Material-Driven Fibronectin Assembly Rescues Matrix Defects due to Mutations in Collagen IV in Fibroblasts

Basement membranes (BMs) provide structural support to tissues and influence cell signaling. Mutations in COL4A1/COL4A2, a major BM component, cause eye, kidney and cerebrovascular disease, including stroke. Common variants in these genes are risk factors for intracerebral hemorrhage in the general population. However, the contribution of the matrix to the disease mechanism(s) and its effects on the biology of cells harboring a collagen IV mutation remain poorly understood. To shed light on this, we engineered controlled microenvironments using polymer biointerfaces coated with ECM proteins laminin or fibronectin (FN), to investigate the cellular phenotype of primary fibroblasts harboring a COL4A2+/G702D mutation. FN nanonetworks assembled on poly(ethyl acrylate) (PEA) induced increased deposition and assembly of collagen IV in COL4A2+/G702D cells, which was associated with reduced ER size and enhanced levels of protein chaperones such as BIP, suggesting increased protein folding capacity of cells. FN nanonetworks on PEA also partially rescued the reduced stiffness of the deposited matrix and cells, and enhanced cell adhesion through β1-mediated signaling and actin-myosin contractility, effectively rescuing some of the cellular phenotypes associated with COL4A1/4A2 mutations. Collectively, these results suggest that biomaterials are able to shape the matrix and cellular phenotype of the COL4A2+/G702D mutation in patient-derived cells.


Abstract
Basement membranes (BMs) provide structural support to tissues and influence cell signaling.
Mutations in COL4A1/COL4A2, a major BM component, cause eye, kidney and cerebrovascular disease, including stroke. Common variants in these genes are risk factors for intracerebral hemorrhage in the general population. However, the contribution of the matrix to the disease mechanism(s) and its effects on the biology of cells harboring a collagen IV mutation remain poorly understood. To shed light on this, we engineered controlled microenvironments using polymer biointerfaces coated with ECM proteins laminin or fibronectin (FN), to investigate the cellular phenotype of primary fibroblasts harboring a COL4A2 +/G702D mutation. FN nanonetworks assembled on poly(ethyl acrylate) (PEA) induced increased deposition and assembly of collagen IV in COL4A2 +/G702D cells, which was associated with reduced ER size and enhanced levels of protein chaperones such as BIP, suggesting increased protein folding capacity of cells. FN nanonetworks on PEA also partially rescued the reduced stiffness of the deposited matrix and cells, and enhanced cell adhesion through β 1mediated signaling and actin-myosin contractility, effectively rescuing some of the cellular phenotypes associated with COL4A1/4A2 mutations. Collectively, these results suggest that biomaterials are able to shape the matrix and cellular phenotype of the COL4A2 +/G702D mutation in patient-derived cells.

Introduction
Basement membranes (BMs) are specialized extracellular matrices (ECM) structures that provide structural support to tissues as well as influence cell behavior and signaling. [1][2][3] Major components are laminins, collagen IV, perlecan and nidogen. [2,4] The vertebrate genome encodes six collagen IV alpha chains, α1(IV)-α6(IV), encoded by the genes COL4A1-COL4A6 with reduced cerebrovascular disease. [8,[27][28][29] However, PBA did not ameliorate the renal defects, [27] with recent evidence from mouse models supporting cell and/or tissue-specific disease mechanisms whereby the contribution of ER stress and matrix defects to disease may differ. [13] This has led to the suggestion that the matrix defects also need to be targeted in order to develop treatments that address the phenotypes in different tissues. [27] However, the effects of COL4A1/4A2 mutations on BM function and characteristics remain poorly understood and, importantly, the effects of the matrix on the behavior and cellular phenotype of COL4A1/4A2 mutant cells remains unknown. Addressing these gaps in our knowledge could identify potential therapeutic targets and/or approaches.
