Control of Adhesive Ligand Density for Modulation of Nucleus Pulposus Cell Phenotype

Abstract

Cells of the nucleus pulposus have been observed to undergo a shift from their notochordal-like juvenile phenotype to a more fibroblast-like state with age and maturation. It has been demonstrated that culture of degenerative adult human nucleus pulposus cells upon soft (< 1 kPa) full length laminin-containing hydrogel substrates promotes increased levels of a panel of markers associated with the juvenile nucleus pulposus cell phenotype. In the current work, we observed an ability to use soft polymeric substrates functionalized with short laminin-mimetic peptide sequences to recapitulate the behaviors elicited by soft, full-length laminin containing materials. Furthermore, our work suggests an ability to mimic features of soft systems through control of peptide density upon stiffer substrates. Specifically, results suggest that stiffer polymer-peptide hydrogel substrates can be used to promote the expression of a more juvenile-like phenotype for cells of the nucleus pulposus by reducing adhesive ligand presentation. Here we show how polymer stiffness combined with adhesive ligand presentation can be controlled to be supportive of nucleus pulposus cell phenotype and biosynthesis.

Introduction

It has been widely observed that age and maturation lead to significant structural and biochemical changes in the intervertebral disc (IVD) (Iatridis 1997, Roughley 2004, Adams 2006, Urban 2003, Wilson 2011) that are strongly linked to early changes in the nucleus pulposus (NP) (Roughley 2004, Adams 2006, Urban 2003, Kepler 2013). In their juvenile state, the notochordally-derived cells of the NP secrete a matrix that is largely distinct from the cells in the adjacent anulus fibrosus (AF) (Risbud 2015, Choi 2011, Trout 1982, Boos 2002). This matrix is highly hydrated (~90% water by wet weight in healthy tissue) and is rich in both non-fibrillar and fibrillar collagens (predominantly collagen II), proteoglycans, and an array of other proteins such as laminins and fibronectin (Choi 2016, Roughley 2004, Iatridis 1996). Degeneration-associated changes in the NP include loss of hydration and tissue stiffening with reported values for NP shear moduli of <1 kPa and 10–20 kPa in the healthy and moderately degenerated NP, respectively (Fearing 2019, Iatridis 1996, Iatridis 1997, Cloyd 2007, Walter 2017). These changes occur with a concomitant loss of sulfated glycosaminoglycans (sGAGs) and collagen II, increases in collagen I, reduced cell density, and phenotypic shifts in the notochordal-like NP cell population toward a more fibroblast-like state (Risbud 2015, Chen 2009). This is further observed with changes in gene expression profiles and biosynthesis of extracellular matrix for cells of the NP, and an associated inability to repair tissue damage or maintain a healthy microenvironment (Urban 2003, Trout 1982).

Control of substrate properties as a means for phenotypic regulation has been explored for a variety of applications ranging from cell differentiation (Huebsch 2010, Engler 2006) to modulation of bioactivity and metabolism (Gilchrist 2011, Enemchukwu 2016, Bridgen 2017, Karimi 2018). Studies such as those by Connelly and co-workers (Connelly 2008) and Burdick and co-workers (Burdick 2002) demonstrate an ability to promote changes in cell spreading and cytoskeletal organization in bone marrow stromal cells (BMSCs) and osteoblasts, respectively, by controlling the density of cell adhesive RGD peptide presented on a substrate. Another study by Maheshwari and co-workers demonstrated that similar control over cytoskeletal regulation could be achieved in immortalized murine fibroblasts not by manipulation of RGD density, but rather through control of its spatial presentation on poly(ethylene oxide) tethers (Maheshwari 2000). While integrin-mediated adhesions have been popularized through the use of the fibronectin-motif RGD (Humphries 2002, Zhu 2018, Phelps 2013) it has been established that cell-matrix interactions are mainly mediated by both integrins and syndecans (Couchman 1999, Morgan 2012, Karimi 2018). Importantly, work by Hozumi and co-workers using human dermal fibroblasts cultured atop substrates functionalized with EF1 and AG73 peptides suggested that concurrent engagement of these integrin- and syndecan-binding domains leads to synergistic adhesive effects (Hozumi 2010). It has further been hypothesized that syndecan-mediated interactions temporally precede integrin-binding interactions, and accordingly facilitate such mechanisms, thus promoting improved binding kinetics (Morgan 2012, Hozumi 2010, Couchman 1999).

