.
O P E N A C C E S S S O U R C E : PNAS
Significance
The mechanical reprogramming of fibroblasts, followed by their redifferentiation into rejuvenated fibroblasts in an optimized 3D collagen matrix, made these cells more contractile and more efficient at synthesizing matrix components including laminin, fibronectin, and collagen-IV. Moreover, the rejuvenated fibroblasts obtained through this approach exhibited a decrease in DNA damage. The rejuvenated fibroblasts derived from this method precisely align into tissue architectures, suggesting its potential application as clinical implants in tissue engineering and regenerative medicine.
Abstract
Over the course of the aging process, fibroblasts lose contractility, leading to reduced connective-tissue stiffness. A promising therapeutic avenue for functional rejuvenation of connective tissue is reprogrammed fibroblast replacement, although major hurdles still remain. Toward this, we recently demonstrated that the laterally confined growth of fibroblasts on micropatterned substrates induces stem-cell-like spheroids. In this study, we embedded these partially reprogrammed spheroids in collagen-I matrices of varying densities, mimicking different three-dimensional (3D) tissue constraints. In response to such matrix constraints, these spheroids regained their fibroblastic properties and sprouted to form 3D connective-tissue networks. Interestingly, we found that these differentiated fibroblasts exhibit reduced DNA damage, enhanced cytoskeletal gene expression, and actomyosin contractility. In addition, the rejuvenated fibroblasts show increased matrix protein (fibronectin and laminin) deposition and collagen remodeling compared to the parental fibroblast tissue network. Furthermore, we show that the partially reprogrammed cells have comparatively open chromatin compaction states and may be more poised to redifferentiate into contractile fibroblasts in 3D-collagen matrix. Collectively, our results highlight efficient fibroblast rejuvenation through laterally confined reprogramming, which has important implications in regenerative medicine.
Fibroblasts are vital constituents of the connective tissue, which provide mechanical strength and maintain tissue homeostasis by promoting extracellular matrix (ECM) remodeling (1, 2). During aging, fibroblasts reduce their actomyosin contractility and matrix remodeling efficiencies (3⇓⇓–6). Transplanting of stem cells and induced pluripotent stem cells (iPSCs) are being seen as potential cellular-therapy models for rejuvenating fibroblast function (7⇓–9). However, these interventions not only rejuvenate, but have been found to acquire genomic mutations that may increase the oncogenic potential of the proliferative fibroblasts, and major efforts are underway to improve the limitations of such methods (10). Therefore, for therapeutic purposes, it would also be ideal to rejuvenate fibroblasts using nongenetic methods.
Recently, we showed that sustained, laterally confined growth of fibroblasts on micropatterned substrates induced their reprogramming into stem-cell-like cells, even in the absence of any genetic or biochemical interventions (11). Such partially reprogrammed cells (PRs) not only exhibited stem-cell-like characteristics, but also retained their differentiation states to some extent, making them a potential model for fibroblast rejuvenation in connective tissues. As a major constituent of connective tissue, collagen-I concentration primarily regulates matrix stiffness and controls cellular processes such as contraction, adhesion, and migration via its interaction with fibroblasts (12). Furthermore, in a three-dimensional (3D) gel, matrix fibers are intertwined into a mesh-like structure, and the porosity of the mesh regulates initial cell spreading and migration through the entangled fibrils (13). Therefore, 3D collagen matrices with appropriate steric (porosity) and mechanical (stiffness) features that closely resemble fibrous connective tissue will be ideal for exploring the fate of reprogrammed cells in a tissue-like microenvironment (14, 15).