Matrix composition, surface topography and the physical properties of the substrate influence the behavior of cells, and engineered biomaterials provide a powerful approach to modulate the microenvironment of cells and investigate their behavior and function in response to this altered environment. [30] In tissues, the ECM is a critical part of the cell environment, and fibronectin (FN) is a major ECM component that binds other matrix proteins, growth factors and cell surface receptors such as integrins, which are also collagen receptors. The biological activity of FN can be controlled by adsorbing it onto polymers with defined chemistry, such as poly(ethyl acrylate) (PEA) and poly(methyl acrylate) (PMA). Adsorption onto PEA results in physiological-like FN nanonetworks that provide better availability of cell and growth factors binding regions, while PMA leads to the formation of globular aggregates, Figure 1. [31][32][33][34][35][36] The ability to adsorb matrix components onto PEA and PMA, therefore, provides a powerful system to present the cell with controlled distribution and conformation of matrix proteins. For FN, the physical and chemical similarity of PEA and PMA provides a robust system to investigate the effect of the FN network as a model of an altered ECM on cells expressing mutant collagen IV. [36] Here, we have uncovered that FN nanonetworks assembled on PEA promote the deposition of collagen IV in primary patient cells carrying a COL4A2 +/G702D mutation. This is associated with an increase in the protein folding capacity of the cell, ameliorated stiffness of the cells and their matrix, and increased cell adhesion through focal adhesions. We show that these effects of FN are via specific integrin-mediated signaling. These data indicate that biomaterials can offer a controlled matrix to modulate the cellular phenotype of collagen IV mutations.
Adsorbed FN assembled into physiological-like nanonetworks on PEA, whilst it retained a globular conformation on PMA, with similar FN amounts adsorbed, Figure 1. To investigate the effects of the controlled matrix on the behavior of COL4A2 mutant cells, we assessed COL4A2 deposition. Image analysis revealed reduced COL4A2 deposition on glass after 1 and 7 days (Figure 2A-C) in MT cells compared to WT, as previously reported. [8,45] To provide the first insight into the secreted collagen IV network of COL4A1 or COL4A2 mutant cells, we also assessed the fractal dimension of the collagen IV network as a descriptor of its interconnected fibrillar organization ( Figure 2E). The collagen IV network of COL4A2 +/G702D fibroblasts on glass is characterized by reduced fractal dimension and a disrupted fibrillar organization; this confirms the "structural" effect of the glycine mutations on the collagen IV network. This was accompanied by a slower growth rate for the MT cells, [6] which were less confluent, with cell size often enlarged and formation of patchy populations indicative of cell death compared to WT cells ( Figure S1).

Figure 2.
Deposition of Col4A2 (green) and LM (red) by control (A) and mutant fibroblasts (B) on PEA and PMA coated with FN. Cells were grown on PEA and PMA substrates-coated with FN 20 µg/ml and on glass for 2 h under serum free conditions; then with serum before fixation at different time points (1 and 7 days). Cells were also simultaneously stained with DAPI (Blue). Scale bars: 50 µm (for all micrographs). Quantification of expressed Col4A2 (C) and LM (D) using integrated density per cell. Fractal dimension analysis of secreted Col4A2 (E). Data presented as mean ± SD, N ≥10; and analyzed with an ANOVA test; *p < 0.05; ***p<0.001. Important statistical significance differences between the WT and the MT cells are indicated, including between the substrates for the MT. WT, wild type control fibroblast cells; MT, COL4A2 +/G702D mutant fibroblast cells.
Strikingly, we observed that MT fibroblasts on FN-coated polymers displayed increased deposition of secreted collagen IV, especially after 7 days of culture ( Figure 2C). This was particularly significant for cells on the FN nanonetworks on PEA, where collagen IV deposition by MT cells was the highest and the fractal feature of the collagen meshwork was recovered ( Figure 2E). This enhanced deposition on FN-coated PEA compared to FN-coated PMA (where FN remains globular) and glass was confirmed by in-cell-western ( Figure S2A-B), ELISA ( Figure S2C) and fluorescent staining without cell permeabilization to avoid detecting intracellular proteins ( Figure S3). It was also independent on the presence of serum in the culture medium ( Figure S4). Increased matrix deposition was also apparent in electron microscopy images, which showed enhanced fibrillar matrix deposition in MT cells on PEA compared to PMA ( Figure S5). Moreover, this response appeared specific for FN-coated PEA, as LM-coated samples did not yield the same effect ( Figure S6). The increased deposition did not extend to other major BM components as LM protein levels decreased on FN-coated polymers ( Figure 2D), and was independent of increased bulk secretion ( Figure S7). We next investigated whether the higher collagen IV levels on FN-coated PEA were accompanied by a reduction in intracellular collagen IV retention. Co-staining was performed with protein disulphide isomerase (PDI), a marker of the endoplasmic reticulum (ER), as ER stress is associated with an increase in ER size and area of the cells. [8] This revealed significantly lower levels for MT cells on PEA-FN compared to MT cells on glass and PMA-FN, suggesting a decrease in ER area in MT cells on FN-coated PEA (Figure 3A and B). Swollen vesicles were also more apparent by electron microscopy in cells cultured on PMA compared to PEA ( Figure   S5).
