Isometric tension has been shown to be important in vitro for the development of fibroblast contractile features such as stress fiber formation ^18-20 and α-SMA expression. ²¹ Mechanical tension alters FN fiber assembly ²² and fibroblast response to growth factors; ^23,24 it also prevents fibroblast apoptosis. ¹⁹ An altered mechanical load of the matrix stimulates fibroblasts to adjust their contractile activity to achieve tensional homeostasis. ²⁵ Moreover, isometric tension modulates signaling mechanisms regulating fibroblast contraction. ²⁶ Most previous studies on the effect of mechanical tension on fibroblastic cells have been restricted to the in vitro use of collagen substrates with different mechanical properties. ²⁷ In this work, we have used an in vivo approach using two different rat models in view of evaluating: the role of mechanical load in myofibroblast modulation (α-SMA expression in particular) and in granulation tissue contractility. In the first model, the contraction of full thickness wound granulation tissue was prevented by splinting with a plastic frame. In the second model, the wall of rat granuloma pouches ⁴ was released from mechanical tension by evacuating the pouch exudate. In both models isometric contraction and expression of the myofibroblast markers F-actin, ED-A FN, α-SMA, TGF-β-RII, and TGF-β1 were evaluated. Our results indicate that mechanical tension is crucial for differentiation and maintenance of the myofibroblast phenotype. In addition, we show that tissue contractility correlates with the expression of α-SMA.

Granuloma pouches were induced in rats by injecting 15 ml of air into the dorsal subcutaneous tissue, followed by 1 ml of 1% croton oil in maize oil. ^28,29 This treatment produces the formation of an abscess-like structure whose wall is rich in myofibroblasts ^4,29,30 and has been used to study the effects of several smooth muscle contraction-inducing and -inhibiting agents on the isometric force development by a myofibroblastic tissue. ^29,30 Pouches were either left intact or released from internal tension by evacuating ~3 ml of the exudate after 14 days using an 18-gauge needle. Evacuation was repeated every 2 days to prevent new development of tension. To control whether the removal of exudate has other effects than tension release, we replaced it with an equal volume of physiological saline in an additional group. To study the role of TGF-β1 in granuloma pouch wall development, we injected daily 14-day-old granuloma pouches with 1 ml of soluble TGF-β-RII (final, 1 μg/ml muTGF-β-sR; gift of Dr. V. Koteliansky, Biogen Inc., Cambridge, MA) ^31,32 for 7 days in another group. Tension-released granuloma pouches were dissected after 15, 16, 17, 18, and 21 days, intact pouches after 14 days and 21 days. Dissected pouches were transferred to ice-cold Krebs-Henseleit solution, bubbling with 95% O₂ and 5% CO₂, and thoroughly washed. Only tissue from the dorsal portion of granuloma pouches was used to standardize experiments. As control we used normal dermis and dorsal fascia covering the muscular layer. Fascia was exposed after full-thickness incision of skin and subcutaneous tissue and removed using fine forceps and scissors. Dermis was prepared by removing the epidermis and subcutaneous tissue with a scalpel blade. Fascia and normal dermis were further treated as described for granulation tissue and granuloma pouch.