Full-length laminins are large heterotrimeric proteins composed of an α, β, and γ domain (Yurchenco 2011, Nomizu 1995, Belkin 2000). These domains are known to interact with each other to formal basal membranes (Figure 1A), while the globular (LG) domains remain exposed and function as the primary site for cell interactions (Yurchenco 2011, Hohenester 2013). Laminins are present in the pericellular matrix of NP cells, and contain many adhesive sites known to interact with NP cells via integrins (including α3, α5, α6, β1, and β4), and syndecans (including syndecans 1 and 4) (Bridgen 2014, Nettles 2004, Morgan 2012, Binch 2016). Studies using disc cells have demonstrated that functionalization of stiff substrates with different full length proteins (collagen, laminin-111) promoted overall increased expression of the degenerative phenotype as observed by decreased metabolic activity, decreased levels of gene expression, and a fibroblast-like morphology (Fearing 2019, Gilchrist 2011, Francisco 2014). However, work has also demonstrated the ability for degenerative NP cells to re-express subsets of the juvenile phenotypic state following culture upon a laminin-111 functionalized soft poly(ethylene glycol) (PEG) substrate (Fearing 2019, Gilchrist 2011, Francisco 2014). Specifically, upon culture on soft PEG-LM hydrogels, it was observed that cells would form rounded multi-cell clusters similar to what is observed in native tissue (Cao 2007), with a concomitant increase in mRNA levels of a panel of genes associated with the juvenile cell state, including ACAN, COL2A1, and GLUT1 (Fearing 2019, Francisco 2014, Hwang 2014). Previous work has suggested an ability to reproduce some features of laminin-induced behaviors (Francisco 2014, Hwang 2014) by using short peptides (Gilchrist 2011, Bridgen 2017, Bridgen 2014). Specifically, work by Bridgen and co-workers (Bridgen 2017) demonstrated an ability to regulate cell adhesion and gene expression profiles in primary human NP cells by coupling single peptides to polyacrylamide hydrogels. While this study identified a number of peptides found within full length laminins based on their reported abilities to engage specific integrins, our interest here is in peptides found in or near the LG domains and reported to bind cells through a variety of integrin and syndecan receptors.

Figure 1: Mechanical testing and schematics.

A) α, β, and γ domains of the full-length laminin protein interact with each other to form basal membranes, while the LG domains remain available for cell interactions. B) Soft (4%, red) and stiff (15%, blue) gels have significantly different shear moduli. C) Incorporation of peptides into the PEG backbone does not impact modulus. D) ~90% coupling efficiency was observed using a fluorometric approach. E) Schematic of PEG backbone functionalization, hydrogel formation, and control of peptide density or substrate stiffness. *** p<0.001, using t-test (B, D), or one-way ANOVA with Dunnett’s multiple comparison’s test (C).

In the present work, we propose a dual-peptide-functionalized hydrogel scaffold with tunable mechanical properties and adhesive ligand presentation for control of NP cell morphology and phenotype. Maleimide-thiol (MAL-SH) coupling chemistry is employed for rapid, efficient, and stable peptide-conjugation and hydrogel formation in a PEG-based system (Darling 2016, Baldwin 2011). Our objective is to develop a stiff peptide-presenting and biocompatible hydrogel with an ability to support restoration of a juvenile cell phenotype for primary NP cells from degenerative human NP tissues.

Materials and Methods

Hydrogel preparation

Maleimide terminated 8-arm star poly(ethylene glycol) (PEG-8MAL, MW 20K, Creative PEGWorks, Durham, NC) was first dissolved in 1X PBS pH 3.25 for improved control of reaction kinetics (Darling 2016). Lyophilized, cysteine terminated IKVAV and AG73 peptides (full sequences for IKVAV and AG73: CSRARKQAASIKVAVSADR, and CGGRKRLQVQLSIRT respectively, GenScript, Piscataway, NJ) were likewise dissolved in 1X PBS pH 3.25. A maleimide-thiol Michael-type addition reaction was employed both for peptide conjugation and hydrogel formation. Peptide solution was first added to the PEG-8MAL solution in order to conjugate peptides to the PEG-8MAL backbone. A small PEG-dithiol (SH-PEG-SH, MW 600, Creative PEGWorks, Durham, NC) crosslinker dissolved in 1X PBS pH 3.25 was then added to initiate hydrogel formation (Figure 1E). Gels were then neutralized with 1X PBS pH 7.4 and allowed to swell to equilibrium volume. Hydrogel stiffnesses were controlled by changing the percentage PEG weight per total gel volume (%w/v).

