One of the defining characteristics of fibrosis in all tissues is the excess deposition of matrix materials. Therefore, we initiated our studies with an examination of matrix production. On the addition of TGF-β1, the constructs became thicker. As seen in
Figure 1A, the thickness of the construct increased as the exposure time to TGF-β1 increased, at both 4 and 8 weeks. The maximum increase, compared with the control (C: 29.5 and 35.2 μm), occurred when TGF-β1 was added the entire time (T1)—62.2 and 111.8 μm (2.1- and 3.2-fold), 4 and 8 weeks, respectively (
Fig. 1A). Of interest, the presence of TGF-β1 for the entire time did not appear to be necessary for the thickness of the construct to increase significantly. A 1-week exposure (T1–1w) resulted in a significant increase in thickness (1.8- and 2.0-fold, 4 and 8 weeks, respectively).
This increase in thickness by TGF-β1 was confirmed when the thick sections of these constructs were compared, as seen in
Figure 1B. In all groups, cells stratified, showed a flat and elongated morphology, produced their own collagen matrix, and aligned with and parallel to the porous membrane (Transwell; Corning). Note, that all groups showed high cell densities at the top and bottom of the construct. This area corresponds to where the cells were most highly aligned (data not shown). Of interest, in
Figure 1B (T1 at 8 weeks), the cell-produced matrix was losing its integrity. As shown by the white arrow, the collagen at the lower half of the construct was highly disorganized, and the fibrils were mainly running in one direction. In addition, the overall fibril length appeared to decrease, cross-banding became less distinct, and fibril diameter appeared to become more variable. The transition to disorganized matrix was seen at 8 weeks in all T1 samples examined.
Examination of the constructs by TEM revealed the cell-matrix interactions, as well as, matrix condition, such as alignment.
Figure 2 shows the constructs after 4 and 8 weeks in culture in all four conditions (C, T1, T1–1w, and T1–4w). The cells appeared elongated and to have produced their own ECM. Collagen fibril orientation can be identified with several changes in direction in lamellae-like structures. Change of direction and fibril organization is one of the main characteristics of a mature cornea. Our TEM data revealed that the fibrils are organized in C, T1–1w, and T1–4w; however, in T1, the fibrils were shorter and the density of the matrix appeared to decrease by 8 weeks. The fibril length appeared to become longer in T1–1w and T1–4w.
We further quantified and compared the fibril diameters at different conditions, as shown in
Figure 3. At 4 weeks, fibril diameters ranged from 30 to 33 μm, sizes comparable to those in vivo (
Fig. 3, week 4). Over time, however, the diameter size increased significantly (
P < 0.05), independent of the presence or absence of TGF-β1 (
Fig. 3, week 8, C and T1). Fibril diameter size and integrity were maintained in T1–1w (
Fig. 3).
We then investigated the expression of specific fibrotic markers, to assess the cell-produced ECM and to gain an understanding of the effect of TGF-β1. One of these, SMA, marks myofibroblasts, a cell type commonly found in fibrotic tissue. The presence of myofibroblasts and fibrosis are often linked to the production of certain matrix components in the stroma, such as type III collagen and EDA-Fn. As seen in
Figure 4 (SMA), few if any myofibroblasts were present in the 4-week construct without TGF-β1 (
Fig. 4A;
Movie S1); however, with the addition of TGF-β1 for the entire 4 weeks in culture (
Fig. 4B; Movie S2), there was a clear increase in positive SMA cells. These cells were found throughout the construct, but there appeared to be more in the top and bottom layers of the construct. Also, as seen in
Figure 4 (Col III), little, if any, type III collagen was present in C (
Fig 4C; Movie S3), whereas, in T1, type III collagen was present at a high level (
Fig. 4D; Movie S4). Of interest, highest levels of type III collagen appeared in the top half of the construct in TGF-β1-treated cells. EDA-Fn was also upregulated after stimulation with TGF-β1 (
Figs. 4E,
4F; Movies S5, S6) and was localized in the topmost layers of the construct. Localization of SMA, EDA-Fn, and type III collagen was very similar between weeks 4 and 8 in all constructs and conditions (data not shown).
The data show that with the addition of TGF-β1, fibrotic markers were upregulated. This finding led us to ask the following questions: For the expression of fibrotic markers, (1) does the construct need TGF-β1 in the construct medium the entire time, and (2) when should the treatment with TGF-β1 take place? Two experiments that mimicked corneal wounding were performed to resolve these questions. As seen in
Figure 5, a 1-week pulse of TGF-β1 (T1–1w) increased the number of positive myofibroblasts (
Fig. 5A), type III collagen (
Fig. 5C), and EDA-Fn (
Fig. 5E) compared with the control (
Figs. 4A,
4C,
4E). However, the 1-week exposure did not have as much of an effect as the 4-week exposure (T1–4w) seen in
Figure 5. As the exposure to TGF-β1 increased, so did the expression of the fibrotic markers—SMA (
Fig. 5B), type III collagen (
Fig. 5D) and EDA-Fn (
Fig. 5F). The localization of fibrotic markers in T1–4w appeared to be close to or similar to that with T1 (
Fig. 4). Whereas, the T1–1w construct gave rather intermediate results. There appeared to be no difference between the 4- and 8-week time points for T1–1w (data not shown); therefore, since T1–4w data were collected only at 8 weeks, T1–1w at 8 weeks was used in
Figure 5.
Finally, due to the increase in thickness of the constructs, Ki67, a marker of proliferating cells, was used to observe whether there was an increase in proliferating cells on the addition of TGF-β1. In
Figure 6, a few proliferating cells were present in C, whereas, in T1, the number of proliferating cells increased considerably. When the percentage of cells expressing Ki67 were quantified (
Table 1), it was found that 2.7% and 1.5% of cells in C were positive at 4 and 8 weeks, respectively. In the TGF-β1 treated cells (T1), 3.4% and 1.6% of cells were Ki67-positive. These values were not significantly different. However, when the total number of cells was determined, it was found that C contained 1.6 and 3.0 million cells, 4 and 8 weeks respectively, while T1 contained 3.7 and 6.7 million. These differences were statistically significant (
P < 0.001). One million cells were originally seeded per construct. No statistically significant difference was recorded in the overall cell density between groups (data not shown).
Overall, it appears that the addition of TGF-β1, even for 1 week (T1–1w), had an effect on the appearance of myofibroblasts and the deposition of both type III collagen and EDA-Fn. As the exposure to TGF-β1 increased so did the effect on the construct. Therefore, there is a direct correlation between the exposure time to TGF-β1 and the expression of fibrotic markers.