Main

The early fetus responds to cutaneous wounding in a fundamentally different manner than does the adult; fetal wounds heal without scars. The mechanisms underlying this difference are unknown, but may reflect a difference in dermal cell phenotypes, extracellular matrices, or cytokine profiles(1–3).

The fetal fibroblast deposits collagen fibrils in a reticular pattern during fetal repair and thus may be considered the effector cell of scarless repair(4). In contrast, in the adult, fibroblasts deposit an abundance of collagen, which is laid down in the extracellular matrix as parallel fibers, a pattern that constitutes a scar. Furthermore, in the adult, wound contraction is mediated by fibroblasts that transdifferentiate into contractile cells and direct wound contraction(1, 2).

Contraction is a vital aspect of adult wound repair because it brings normal skin at the wound edges into closer approximation. However, whether wound contraction plays a role in the fetal period is not clear(1). To study wound contraction in vitro, different models using fibroblasts cultured in collagen matrices have been developed. These fibroblast-laden collagen gels can be either anchored to the culture substratum or free floating on culture media. However, the anchored model has been shown to more closely resemble granulation tissue, whereas the floating model more closely resembles the normal resting dermis(5). Therefore, the anchored collagen gel contraction model was used in this study to compare the differences in contractile properties of early fetal, late fetal, and adult skin fibroblasts. Because early fetal wounds heal without scarring, we postulated that fetal fibroblasts may lack the necessary signals, possibly from peptide growth factors, to induce a contractile phenotype, and hence the mechanism of wound repair may be qualitatively different from that of the adult(6–8).

The cytokine TGF-β promotes wound contraction(9, 10), and the relative lack of TGF-β in fetal wounds has been implicated as a major cause of scarless repair(11–13). TGF-β is secreted by many different cell types that participate in the tissue repair process, including macrophages, platelets, and fibroblasts. Fetal wounds do not elicit a significant inflammatory reaction, and therefore platelets and macrophages, which are major sources of TGF-β in the adult wound, are significantly reduced or absent in fetal wounds(14). Furthermore, when exogenous TGF-β is added to fetal wounds, both contraction and scarring are seen(10, 15). Therefore, the local TGF-β expression is thought to be impaired or repressed in fetal wounds, due to the decrease in potential external sources of TGF-β, such as macrophages(16, 17). However, TGF-β is highly expressed by fibroblasts in the adult wound and induces them to produce large quantities of collagen(18–20). In contrast, the expression of TGF-β by the fetal fibroblast is unknown.

TGF-β is secreted from cells in a latent form that is unable to directly bind and activate its signaling receptors. The activation of TGF-β requires the cleavage of the latency activated peptide from the latent TGF-β molecule(21). The proteolytic enzyme plasmin can perform this function(17, 21, 22).

Our study compared the levels of active TGF-β and plasmin in culture supernatants from early fetal, late fetal, and adult mouse skin fibroblasts, grown on either plastic or integrated into an anchored type I collagen gel matrix. In addition, it compared the degree of collagen gel contraction by these fibroblasts. The purpose of our study was to explore whether there is a developmental regulation of the contractile properties, TGF-β expression and activation, and plasmin formation in murine skin fibroblasts.

METHODS

Animals. All animal experiments were performed in accordance with institutional guidelines set forth by the Childrens Hospital Los Angeles animal care committee. Timed pregnant CD-1 mice (Charles River Laboratories, Hollister, CA) were housed in individual cages at the Childrens Hospital Los Angeles vivarium and fed food and water ad libitum. The fetal mice were delivered by cesarean section at gestational ages E15 or E17, the skin was removed and placed in ice-cold DMEM (Sigma Chemical Co., St. Louis, MO) supplemented with penicillin 100 U/mL, streptomycin 100 μg/mL, and Fungizone 2.5 μg/mL (Sigma Chemical Co.). The maternal skin was shaved, prepared with betadine and 70% ethanol, harvested from the back and abdomen, and then placed in iced DMEM + antibiotics. Four separate isolations of fetal and adult mouse fibroblasts were made.