Substrate characterization

For mechanical testing, PEG-peptide hydrogels were synthesized as described above. The gels were then cut into discs 8 mm in diameter and roughly 2 mm thick using a disc punch. All samples were tested in oscillatory shear using an AR-G2 Rheometer (TA Instruments, New Castle, DE). Samples were placed on a pre-heated plate, allowed to reach 37°C, and subjected to a pre-loading step of 0.015 N. Following, a 10% compressive strain was applied, and the samples were allowed a 2-minute conditioning step for relaxation. Samples were then subjected to oscillatory torsional strains (1 – 10 rad/s with a constant shear strain (γ) of 0.01), and complex shear modulus (|G*|) was reported for all samples at an angular frequency of 10 rad/s. To determine MAL-SH coupling efficiency, 775 μM SH-PEG-FITC (MW 10K, BiochemPEG, Watertown, MA) was added to the 8-arm PEG-MAL and allowed to react for 1 hour at room temperature in 1X PBS pH 7.4. The conjugated PEG-8MAL-FITC solution was then passed through a 10K filter spin column four times in order to remove any unconjugated FITC. Fluorescence was then measured (Ex/Em 488/525) using a plate reader (PerkinElmer EnSpire Multimode Reader, Waltham, MA) in order to calculate the concentration of conjugated PEG-8MAL-FITC.

NP cell isolation

Primary human adult NP cells (n≥3, both sexes, ages 35–72) were isolated from to-be-discarded tissue of patients undergoing surgical treatment for degeneration-associated complications (exempt from IRB review, Washington University Institutional Review Board) as previously described (Bridgen 2017). Briefly, NP fragments were identified, separated from the surrounding AF and cartilaginous tissue, and digested for 2–4 h at 37°C and 5% CO₂ in digestion medium containing 0.4% type 2 collagenase (Worthington Biochemical, Lakewood, NK) and 0.2% pronase (Roche, Basel, Switzerland). The digestion medium was then passed through a 70 μm filter. The remaining cell suspension was spun down for 10 minutes at 400 rcf, and the resulting cell pellet was resuspended and cultured in Ham’s F12 medium (Thermo Fisher Scientific, Waltham, MA) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin. Cells were passaged at least one time prior to usage and used prior to passage 4.

Morphological analysis

To study the dependence of ligand type on cell-matrix interactions, stiff (15% PEG w/v) gels were made with a total of 400 μM peptide in wells of a 16 well chamber slide (Nunc Lab-Tek Chamber Slide Systems™) using either IKVAV, AG73, or equimolar amounts of both. Primary adult human NP cells were seeded on the constructs at a density of 8,000 cells/well and allowed to attach for 24 hours at 37°C and 5% CO₂. Following the incubation period, cells were fixed with 4% paraformaldehyde (PFA) and stained with Alexa Fluor™-488 Phalloidin (Invitrogen™, Carlsbad, CA) for F-actin detection and 4′,6-diamidino-2-phenylindole (DAPI, 2 μg/mL, Sigma-Aldrich, St. Louis, MO) as a nuclear counterstain. Measures of cell body circularity, spread area, and clustering were obtained using the CellProfiler™ software. Data was stratified into single cells, small clusters (<5 cells), and large clusters (n>5 cells) in order to profile phenotypic differences due to cluster size.

Percent cell adhesion

In order to identify adequate working concentrations of total peptide, stiff hydrogel constructs were conjugated with peptide densities ranging from 0 μM (i.e. nonfunctionalized) to 400 μM peptide. Cells were seeded on these constructs at 8,000 cells/well in 16-well chamber slides and incubated for 24 hours at 37°C and 5% CO₂. Cells were then lysed, and percent cell adhesion was determined using the Cell-Titer GLO® (Promega, Madison, WI) plate reader assay following manufacturer protocols in a PerkinElmer EnSpire Multimode plate reader (Waltham, MA) and normalized to the number of cells seeded on the substrate.

