direction of migration, which finally is followed by RhoA-mediated contraction of the cell rear.^19–21 An important study compared cell migration on and within 3D matrices, revealing a completely different mode of migration inside the matrix versus the upper flat surface of the matrix.²² The authors discovered that nonpolarized RhoA activation was sufficient to drive migration within the hydrogel. Petrie et al. termed the new form of migration “lobopodial migration,” owing to the cylindrical protrusions that arise in the cell membrane formed from the increased intracellular pressure.²² The change from a cdc42, Rac 1, RhoA-driven migration sequence to one where RhoA is a dominate factor may be an indication of the dimensionality of the cell environment. Indeed, we found the contribution of small GTPases to migration outcomes to vary based on the fiber curvature.¹⁴

Owing to the lower stiffness of cells than most polymer nanofibers, cells can not only interact with nanofibers via integrins in focal adhesions, but also via cell membrane wrapping. Information on both the adhesion and other aspects of cell attachment and migration are critical to our understanding of bioinstructive cues²³ in scaffolds and their eventual application in regenerative engineering. To increase this understanding, we utilized image processing of vinculin localization, quantification of whole cell-active RhoA, quantification of active signaling and structural proteins, electron microscopy of cell cross sections, and high temporal resolution time-lapse microscopy.

II. EXPERIMENT

A. Substrate preparation

Electrospun nanofiber scaffolds were fabricated from high molecular weight polymethylmethacrylate (PMMA) (HMW—PMMA, 996 kDa) dissolved in DMF:THF (60:40) and collected on a target using a 15–20 cm working distance, ∼15 kV driving voltage, 0.7–2.5 ml/h flow rate, and an 18- or 30-gauge needle (adjustments used to control quality of fibers). Fibers were collected on glass coverslips that were first spin-coated with a layer of cell adhesion-resistant poly(2-hydroxyethyl methacrylate) (pHEMA). A prolonged electrospinning time was used for smallest fibers to overcome embedding in the pHEMA layer. Coverslips spin-coated with HMW-PMMA were used as flat surface controls.

Collector electric field manipulation created straight fibers with predominate orientation as explained in our previous publication.¹⁴ Polymer concentration was the driving force for curvature control where 1.54%, 3.1%, 3.7%, 6%, and 8.3% HMW-PMMA by weight was used for curvatures of 41.41, 10.45, 8.36, 4.56, and 1.65 μm⁻¹ respectively, corresponding to diameters 48, 191, 239, 438, and 1212 nm, respectively (field aligned fibers), and 3%, 6.5%, 7.5%, 10%, and 12% HMW-PMMA by weight was used to fabricate scaffolds with target curvatures of 10, 4, 2.5, 1.333, and 1 μm⁻¹, respectively, corresponding to diameters of 200, 500, 800, 1500, and 2000, respectively (field and kinetic aligned fibers). Notation of the fiber set used for each dataset has been made in the Results section at the first mention of a figure.

B. Cell culture

Human mesenchymal stem cells (hMSCs) were cultured in αMEM (Gibco 12571071) with 10% fetal bovine serum (FBS, Atlanta Biologicals, Flowery Branch, GA) and 1% penicillin/streptomycin (P/S, Corning, Corning, NY, USA, Part#: 30-002-CI). hMSCs were used for migration as well as biochemical studies between passages 7 and 10. All cells were cultured at 37 °C in a humidified incubator with 5% CO₂. Substrates were placed under the UV lamp in the biosafety cabinet for 20 min to deactivate microbial contamination. Cells were then seeded onto the substrates at ∼5 × 10⁴ or 1 × 10⁵ cells per construct (22 × 22 mm) for imaging or cell lysate collection, respectively.