Fibroblasts. Skin fibroblasts were isolated by mincing the skin into 3-mm cubes. The skin pieces then underwent enzymatic digestion at 37°C in a mixture of 0.5% Dispase II (Boehringer Mannheim, Indianapolis, IN) and 1000 U/mL collagenase (Worthington Biochemical Corp., Freehold, NJ) in PBS, with gentle mixing for 20 min for the fetal skin and 2 h for the adult skin. The digested skin was passed through a 160-μm Nitex filter, and the enzymes neutralized with DMEM + 10% FCS (Sigma Chemical Co.). The resulting suspension was centrifuged at 450 × g for 5 min, and the cells were resuspended in DMEM + 10% FCS with antibiotics. The latter step was then repeated to ensure that all remaining digestive enzymes were removed from the cell suspension. The cells were once again collected and plated at a density of 1 × 106 cells/cm2 in culture flasks (Falcon, Franklin Lakes, NJ). The fibroblasts were placed in an incubator for 1 h at 37 °C supplemented with 5% CO2 and observed over this time course for attachment to the culture flask. Once the fibroblasts had attached to the culture flask, the supernatant was removed, and the culture flask was washed vigorously with PBS to remove any unattached cells. DMEM + 10% FCS was added, and the cells were cultured overnight. The next morning the cells were washed with PBS three times, removed from the plates with trypsin, and resuspended in DMEM + 10% FCS. This method isolated approximately 1 × 106 fibroblasts per 3 g of adult skin and 0.5 × 106 fibroblast per g of fetal skin. The cells were centrifuged as described above, and resuspended in DMEM at a concentration of 1 × 106 cells/mL. The cells were plated on either a plastic culture dish, on a type I collagen-coated plate(200 μL of Vitrogen/cm2, Collagen Corp., Palo Alto, CA), or placed in a collagen gel as described below at a density of 0.8 × 105 cells/cm2. The cells were cultured in DMEM, DMEM + 10% FCS, or DMEM + 10% Ultroser G (IBN Biotechnics, France) for a 20-h time period. Ultroser G is a chemically defined, commercially available serum substitute that contains serum proteins, vitamins, and the following growth factors and hormones in undisclosed amounts: epidermal growth factor, fibroblast growth factor, PDGF, IGF, and FSH. Ultroser G does not contain TGF-β.

Preparation of collagen gels. Collagen gels were prepared using Vitrogen (Collagen Corp.), a commercial preparation of predominantly type I collagen, according to the methods described previously by Guidry and Grinnell(23) and Nakagawa et al.(24). Briefly, collagen was adjusted to physiologic ionic strength and pH with 10 × minimum essential medium (MEM; Sigma Chemical Co.) and 0.1 M NaOH while being maintained at 4 °C. The final collagen concentration was 1.5 mg/mL. Fibroblasts were incorporated into the reconstituted collagen to a final concentration of 1 × 106 cells/mL. Samples (80 μL) of the reconstituted collagen or collagen-fibroblast suspension were placed in a 48-well culture plate(Falcon). Each sample filled an area outlined by an 8-mm diameter circular score mark within the well. The plates were placed in a 37 °C incubator for 45 min to allow collagen to form fibrils. Samples were then covered by 0.5 mL of DMEM, DMEM + 10% FCS, or DMEM + 10% Ultroser G. Initial measurements of gel thickness were taken, and the gels were placed in culture for a 20-h time period.

Measurement of collagen gel thickness. Gel thickness was measured using a digital dial micrometer as described by Tuan and Grinnell(25). These measurements were made at time 0, and after 20 h. The percent of gel contraction was calculated based on the difference between the initial measurement and the 20-h time interval gel measurement.

Measurement of TGF-β activity. TGF-β activity was measured using the PAI-1 luciferase activity assay in mink lung epithelial cells stably transfected with an expression contruct containing a truncated PAI-1 promoter fused to a firefly luciferase reporter gene(26). Addition of active TGF-β to these cells results in a dose-dependent expression of luciferase. Transfected mink lung epithelial cells were plated in 96-well plates at 1.6 × 104 cells per well in DMEM with 10% FCS and allowed to attach for 4 h. The serum-containing medium was then removed and replaced with conditioned medium from fibroblasts, controls, or with TGF-β standards diluted in the appropriate medium. The controls for each group consisted of 1) DMEM, 2) DMEM + 10% FCS, or 3) DMEM + 10% Ultroser G medium alone placed on plastic, a collagen coat, or a collagen gel treated in the same manner as the experimental fibroblast groups. The fibroblast culture supernatants and controls were diluted with DMEM + 0.1% BSA as required for assay. After 18-h incubation with the conditioned medium, the cells were washed with PBS and lysed at 4 °C with lysis buffer provided in the enhanced luciferase assay kit from Analytical Luminescence Laboratories (Ann Arbor, MI). The lysates were assayed for luciferase activity using the substrates provided in the kit and a scintillation spectrometer equipped with a single photon monitor. Samples were analyzed in triplicate. For measurement of total TGF-β activity, aliquots of the conditioned medium were heated at 80 °C for 5 min and cooled to room temperature before assay.