The protein folding capacity of the cells is determined by the levels and activities of the protein folding machinery including chaperones such as BIP; for example, levels of BIP are associated with levels of protein secretion in yeast. [46] To investigate whether the decrease in ER area in the MT cells on FN-coated PEA was associated with increased protein folding capacity, the protein levels of chaperones BIP and calnexin were measured by western blot ( Figure 3C).
Interestingly, this revealed elevated BIP protein levels in both cell lines when grown on PEA-FN compared to PMA, indicating a mutation-independent effect. This suggests that PEA-FN may lead to an increase in the protein folding capacity of the cells.

Elastic properties of COL4A2 +/G702D fibroblasts and of their ECM
The characteristics of the matrix influences cell behavior and cell function including those of vascular cells. [47] However, the effect of any collagen IV mutation on the characteristics of the basement membrane remains unclear. To address this, the effect of the COL4A2 +/G702D mutation on the biomechanical properties of the cells and their matrix was analyzed via atomic force  Young's modulus of WT and MT cells on the different substrates measured via AFM force mapping; cells were cultured for 7 days and then indented with a cantilever mounted with a 4.83 µm silica bead (A). Young's modulus of ECMs obtained using the Hertz model on at least 20 measurements (N ≥20) taken from the points indicated by the yellow arrows in the AFM images C and D (B). AFM height images (first column) and 3D reconstruction (second column) of ECMs after decellularization of WT (C) and MT (D) cells; cells were cultured on FN-coated PEA and PMA for 7 days and then decellularized using 20 mM ammonium hydroxide (NH 4 OH) solution, leaving the ECMs intact; then, AFM quantitative imaging was carried out in DPBS using a pyramidal tip. The color scale of the 3D reconstruction represents the local Young's modulus of the ECMs, calculated using the Hertz model. All data are presented as mean ± SD, and analyzed with an ANOVA test; **p<0.01, ***p<0.001, ****p<0.0001. WT, wild type; MT, COL4A2 +/G702D cells.
Most interestingly, the measurement of the elastic properties of the decellularized matrix  and count (K) performed with ImageJ and an N =3. Data presented as mean ± SD, N ≥12, and analyzed with an ANOVA test; *p<0.05, **p<0.01, ***p<0.001. Important statistical significance differences between the WT and the MT cells are indicated, including between the substrates for the MT. WT, wild type; MT, COL4A2 +/G702D fibroblasts.
To quantify the maturation level of the FAs on the different surfaces, FA count and size (defined as the length of the major axis of the FA plaque) were analyzed (process detailed in Figure   S10). The number and size of FAs for MT cells were significantly higher on PEA-FN compared to glass and PMA-FN ( Figure 5D and E), whilst no differences were found for WT cells.
Strikingly, treatment with blebbistatin only affected the MT cells cultured on PEA-FN by reducing the number and size of the FAs to the same levels as the other surfaces ( Figure 5D and E), indicating myosin II-regulated adhesion. [48] It is also worth noting that ligand availability, as defined by the concentration of the FN coating solution, affected MT cell behavior, in terms of cell size and FA number only on PEA-FN whilst it had no effect for MT cells cultured on PMA-FN ( Figure S9D and E).