C. Cell lysis

Cells were washed immediately prior to lysing with phosphate-buffered saline without calcium and magnesium. Samples were washed for 1 min in a cytoskeletal stabilizing buffer (CSB, see components in Table 1 in the supplementary material)⁸² to remove unbound proteins (see Fig. 4 in the supplementary material,⁸² demonstrating the effect of using CSB on the tubulin content in the lysate). Scaffolds or other surfaces were then submerged in the M-PER lysis buffer (ThermoFisher) or a detergent-based lysis buffer made according to components detailed in Table 2 in the supplementary material,⁸² subjected to vortexing to mechanically disrupt cells and then placed at −80 °C to complete the cell lysis and preserve the sample until analysis. Samples were thawed, collected into centrifuge tubes, and spun at 12 000 × g to prepare the sample on the day of analysis. The samples were then kept at −80 °C for long-term storage.

D. Immunostaining

Samples were fixed in 4% paraformaldehyde for 15 min following a wash in cold CSB to remove unbound proteins for 1 min. Blocking with 5% BSA in PBS for 45 min was followed by application of the primary antibody for 1 h. Three washes in PBS were followed by the secondary antibody for 1 h and another set of three washes. If DAPI or Phalloidin were applied, they were incubated on the samples for 5 and 10 min at 1:10 000 and 1:5000, respectively, between the final three washes. For FRET probes, cells were fixed, washed, and directly mounted.

E. Image quantification

Unless otherwise specified, ImageJ (NIH, Bethesda, MD) was used for image quantification. Focal adhesion dimensions were quantified using CellProfiler (www.cellprofiler.org) open-source software,²⁴ which allows the same “pipeline” of processing modules to be applied to a large group of images. MS Excel was then used to calculate averages per cell or per image from the CellProfiler output.

F. Live-cell imaging

Cells were time lapse imaged in supplemented FluoroBrite DMEM (ThermoFisher) media (details below). Cells and scaffolds were held in P30 culture dishes custom prepared with a glass coverslip attached to the dish covering an 18 mm diameter circular hole cut using a heated metal cork borer. The plate was sealed with UV activated optically clear Norland Optical Adhesive 68. The 22 × 22 mm coverslip was mounted either in or to the bottom of the dish depending on the final use and dish size. A heated stage with a layer of white mineral oil over the media was used. Images were taken at intervals of 90 s (unless noted otherwise) with differential interference contrast (DIC) illumination using a 10× long-working-distance water dipping objective.

Live cell imaging was done in FluoroBrite DMEM (Life Technologies A18967-01) supplemented with 15% FBS, 1% Glutamax (Life Technologies 35050-061), and 25 μl/ml HEPES (AMRESCO J848). Cells were allowed to attach in the humidified incubator prior to the collection of time-lapse images. For each set of experiments (i.e., one of each curvature and a flat control: 6 conditions), cells were first divided at decreasing densities into P60 dishes, from which cells would be lifted and plated at consistent times before that sample would be imaged to control for passage and density effects on migration speed. Unless noted otherwise, about ∼12 h postseeding in normal growth media, media were changed to live cell imaging media. The plate was then placed on the heated stage warmer (BioPTECH, Butler, PA, USA), with inhibitors added at this time if applicable. Contact was formed with the water dipping objective followed by application of a layer of white light mineral oil (AMRESCO PN J217) to prevent evaporation.

G. Motility quantification

DIC images for cell migration velocity measurements were collected with a 10× objective at 90 s intervals for approximately 12 or 24 h. During postanalysis, a 6-h time window was selected with a starting point that ensured the time between plating and the quantified window was consistent for the set of acquisition. Cells were selected to reduce confounding factors including migration out of the field of view, interaction with other cells, detachment from nanofibers, and cell division. Cells were tracked with the Manual Tracking ImageJ plugin (NIH, USA). Because automated cell tracking techniques do not work well for cells that are not labeled, manual mouse clicks were positioned to estimate the center of mass at each time point for each cell in their unmanipulated state, allowing for frequent imaging over a long period. Most migration data were presented after standardizing to flat surface migration velocities for that experimental set. Therefore, if one batch of cells exhibited a higher migration activity than another, the effect would have been equalized in the final analysis. Graphs were presented at a log base 2 scale so that a distance above 1 is the same interpretation as the equal distance below 1 (i.e., 4-fold and 0.25-fold). Migration velocity was calculated directly from the Manual Tracking ImageJ plugin output.