Plasmin assay. The plasmin assay was performed exactly as described by Khalil et al.(27). Briefly, plasmin was quantitated by measuring the increase in absorbance at 405λ due to cleavage of the plasmin-specific chromogenic substrate N-p- tosyl-gly-pro-lys-p-nitroanilide (Sigma Chemical Co.). The analysis of all experimental and standard curve samples were performed in 96-well flat-bottomed microtiter plates (Falcon) using a Thermomax Microplate Reader (Molecular Devices Corp., Sunnyvale, CA). The standard curve was made by incubating 2 mM N-p-tosyl-gly-pro-lys-p-nitroanilide in the presence of a range of concentrations of plasmin (Sigma Chemical Co.) from 1 × 10-4 U/mL to 1 × 10-2 U/mL. To measure plasmin generated by the skin fibroblasts, 100 μL of either fibroblast conditioned medium or control medium were incubated with 100 μL of the plasmin substrate. The control media were: 1) DMEM alone, 2) DMEM+ 10% FCS, or 3) DMEM + 10% Ultroser G cultured 20 h in the absence of fibroblasts. All samples were tested in triplicate. The samples were incubated for 5 h at 37 °C. Absorbance was measured, and the background absorbance from the control medium was subtracted from all experimental and standard curve samples. The quantity of plasmin activity present was then derived from the values obtained in the standard curve and presented as units of plasmin activity per 106 fibroblasts.

Cell counts. The cell counts were performed by digesting the collagen gels and removing the cells from either a plastic culture dish or a collagen-coated culture dish with a solution of 2.5% Dispase and 0.25% collagenase at 37 °C for 30 min. Cells for each condition were pooled, and an aliquot was taken for counting on a hemocytometer.

Cell identification. During each isolation 0.4 × 105 cells were placed in each well of an 8-chamber slide. The chamber slides were incubated for 20 h in DMEM, DMEM + 10% FCS, or DMEM + 10% Ultroser. After 20 h the medium was removed, and the cells were washed with PBS three times. The cells were then fixed to the slide with ice-cold 5% acid alcohol for 10 min. The cells were identified by immunostaining with anti-vimentin goat anti-mouse (ICN, Irvine, CA), anti-pancytokeratin monoclonal mixture (Sigma Chemical Co.), or anti-PECAM-1 rat anti-mouse(Pharmingen, San Diego). PECAM-1 is a constituent of the endothelial cell intercellular junction and is used in this cell isolation as a marker of endothelial cells(28). A fetal mouse umbilical cord preparation fixed in acid alcohol was used as a positive tissue control. Negative controls included a reagent control and an immune serum control. Zymed (South San Francisco, CA) Histostain SP kits were used to generate a signal for each antibody. The staining revealed a greater than 99.9% pure fibroblast cell population (vimentin+, cytokeratin-, PECAM-1-), with the greatest contaminant for all three age groups being keratinocytes. There was a range of 0 to 4 keratinocytes counted per 400,000 cells plated.

Statistics. Statistical analysis was done using ANOVA and Tukey's multiple comparison tests. Statistical significance for differences between groups of data were accepted with p < 0.05. The contraction, total, and active TGF-β data were analyzed from four different isolations of E15, E17, and adult fibroblasts. The actual values of total TGF-β were logarithmically transformed using natural log (ln) of(TGF-β + 1) before analysis to compare the values from the three different groups on the same numeric scale. Three of the four isolations were used for the plasmin assay.

RESULTS

Fetal (E15) skin fibroblasts contract collagen gels less than late fetal (E17) or adult skin fibroblasts. Collagen gel contraction(-X ± SEM) by E15, E17, and adult fibroblasts cultured in each of three culture media, DMEM, DMEM + 10% FCS, or DMEM + 10% Ultroser G, are presented in Figure 1. In all three media tested the E15 fibroblasts contracted a collagen gel significantly less than the E17 and adult fibroblasts (ANOVA n = 126, p < 0.0001).