In this paper, we describe a unique method of fibroblast rejuvenation, which involves partial reprogramming of fibroblasts by growing them under lateral confinement, followed by their redifferentiation into fibroblast-like cells by embedding them in a 3D collagen-I matrix. We optimized an appropriate 3D collagen matrix density for the redifferentiation process. Here, we demonstrate fibroblast rejuvenation by revealing enhanced actomyosin contractility, collagen remodeling, and matrix protein deposition in redifferentiated cells compared to parent fibroblasts. RNA sequencing (RNA-seq) reveals a shift in transcriptome from a fibroblastic to an intermediate reprogrammed state following lateral confinement, which shifts back to the fibroblastic transcriptome (enhanced expression of genes related to contractile cytoskeletal pathways) upon redifferentiation in the collagen matrix. Importantly, we show an amelioration of DNA damage, which was facilitated by an increase in laminA levels in the nucleus upon rejuvenation. In terms of changes to nuclear architecture, we reveal that the comparatively open chromatin compaction state (chromatin poise state) induced by partial reprogramming was more likely to differentiate into contractile fibroblasts in response to ECM cues present in the 3D collagen matrix than the parental fibroblasts. In summary, we suggest that the mechanically induced partial reprogramming approach described here can overcome the shortcomings of conventional rejuvenation methods, including generation of short-lived or oncogenic fibroblasts, and therefore could have potential implications in the field of regenerative medicine.
Results
Redifferentiation of Fibroblasts from Partially Reprogrammed Spheroids Depends on 3D Collagen Matrix Density.
In our previous study on mechanically induced nuclear reprogramming in the absence of exogenous biochemical factors, we found that mouse embryonic fibroblasts undergoing laterally confined growth for 6 d on micropatterns started to acquire partial stem-cell-like gene expression (11). These 6-d-old spheroids were embedded in collagen gels and cultured for at least 2 d (Fig. 1A). Within a few hours, cells originating from the spheroids progressively invaded the collagen matrix and migrated either individually as unicellular sprouts (single cells) or collectively as complex, capillary-like structures (Fig. 1B and SI Appendix, Fig. S1). In addition to cell invasion, morphological modifications in the spheroid core itself were detected. While the spheroids initially appeared as a compact structure, subsequent cell migration induced spheroid expansion and led to breaches in the spheroid core. Since cell/substratum adhesion is known to govern cell-sprouting parameters in the 3D matrix (13), we next investigated the influence of physical properties of the 3D matrix, such as its porosity and stiffness (usually combined into a single parameter known as matrix density) on cell-sprouting patterns. We compared cell sprouting in the matrix as a function of matrix density obtained through varying collagen concentration from 0.5 to 2 mg/mL. By using AngioTools analysis (16) to quantify cell sprouting from the spheroids on matrices of varying densities, we found that average sprouting length in these cells showed a biphasic dependence on matrix density (Fig. 1 C and E). Similar to sprouting length, cell contractility, as measured by actin levels, also showed a similar biphasic trend depending on matrix density (Fig. 1F). Addition of cytochalasin D (which depolymerizes actin) and blebbistatin (which inhibit myosin II contractility) led to a significant reduction in average sprouting length, as well as a drop in collagen fiber stiffening (SI Appendix, Fig. S1). 1We next investigated the effect of varying matrix density on cell-proliferation rates, using a 5-ethynyl-2′-deoxyuridine (EdU) incorporation assay to quantify DNA synthesis. By measuring the percentage of EdU-positive cells following cumulative incorporation (16 h), we found that proliferation rates were significantly affected by matrix density (Fig. 1 D and G). Furthermore, we compared the effect of changing collagen density on both redifferentiated fibroblasts (RFs) and fibroblasts clump in collagen gel (FCGs). We embedded the control fibroblasts in three different collagen densities, i.e., 0.5, 1, and 2 mg/mL (SI Appendix, Fig. S2A). We observed that the sprouting efficiency and contractility (actin mean intensity) was increased with the increasing collagen densities (SI Appendix, Fig. S2 B and C). The proliferation of these FCGs showed a biphasic trend with collagen densities similar to RF cells (SI Appendix, Fig. S2D). Importantly, the sprouting length of FCG was relatively smaller than the RF condition in all three collagen densities, suggesting that the RF cells could have higher cell-matrix contacts. The 1 mg/mL collagen concentration was therefore selected as the optimal matrix density in all subsequent studies. All these results suggest that an optimal mechanical state of the 3D collagen matrix results in the redifferentiation of PRs into fibroblasts.