Immunocytochemistry

In order to evaluate the ability to use stiff substrates to mimic the behavior of soft materials, both soft (4% PEG w/v) and stiff (15% PEG w/v) constructs were functionalized with either 50, 100, 200, or 400 μM peptide using equimolar amounts of IKVAV:AG73 as previously described. Markers were selected following recommendations from the Spine Research Interest Group as published in the 2015 consensus paper (Risbud 2015). Primary adult human NP cells were seeded at a density of 8,000 cells/well in wells of a 16-well chamber slide and cultured for 24 hours at 37°C and 5% CO₂. Cells were then fixed in 4% PFA for 10 minutes, rinsed with 1X PBS (+Ca, +Mg), and permeabilized with 0.2% TritonX-100 (Sigma-Aldrich, St. Louis, MO). Constructs were blocked with 3.75% bovine serum albumin (MilliporeSigma) and 5% goat serum (Thermo Fisher Scientific) and immunolabeled with either mouse-anti-N-Cadherin (1:150, Sigma-Aldrich), rabbit-anti-BASP1 (1:150, Abcam, Cambridge, United Kingdom), mouse-anti-PanCytokeratin (1:200, Sigma-Aldrich), rabbit-anti-Paxillin (1:100, Abcam), rabbit-anti-phospho-myosin light chain (pMLC, 1:100, Cell Signaling Technology, Danvers, MA), rat-anti-YAP-TAZ (1:100, Santa Cruz Biotechnology, Dallas, TX), or Alexa-conjugated phalloidin (1:250, Invitrogen). The proper isotype controls were used for each antibody. AlexaFluor™ (Invitrogen) secondary antibodies were applied using a dilution of 1:250, and cells were counterstained with DAPI (2 μg/mL, Sigma-Aldrich). F-actin fiber orientation was obtained by staining cells with DAPI and phalloidin, then analyzing the confocal images using the OrientationJ ImageJ plugin, where an output of one indicates strong fiber orientation in the same direction and a value of zero would reflect no fiber-alignment. For quantification of protein expression from immunostained samples, a minimum of three fields of view from no less than three independent human donors were analyzed. In each sample, cell outlines were traced, and mean fluorescence intensity signal was quantified by obtaining the raw signal intensity, normalizing to the cell area, and subtracting background (ImageJ, National Institutes of Health, Bethesda, MD). Ratio of nuclear-to-cytosolic YAP localization was likewise analyzed using ImageJ. Paxillin data was quantified following the protocols outlined by Horzum et al. (Horzum 2014).

Gene expression

Gene expression was assayed using qPCR on an Applied Biosystems™ StepOnePlus™ Real-Time PCR System (Software v2.3, Foster City, CA). As above, markers were selected following the recommendations set in the 2015 consensus paper (Risbud 2015). Briefly, 100,000 primary adult human NP cells were seeded on the appropriate gels made in wells of a 4-well chamber slide in duplicate and cultured for 4 days at 37°C and 5% CO₂. Following the incubation period, the cells were lysed using RLT buffer (Qiagen, Hilden, Germany) + 1% mercaptoethanol and stored at −80°C until ready for RNA isolation. Isolation was done using the QIAGEN ™ Mini kit following manufacturer instructions. Briefly, samples were homogenized using a QIAshredder™ column and passed through an RNeasy spin column in order to bind RNA to the membrane. The sample was then washed, and DNA was digested using DNAse I. RNA was eluted with RNAse-free water. RNA concentration and purity were determined using the 260/280 ratio in a NanoDrop™ system (ThermoFisher Scientific, Waltham, MA). RNA was then converted to cDNA using the iScript cDNA Synthesis Kit (BioRad, Hercules, CA). qPCR was used to detect amplification of aggrecan (ACAN), collagen 2 (COL2A1), N-Cadherin (CDH2), and glucose transporter 1 (GLUT1) (Supplementary Table 1, Applied Biosystems) using the ΔΔCt method, with the first Δ being normalization of the gene to housekeeping genes 18S and GAPDH, and the second Δ being normalization to a control substrate, in this case tissue culture polystyrene (TCPS, Figure 5) or soft PEG-laminin hydrogels (Figure 6).

Figure 5: Stiff 100 μM substrates promote a more juvenile-like phenotype.

A) qPCR for a panel of genes associated with the juvenile NP phenotype. B-C) Quantification of protein expression and representative images of a panel of proteins associated with the juvenile NP phenotype. n≥3 human donors. * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001. One-way ANOVA with Tukey’s multiple comparison’s test (A, B). Scale bar is 50 μm.

Figure 6: Stiff 100 μM substrates promote a more juvenile-like phenotype.

A) qPCR for a panel of genes associated with the juvenile NP phenotype. B-C) Quantification of protein expression and representative images of a panel of proteins associated with the juvenile NP phenotype. n≥3 human donors. * p<0.05, ** p<0.01, *** p<0.001, using one-way ANOVA with Tukey’s multiple comparison’s test (A, B). Scale bar is 50 μm.

Biosynthesis assays

sGAG production was assayed using 1,9-dimethyl-methylene blue zinc chloride (DMMB, Sigma-Aldrich). Briefly, primary human NP cells were seeded on gels made in wells of a 96 well plate and cultured for 4 days at 37°C and 5% CO₂. The entire cell-gel construct was then digested in papain buffer (5 mM EDTA, 5 mM L-Cysteine, 125 μg/mL papain in PBS) overnight at 60°C. sGAG concentration in the digested constructs + supernatant media was determined against chondroitin sulfate (Sigma-Aldrich) standards and normalized to DNA content obtained using the Quant-iT ™ PicoGreen® dsDNA kit (Invitrogen) following manufacturer recommendations. Hydrogel constructs without cells were used as negative controls to correct for any colorimetric interference introduced by the gels.