Figure 1
figure 1

Early fetal skin fibroblasts (E15) contract collagen gels significantly less than late fetal (E17) and adult fibroblasts in all three media types DMEM, DMEM + 10% FCS, or DMEM + 10% Ultroser (ANOVA,n = 126, p < 0.0001). (*) Tukey's multiple comparison tests E15 < E17 = adult (p < 0.05). Each bar represents -X ± SEM.

Full size image

DMEM + 10% FCS or 10% Ultroser G stimulated contraction more than DMEM alone. We found that fibroblasts of all ages contracted the collagen gel in DMEM, a medium with essential nutrients, but no growth factors. However, the contraction was significantly enhanced in DMEM + 10% FCS or DMEM+ 10% Ultroser G (ANOVA, n = 126, p < 0.0001). Although the addition of FCS and Ultroser G to DMEM induced fetal (E15) skin fibroblasts to contract better than they did in DMEM alone, they were still not stimulated to contract to the same degree as either E17 or adult fibroblasts (Tukey's multiple comparison test, p < 0.05). The adult skin fibroblasts contracted better in the Ultroser G-supplemented medium than in the FCS-supplemented medium.

Total and active TGF-β secretion by E15, E17, and adult skin fibroblasts. The statistically transformed (natural log of TGFβ + 1) values for total and active TGF-β (-X ± SEM) at each developmental age, for fibroblasts grown in DMEM, DMEM + 10% FCS, or DMEM+ 10% Ultroser G are shown in Figure 2, A and B. The differences in the secretion of total TGF-β between age groups (ANOVA, n = 95, p = 0.01) and medium (ANOVA,n = 95, p < 0.001) were statistically significant. The differences in secretion of active TGF-β between age groups (ANOVA,n = 89, p = 0.001) and media (ANOVA, n = 89,p < 0.001) were also statistically significant. Tukey's multiple comparison analysis were used to detect significant differences between each combination of age and media.

Figure 2
figure 2

The secretion of total TGF-β (A) and active TGF-β (B) by E15, E17, and adult fibroblasts cultured in DMEM, DMEM + 10% FCS, or DMEM + 10% Ultroser is shown. Each vertical bar represents -X ± SEM TGF-β transformed by the equation natural log (ln) of (TGF-β + 1). (*) Tukey's multiple comparison analysis of age and media p < 0.05.**p < 0.05 comparison of media of DMEM alone, vs DMEM + 10% FCS, and vs DMEM + 10% Ultroser.

Full size image

In DMEM alone, in the absence of exogenous growth factors, the E15 skin fibroblasts expressed significantly less total and active TGF-β(p < 0.05) than either the E17 or adult fibroblasts. Moreover, the level of expression of total and active TGF-β was similar between the E17 and adult fibroblasts grown in DMEM.

In DMEM + 10% FCS, the E15 and E17 fetal skin fibroblasts were stimulated to express more active and total TGF-β than they did in DMEM alone. The E17 fibroblasts secreted significantly more active TGF-β in response to serum than did either the E15 or adult fibroblasts (p < 0.05). In contrast, the secretion of total and active TGF-β was suppressed in adult skin fibroblasts grown in DMEM + 10% FCS, Thus, although the addition of serum causes the up-regulation of total and active TGF-β in both the early and late fetal fibroblasts, it suppresses the secretion of both active and total TGF-β in the adult fibroblasts.

In the presence of DMEM + 10% Ultroser G, the secretion of both active and total TGF-β was markedly down-regulated in all three groups of fibroblasts. This difference was statistically significant with p< 0.05.

The culture substrata did not affect TGF-β secretion. The secretion of TGF-β by E15, E17, or adult fibroblasts was not significantly different whether they were plated on plastic, collagen-coated plastic, or in a collagen gel, neither among the age groups nor in any of the different media. Thus, cell culture substrata per se did not confer any significant difference on TGF-β secretion in any of the analyses performed.

