Fibroblast viability and phenotypic changes within glycated stiffened three-dimensional collagen matrices

Abstract

Background: There is growing interest in the development of cell culture assays that enable the rigidity of the extracellular matrix to be increased. A promising approach is based on three-dimensional collagen type I matrices that are stiffened by cross-linking through non-enzymatic glycation with reducing sugars.

Methods: The present study evaluated the biomechanical changes in the non-enzymatically glycated type I collagen matrices, including collagen organization, the advanced glycation end products formation and stiffness achievement. Gels were glycated with ribose at different concentrations (0, 5, 15, 30 and 240 mM). The viability and the phenotypic changes of primary human lung fibroblasts cultured within the non-enzymatically glycated gels were also evaluated along three consecutive weeks. Statistical tests used for data analyze were Mann-Whitney U, Kruskal Wallis, Student's t-test, two-way ANOVA, multivariate ANOVA, linear regression test and mixed linear model.

Results: Our findings indicated that the process of collagen glycation increases the stiffness of the matrices and generates advanced glycation end products in a ribose concentration-dependent manner. Furthermore, we identified optimal ribose concentrations and media conditions for cell viability and growth within the glycated matrices. The microenvironment of this collagen based three-dimensional culture induces α-smooth muscle actin and tenascin-C fibroblast protein expression. Finally, a progressive contractile phenotype cell differentiation was associated with the contraction of these gels.

Conclusions: The use of non-enzymatic glycation with a low ribose concentration may provide a suitable model with a mechanic and oxidative modified environment with cells embedded in it, which allowed cell proliferation and induced fibroblast phenotypic changes. Such culture model could be appropriate for investigations of the behavior and phenotypic changes in cells that occur during lung fibrosis as well as for testing different antifibrotic therapies in vitro.

Fig. 1

Increased collagen type I fiber reflection fluorescence with non-enzymatic glycation. Collagen type I fiber conformation observed with CRM (laser ex/em 568/568 nm) at different ribose (R) concentration over a period of twenty-one days. a. In post-glycated DMEM matrices, collagen type I fibers formed aggregates and reflected more fluorescence with elevated R concentrations. b. In post-glycated PBS matrices, collagen type I fibers were short, thin and homogeneous, with no increase of reflected fluorescence with higher R concentration (240 mM). a and b. Reflection fluorescence of non-glycated matrices (controls) was done to the reflective properties of the collagen. Time seems to not affect the reflection fluorescence of collagen. The increase in CRM signal in DMEM gels could also be contributed by the increased aggregation of collagen fibers observed in these gels, perhaps because the presence of glucose. The scale bar corresponds to 40 μm. CRM = Confocal reflection microscope; DMEM = Dulbecco’s modified Eagle medium; PBS = Phosphate-buffered saline

Fig. 2

The formation of advanced glycation end products (AGEs) increased with collagen glycation. AGEs formation was evaluated in non-cellular post-glycated DMEM (a1-3) and PBS (b1-3) matrices based on autofluorescence (ex/em: 360/440 nm wavelengths). The matrices were elaborated using different concentrations of fetal bovine serum (FBS), 0 % (a1, b1), 1 % (a2, b2) and 10 % (a3, b3). The media was changed every 2 days beginning on the 5^th day to allow collagen cross-linking to occur. AGE formation in both type of matrices (DMEM and PBS), was significantly greater with higher ribose concentrations (240 mM) (p < 0.01) and at the 7^th day (p < 0.01). After that, less autofluorescence was observed for all the matrices suggesting less AGEs’ formation. Interestingly, DMEM matrices (2A) were more autofluorescent than PBS matrices (2B) at days 1 and 7 (p < 0.05 in practically all conditions). Nevertheless, PBS matrices became more autofluorescent than DMEM ones after the 14^th day (p < 0.05 in practically all conditions for FBS 1 % and 10 %). The experiments were repeated three times with similar results obtained. * p < 0.05

Fig. 3

Collagen post-glycation increased the matrix stiffness. The stiffness of the non-cellular post-glycated matrices was measured using atomic force microscopy at days 7, 14 and 21. The values were normalized using that of the non-glycated matrices, i.e. control matrices (0 mM of ribose (R)) at the 7^th day. The black arrows indicate the R concentration that rendered the collagen gels stiffer than the non-glycated gel at the 7^th day. a. DMEM matrices. Collagen gels stiffened in a R dependent manner at the 14^th day after glycation at 30 and 240 mM of ribose respect to the control non-glycated gels (p < 0.01). b. PBS matrices. The collagen gels stiffened with 240 mM of ribose at the 21^st day after glycation (p < 0.01). a and b. This different stiffening dynamics suggests different cross-linking rates between these gels. The glucose present in DMEM could play a role in the stiffness developed, which could render a higher basal level of glycation before ribose treatment. Accordingly, the non-glycated DMEM gels (controls) showed greater stiffness than the glycated matrices at the 7^th day. Phenomenon not observed in PBS matrices. The highest R concentration (240 mM) led to the stiffest conformation of both types of matrices (p < 0.01). p-values for 30 mM of ribose: + (0.01 < p < 0.05), ++ (p < 0.01); p-values for 240 mM of ribose: *(0.01 < p < 0.05), ** (p < 0.01)