Results

Substrate characterization

Mechanical testing of the hydrogel constructs indicated moduli of ~500 Pa for the soft (4% PEG w/v) substrates, and >10kPa for the stiff (15% PEG w/v) substrates (Figure 1B), indicating significant differences in stiffness with biological relevance, and similar in magnitude to those previously reported to elicit differential behaviors in NP cells (Bridgen 2017). Inclusion of peptides at all concentrations tested did not lead to significant changes in hydrogel stiffness (Figure 1C). Coupling efficiency through the quantification of a fluorescent FITC tag indicated that >90% of the added SH-PEG-FITC were bound (Figure 1D). These results were further confirmed through the use of a maleimide quantification kit following manufacturer recommendations (data not shown).

Co-presentation of syndecan- and integrin-binding peptides promotes improved cell attachment and morphological differences

Hydrogels with IKVAV alone, AG73 alone, or equimolar amounts of IKVAV:AG73 were made directly in wells of a 16 well chamber slide. It was observed that co-inclusion of the integrin- and syndecan-binding domains led to significant increases in adherent cell numbers compared to either single peptide system (Figure 2A). Furthermore, both AG73-only and IKVAV-only substrates promoted attachment predominantly as single cells, while equimolar inclusion of AG73 and IKVAV was able to promote increased cluster formation (defined as bodies containing >5 nuclei, Figure 2B). Distinct morphologies were observed on the different substrates. Cells on AG73-only gels displayed a more rounded, less spread phenotype, while cells cultured atop the IKVAV-only substrates exhibited a more fibroblast-like phenotype with strong protrusions being observed (Figures 2C, D, E).

Figure 2: Equimolar inclusion of IKVAV and AG73 leads to improved adhesion and morphological differences.

A) Percent cell attachment in single- and dual-peptide coupled substrates. B) Characterization of cell clustering. C-D) Cell spreading and circularity behaviors of single cells and large clusters upon culture on peptide functionalized substrates. E) Representative immunostaining images of primary human NP cells, green is actin, blue is nuclei, scale bar is 20 μm. n≥3 human donors. * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001, using a one-way ANOVA with Tukey’s multiple comparison’s test (A, C, D).

Ligand density controls morphological behaviors in both soft and stiff hydrogel systems

Given the synergistic effects on cell adhesion and morphology previously described, we sought to investigate the roles of substrate stiffness and ligand density on cell behaviors. Hydrogels were synthesized at both ~500 Pa (4% PEG w/v, “soft”) and ~10 kPa (15% PEG w/v, “stiff”), each with equimolar amounts of IKVAV and AG73 peptides. At 50 μM, less than ~40% cell adhesion was observed at either stiffness. This increased to ~75% at 100 μM peptide. No significant differences were observed in cell adhesion with increasing peptide density above 100 μM, with significance only being observed between all peptide densities and the 50 μM conditions (Figure 3A, Supplementary Figure S1). A trend toward increased cluster formation was also observed as a function of peptide density (Figure 3B). Increasing ligand density led to significantly increased cell spreading and decreased circularity (Figure 3C, D). Specifically, these data suggest significant differences between the 100 μM and 400 μM conditions at both stiffness levels. To test the hypothesis that stiff low peptide systems can elicit responses similar to those observed in the soft conditions, these data were re-plotted to compare all soft substrates to the stiff 100 μM condition. Here, we observed that the decreased peptide density stiff system exhibited little to no differences to the soft constructs (Figure 3G). Cells cultured upon stiff substrates at 400 μM total peptide were observed to undergo a morphological shift toward a more fibroblast-like spread morphology, with higher degrees of cell spreading and significantly decreased circularity. More aligned fibrillar actin structures became evident on stiff substrates at 400 μM peptide, while stiff 100 μM and soft substrates exhibited more cortical actin (Figures 3E, F). The stiff 400 μM condition along with TCPS promoted the formation of paxillin-positive mature focal adhesions (FAs), while no other substrate (soft 400 μM, soft PEG-LM, or stiff 100 μM) led to these observations (Figure 4A, D, Supplementary Figure S2). We further observed a correlation between cell area, circularity, and paxillin presence, with more spread and less rounded cells promoting higher degrees of paxillin presence (Supplementary Figure S2). A significant increase in pMLC positive cells on stiff 400 μM compared to stiff 100 μM substrates further suggests a shift towards a more fibroblast-like contractile phenotype (Figure 4C, D).