For confocal laser-scanning microscopy tissue strips were snap-frozen in precooled liquid isopentane and embedded in OTC™ resin (Miles Scientific, Naperville, IL). Cryosections of 4 μm were stained for α-SMA (anti-αSM-1, IgG2a mAb), ⁶ F-actin (phalloidin-Alexa 488; Molecular Probes, Eugene, OR), ED-A FN (IST-9, IgG1 mAb; gift from Dr. L. Zardi, National Institute for Cancer Research, Laboratory of Cell Biology, Genoa, Italy), ^33,34 desmin (D33, IgG1 mAb; DAKO), smooth muscle myosin heavy chain (SMMHC, rb polyclonal Ab ³⁵ or BT-562, rb polyclonal Ab; Biomedical Technologies, Stoughton, MA), and non-muscle myosin heavy chain (NMMHC, rb polyclonal Ab) ³⁵ and DNA [4,6-diamidino-2-phenylindole (DAPI)]. As secondary antibodies goat anti-mouse IgG2a tetramethylrhodamine B isothiocyanate-conjugated, goat anti-mouse IgG1 fluorescein isothiocyanate-conjugated, and goat anti-rabbit tetramethylrhodamine B isothiocyanate-conjugated (Jackson ImmunoResearch Laboratories, West Grove, PA) were used. All slides were mounted in buffered polyvinyl alcohol. Images were taken by means of confocal laser-scanning microscopy (LSM 410, Zeiss) using a Plan-Neofluar ×10/0.50 objective or a ×63/1.4 Plan-Neofluar oil immersion objective with a digitally extended focus depth of 1 μm, reconstructed from five optical sections at 0.2-μm z-resolution. All digital images were processed for printing using Adobe Photoshop and printed with a digital Fujifilm Pictrography 4000 printer (Fujifilm, Tokyo, Japan).

To quantify F-actin by fluorescence image analysis 4-μm sections of granulation tissue strips were fixed and triple-stained for F-actin (red), desmin (green), and cell nuclei (blue) as described above. For a given region of interest digital images with a resolution of 515 × 512 pixels were taken at red-, green-, and UV-light excitation, respectively, using a Plan-Neofluar ×20/0.50 objective (Zeiss), color-camera (Coolview, Photonic Science Ltd.), and image grabber software (ImageAccess V2.04K; Imagic Bildverarbeitung AG, Glattbrugg, Switzerland). Constant gain values and integration times were used for the digital camera to obtain similar exposure times for all specimens. To determine the level of F-actin, images were then automatically processed by KS400 software (Zeiss). First, the three 24-bit Red · Green · Blue-images per region were converted to 8-bit gray value images with pixel intensity values ranging from 0 (black) to 255 (white). Desmin staining, specific for vascular smooth-muscle cells, was automatically selected by a thresholding process and corresponding pixel values were set to 255. To exclude vessels from F-actin quantification, desmin-positive areas were then subtracted from F-actin and DAPI images. Second, the background of the desmin-corrected F-actin image was calculated from a manually chosen F-actin-negative area and subtracted from the mean pixel intensity of the whole image, thereby quantifying the level of F-actin expression in the region of interest. Finally, the F-actin expression level was related to the number of cells in the region, represented by the number cell nuclei calculated from the desmin-corrected DAPI image. Tissue was collected from three animals per experimental condition, five sections were prepared per animal, and five regions of interest were analyzed per section.

Figure 1. Isometric contraction of wound granulation tissue. Maximum contraction of tissue strips taken from 3- to 12-day-old wound granulation tissue and normal dermal tissue (0 days) was quantified after stimulation with ET-1 (A) and AT-II (B). Tissue from splinted (more ...)

Figure 2. Mechanical tension enhances expression of F-actin and ED-A FN in wound granulation tissue. A: Sections of 3-day-old (Aa, Ae), 6-day-old (Ab, Af), 9-day-old (Ac, Ag), and 12-day-old (Ad, Ah) granulation tissue were double-stained for F-actin (green) and (more ...)

α-SMA was not expressed in fibroblasts of normal dermis (not shown) and of 3-day-old granulation tissue (Figure 3, A and E) , and was not detected in fibroblasts of 6-day-old unsplinted granulation tissue (Figure 3B) . In unsplinted wounds, it was first detected after 7 days (not shown), was strongly expressed after 9 days (Figure 3C) , and expression was clearly reduced after 12 days (Figure 3D) . In contrast, splinting induced α-SMA expression by granulation tissue fibroblasts already at 4 to 6 days after wounding (Figure 3F) ; the level of α-SMA expression increased in splinted tissue to a maximum at 9 days (Figure 3; E to G ) and remained at an apparently similar level at 12 days (Figure 3H) . Three-day-old to 12-day-old splinted wounds exhibited increased expression of both TGF-β1 and TGF-β-RII; however, no clear differences were observed between splinted and unsplinted wound tissue by means of immunohistochemistry (not shown). Desmin and SMMHC were not expressed by granulation tissue fibroblasts, and expression of NMMHC did not change in any condition as tested by means of immunofluorescence and Western blotting (not shown).