H. Western blot

Two-stage gels were cast on the day of running electrophoresis using a 6%–10% resolving gel. Ammonium persulfate (APS, Sigma) and tetramethylethylenediamine (TEMED, Sigma) were used as initiators for gelation. The resolving gel was allowed to polymerize for 1 h followed by another hour for stacking gel polymerization. A layer of isopropanol was applied on the resolving gel while it polymerized and then was removed before adding the stacking gel prepolymer. The buffer used during electrophoresis contained 10% SDS, while the buffer used during transfer to polyvinylidene difluoride (PVDF) membranes contained only 5% (by vol.) methanol formulated for larger proteins. Protein denaturation was accomplished with β-mercaptoethanol at 95 °C for 10 min. Electrophoresis was run at 120 V for 3–5 h depending on the movement of the front. Depending on the protein targets, transfer to the PVDF membrane was accomplished with variable settings including 80 V for 80 min and 240 A for 2 h. The membrane was rinsed once in the TBST buffer for 5 min and then blocked overnight in 5% BSA in TBST at 4 °C.

I. Immunoblot staining

The day after electrophoresis and transfer, the membrane was stained. Membranes were first allowed to bind the primary antibody in 5% BSA in TBST solution for 1 h at room temperature. After three washes in TBST, the secondary antibody was added in 5% BSA in TBST for 1 h followed by three more washes in TBST. Fluorescent secondary antibodies that are compatible with the LICOR Odyssey imaging system were used. Imaging parameters were adjusted based on the intensity of bands in the particular membrane. The band quantification of the LICOR software was used with export to table function to analyze.

Secondary antibody fluorophore brightness was periodically tested by binding dots representing each aliquot to a dry membrane. Secondary antibody recognition of primary antibodies was tested in a similar way by binding the primary antibody to a dry membrane, hydrating the membrane, blocking, and proceeding with the standard secondary antibody protocol.

J. FRET sensors

The cDNA for the RhoA FRET sensor (pTriEx-RhoA FLARE.sc Biosensor WT) was a gift from Klaus Hahn (Addgene plasmid no. 12150; http://n2t.net/addgene:12150; RRID:Addgene_12150). The cDNA for the VINC tension sensor was a gift from Martin Schwartz (Addgene plasmid no. 26019; http://n2t.net/addgene:26019; RRID:Addgene_26019). The bacteria were grown on penicillin selection plates. Small colonies were then expanded in liquid culture. cDNA was harvested and then purified using a DNA purification kit (Plasmid DNA MiniKit I, EZNA part no. D6943-02).

Cells were plated at least one day before transfection and allowed to grow to approximately 60%–80% confluence. On day 0, the DNA was transfected into the cells. Transfection was done utilizing Lipofectamine LTX or Lipofectamine 3000. The manufacturer supplied protocol was used. Briefly, the appropriate amount of DNA or RNA was mixed with plus reagent in Optimem media. Lipofectamine was also diluted in Optimem. The diluted DNA was added to the diluted Lipofectamine and incubated for at least 5 min. The mixture was then added to the cells in fully supplemented growth media, allowed to incubate for loading into the cells overnight. Media were changed the next day. Cells were then replated to the substrate of interest between day 2 and 4.

K. FRET image acquisition

Fixed cells were mounted with fluorescent protecting mounting media and then imaged after sitting overnight at room temperature. Images were collected with a 40× objective using equal exposure times for all channels. The donor, FRET, and acceptor were imaged using three separate cubes for excitation-emission: CFP-CFP, CFP-YFP, and YFP-YFP (Chroma). Exposure times of multiple seconds were used to ensure that the dynamic range of the camera was utilized. Processing was completed on TIFF images exported from the Leica software.

L. FRET image processing

A custom suite of the MATLAB code was used to process the image sets, which has been described previously.²⁵ Table 3 in the supplementary material⁸² lists the user-defined parameters utilized in the algorithm. These parameters were set permissively, and then a manual selection of focal adhesions was performed. The analysis using a custom written set of algorithms was also performed (code available upon request). Dark current, shading, and image drift images were collected on the microscope setup and processed by a parameterization module of the MATLAB suite, which produced inputs for image processing.