Figure 3: Ligand density controls morphological behaviors in both soft (red) and stiff (blue) hydrogel systems.

A-B) Increasing peptide density leads to increased cell adhesion, and trends towards increased cluster formation. C-D) Observe increases in cell spreading and paralleled decreases in cell circularity for both soft and stiff hydrogels. E-F) Stiff 400 μM substrates lead to increased actin fiber alignment, while soft substrates and stiff 100 μM lead to more cortical actin. Dashed lines show values of cells upon TCPS (gray) and 4% PEG-LM (green) for reference. For immunostaining, green is actin, blue is nuclei, scale bar is 20 μm. Bottom row shows color survey of actin alignment with fibers of equal orientations represented in similar colors. G) Stiff 100 μM substrates demonstrate no significant differences in cell spreading and circularity versus soft gels at 100, 200, or 400 μM. n≥3 human donors. * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001, using one-way ANOVA with Tukey’s multiple comparison’s test (A, C, D, F), or Dunnett’s multiple comparison’s test (G).

Figure 4: Stiff 400 μM gel promote a more contractile phenotype.

A) Quantification of paxillin rich focal adhesions per condition. B) Simple linear regression model with 95% confidence bands for stiff 100 and 400 μM substrates suggests correlation between cell spreading and paxillin presence. C) Immunopositivity of pMLC in primary adult human NP cells. D) Representative immunostaining of paxillin and pMLC. n≥3 human donors. * p<0.05, *** p<0.001, using one-way ANOVA with Tukey’s multiple comparison’s test (A), or t-test (C). Modified ANCOVA used for testing difference of slopes (B). Scale bar is 20 μm.

It has been previously suggested that stiff substrates promote the observed fibroblast-like behavior (Fearing 2019, Gilchrist 2011, Bridgen 2017, Francisco 2014). Interestingly, we observed that degenerative primary NP cells cultured upon stiff substrates at 100 μM total peptide underwent a shift back towards a more “juvenile” rounded morphology, exhibiting lower degrees of spreading, decreased actin alignment, and increased cluster circularity (Figure 3E, G), with increased cytoplasmic vacuole presence becoming apparent (Supplementary Figure S4). Cytoplasmic vacuoles are a consistent marker of the notochord and notochordally derived NP cells which have been observed to be lost as NP cells become more chondrocyte-like with tissue maturation (Trout 1982, Hong 2018). When comparing the stiff 100 μM data to the stiff 400 μM data, we observed significant differences between the slopes of the simple linear regression fits of total spread area versus total paxillin area, suggesting a correlation between cell spreading and paxillin presence (Figure 4B). The behaviors observed in stiff 100 μM systems were consistent with those observed in soft substrates at any peptide density. The data suggests that stiff hydrogel systems with decreased adhesive-ligand presentation are able to guide degenerative cells toward a more juvenile-like morphology in a similar manner as the soft peptide-coupled substrates.

Decreasing peptide density on stiff hydrogel substrates promotes altered gene and protein expression levels versus high peptide density stiff constructs and TCPS

Primary adult human NP cells were cultured upon either stiff hydrogels functionalized with 100 or 400 μM total peptide (equimolar amounts of IKVAV:AG73), or upon TCPS. Changes in mRNA levels of a panel of genes (Supplementary Table 1) associated with the juvenile phenotype of NP cells were studied using qPCR. It was observed that the stiff 100 μM condition led to higher mRNA levels of all the pro-juvenile genes studied (ACAN, COL2A1, CDH2, GLUT1) than the stiff 400 μM system and TCPS (Figure 5A). Quantification of immunocytochemistry of proteins associated with the juvenile phenotype also suggested significantly increased protein expression levels in cells cultured atop stiff 100 μM than both stiff 400 μM and TCPS (Figure 5B, C).

Stiff low peptide systems promote gene and protein expression levels similar to those observed on soft peptide-functionalized and full-length laminin functionalized constructs