The increased size and number of FAs for MT cells on PEA-FN correlated with higher adhesion strength, measured using a spinning disk hydrodynamic shear assay, Figure 5F. [39,49,50] Detachment profiles (adherent fraction f versus shear stress τ) were fitted to a sigmoid curve to obtain the shear stress for 50% detachment (τ 50 ), which is defined as the adhesion strength ( Figure 5G). [51] MT cells showed statistically higher adhesion strength on PEA-FN compared to glass and PMA-FN. This was reduced when contractility was inhibited using blebbistatin ( Figure 5H, Figure S11), in accordance with FA analyses results ( Figure 5D and E) and supporting the role for myosin II-regulated adhesion on PEA-FN. Studies at increasing FN coating solution concentrations confirmed a direct relationship between ligand density and adhesion strength, which increased for PEA-FN ( Figure S9G and S12). [39,50,51] The role of cell cytoskeleton in adhesion to PEA was also confirmed via electron microscopy, which showed enhanced microfilament organization within cells cultured on PEA-FN compared to PMA-FN ( Figure S7).
Considering that cell adhesion to FN is mainly mediated by α 5 β 1 and α V β 3 integrins, [52] we explored their role in the stronger adhesion of MT cells to the FN nanonetworks on PEA compared to PMA-FN or glass. Both integrins were expressed by both cell types adhering on FN-coated substrates ( Figure 5A-B and Figure S13). In particular, β 1 staining showed wellpronounced clusters resembling FA contacts for the WT cells on all the surfaces. In contrast, it was rather dispersed throughout the MT cells ( Figure 5A As myosin II is important for the FA recruitment of focal adhesion kinase (FAK), [53] we explored the potential role of FAK via Western Blotting and immunofluorescent staining. Well-  [34,35] Besides integrin receptors, the disk-shaped receptor tyrosine kinases discoidin domain receptors (DDRs) [54] also bind collagen. [55] DDR1 is a tyrosine kinase transmembrane receptor that binds collagen IV and other collagens and regulates cell adhesion, differentiation, migration and proliferation. [54,55] As the effects of collagen IV mutations on DDR1 expression are unknown, we performed immunostaining which revealed DDR1 expression on the membrane surface for both cell types on all surfaces ( Figure S14D). Intriguingly, western blotting detected higher DDR1 protein levels for the MT cells on glass compared to the FN-coated substrates, and statistically higher than WT cells ( Figure S14E-F). This suggests an inverse correlation between DDR1 protein levels and amount of collagen IV deposition, whereby reduced collagen deposition due to the COL4A2 mutation leads to upregulation of DDR1. In particular, mouse ΔRGD FN (FN lacking the RGD motif) and mouse syn FN (FN with a mutation on the DRVPPSRN synergy binding site) were coated onto PEA, forming FN nanonetworks lacking the ability to bind both α 5 β 1 and α v β 3 (ΔRGD FN) or to reinforce binding to α 5 β 1 (syn FN), respectively ( Figure S15A). [56,57] Immunofluorescent staining confirmed higher deposition and fibrillar organization of collagen IV by MT fibroblasts only occurred on PEA coated with wild type FN, as both mutations reverted the rescue effect of the wild type FN nanonetworks, Figure 6. This suggests a major role for α 5 β 1 integrin. Interestingly, the unavailability of the cell binding and of the synergy domain on the FN nanonetwork only affected the behavior of MT cells, whilst WT fibroblasts were unaffected ( Figure S15B).
Moreover, the addition of blebbistatin to the media during the culture also ablated MT cells collagen IV deposition by inhibiting cell contractility, Figure 6.