Figure 3. Wound splinting promotes early expression of α-SMA in granulation tissue. Sections of 3-day-old (A, E), 6-day-old (B, F), 9-day-old (C, G), and 12-day-old (D, H) granulation tissue were double-stained for α-SMA (red) and desmin (green). (more ...)

Analysis of myofibroblast marker expression in wound granulation tissue by Western blotting (Figure 4) essentially confirmed immunolocalization results. In unsplinted tissue ED-A FN expression was continuously increased from 3 days to 9 days (630%) and exhibited a reduction by 160% between 9 days and 12 days (Figure 4, A and B) . Splinting enhanced the amount of ED-A FN (Figure 4A) with the largest difference at 3 days (Figure 4B , 347%). Expression of α-SMA moderately increased from 3-day to 6-day unsplinted tissue (195%), steeply increased from 6 days to 9 days (400%), and was reduced by 145% between 9 days and 12 days (Figure 4, A and C) . Splinted tissue exhibited increased α-SMA expression levels at 6 days (Figure 4C , 420%) and showed similar α-SMA expression at 9 days compared to unsplinted granulation tissue. Splinting maintained high α-SMA expression levels at 12 days in contrast to the decreased expression in unsplinted wound tissue (Figure 4C) . TGF-β-RII expression increased from 0 to 12 days after wounding with no significant difference between splinted and unsplinted granulation tissue (Figure 4, A and D) . To relate changes in myofibroblast marker expression to enhanced tissue contractility during the evolution of unsplinted and splinted wound granulation tissue, we calculated correlation coefficients. Highest correlation was observed between the temporal course of ET-1-stimulated tissue contraction (Figure 1A) and expression of α-SMA (r ² = 0.98). Lower correlation was calculated for TGF-β-RII (r ² = 0.92), ED-A FN (r ² = 0.90), and F-actin (r ² = 0.84), no correlation for β-actin (r ² = 0.69), NMMHC (r ² = 0.61), desmin (r ² = 0.57), and vimentin (r ² = 0.55).

Figure 4. Quantification of myofibroblast markers in granulation tissue by means of Western blot. A: Expression of β-actin, vimentin, and the myofibroblast markers TGFβ-RII, ED-A FN, and α-SMA was investigated. Extracts of 3- to 12-day-old (more ...)

To investigate further the chronological relationship between tissue contraction and myofibroblast modulation after release of mechanical tension, we used the granuloma pouch model. ²⁹ Tension was released in granuloma pouch tissue at 14 days by evacuating the exudate and isometric contraction was measured 1 to 7 days after release (Figure 5) . Contraction of 21-day control pouch tissue was maximally stimulated by ET-1 (10⁻⁷ mol/L) and by the following agonists in percentage of ET-1-response: AT-II (10⁻⁵ mol/L), 60%; serotonin (10⁻⁴ mol/L), 48%; potassium (4 × 10⁻³ mol/L), 45%; and histamine (10⁻⁴ mol/L), 6%. Results are presented for ET-1 (Figure 5A) and AT-II (Figure 5B) . Tissue contractility was higher in intact 21-day-old compared to 14-day-old pouches (Figure 5 , dashed lines), confirming previous observations. ^30,38 Release of mechanical tension at 14 days decreased pouch tissue contraction beginning 2 days after release, exhibited a minimum at 3 days after release (14d+3dR), and kept this level until 7 days. Compared to intact pouches of the same age (21 days), tension release resulted in a significantly lower tissue contraction (38% for ET-1 and 46% for AT-II). As a control, we substituted the evacuated exudate with an identical volume of physiological saline for 7 days (Figure 5 , 14+7dS). Saline-injected pouch tissue contraction did not differ from intact pouches of the same age (21 days).