Fig. 1 Redifferentiation of fibroblasts from partially reprogrammed spheroids depends on the 3D collagen matrix density. (A) Schematic representation of the effect of geometry-driven laterally confined growth of fibroblasts on reprogramming, followed by their redifferentiation within the embedded 3D collagen matrix. (B) Phase-contrast images of NIH 3T3 mouse fibroblast cells cultured on micropatterns for up to 6 d, spheroids of PRs, and RFs undergoing sprouting in 3D collagen matrix. (Scale bars, 100 µm.) © Sprouting efficiencies of cells from 6-d-old spheroids on collagen matrices of varying densities. Cells were stained with phalloidin to label actin. (Scale bars, 500 µm.) (D) Proliferation rates of sprouting cells at varying matrix densities using an EdU incorporation assay. (Scale bars, 50 µm.) (E–G) Sprouting efficiencies, contractility, and proliferation levels at varying matrix densities measured by average (Avg.) vessel length, mean actin intensities, and percentages of EdU-positive cells in each field of view, respectively. For E and G, n > 5 fields of view were randomly measured in each condition. For F, number of cells, n = 3,304, 4,786, and 1,015 for 0.5, 1, and 2 mg/mL conditions, respectively. Error bars represent ± SD. *P < 0.05; ***P < 0.001. Two-sided Student’s t test was used.
A Shift in Transcription Profiles Is Accompanied by Enhanced Cytoskeletal Gene Expression: RNA-Seq.
In order to characterize the gene-expression profiles in RFs and compare them with other control conditions, including PRs, fibroblasts grown in clumps (FCs), and FCGs, RNA-seq experiments were performed. Thousands of genes, including key pluripotency markers Bmp4, Cdx2, Fgf4, Gdf3, Nanog, Nodal, Nt5e, Sall4, and Sox2, were solely up-regulated in the PR cells (Fig. 2 A and B). When the gene-expression profiles in these four conditions were analyzed, two drastically different cell states were revealed by principal component analysis (PCA) (Methods and Fig. 2C). PR cells shifted away from the parental fibroblast-like state (FC) to a stem-like state, as a result of lateral confinement (11). Embedding these PRs in the 3D collagen environment led their gene-expression profiles to return to the parental 3T3 fibroblast-like state in RF cells. We observed a difference in gene-expression profiles between cells undergoing reprogramming and original fibroblasts (both cultured on a 3D collagen matrix), which is supported by a Venn diagram showing the number of up-regulated genes in different comparisons (Fig. 2D). Based on these comparisons, we identified two groups of genes, which were either selectively overexpressed (23 genes) or down-regulated (53 genes) in RF compared to all of the other conditions. Further, of all of the down-regulated genes, we found that a specific gene, Follistatin (Fst), which is a common marker for aging (17), was expressed at significantly lower levels in RFs than in all other conditions (Fig. 2E). Such a significant decrease in Fst expression in RFs indicates the rejuvenation of fibroblasts through the reprogramming process. In order to confirm the difference between the Fst protein level in RF and FCG conditions, we performed Western blotting. Consistent with the RNA-seq result, the Fst protein level was lesser in the RF condition than the FCG condition (SI Appendix, Fig. S6 B and C). Genes up-regulated in RFs formed a molecular interaction network, which was characterized by several connection nodes around proteins such as Rab25, Cdc42bpa, Rhoj, and Iqgap1 that enhance cell migration and cell contractility (SI Appendix, Fig. S3A) (Methods). The expression of selected genes regulating cell contractility was up-regulated in RFs compared to FCGs (Fig. 2F and SI Appendix, Fig. S3B). Further, in agreement with the RNA-seq profile, the increase in messenger RNA (mRNA) levels of selected contractility-related genes was validated by qPCR assay (Fig. 2G). These experiments show that PR cells can be redifferentiated into a fibroblast-like (RF) state by embedding them into a 3D collagen matrix, and these cells are characterized by elevated expression of contractility- and rejuvenation-related genes.
Fig. 2 A shift in transcription profile accompanied with enhanced cytoskeletal genes’ expression. (A) Heatmap showing fold change (log2) in global transcription profiles between PR, RF, FCG, and FC cells. FDR (adjusted P value) < 0.1. (B) Heatmap showing log2 fold change in the differential day 6 partially expression of selected genes (compared to the PR sample). FDR (adjusted P value) < 0.1. © PCA showing the shift in cell states through reprogramming and reverting back through redifferentiation. FDR (adjusted P value) < 0.01 and |log2 fold change| > 2. (D) Venn diagram showing the number of up-regulated genes in 14 (2n – 2; n is four conditions) comparisons. FDR (adjusted P value) < 0.1. (E) Bar plot depicting changes in the expression of the representative aging gene Fst. The error bars represent ± SD. *P < 0.1. (F) Heatmap showing the log2 fold change in the differential expression of selected cytoskeletal-related genes (compared to FCG). P value (not adjusted) < 0.05. (G) mRNA levels of selected cytoskeletal genes obtained by qRT-PCR, normalized with respect to FCG.