Discussion
It is known that material properties, such as stiffness, topography and chemistry, can alter cell phenotype and drive cellular processes such as cell migration, cell signaling and (stem) cell differentiation. [58][59][60] These processes involve the secretion of new ECM at the cell-material interface, whose composition and nature are modulated by the properties of the biomaterials on which cells grow. Here, we demonstrate that engineered biomaterials have the potential to modulate matrix defects due to mutations in collagen IV in fibroblasts. The COL4A2 +/G702D mutation and other COL4A1/4A2 mutations result in lower deposition and incorporation of collagen IV into the BM. [8,45] Here, results show that FN nanonetworks assembled on the surface of PEA increase the deposition of COL4A2 in COL4A2 +/G702D mutant cells, with similar levels to WT (normal) cells. This phenomenon is shown by several approaches, immunofluorescence (including without permeabilization, to rule out the possibility of staining intracellular COL4A2 and after a decellularization assay), in-cell western and ELISA. We also demonstrate reduced ER stress (reduced ER area -PDI) in MT cells triggered by FN-PEA; this is accompanied by a likely increase in protein folding capacity, with increased levels of molecular chaperone BIP, Figure 7. these FAs contain higher density of β 1 integrins, which is the main receptor involved during the initial cell interaction with the material-driven FN network and physiological FN matrices. [34] FN adsorption on PEA unfolds the protein leading to the availability of domains that promote FN-FN interactions, such as FNI 1-5, enabling self-assembly into fibrils, recapitulating aspects of cell-mediated FN fibrillogenesis. [35,36] This material-driven FN matrix also favors enhanced exposure of the integrin-binding region FNIII 9-10 along with the growth factor binding region FNIII [12][13][14] . [31,33,60] FN assembled on PEA is recognized by β 1 integrins in COL4A2 +/G702D fibroblasts, Figure 5. Higher expression of β 1 has been previously correlated to α 5 β 1 integrinmediated adhesion to FN matrices, as it happens here for MT cells on FN-PEA, Figure 5. [51] The role of β 1 in rescuing the deposition of COL4A2 in MT cells is demonstrated by the use of mutant fibronectins, Figure 6. Binding of α 5 β 1 involves simultaneous availability of the synergy DRVPPSRN and RGD sequences within FN. [34] FNs lacking the RGD peptide or the synergy sequence are also assembled into nanonetworks on PEA ( Figure S15A), but they do not lead to enhanced matrix deposition of COL4A2 in MT cells, Figure 6. [33,36,61] Interestingly, our data are in agreement with recent in vivo data in C. elegans supporting the role of integrin in promoting the incorporation of collagen IV into basement membranes from secreted proteins, independently of actual increasing in collagen secretion. [62] Also, we show that the contractile machinery of the cells (ROCK) is activated by binding of β 1 integrins to FN networks on PEA as the use of blebbistatin reduces adhesion strength only on PEA ( Figure 5G), and further ablates the rescue of COL4A2 deposition, Figure 6. This ability of FN-assembled on PEA to trigger cell contractility has been previously demonstrated in cell differentiation processes. [35] Mechanical properties, especially stiffness, of cells and their surrounding ECM play important roles in many biological processes including cell growth, motility, division, differentiation, tissue homeostasis, stem cell differentiation, tumor formation and wound healing. [63] Changes in stiffness of live cells and ECM are often signs of changes in cell physiology or diseases in tissues. [64] Stiffness analysis via AFM showed that the MT cells and their ECMs were 10-fold softer than the WT cells, directly indicating for the first time effects on matrix stiffness due to collagen IV mutations, Figure 4A. However, MT cells were significantly stiffer on FN assembled on PEA, than on glass and FN-coated PMA. Cell stiffness is related to the networks of F-actin and intermediate filaments inside the cells, [39,49,65,66] which are generally observed in a lower amount in the MT than in the WT cells, but not on PEA-FN ( Figure 5 and Figure   S5). In addition to this, the ECM secreted by MT cells is also stiffer on PEA-FN ( Figure 4B), which can be related to the higher amount of deposited COL4A2 on PEA-FN forming interconnected fibrillar networks, Figure 2. [67] It is interesting to note that variants in COL4A1, the obligatory protein partner of COL4A2, are genetically associated with vascular stiffness. [17] This study has used patient-derived fibroblasts but can be extended to other type of collagen producing cells in relevant tissues and organs. Analysis into different cell types would include extensive genome editing, as altering collagen IV levels, for example by transfecting expression construct driving the COL4A2 G702D mutation, may cause defects of itself, hampering analysis.

Conclusion
We show that biomaterials alter the behavior of COL4A2 +/G702D mutant cells by overcoming some of the cellular and matrix defects caused by the mutation. Indeed, physiological-like FN nanonetworks assembled on a specific polymer chemistry (PEA) trigger contractility-dependent strengthening of MT cell adhesion through enhanced recruitment of β 1 integrins, leading to increased protein folding capacity, increased collagen IV deposition and partial rescue of the mechanical properties of the secreted matrix. Collectively, our results suggest that enhanced integrin signaling, controlled here via biomaterial engineering, influences aspects of the matrix and cellular phenotype of the COL4A2 +/G702D mutation in primary patient cells. These data enhance our understanding of the biological consequences (function/behavior) of COL4A2 mutations and highlight avenues for potential therapeutic approaches, which are critical to developing personalized therapeutic strategies for intracerebral hemorrhage and other pathologies due to collagen IV mutations.