Plasmin activity levels in conditioned media from E15, E17, and adult skin fibroblasts were significantly different. The levels of plasmin activity detected (-X ± SEM) in the conditioned media from E15, E17, and adult fibroblasts are shown in Figure 3. The data were not normally distributed, and therefore Wilcoxon scores were used to analyze the data and the Kruskal-Wallis χ2 approximation test was used to determine significance. There was a significant difference in the production of plasmin between the different age groups(n = 68, p = 0.033), and in the different media conditions(n = 68, p = 0.006). The formation of plasmin by E15 skin fibroblasts was almost negligible in all three media, whereas the E17 and adult fibroblasts generated significant quantities of plasmin in all three media.

Figure 3
figure 3

The generation of plasmin by E15, E17, and adult skin fibroblasts in DMEM, DMEM + 10% FCS, DMEM + 10% Ultroser are shown. The plasmin levels in the conditioned media from E15 skin fibroblasts in all three media were almost negligible. The E15, E17, and adult skin fibroblasts generated statistically different quantities of plasmin by Kruskal-Wallis analysis, n = 68, p = 0.033.

Full size image

TGF-β activation correlates with the formation of plasmin by E15, E17, and adult skin fibroblasts. We used a multivariate ANOVA model to correlate the production of active TGF-β and plasmin. There was a weak, but significant, positive correlation between the production of active TGF-β and plasmin (r = 0.44, n = 68, and p< 0.001).

DISCUSSION

Wound contraction is a vital component of wound repair, but in the extreme may lead to excessive scar formation and pathologic wound contracture(1). In fetal wound repair the skin heals without a scar(1–3). We have determined that early fetal murine fibroblasts do not contract a collagen gel to the same extent as late fetal and adult fibroblasts, indicating that the contractile properties of early gestation fetal mouse fibroblasts are qualitatively different from late gestation fetal skin and adult skin mouse fibroblasts. We studied skin fibroblasts freshly isolated from normal mouse skin cultured 20 h to determine the functional developmental differences between early fetal skin fibroblasts, late fetal skin fibroblasts, and adult skin fibroblasts, while minimizing the confounding effects of prolonged in vitro culture.

Numerous studies on fetal wound healing and the factors that alter or contribute to scarless repair have been performed in the hope of developing strategies to alter adult wound repair and reduce scarring(1, 2, 5). These previous studies involve inducing wounds in the fetal environment, placement of sponges laden with factors that allow fibroblast ingrowth, and the investigation of properties of fetal fibroblast cell lines(3, 12, 13, 15, 18, 29–34). These manipulations, however, may have selected certain populations of fibroblasts and therefore may not reflect the properties of the entire skin fibroblast population. Furthermore, fetal cells differentiate with passage in culture(7).

Ihara et al.(33) examined both in vivo and in vitro wound healing in fetal rat skin wounds, and found and ontogenic transition in the wound healing process between 16 and 18 d of gestation. Whitby and Ferguson(34) examined the extracellular matrix of lip wounds in fetal, neonatal, and adult mice by immunohistochemistry. They found that E14 mouse lip wounds established a reticular pattern of collagen fibers more rapidly than did E16 and E18 lip wounds. However, a reticular collagen pattern was eventually established in the E16 and E18 fetus, but normal dermal structures were not restored. Estes et al.(7), found that fibroblasts from early fetal sheep dermal wounds did not express α-smooth muscle actin, a marker of myofibroblast transdifferentiation in vivo. However, after multiple passages in vitro in the presence of serum, these early fetal sheep skin fibroblasts did express α-smooth muscle actin. Taken together with our new results, these findings support the hypothesis that early fetal fibroblasts are inherently different from late fetal fibroblasts. In our study this transition in the fibroblast phenotype occurred between E15 and E17 in the mouse.

Our studies on the generation of TGF-β and plasmin by skin fibroblasts further support a transition in phenotype between early and late fetal fibroblasts. We found that, like the contractile properties, the expression of TGF-β was also developmentally regulated. In the absence of external growth factors in the media, the early (E15) fetal fibroblasts did secrete significantly less total and active TGF-β than did E17 skin fibroblasts, the latter secreting similar amounts of both active and total TGF-β to adult skin fibroblasts. The finding that the fibroblast-conditioned media, in the absence of external growth factors, contained active TGF-β, was interesting in that TGF-β is secreted as an inactive TGF-β precursor, which is noncovalently bound to the latency activation peptide(17). Thus, we deduced that these freshly isolated skin fibroblasts could activate TGF-β.