RFs Are Characterized by Enhanced Contractility and Matrix Remodeling.
In order to characterize these RFs in vitro, we compared the gel-contraction abilities of cells derived from PR spheroids or from FCs, both of which were equally embedded in the 3D collagen matrix. Representative images and quantitative analysis of gel-area reduction revealed that the amount of collagen-gel contraction by the RFs was higher compared to FCGs (Fig. 3A). Fibroblasts exhibit contractile actin bundles (18), and, therefore, we compared actin and phosphorylated myosin light chain (pMLC) global intensities in these two cell types embedded in the collagen matrix. In agreement with the RNA-seq results, Fig. 3 C–E and SI Appendix, Fig. S4 clearly show that the RFs exhibited enhanced actomyosin contractility compared to control fibroblasts (FCGs). Fibroblasts are known to exert mechanical forces on the ECM surrounding them. Hence, we studied fibroblast-induced reorganization of the matrix by visualizing immunostained collagen fibers and qualitatively evaluating the effect of different inhibitors on collagen-fiber remodeling. Both RFs and FCGs embedded in fibrillary collagen matrix were able to remodel collagen fibrils into thicker bundles (Fig. 3F). Remodeling of collagen fibrils into thick bundles was observed within the fibroblast-populated collagen gel. We observed that collagen fibrils rearranged thicker around RFs compared to control FCG samples. In addition, to check the expression level of collagen–cross-linking molecules Lox in RF and FCG cells, we plotted the Lox mRNA level from the RNA-seq data. RF cells showed higher Lox expression compared to FCG cells (SI Appendix, Fig. S6D). This result suggests the enhanced remodeling properties of RF cells compared to FCG. However, collagen fibrils around RFs exhibited very less or no remodeling in samples treated with 25 µM Y-27632 (a RhoA-kinase inhibitor) and 4 µM cytochalasin D (inhibitor of actin polymerization) (SI Appendix, Fig. S1). Matrix assembly and remodeling are usually promoted by ECM glycoproteins that bind to cell-surface receptors, such as fibronectin (FN) dimers binding to integrins. Fibroblasts deposit FN to the matrix along the way of migration. Immunostaining experiments showed that both the fibroblasts migrated within the 3D collagen matrix; however, RFs deposited more FN along their migration trails compared to control FCGs (Fig. 3G and SI Appendix, Fig. S5). In addition, these RFs also expressed several other ECM-related genes, including Lama1 (laminin, alpha 1), Fn1 (fibronectin 1), Col4a1, and Col1a1 at higher levels than in controls, as quantified by qPCR assay (Fig. 3H). Further, we performed Western blotting analysis to confirm the difference between the Fst protein level in the RF and FCG conditions. Consistent with the RNA-seq result, the Fst protein level was lesser in the RF condition than the FCG condition (SI Appendix, Fig. S6 A and C). Fibroblasts embedded in the 3D ECM are mechanically supported by the ECM and, in turn, exert forces onto the ECM through cell–ECM contacts (1, 19). A temporal quantitative measurement of the contractile forces exerted by these two fibroblast types was done by using 3D traction force microscopy (TFM) during fibroblast sprouting (20). A color map based on the measurements indicated that RFs exerted comparatively higher traction stress during the initial 12 h of the sprouting phase (Fig. 3I). Vector arrows are indicated in the direction of force at each small window. During sprouting of cells from spheroids, the peak strain energy exerted by cells varied between spheroids, with a maximum energy of 450 and 220 pJ exerted by RFs and FCGs, respectively (Fig. 3J and SI Appendix, Fig. S7A). Considering the nonlinear properties of the collagen, we also measured the traction force of FCG and RF cells seeded on a two-dimensional (2D) FN-coated soft polydimethylsiloxane (PDMS) substrate (described in Methods). Consistent with the 3D TFM results, we observed that the RF cells showed higher 2D traction compared to the FCG cells (SI Appendix, Fig. S7 B and C). All together, these results support that augmented contractility and enhanced matrix remodeling are characteristics of the RFs.