Soft laminin-conjugated PEG hydrogels (PEG-LM) have been previously demonstrated to promote expression of a more juvenile phenotype in degenerative NP cells over other substrate types such as stiff PEG-LM, LM-coated TCPS, and collagen gels (Fearing 2019, Francisco 2014, Bridgen 2014). Thus, soft PEG-LM was chosen as the positive control for these studies. It was found that the stiff 100 μM and soft 400 μM PEG-peptide systems had similar to or higher levels of gene expression than those observed in the soft PEG-LM control (Figure 6A). Matrix deposition genes ACAN and COL2A1 were significantly upregulated in both PEG-peptide systems versus soft PEG-LM. Expression of CDH2 was highest in the stiff 100 μM system, and GLUT1 expression was modestly higher in both PEG-peptide-gels when compared to soft PEG-LM, although these differences did not reach statistical significance due to high patient-to-patient variability. Levels of protein expression in PEG-peptide systems were also similar to or higher than those observed in the soft PEG-LM control (Figure 6B, C). A modest decrease in cell viability was observed in both stiff 100 μM and soft 400 μM peptide systems versus soft PEG-LM, although levels of sGAG production were similar to that of soft PEG-LM (Supplementary Figure S3). A limitation of this dataset is that in order to focus on the goal of understanding the stiff low peptide density system behaviors, sGAG production was not measured at every substrate stiffness and peptide density. The question about the means by which the peptide-presenting gels promote similar effects to the full-length laminin is one we have begun to consider in this study. While ligand accessibility, cell shape, and adhesive domain type and concentration may all play a role in the observed substrate effects, additional experiments beyond the scope of this work are needed to fully elucidate the mechanism.

Discussion

Previous studies of varying cell types have suggested that mechanical cues of microenvironment can be used to modulate cell phenotype (Fearing 2019, Engler 2006, Gilchrist 2011, Enemchukwu 2016, Mao 2016). Recent work in IVD cells has further suggested that cell shape may be an important regulator of phenotype, as observed by close interactions between cell shape and regulatory activities of YAP, TEAD, SRF, and other factors such as matrix deposition and gene expression profiles (Fearing 2019, Bonnevie 2019). Specifically, previous work has reported an ability to induce phenotypic shifts in degenerative NP cells toward a juvenile-like state through culture upon soft PEG-LM substrates, while their stiffer counterparts led to a more degenerative-like phenotype (Fearing 2019, Gilchrist 2011, Hwang 2014, Francisco 2014). The current work supports and further builds upon these findings by demonstrating tunable peptide-conjugated hydrogel constructs able to promote phenotypic shifts both by control of bulk substrate stiffness, and by control of adhesive ligand presentation. A major finding of this work was the ability to promote increased expression of juvenile markers of NP cells by controlling the adhesive ligand presentation rather than the stiffness of the underlying scaffold.

It has been previously hypothesized that large degrees of crosstalk and co-dependence exist between different cell-adhesive domains such as the integrin-binding and syndecan-binding domains studied here (Morgan 2012, Hozumi 2010, Roper 2012). Functionalization of hydrogels with either integrin-binding or syndecan-binding peptides alone led to different morphologies and adhesion profiles. The co-presentation of the integrin- and syndecan-binding domains led to improved cell adhesion and morphologies that were an intermediate between the two single peptide outcomes. Prior work on integrin-syndecan crosstalk has suggested that the long heparan sulfate chains of syndecans, which may extend up to 500 nm, enable interactions between the cell and its surroundings (Roper 2012). These syndecan-mediated interactions appear to facilitate integrin-mediated adhesions, leading to cell-membrane protrusions that enable integrin-rich domains at the polymerizing tips of actin that initiate formation of the stronger cell-matrix contacts (Roper 2012, Kam 2002). Importantly, work carried out using the A375-SM human melanoma cell line, which looked at pathway activation in different combinations of syndecans and integrin dimers, elucidated that crosstalk between integrins and syndecans is highly specific, with only particular combinations of receptor pairs leading to pathway activation (Mostafavi-Pour 2003). The integrin-binding peptide used in our studies, IKVAV, is thought to bind through a number of integrin dimers containing the chains α3, α4, α6, and β1 depending on cell type (Frith 2012, Kikkawa 2013), although there exist discrepancies on this matter in the literature. Similarly, AG73 has been proposed to bind through syndecans 1, 2, and 4 also depending on cell type (Kikkawa 2013). We thus believe that the broad range of adhesive domains that can be engaged by the IKVAV-AG73 pair studied here likely plays a role in enabling the observed synergistic effects upon dual peptide coupling.

The FA adapter protein paxillin has been suggested to be responsible for recruitment of structural and signaling molecules to adhesive sites (Lopez-Colome 2017). The focal adhesion kinase (FAK)-paxillin complex has further been suggested to play a role in Rho-family GTPase activity and can lead to actin regulation (Bachir 2017). Integrin engagement with the extracellular matrix leads to paxillin phosphorylation and promotes the assembly of FAs (Zaidel-Bar 2006). Syndecans, specifically syndecan-4, have also been observed to directly interact with paxillin via the cytoplasmic syndesmos proteins (Denhez 2002). Recent work looking at substrate stiffness and its regulation of NP focal adhesion formation suggested that inhibition of FAK led to decreased actin bundling and increased gene expression of ACAN, COL2A1, and MMP13 amongst others (Krouwels 2018). In our study, stiff 400 μM substrates as well as TCPS led to the formation of paxillin-rich focal adhesions and strongly aligned F-actin filaments, while soft gels or stiff 100 μM substrates did not promote cell spreading, stress fibers, or mature focal contacts. Because both integrins and syndecans have direct interactions with FA proteins, specifically paxillin, decreasing the availability of adhesive domains (e.g. 400 μM to 100 μM) may lead to the observed inhibition of FA formation and the subsequent observed effects on cell shape. The data presented here thus suggest that cell shape may play a role in the phenotypic state of NP cells.