The serine protease plasmin activates latent TGF-β in conditioned media from fibroblast cell lines(35). Plasmin is a cleavage product of plasminogen and requires the enzymatic action of either urokinase or tissue-type PA for its activation. The enzymatic actions of urokinase PA and tissue PA are inhibited by PAI(36). Plow et al.(37) examined a fetal lung fibroblast cell line and found that these cells express both plasminogen receptors and urokinase PA receptors, which bound plasminogen and urokinase PA, respectively, in a dose-dependent manner. These fibroblast receptors could also bind the active plasmin molecule, and thus protect the active plasmin from the serum inhibitor α2-antiplasmin. The latter mechanism would allow proteolysis to occur, even in a profibrotic environment. Thus fibroblasts possess the cellular mechanisms that can generate active plasmin. In our future studies we hope to examine our freshly isolated skin fibroblasts for plasminogen receptors, urokinase PA receptors, and PAI-1 expression to further elucidate the cause for the marked differences in plasmin formation between the early fetal, late fetal, and adult skin fibroblasts.

In our study we examined the end result of plasminogen activation, the formation of plasmin by the early, late, and adult fibroblasts. These cells were exposed to serum during isolation, and thus would have had the opportunity to bind the precursor of active plasmin, plasminogen. We found that early fetal mouse skin fibroblasts generated nearly undetectable levels of plasmin compared with late fetal and adult mouse skin fibroblasts. Moreover, the generation of plasmin was only weakly correlated with the production of active TGF-β. Thus, plasmin may not be the actual physiologic activator of TGF-β, but may be an intermediate factor required for the eventual production of active TGF-β. This would suggest a potential scenario in which fetal skin fibroblasts would not only secrete TGF-β peptides at low levels, but would also fail to activate them due to a lack of plasmin.

Proteases other than plasmin have been implicated as possible activators of TGF-β. The endoprotease proprotein convertase 5 has been implicated in the hormone bioactivation of Müllerian-inhibiting substance, a member of the transforming growth factor superfamily(38). The mammalian proprotein convertase enzymes are related to both the yeast kexin and the bacterial subtilisin endoproteases. These enzymes are implicated in hormone bioactivation via proteolytic processing after dibasic amino acid cleavage recognition sites. The convertase enzymes are evolutionarily conserved, and have been shown to be important during development(38). Thus, a proprotein convertase may be the actual physiologic activator of TGF-β, and plasmin may play a role in one of the intervening steps leading to its activation.

Although TGF-β has been shown to augment skin fibroblast contraction, its presence is evidently not essential for contraction to occur, because both fetal and adult skin fibroblasts can contract collagen gels in Ultroser G, a serum substitute that lacks TGF-β. Furthermore, the expression of TGF-β by early fetal, late fetal, and adult skin fibroblasts was suppressed in the presence of Ultroser G.

Further studies will be necessary to determine the role of growth factors other than TGF-β in skin fibroblast contraction. Studies of the effects of growth PDGF, however, have been performed. These studies reveal that PDGF augments collagen gel contraction(39) and induces fetal wounds to scar(40). The serum substitute Ultroser G contains PDGF and thus may explain our findings of augmented contraction in medium supplemented with 10% Ultroser. Moreover, we speculate that the down-regulation of both active and total TGF-β secretion by the skin fibroblasts cultured in media supplemented with Ultroser G may also be related to an effect of the exogenous administration of PDGF.

Our studies revealed that the substratum on which we cultured the skin fibroblasts (plastic, collagen-coated plastic, and collagen gel) had no effect on the secretion of TGF-β. Thus, it appears that the cytokine environment had a greater influence on the expression of TGF-β than the substratum on which the cells were placed.

In conclusion, early fetal mouse skin fibroblasts contract a collagen gel and secrete and activate TGF-β to a lesser extent than late fetal and adult skin fibroblasts. Furthermore, early fetal skin fibroblasts generate less plasmin than do late fetal or adult skin fibroblasts. We speculate that the fetal skin fibroblast undergoes a developmental transition that causes wounds in mouse to contract at or after E17. We further speculate that this developmental transition is influenced by growth factor-fibroblast interactions and coincides with the emergence of the ability of the fibroblasts to generate plasmin and activate TGF-β, although TGF-β is evidently not the sole factor that can mediate gel contraction. Finally, extracellular matrix-fibroblast interactions do not appear to play a significant role in the short-term modulation of fibroblast expression of TGF-β and plasmin.