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

Pulmonary extracellular matrix (ECM) is a complex mesh of proteins, proteoglycans and glycosaminoglycans, which contains the cells and participates in tissue’s homeostasis and repair. Active lung fibroblasts, myofibroblasts and other cell types, are normally surrounded by this fibrous three-dimensional (3D) extracellular matrix and are mainly responsible for the synthesis, secretion and degradation of the ECM components. Moreover, these cells are responsible for the correct turnover of ECM proteins, preserving the lung architecture and function [1]. However, most in vitro studies of fibroblasts behavior have been performed using conventional two-dimensional (2D) cultures, which lack the structural three-dimensionality provided by the ECM in the original tissue. This limitation is important because it can cause the loss or change of important tissue-specific cell functions due to the lack of essential ECM signals [2, 3]. A suitable alternative to overcome this limitation is 3D cultures, in which cells are embedded within ECM gels. Studies using 3D cultures have shown that the ECM influences cell behavior through mechanical interactions and biochemical signals [3‐6]. Moreover, these studies have clearly shown that cell functions and phenotypes depend on the structure of the microenvironment, such as the ECM [2‐4] and have revealed the profound impact of extracellular forces on the behavior of fibroblasts and other cell types [3, 7, 8].

On the other hand, fibrotic lung diseases, such as idiopathic pulmonary fibrosis (IPF), are characterized by an excessive ECM deposition and the expansion of the fibroblast/myofibroblast population leading to an increased parenchyma stiffness and retraction, with final lung destruction and patient death [9‐12]. In this context, the increased matrix stiffening observed in the lung fibrotic process may be a critical fibrogenesis driving factor [13, 14]. Unfortunately, suitable 3D culture assays for the study of fibrotic lung diseases and other aging disorders, primarily mimicking matrix stiffness are scanty.

Therefore, there is growing interest in developing 3D culture assays that enable increasing the rigidity of the ECM mimicking the situation observed during fibrotic processes in vivo. In some 3D culture assays based on collagen gels, the rigidity of the ECM is modified by changing the ECM concentration [15]. Alternatively, ECM stiffness can be changed by increasing the cross-linking through non-enzymatic glycation with reducing sugars, which cause the accumulation of the final products of glycation, i.e., the advanced glycation end products (AGEs) [3, 5, 16‐24]. Interestingly, an excess of AGEs in aged human tissues has been observed to stiffen tissues in different degenerative diseases [25] and may accelerate protein oxidation, altering their structure and threatening their function [23, 24]. Therefore, various physical characteristics of the fibrotic lung tissue, such as reduced elastic recoil [2], decreased matrix degradation [21, 22, 26, 27], oxidative ECM protein damage [28], may be related to the accumulation of AGEs, contributing to the acceleration of the fibrotic process. However, the effects of non-enzymatic glycation on collagen gels and the resulting behavior of lung fibroblasts remain largely unknown.

Thus, the aims of this study were: 1) to develop an experimental culture model system that recreate the biomechanical and three-dimensional conditions of fibrotic lungs, and 2) to examine the cellular viability and biomechanical changes in the ECM, including AGE formation and stiffening, in these non-enzymatically glycated collagen gels. This model may facilitate the experimental study of the complex cell-extracellular pro-fibrotic interactions, such as those occurring in fibrotic lung diseases, as IPF.

Lung tissue fibroblasts are normally surrounded by a 3D ECM and the interaction between fibroblasts and ECM proteins could be crucial in tissue homeostasis and disease. However, most in vitro studies in lung fibrosis have been performed with fibroblasts growing in a plate (2D) or in 3D based on polyacrylamide hydrogels [8, 38, 39] or collagen type I gels [17, 18, 27, 40] with fibroblasts growing on top. Few studies have analyzed the viability of fibroblasts growing within type I collagen matrices [7, 41]. Our study was designed to modify the 3D collagen matrix via non-enzymatic glycation and to analyze the impact of such matrices on fibroblast phenotype and viability growing inside the gel. To our knowledge, there are no previous reports of analyzing fibroblasts viability in glycated DMEM matrices by using two corroborating cell growth methods. We used ribose as a reducing sugar rather than glucose due to the ability of ribose to modify faster the physical and chemical properties of type I collagen gels [21]. Very few studies have used post-glycated collagen matrices with cells growing inside, and none in the field of lung fibrosis [19, 21]. Our novel 3D culture model displayed collagen enrichment and an aging/oxidative microenvironment with increased stiffness, which may be useful for the study of cell behavior and phenotypic changes that depend on pro-fibrotic ECM conditions. Our study points up the best three-dimensional ribosilated matrices conditions to allow primary lung fibroblast viability and growth. Furthermore, the results demonstrate that the 3D collagenated matrices induce fibroblast-to-contractile phenotype differentiation and an increase of TNC synthesis. We observed structural differences in the collagen fibers depending on the medium used in the preparation of the matrices (DMEM or PBS), which may be attributable to the presence of glucose in DMEM. Our results showed that once the matrices have been glycated, the higher ribose concentrations (240 mM) increased the reflectance fluorescence of collagen type I in DMEM matrices (Fig. 1) and the AGEs production (Fig. 2) in agreement with previous published results [20, 21, 27]. However, the ribose-dependent increase in CRM signal in DMEM gels could also be contributed by the increased aggregation of collagen fibers observed in these gels, which could locally enhance the reflective properties of the collagen. In contrast, such structure-related differences observed by CRM may not contribute to the autofluorescence measured by the plate reader or to the stiffness results.