Imaging occurred on a wide-field fluorescence microscope with three separate filter pairs, one for the donor (CFP), one for the acceptor (YFP), and one for the donor excitation paired with acceptor emission. The exposure for each channel was equal. The intensity of the excitation source lamp was kept to 25% of maximum for live cell imaging. Because the donor and acceptor are conjugated to the same molecule, the molar ratio of the donor and acceptor in any given volume of the cell will be the same, leaving the ratio of FRET to donor the central equation required for calculating the FRET ratio. The image processing procedure is summarized as follows:

• (1)The variances in the light path that show in the image as shading were corrected in each channel by subtracting an image in a blank space with no objects in focus.
• (2)The background was established by collecting images of the plate or slide where there are no cells or constructs present.
• (3)FRET ratio was calculated as the ratio of the FRET (acceptor emission when the donor is excited) to the donor (excitation and emission) for each pixel.
• (4)A heat map based on the FRET ratio values was generated.

M. Statistics

Raw data were organized and mathematical transformations performed in Microsoft Excel or MATLAB, while the statistical analysis and graphing were conducted in GraphPad Prism (v8, San Diego, CA, USA). For experiments where different timepoints or treatment with inhibitors were compared, the ordinary two-way ANOVA was used (sphericity not assumed) with Tukey's multiple comparisons test applied with individual variances computed for each comparison.

III. RESULTS AND DISCUSSION

Electrospun fibers were constructed with the same parameters as in our previous work. Briefly, PMMA was dissolved at decreasing concentrations to increase the curvature of fibers with electric field and kinetic collector manipulations to increase the alignment.¹⁴ To increase throughput and the accuracy of attached cell measurements,²⁶ we utilized the automated cell morphology analysis tool CellProfiler.²⁴ When presented with low- and mid-range curvature fibers, the FA major and minor axes were increased compared to the flat surface control (Fig. 1, field and kinetic aligned fibers). However, as the fiber curvature continued to increase and the width of the adhesive area for focal adhesion decreased, both the width and the length of the vinculin positive area declined (2130/798, 2435/809, 2853/858, 3970/1046, 3598/1324, 2021/866; mean major/minor axis length for 0, 1, 1.3, 2.5, 4, and 10 μm⁻¹, respectively). Interestingly, although nanofibers would not limit the length of the FA, these data suggested another limitation to the length of FAs. We confirmed this trend by looking at the major axis length on two overlapping sets of nanofibers with different analysis techniques (see Fig. 1 in the supplementary material,⁸² field aligned fibers, and field and kinetic aligned fibers).

FIG. 1.

Focal adhesion major and minor axis dimensions increase until an inflection point in nanofiber subcellular curvature. (a) The focal adhesion major and minor axis as determined by an automated image analysis. *Major axis significant compared to flat (p < 0.0001 by two-way ANOVA), minor axis significant compared to flat (p < 0.001 by two-way ANOVA). N = 89, 34, 46, 126, 103, and 40 focal adhesions for 10, 4, 2.5, 1.34, 1, and 0 (flat) μm⁻¹, respectively, from 4, 2, 2, 4, 3, and 2 fields of view, respectively. Major axis denoted by squares, minor axis denoted by triangles. (b) Representative images of cells attached to the indicated substrate and stained for vinculin. Scale bar is 20 μm.