Fig. 3 RFs are characterized by enhanced contractility and matrix remodeling. (A) Representative images of temporal collagen-gel contraction by matrix-embedded RFs and FCGs. (B) Normalized gel area plot representing the gel-contraction efficiencies of these two types of cells. Error bars represent ± SD. The P values represent the adjusted P values obtained by Bonferroni adjustment methods. *P < 0.05; **P < 0.01; ***P < 0.001. Two-sided Student’s t test was used. © Representative actin and pMLC immunofluorescence micrograph of RFs and FCGs embedded in 1 mg/mL collagen matrix. (Scale bars, 20 µm.) (D and E) Corresponding box plots for cellular mean intensity of actin and pMLC; n = 81 and 67 for FCG and RF conditions, respectively. ***P < 0.001. Two-sided Student’s t tests were used. (F) Representative fluorescence micrographs of immunostained collagen matrix in these two conditions. Corresponding phalloidin-stained actin images represent cells within the matrix. (Scale bar, 100 μm.) (G) Representative immunofluorescence micrographs of extracellular FN deposited on to the matrix in these two conditions. In merged images, the nucleus is labeled in blue, FN in green, and actin in red. (Scale bar, 50 μm.) (H) mRNA levels of selected ECM-related genes obtained by qRT-PCR, normalized with respect to FCG. Error bars represent ± SD. *P < 0.05; **P < 0.01. Two-sided Student’s t test was used. (I) Representative temporal traction force maps quantifying forces exerted on the matrix during sprouting of these two cell types. (J) Corresponding maximum strain energy plots during sprouting of these two cell types. Error bars represent ± SD. **P < 0.01. Two-sided Student’s t test was used.
Rejuvenation through Redifferentiation of Partially Reprogrammed Fibroblasts Ameliorates Age-Associated Phenotypes.
In order to investigate whether aging-associated phenotypes improve following rejuvenation, we next analyzed the level of DNA damage in these cells. Interestingly, the number of foci containing histone gH2AX, a marker of nuclear DNA double-strand breaks associated with aging (21), were significantly reduced in RFs compared to FCGs (Fig. 4 A and B). Lateral confinement induced PR cells to accumulate significantly fewer gH2AX foci compared to FCs. Sprouting of FCs induced by constriction of pores in the 3D collagen matrix resulted in an increase in gH2AX foci, whereas the change in the number of gH2AX foci in RFs was decreased compared to that in PR cells (Fig. 4 A and B). Cell migration through constricting pores can lead to accumulation of DNA damage, which is dependent on its nuclear lamina levels (22). By using qPCR to quantify Lmna gene regulation, we found a decrease in Lmna mRNA levels in sprouted cells derived from FCs compared to FCGs (Fig. 4C). Interestingly, Lmna mRNA levels increased during redifferentiation in the RFs. In agreement with the qPCR data, immunofluorescence data showed a significant increase in LaminA levels in the RF compared to the FCG condition (Figs. 4 D and E and SI Appendix, Fig. S8). In addition, an increase in the number of gH2AX foci in Lamna−/− RFs suggests that higher LaminA levels in wild-type RFs may act to shield their nuclei from accumulating DNA damage during migration through constricted pores in the collagen matrix (Fig. 4 A and B). The nuclear lamina can regulate cellular contractility, and vice versa (23). Therefore, we next investigated the relationship between LaminA changes in rejuvenated cells and contractility. We found that the actin level was significantly increased in RFs compared to FCGs, yet when lmna−/− cells were used, RFs exhibited decreased actin compared to FCGs (Fig. 4 F and G). However, we observed an increase in pMLC levels in Lmna−/− RFs, although not as high as in wild-type RFs (Fig. 4 F and H). Collectively, these results demonstrate that short-term, in vitro induction of fibroblast reprogramming through lateral confinement of 3T3 cells, followed by their subsequent redifferentiation can ameliorate phenotypes associated with physiological aging (e.g., accumulation of DNA damage and nuclear-envelope defects) by increasing LaminA levels in their nuclei.
.../...
.