Figure 5. Loss of mechanical tension decreases granuloma pouch tissue contraction. Maximum contraction of tissue strips from granuloma pouches was quantified after stimulation with ET-1 (A) and AT-II (B). Pouches were either left intact for 14 days and 21 days (more ...)

Alexa 568-phalloidin (Figure 6 , insets) revealed an intense F-actin staining in 14-day-old granuloma pouch fibroblasts, that was clearly decreased already 2 days after release (Figure 6B) ; this decrease was more important at 7 days after release (Figure 6C , inset). ED-A FN and α-SMA were co-localized and their expression increased in 21-day-old compared to 14-day-old intact granuloma pouch tissue (Figure 6, A and D) . Three days after tension release α-SMA expression was reduced (Figure 6B) ; 7 days after tension release, ED-A FN expression was also reduced and α-SMA was hardly detectable (Figure 6C) . TGF-β1, previously shown to be expressed in granuloma pouch tissue, ^39,40 and TGF-β-RII were both predominantly expressed in the external collagen-rich layer of the pouch tissue. TGF-β1 and TGF-β-RII expression patterns in 7-day-released pouches did not differ significantly from 1- to 5-day-released tissue or 21-day intact pouch tissue, but the thickness of the collagen-rich layer decreased progressively during the 7-day period of tension release to ~70% compared to 21-day intact pouches (not shown).

Figure 6. Loss of mechanical tension reduces myofibroblast differentiation in granuloma pouch tissue. Tissue sections of intact (A, D) and tension-released granuloma pouches (B, C) of different ages were double-stained for α-SMA (red) and ED-A FN (green) (more ...)

As determined by Western blotting, α-SMA was significantly reduced 2 to 3 days after release, compared to 14-day-old intact pouch tissue, whereas decrease of ED-A FN expression started 4 days after release (Figure 7) . TGF-β-RII expression was only moderately diminished throughout the 7-day period after release and no change was observed for β-actin and vimentin expression (Figure 7A) . Compared to 21-day-old intact pouch tissue, 7-day tension release reduced α-SMA expression to 51%, ED-A FN to 47%, and TGF-β-RII to 81% (Figure 7B) . In contrast to wound granulation tissue fibroblasts, which did not express desmin and SMMHC, low expression of both proteins was observed in fibroblastic cells of 14-day and 21-day intact granuloma pouches. Tension release did not change desmin and NMMHC expression and only moderately diminished expression of SMMHC 7 days after release as evaluated by Western blotting and immunofluorescence (not shown). When removed exudate was replaced by physiological saline, no changes in protein expression were detected (Figure 7A , 14d+7dS).

Figure 7. Quantification of myofibroblast markers in granuloma pouch tissue by means of Western blot. A: Protein expression in intact granuloma pouches (14d, 21d) was compared to 14-day-old pouches, which were subsequently released from mechanical tension for 1 (more ...)

Injection of soluble TGF-β-RII into the granuloma pouch for 7 days decreased tissue contraction to 49% after stimulation with ET-1 (Figure 5A , 14d+7dT) and to 48% after AT-II stimulation (Figure 5B , 14d+7dT) compared to 21-day-old control pouch tissue. It reduced the expression levels of TGF-β, TGF-β-RII, ED-A FN, F-actin, and α-SMA significantly compared to 21-day-old control pouch tissue as demonstrated by Western blotting (Figure 7A) and immunofluorescence (not shown).