The intensity of focal adhesion molecule fluorescence varies within single focal adhesions.^27–29 We measured the distance between the maximal vinculin intensity and the geometric center of the projected FA area. Assuming that a higher fluorescence intensity correlates to a greater spatial density, the position of highest intensity relative to the area of the FA would reveal information about the structure of the FA. This distance ranged between 500 and 1000 nm and was higher than the flat surface for all nanofiber curvatures tested. The distance significantly increased with the fiber curvature until the same curvature as the major axis length reached a maximum [Figs. 1(a) and 2(a), field and kinetic aligned fibers]. This trend was conserved across multiple cell types. Although the absolute distance from the center decreased after the inflection point, when examined as the fraction of the major axis length, this distance trended greater than the flat surface (∼14%) and appeared to approach a maximum of 18% [Figs. 2(b) and 2(c)]. This relative shift in the vinculin maximum intensity may be indicative of a change in force applied by the cell to focal adhesions. At the protein level, the expression of vinculin in cells on nanofibers decreased on the higher curvature nanofibers in contrast to other cytoskeletal proteins α-tubulin and α-actinin [Figs. 3(a) and 3(b), field and kinetic aligned fibers], suggesting curvature-based regulation. Since vinculin tension stimulates FA growth and, therefore, greater cellular vinculin content, it was possible that the observed trend in vinculin expression correlates to vinculin tension. Cells were transfected with a FRET-based tension sensor, where the donor and acceptor molecules are linked by a tension-sensitive motif.³⁰ Thus, FRET inversely correlates with tension on this linker molecule. A trend of increased FRET on fiber substrates compared to the flat surface when comparing transfected cells was observed [Figs. 3(c)–3(e)]. The increased FRET observed on nanofiber substrates indicates a decreased tension in the FA and the cytoskeleton, suggesting that force is distributed to other force-bearing complexes when cells are migrating on electrospun nanofibers. Also, lower expression of vinculin on nanofiber substrates may be, at least in part, caused by the feedback loop from decreased FA tension.

FIG. 2.

Maximum intensity of vinculin shifts with respect to focal adhesion center as subcellular curvature increases. The distance of the maximum intensity of vinculin from the geometric center of adhesion on straight fibers, shown in (a) nanometers and (b) as a fraction of major axis length. (c) Representative images of focal adhesions, shown with intensity variation heat mapped (top) and as shades of gray (bottom). (a) and (b) ±SEM shown, N = 9, 7, 7, 7, 7, and 6 fields of view for 10, 4, 2.5, 1.34, 1, and 0 (flat) μm⁻¹, respectively, adjusted p-values: *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 by two-way ANOVA.

FIG. 3.

Modulation of cytoskeletal protein abundance and focal adhesion tension by nanofiber morphology. (a) and (b) Vinculin displays a greater decrease in nanofiber substrates compared among structural proteins. ±SD shown, n = 3 lanes per substrate. (c)–(e) Extension of mechanosensitive vinculin molecule as a function of the substrate. (c) and (d) Representative images of vinculin-tension sensor transfected cells showing the relative tension across the vinculin molecule. (e) FRET ratio measured on flat and fiber substrates (n = 161 and 90 for flat and fibers, respectively).

Knowing that the cell membrane may appear distinct in cells experiencing cytoskeletal tension compared to cells where intracellular pressure is dominate, we examined fiber-attached cells by electron microscopy. Cross sections of a cell attached to several nanofibers demonstrated invaginations into the membrane that matched expected sizes of nanofibers in diameter [Fig. 4(a), see Fig. 2 in the supplementary material],⁸² suggesting that nanofiber-attached cells have intracellular pressure as a dominate force.

FIG. 4.

Modulation of focal adhesion activation state by nanofiber subcellular curvature. (a) Cell attached to several fibers imaged with electron microscopy, open arrowhead: ∼6.67 μm⁻¹ curvature, filled arrow heads: ∼2–3.33 μm⁻¹. (b) Focal adhesion kinase (FAK) and Src phosphorylation patterns vary with the curvature of nanofibers.