Phosphorylated myosin light chain (pMLC) is known to be associated with actin filament formation, cellular stiffness, and migration, and is essential for the initiation of cellular contraction (Ikebe 1985, Ikebe 1988, Hirano 2016). We observed that upon culture on our stiff 400 μM peptide-conjugated substrates, over 30% of the cells stained positive for pMLC, while fewer than 10% of the cells on stiff 100 μM substrates stained positive. Interestingly, not all cells within a cluster stained positive for pMLC, an observation that was repeated in focal adhesion formations (Figure 4D). These results may suggest that on stiff 400 μM peptide-conjugated substrates, cells on the periphery of clusters continuously probe their surroundings leading to high degrees of spreading and the observed immunostaining patterns. However, on the stiff 100 μM substrates, the reduced presentation of adhesive ligands and consequent inhibition of focal contacts and stress fiber formation prevents this from occurring.

For cells cultured on stiff 400 μM peptide-conjugated substrates, cells were observed to spread and simultaneously display higher degrees of nuclear translocation of YAP; in contrast, cells on stiff 100 μM peptide-conjugated substrates as well as all soft substrates tested exhibited a more rounded and less spread morphology along with lower levels of nuclear translocation and thus more cytosolic YAP (Supplementary Figure S5). These findings are consistent with the prior observations for IVD cells that cell shape may be a dominant driver of YAP/TAZ activation and translocation to the nucleus, not substrate stiffness alone (Bonnevie 2019, Fearing 2019). Indeed, a recent study showed that in the absence of distinct morphological changes, a targeted YAP siRNA knockdown was not enough to elicit effects on transactivation of the downstream targets TEAD or SRF that regulate fibroblast-like phenotypes in many cell types; thus, YAP alone may not be a true indicator of how cells sense or transduce stimuli in their microenvironment (Fearing 2019). Although further assessment of upstream signaling molecules may be necessary, our observations of elevated nuclear localization of YAP in combination with highly spread morphologies may still be indicative of an altered mechanoresponsive state.

All substrates in this study promoted cell-cell contacts, as observed by higher numbers of cell clusters on all peptide-conjugated PEG hydrogels. It has previously been observed that as the number of one type of interaction (e.g. cell-cell contacts) increases, the number of other contacts (e.g. cell-matrix) is reduced (Cosgrove 2016, Weber 2011). Interpreting this observation in light of our study findings for NP cells, we conclude that both soft and stiff-low peptide presenting substrates inhibit cell-matrix interactions and drive cells to higher levels of cell-cell contacts. The induced decrease in cell-matrix interactions is likely responsible for the reduced ability to form mature focal adhesions and stress fibers. We and others have previously shown N-cadherin to be a key player in modulating cell-cell contacts that drive the re-expression of “juvenile” phenotypes for cells of the NP (Hwang 2014, Yuan 2018). Indeed, the ability for a peptide-hydrogel system to promote increased expression of N-cadherin in low-expressing cells (i.e. pathological adult NP cells), may be a key driver of healthy and functional cell-cell contacts that promote the appropriate cell phenotype. Moving forward, it may be beneficial to support integrin-, syndecan- and cadherin-mediated interactions through the functionalization of hydrogel systems with adhesive domains derived from N-cadherin, as has been explored by Burdick and co-authors for chondrocytes (Cosgrove 2016, Bian 2013).

Conclusions

In the present study, we demonstrate that close control of peptide selection and peptide presentation can lead to a more juvenile NP cell phenotype regardless of substrate stiffness. The absence of mature focal contacts and stress fibers observed in the substrates that promoted the most juvenile-like phenotype may further indicate that reduced cell contractility and the spatial control of cell-matrix interactions may be associated with the observed phenotypic shifts. Taken together, our results suggest that human NP cell phenotype may be directly regulated by cell shape as observed by the changes in protein and gene expression levels associated with specific cell morphologies, and that these morphologies may be observed regardless of substrate stiffness by modulation of ligand presentation.