Fig. 4

Post-glycated matrices support fibroblast viability for long periods. Primary human lung fibroblasts cultured within different glycated collagen DMEM matrices over a period of 21 days. a. Fluorescence determination using alamarBlue assay. b. LIVE/DEAD viability/cytotoxicity stained cells observed using confocal reflection microscopy. The viable cells are stained green, and the non-viable cells are stained red. a and b. Cell death occurred under all conditions at the highest ribose (R) concentration (240 mM), independently of the use of serum (FBS). The use of FBS increased the proliferation rate under the rest of the conditions (p < 0.01). a1 and b1 (0 % FBS). No cell proliferation was observed, but cells were alive during the 21 days. a2 and b2 (1 % FBS). A gradual increase in cell proliferation was observed under all conditions except R 240 mM, but the rates for R 5 mM (p < 0.05) and the control matrices (p < 0.01) were significantly different only between the 7^th and 21^st days. a3 and b3 (10 % FBS). An improved proliferation rate was observed with the use of 10 % of serum, which was significantly for the control (non-glycated) only between the 7^th and 14^th days (p < 0.05). Gel contraction was observed in ≤15 mM R conditions between the 14^th and the 21^st day. No gel contraction was observed at 30 mM R (explanation in discussion). It was not possible to determine the fluorescence in the controls and in matrices glycated using 5 mM R at the 21^st day because of gel contraction and cells growing on the bottom of the well (a3). The scale bars correspond to 200 μm. The experiments were repeated three times, with similar results obtained

Fig. 5

Post-glycated matrices induce a contractile phenotype. Primary human normal lung fibroblasts cultured within three-dimensional glycated DMEM matrices with 10 % of serum were analyzed to determine the level of alpha-smooth muscle actin (α-SMA) gene expression at the 7^th, 14^th and 21^st day of culture. α-SMA gene expression was observed in the glycated matrices prepared with 5 mM, 15 mM ribose (R) and in the non-glycated matrices (controls) on the 21st day (only statistically significant (p < 0.05) for 5 mM). This phenomenon corresponded to the occurrence of gel contraction. In contrast, α-SMA gene expression and gels contraction were not observed in the 30 mM R glycated matrices on the 21^st day. Interestingly, a progressive increase of α-SMA gene expression was observed with a gradual increase of R on day 7^th, suggesting that the differentiation of fibroblasts to a contractile phenotype depends not only on time but also on the ribose concentration, which is associated with the extent of collagen cross-linking. The data are presented as percentages respect to the values on the 1^st day. The experiments were repeated three times, with similar results obtained. * p-value < 0.05

Fig. 6

α-SMA protein detection in the 3D matrices. alpha-smooth muscle actin (α-SMA) protein expression in the primary human normal lung fibroblasts was evaluated using western blot. Fibroblasts were cultured within 3D glycated DMEM matrices containing 10 % serum. a. Western Blot. α-SMA band, 42 kDa; α-tubulin band, 50 kDa. Albumin band (66 kDa, due to FBS and visible in cellular and acellular matrices blots’, probably because of the difficulty in rinsing all the media from the matrices prior to protein extraction). The experiments were repeated three times. a1. An insignificant amount of α-SMA was detected in all the cellular matrices. a2 and a3. Higher amounts of α-SMA were observed in all the cellular matrices on the 14^th and 21^st days of culture. a1, a2 and a3. No tubulin or α-SMA bands were detected in the non-cells matrices. b. Densitometric analysis of α-SMA levels (Ratio of α-SMA to α-tubulin). The results were obtained from three independent experiments. *p < 0.05, **p < 0.01. A gradual decrease of α-SMA was observed at higher ribose concentrations on the 7^th day of culture (p > 0.05). Contractile phenotype was most strongly detected on the 14^th and 21^st days of culture. It was significantly higher at all ribose concentrations, also for controls without ribose, compared with the level on the 7^th day. It suggests that the microambient ‘per se’ could induce a phenotype change and that cells contribute to the matrix contraction

Fig. 7

3D matrices induce Tenascin-C synthesis in lung fibroblasts. Levels of Tenascin-C (TNC) were measured from the supernatants of the 3D DMEM matrices by ELISA. The results were obtained from three independent experiments. Fibroblasts were cultured with 10 % of FBS. The synthesis of TNC was induced in a time-dependent manner, observing significant statistical differences between the 1^st and the 14^th and 21^st days of culture (p < 0.05 for both comparisons). Although no differences were appreciated between the 14^th and the 21^st days of culture, a higher level of TNC was observed with 30 mM of ribose at the 21^st day (p = 0.08). No statistical differences were observed between glycated and non-glycated matrices. These results suggest that the 3D microenvironment could be enough to produce changes in the secretion of TNC by normal fibroblasts in a time-dependent manner