FA signaling partners FAK and Src were assayed to understand how activation of the FA is affected by alterations in the FA structure on nanofibrous substrates. We utilized phosphorylation-specific antibodies on western blots to quantify one Src and three FAK activation sites in cells attached to nanofiber substrates. Phosphorylation at the 397 site on FAK was similar to the flat surface until higher curvatures where it decreased to nearly half of the flat substrate attached cells [Fig. 4(b), field aligned fibers]. Activation of FAK at 576/577 and Src 416 were below and above the flat surface control at low curvatures, but then followed a nearly identical trend from below the flat surface to above the flat surface at higher curvatures [Fig. 4(b)]. In contrast, FAK phosphorylation at 925 was the only activation site found to be greater than the flat surface control on all nanofiber curvatures tested and ranged between 2 and 8 times greater than the flat surface control [Fig. 4(b)]. Thus, the FA organizational and activation state is affected not only by the attachment to a nanofiber substrate but specifically by changes in the subcellular curvature.

The RhoA/ROCK signaling is a key pathway in response to the extracellular environment.^31–33 Varying levels of RhoA expression in whole mounted cells transfected with a FRET-based RhoA protein³⁴ were found [Fig. 5(a), field aligned fibers]. These differences were not significant, yet we wondered if there was any functional implication of these changes. We found that measuring human mesenchymal stem cell migration velocity during a range of 14–20 h of being attached to nanofibers with the ROCK inhibitor Y27632 resulted in significantly reduced migration velocity on two fiber curvatures [Fig. 5(b), see Fig. 3 in the supplementary material,⁸² field aligned fibers]. An hourly analysis suggested that much of this difference occurred during the 18th hour [Fig. 5(c)]. Therefore, we found that RhoA-mediated signaling was a driving force for nanofiber-attached human cell migration.

FIG. 5.

RhoA ROCK pathway stimulates random migration on nanofiber substrates. (a) Variation in whole cell-activated RhoA derived from fluorescence resonance energy transfer by microscopy of substrate attached cells. (b) Human mesenchymal stem cell migration velocities verses curvature during a 6-h imaging window (14–20 h postcell seeding). (c) This effect is shown for 1-h windows (black region in each stopwatch shaped chart covers the 1-h window that was averaged in the graph). All groups (treated and untreated) were normalized to the untreated flat group that displayed a velocity of 89.1 μm/h. ±SEM shown, n = 5 cells per condition, adjusted p-values: *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 by two-way ANOVA. Significant comparisons that occurred between fiber sizes are shown in supplemental figures (See Fig. 3⁸², field aligned fibers).

Cells contain multiple systems that combine to form functions. In this work, we looked at nanofiber attached cell characteristics toward a systems’ approach to understand human cell behavior on nanofibers. We observed multiple systems to vary with attachment to fibers of a range of subcellular curvatures when compared to a flat surface control (Fig. 6). Our findings corroborate an existing body of literature showing extracellular environment shape reflected in the shape of attached cells, the morphology of the cytoskeletal components and associated functions,^35–44 cell differentiation,⁴⁵ and the age of attached cells.⁴⁶

FIG. 6.

Graphical summary of findings figure summarizes the changes in size and intensity of vinculin-positive focal adhesions, vinculin expression, vinculin tension, cross-sectional cell shape, and migration relation to RhoA.

Since focal adhesion is a major signaling hub that is in close proximity to the extracellular environment, we wanted to investigate how it adapts to the nanofiber curvature. It may adapt in several ways, including the following. One, the size of focal adhesion may be dependent on physical dimensions and topology of the adhesive substrate. Two, the focal adhesion may adapt to the stiffness of the substrate. Three, the dynamic collection of molecules that construct the FA may adapt to temporal changes in the substrate. Our electrospun PMMA nanofiber substrates are nondegradable, and because the fibers are attached to a glass coverslip, they are not freely manipulated by migrating cells, resulting in functionally similar stiffness across curvatures. Only minimal temporal effects occurring in the deposition of proteins from the media are expected on PMMA substrates, primarily happening before our measurements and observations occurred. Furthermore, because all fiber and flat surface⁴⁷ samples are made with PMMA, both of which were smooth under SEM,⁴⁸ the surface chemistry and protein adsorption⁴⁹ should be similar.⁵⁰ Therefore, we expected any changes in focal adhesion parameters to arise from dimensional and topological factors.