The interest of using glycated collagen gels was to resemble the increased stiffness observed in aged and fibrotic lungs [2, 42]. Furthermore, the use of the post-glycation method (instead of pre-glycation technique) aimed to approach to the collagen cross-links, which could occur in vivo. Some studies have used non-enzymatic glycation to create stiffer scaffolds with increased mechanical rigidity but none concerned to lung fibrosis [20, 21, 27, 43]. Our results demonstrated that the glycation process stiffened the collagen matrices in a concentration-dependent manner, occurring earlier in the DMEM matrices than in the PBS matrices. To our knowledge, we provide first evidence that glycation-induced collagen gel stiffening depends not only on ribose concentration, but also on the type of substrate solution used (DMEM or PBS). However, these different stiffening dynamics, suggests different cross-linking rates between those gels (DMEM compared to PBS), perhaps attributed to the glucose already present in DMEM, which could render a higher basal level of glycation before ribose treatment and to the diameter and length of the cross-linked collagen fibrils [42]. In support of this hypothesis, the basal stiffness values in control gels (0 mM of ribose) at day 7^th were higher in DMEM than in PBS gels. Our stiffness values were similar to those reported by Roy et al. [20], and slightly higher than those from other authors without cells [27, 43], most likely because of differences in the experimental conditions, such as the method of measure. In contrast, when cells were cultured within the glycated matrices, an elevated Young modulus was observed [19, 21]. Interestingly, a recent study by Booth et al. [14] showed an increased stiffness of decellularized fibrotic lungs compared with normal lungs, plausibly attributed to the substantial and varied amounts of pro-fibrotic ECM proteins present in the former in addition to collagen. Thus, it is conceivable that cells embedded into the collagen gels would also upregulate the expression of other interstitial proteins, increasing the rigidity of the gel, which will be the focus of future in vitro studies [3, 42]. As a preliminary data, the fibroblasts growing inside our 3D collagen matrices increased the synthesis of TNC, a pro-fibrotic ECM protein. This phenomenon occurred in a time-dependent manner, independently of the ribose concentration. However, the level of TNC was higher in 30 mM ribosilated matrices at the 21^st day, suggesting that TNC synthesis by normal fibroblasts was dependent on the cell-type I collagen interactions and also on the indirect effect of ribose on collagen (stiffness, AGEs or collagen cross-links). These observations leads to further studies in order to better explain this effect.

Importantly, the non-enzymatic glycation of collagen resulted in the formation of AGEs, which are related to aging and have been observed to stiffen tissues in different degenerative diseases [22‐25]. With age, collagen becomes less soluble [26], more cross-linked and more glycosylated [5]. In studies of aged collagen fibers from different tissues, such as skin and lungs, the presence of AGEs was detected indirectly as an increase in the collagen autofluorescence [31, 32]. Our results showed that the change in the autofluorescence of glycated collagen, measured with the specific wavelength of fluorescent AGEs as reported elsewhere [20, 21, 27, 28, 31, 32], was dependent on the formation of AGEs (Fig. 2). Furthermore, we have shown that the collagen autofluorescence was affected by the ribose concentration in a dose-dependent manner, with fluorescence increasing when the highest ribose concentration was used. In support of our findings, other groups have reported that the fluorescence of glycated collagen matrices gradually increased with increasing concentrations of reducing sugars independently of the carbohydrate used for glycation [20, 21, 27].