Experimental Section
Preparation of polymer surfaces and protein adsorption. PEA and PMA polymers were Mouse plasma fibronectins carrying mutations in the FNIII 10 module were also used. ΔRGD FN (FN without the RGD sequence) and syn FN (FN with a mutation in the synergy sequence DRVPPSRN>DAVPPSAN) were generated and purified as previously reported. [37,38] The amount of adsorbed protein was calculated via depletion assay using the bicinchoninic acid working reagent (Thermo Fisher Scientific, Waltham, MA) as previously reported. [36] Cell culture. Primary dermal fibroblast harboring COL4A2 +G702D mutation (MT) and wild type and kept in an incubator at 37ºC with 5% CO 2 . L-ascorbic acid 2-phosphate sesquimagnesium salt hydrate (0.25 mM; Sigma) was administered before cells were fixed to standardize collagen expression and post-translational modifications. [8] Cell adhesion strength measurements. Cell adhesion strength to adsorbed FN on the polymer surfaces was measured using a spinning disk device. [39] Substrates (25 mm coverslips) were incubated with 2 or 20 µg/mL human FN for 30 min at RT and then incubated with 1% BSA for 30 min at RT. Cells (10,000 cells/cm 2 ) were seeded onto substrates and allowed to uniformly attach for several hours (as indicated in figure captions) in media with/without serum in the incubator. Some substrates were treated with 10 µM blebbistatin (B0560-Sigma-Aldrich) (inhibitor of myosin II). Samples were mounted on the spinning disk device, the chamber apparatus was filled with DPBS with 2 mM glucose, and the disk was spun for 5 min at a constant speed with controlled acceleration rates at RT. After spinning, cells were immediately fixed in 3.7% PFA, permeabilized with 1% Triton X-100, and stained with ethidium homodimer (Molecular Probes, Eugene, OR, USA), a DNA-specific fluorescent probe. Cells were counted automatically at specific radial positions using a Nikon TE300 equipped with a Ludl motorized stage, Spot-RT camera, and Image-Pro analysis system, at ×10 magnification. Sixty-one fields Samples were washed 3x, then mounted with Vectashield with DAPI to stain the nuclei and visualized using an epifluorescence microscope. Images were taken and channels merged using ImageJ (1.47v).
Fractal dimension analysis of stained secreted collagen IV was carried out using the ImageJ Fractal box count analysis tool, using box sizes of 2, 3, 4, 6, 8, 12, 16, 32, and 64 pixels after thresholding and binarization ( Figure S16). Quantification of integrated density was done using ImageJ; the integrated density of each picture was normalized using the number of cells (counted using DAPI) to obtain the integrated density per cell.
Focal adhesions were quantified using the focal adhesion analysis server. [40] Focal complexes, dot-like complexes shorter than 1 µm, were discarded from the analysis. [41] Only isolated cells were used to avoid altered area and roundness values that overlapping cells would have produced. Images were analyzed with ImageJ coupled with an in-house macro processor, and the values of each condition were compared.
Western blotting. Western blotting was performed as previously reported. [26] Protein extracts were prepared using RIPA buffer containing EDTA protease (Roche Applied Science) and the Young's modulus. [42,43] The stiffness of the ECMs secreted by the cells on the surfaces was also measured via AFM, using the quantitative imaging (QI) mode. To do so, the samples were Electron microscopy analysis. Cells were cultured on FN-coated PMA and PEA for 7 days and fixed in 2% glutaraldehyde prepared in 100 mM phosphate buffer (pH 7.0). Further processing was performed as previously described. [44] Statistical analysis. All images were analyzed using ImageJ software (v1.48). The data were statistically analyzed using GraphPad Prism 6 (GraphPad software, La Jolla, CA). Where relevant, one-way or two-way ANOVA tests were performed using a Bonferroni or Tukey posthoc test to compare all columns, and the differences between groups were considered significant for *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, and ****p ≤ 0.0001. Error bars represent a standard deviation.

Supporting Information
The following is the supporting data related to this article: supporting methods and 16 supporting figures. All the original data related to this manuscript are within the depository of the University of Glasgow with https://doi.org/10.5525/gla.researchdata.720.