We stained for the FA protein vinculin in cells attached to nanofiber substrates with controlled subcellular curvatures. Interestingly, the major and minor axes as well as the position of highest vinculin intensity were proportionally changing with curvature until ∼2.5 μm⁻¹. When looking at three curvatures (0.83, 1.43, and 2.5 μm⁻¹), Meehan and Nain also found a maximum focal adhesion length on ∼2.5 μm⁻¹ fibers (range of approximately 6–14 μm).⁵¹ It is known that interfocal adhesion arrangements of integrins change with time as the extracellular environment changes,^52,53 pointing to observed changes in intensity reflecting FA structural changes. To understand how the structural changes may be affecting the activation state of the FA, we looked at phosphorylation patterns on FAK and Src. Most striking were the changes in phosphorylation at FAK 925, increasing between 2- and 8-fold greater than the flat surface on all curvatures. Phospho-Src415 and pFAK397 were increased compared to flat, and pFAK576/577 was below flat surface control until ∼10 μm⁻¹ when all three decreased compared to flat and then diverged as the curvature increased. Examination of pFAK397 on ∼7.7 μm⁻¹ fibers,⁵⁴ FAK and Src activation on electrospun fibrin,⁵⁵ pFAK397 on isotropic and anisotropic nanofibers,⁵⁶ as well as on nano- and microfibers⁵⁷ have been reported. However, this may be the first report of FAK and Src activation states across multiple curvatures.

Using molecular sensors of force are among the only options for measuring force in cells attached to substrates such as rigid nanofibers that do not lend themselves to techniques such as traction force microscopy. FRET biosensors have been utilized in mechanobiology studies by constructing modified versions of a collection of proteins to probe different functions including FAK,⁵⁸ Src,⁵⁹ and Spleen tyrosine kinase.⁶⁰ To understand the magnitude of force experienced in focal adhesion, we selected a modified version of vinculin that includes a force-sensitive linker connecting two fluorophores that are a FRET pair. After a careful design process, Grashoff and Hoffman et al. found that the appropriately calibrated system was sensitive to pN forces and estimated the force in a stable FA to be approximately 2.5 pN.³⁰ Many important advances in our knowledge of forces including that required to construct a podosome,⁶¹ actin network force distribution upon stress-fiber ablation,⁶² and regulation of FA composition in a force dependent manner⁶³ have been made using this sensor. In our experiment, the tension across FA trended lower in human cells attached to fibers, suggesting the distribution of the force to other tension-bearing structures or a lower tension in the cell when attached to nanofibers compared to a flat surface. In murine cells, the cell area was two- to fourfold higher on similar nanofibers than on the flat surface control, suggesting a mechanism of cell tension that may be less dependent on vinculin than those previously reported.⁶⁴

Tension in the FA and cytoskeleton has been linked to propensity to migrate. Interestingly, in our studies, human cells were less stimulated to migrate by curvature than murine cells.¹⁴ The effect of ROCK inhibition was indeed larger than the effect of subcellular curvature. If combined with the findings of Yevick et al. that 0.025 μm⁻¹ and higher curvature fibers stimulated migration that tended to be faster than cells on a flat track,⁶⁵ the curvature across a large range stimulates migration compared to flat surfaces, depending on the cell type. Indeed, Guetta-Terrier et al. studied cell migration on aligned suspended fibronectin-coated PCL nanofibers and observed increased velocity on the fibers compared to a flat surface. Comparing migration on a micropatterned line that offered similar adhesive area (4 μm wide line and 0.77 μm⁻¹ curvature fibers), revealed an increase in directional migration on fibers despite similar velocities.⁶⁶

Published findings from the Brown Lab demonstrate that POR1 can inhibit active Rac1 if it is not bound to the membrane, an effect that was greater on very high curvature fibers (10 μm⁻¹) and to a lesser extent as the curvature decreased (3.33 μm⁻¹);⁶⁷ however, a mechanism for geometry sensing on lower curvature fibers (1 μm⁻¹) remains to be elucidated. It is possible that focal adhesion adapts to the curvature causing multiple changes that affect phenotypic changes. In this work, we found that at low curvatures, the distance between the intensity peak and the geometric center of focal adhesion and the effect of ROCK inhibition were notable, suggesting pathways that are involved in reacting to low curvature. Since low curvature fibers exhibit properties that can be desirable in certain contexts, understanding this curvature range is important for regenerative tissue engineering.