On the other hand, a peak of autofluorescence was always observed at 7th day and then the autofluorescence decreased (days 14^th and 21^st), probably because less new AGEs formation after day 7^th. It could be a consequence of the technical approach, since the media was not changed until the 5th day after post-glycation and then it was changed every two days. This phenomenon could be attributed to the rinse of the matrices with the changes of the media, which would attenuate the formation of new cross-links between the glucose and the collagen. Another possibility would be that the cross-links could be higher at the beginning of the post-glycated matrices reaction. Additionally, our experiments of autofluorescence in PBS and DMEM matrices revealed different cross-linking rates. As DMEM matrices (Fig. 2a) showed more autofluorescence than PBS matrices (Fig. 2b) at days 1 and 7, less autofluorescence from DMEM compared to PBS matrices was observed since the 14^th day. These different cross-linking rates could explain the differences in stiffening dynamics between those gels (DMEM and PBS). In agreement with these dynamics, fold stiffness in DMEM gels remained rather unaltered after day 14th. In contrast we observed a moderate rise in fold gel stiffness in PBS gels to the 21st day.

One of the main findings in the present study was the demonstration of the profound effect of the physical and chemical changes of the three-dimensional matrices on the fibroblast phenotype and cell growth. We demonstrated that fibroblasts embedded in the matrices containing low ribose concentrations (≤30 mM) were able to proliferate and that the better growth rate was observed using FBS, whereas significant cell mortality was observed in matrices with highest ribose concentration (240 mM) independently of the use of FBS (Fig. 4), most likely because of the increased osmosis [19]. Remarkably, we detected a contractile cell phenotype expressing alpha-smooth muscle actin (α-SMA). This occurrence of the contractile phenotype was related to the contraction of glycated gels using 5 and 15 mM of ribose and to that of the non-glycated matrices between the 14^th and 21^st days, when 10 % of serum was used (Fig. 4 b3). Remarkably α-SMA was induced without adding TGFβ1, suggesting that the microenvironment ‘per se’ modifies cell phenotype by increasing cell contractility. Therefore, the experiments of expression and detection of α-SMA as a contractile phenotype marker (Figs. 5 and 6, respectively), were performed only in the matrices where the gel contraction was observed, (i.e. when 10 % of FBS was used), to verify if the cells embedded into the matrices could contribute to their contraction. Therefore, it is important to emphasize two messages: 1) the induction of α-SMA expression and protein level is modified by the microambient and 2) cells contribute to the matrix contraction.

The mechanisms by which cells contract the collagen gels are unclear. Based on previous studies, it is conceivable that as matrix stiffness increases, the fibroblasts develop proportional isometric tension, acquiring prominent actin stress fibers that cause the contraction of the matrices [3, 5, 7, 17, 44, 45]. Nevertheless, fibroblasts that do not surmount the tension of the matrix also develop actin stress fibers, but they do not contract the lattice [7, 17]. This attribute is relevant to the differences found in α-SMA detection and the contraction of our gels depending on the ribose concentration, as occurs with 30 mM of ribose [7, 44]. In those matrices, the contractile phenotype, α-SMA positive (Fig. 6 a2 and a3), did not contract the gels (Fig. 4 b3), probably because the cells did not surmount the tension present in the matrix.

Thus, matrix contraction depends on the presence of fibroblasts [7, 15, 45, 46], the use of serum [7, 44, 46], and the glycation cross-linking [17]. Additionally, the degree of contraction also depends on the initial collagen concentration [15], the number of cell passages [7, 15, 46] and the temperature [46].

Even though, we have not performed stiffness experiments when cells are seeded into the matrices, our results suggests that the effect of the matrix on the cells could be because of the matrix biomechanics’ microenvironment, the ECM protein cross-links or even either microenvironment factors that can be produced under these conditions. In support of this interpretation, other authors have reported that an increase in the ECM protein concentration, as well as the matrix cross-linking and the mechanical interactions between the ECM and the cells, promote cell proliferation, the acquisition of a contractile phenotype and the modulation of gel contractility, which they attribute to the matrix stiffness [4‐8, 15, 17, 38, 40, 42, 43].

Table 1. Summary results from the post-glycated three-dimensional collagen matrices. PBS: phosphate-buffered saline; DMEM: Dulbecco’s modified Eagle medium; ex/em: excitation/emission; nm: nanometers; KPa: KiloPascals; SD: Standard deviation; mM: miliMolar; N/A: Not Available; [Ribose]: concentration of ribose; ↓: decreased; ↑: elevated; ≤: minor or equal; (*) p < 0,05.