The existence of cellular membranes wrapping around fibrous substrates has been observed;^67,68 however, there are two ways in which this could happen. In a passive mechanism, intracellular pressure (driven by diffuse RhoA activation as observed in lobopodial migration) would push the cell membrane around a fiber (or any shape), and in the case of a cell on top of a fiber in vitro, the force of gravity would also work to pull a cell around a fiber. Conversely, in an active mechanism, force generation (driven by Rac-1 and cdc-42 lamellipodial type mechanisms) at the fiber would cause the membrane to extend around the fiber circumference. In the active membrane wrapping scenario, the cell membrane will be stretched by the forces applied at the leading edge (changes in cell membrane physiology contribute to local biochemistry, reviewed in Ref. 69). Interestingly, Mukherjee et al. recently reported an active membrane phenomenon termed coiling, where rapid activity occurs at the leading edge of cell protrusion exploring intersecting fibers.⁷⁰ Our results in both human and murine cells suggest that small GTPases are active participants in cells migrating on nanofibers, with Y27632 producing decreases in migration velocity in both cell types; however, it is not clear to what extent these signaling partners are participating in the membrane wrapping we have observed. Several groups have found that vinculin tension depends on Rho/ROCK (in ZA junctions)⁷¹ and myosinII,⁷² suggesting a link between our observations that should be investigated in the future.

Type 1 collagen organizes into fibers that have a curvature of ∼10 μm⁻¹,⁷³ while elastin-based fibers are approximately 1.25 μm⁻¹.^74–76 Therefore, the effect of fiber diameter in natural ECM is intimately linked to protein composition. However, collagen-binding integrins are thought to rely heavily on the helix structure of collagen^77–79 This raises the possibility that some integrins are more likely to bind based on shape rather than the molecular sequence. Although not conducted for different diameter fibers, we found that, relative to tubulin, PMMA fibers caused a slight decrease in integrins alpha5, beta3, and beta5 as compared to the flat surface control (data not shown). Subcellular curvature may exert an effect on integrins that requires more investigation.

IV. CONCLUSIONS

In this work, we sought to understand more about cell attachment and migration on nanofiber substrates as a tool for cell biologists with the ultimate goal of designing better regenerative engineering scaffolds to improve human health. Beginning with focal adhesion, the investigation then led to protein expression and the activation state, the shape of attached cells, and the migratory phenotype under ROCK inhibition. A maximum of focal adhesion size and shift in the vinculin intensity both occurred around 2–3 μm⁻¹. As fibers increased in curvature, measurements trended back toward flat surface values, except for migration velocity and phosphorylation at FAK position 925. Our results suggest a composite of migration modes, utilizing focal adhesions, but also having RhoA as a major driver, similar to other 3D migration findings. When designing nanofiber scaffolds, our results suggest that certain subcellular curvatures can be selected to obtain desired outcomes, such as larger focal adhesions or greater dependence of migration on Rho/ROCK signaling. Carefully selected nanofiber diameter may even affect the ability of contaminating bacteria to colonize a construct⁸⁰ and the intensity of implant fibrosis.⁸¹ We conclude that nanofiber-attached cell migration is unique and depends on the curvature of fibrils as a design input.

AUTHOR DECLARATIONS

Conflict of Interest

The authors have no conflicts to disclose.

Ethics Approval

Ethics approval is not required.

Author Contributions

Daniel T. Bowers: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Software (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). Mary Elizabeth McCulloch: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (supporting); Writing – review & editing (supporting). Justin L. Brown: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Funding acquisition (equal); Investigation (equal); Methodology (equal); Project administration (equal); Resources (equal); Supervision (equal); Validation (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal).

DATA AVAILABILITY

The data that support the findings of this study are available from the corresponding author upon reasonable request.