It is accepted that mechanical tension plays an important role in the modulation of cultured fibroblasts into myofibroblasts by favoring the formation of stress fibers and the expression of α-SMA. ²⁷ Here, we demonstrate that mechanical tension also influences myofibroblast differentiation in vivo. Fibroblasts populating the granulation tissue of splinted wounds exhibit earlier formation of stress fibers, expression of ED-A FN and of α-SMA compared to nonsplinted wounds. Inversely, release of tension in wound granulation tissue and in granuloma pouch leads to a sequential loss of stress fibers, α-SMA, and ED-A FN. Moreover, our study indicates a correlation between the level of α-SMA expression and contractility of the tissues obtained during the different experimental situations. The acquisition of contractile activity by granulation tissue on stimulation with smooth muscle agonists is a well-accepted phenomenon and has been suggested to represent a useful indication of the retractile potential of granulation tissue. ^29,30,41 Acquisition of an increased contractile activity by wound splinting ⁴² has been correlated to the induction of microfilament bundles in fibroblasts. ¹⁸ Our results corroborate these observations and establish a correlation between changes in tissue contractility and cytoskeletal protein expression and/or organization after mechanical load modulation.

In both splinted and unsplinted wounds, the granulation tissue strip contractile activity evolves according to three phases, as depicted schematically in Figure 8 . During the first phase (red) the contractility of granulation tissue increases slowly, whereas during the second phase (blue), the increase is steep; the third phase (green) corresponds to a decrease of contractile activity. In the first phase there is an increase of F-actin and of ED-A FN expression that is clearly more important in splinted wounds. The onset of high contractile activity correlates always with the de novo expression of α-SMA, assembled into stress fibers. It is noteworthy that the first phase is shorter and the second phase is longer in splinted compared to unsplinted wounds and corresponds to the maximal level of α-SMA expression. Interestingly, the maximal level of contractile capacity is similar in both situations. Finally, a decrease in contractile activity takes place more slowly in splinted than in unsplinted wounds. The observed correlation between α-SMA expression and granulation tissue contractility is in agreement with the previously reported correlation between α-SMA expression and contraction of fibroblast-populated anchored collagen gels. ^8,9 More recently we have shown that 3T3 fibroblasts exhibit a significantly higher contractile activity after transient and/or stable transfection with α-SMA cDNA compared to fibroblasts transfected with β-actin, γ-actin, or even α-cardiac actin cDNA in the absence of any change of myosin heavy chain expression. ⁷ All these data are compatible with the assumption that α-SMA expression alone plays a role in the enhancement of fibroblast contractility in vitro. Our results allow to hypothesize that also in vivo α-SMA expression plays a crucial role in the development of granulation tissue contractile capacity. In all situations, increased tension results in a greater expression of α-SMA and an increased force production, whereas decreased tension has the opposite effect. ^2,21,43,44 Although in vivo and in vitro data indicate that α-SMA expression is stimulated by mechanical stress, it remains unclear how fibroblasts translate the physical signal into protein expression. In cells such as cardiomyocytes, endothelial and smooth muscle cells, mechanical stress was shown to stimulate the production and secretion of growth factors that mediate stress-induced cell responses. ⁴⁵ In particular, expression of TGF-β1, the major inducer of α-SMA expression ^13,14 and contractility in myofibroblasts, ^8,21 is up-regulated in mechanically stressed mesangial cells. ⁴⁶ Under our conditions the expression of TGF-β1 and of its receptor TGF-β-RII was gradually increased, but similarly in splinted tissue and unsplinted tissues. Nevertheless, blocking TGF-β1 by an excess of soluble receptor without changing the mechanical load reduced α-SMA expression and tissue contraction in our experimental conditions, implying a role of TGF-β1. Enhanced mechanical stress increases growth factor sensitivity ^23,24,47 and loss of mechanical tension reduces the growth factor response of cultured fibroblasts. ^48,49 It remains to be examined whether TGF-β1 activation changes in our experimental situations.

Figure 8. Evolution of granulation tissue contractility. The chronological evaluation of granulation tissue contractile activity distinguished three phases characterized by: 1) slow increase (red), 2) steep increase (blue), and 3) decrease (green). These phases (more ...)

Changes in mechanical stress may directly affect the level of extracellular matrix proteins. ^50-52 Recently, it was shown that during wound healing ED-A FN expression ^53,54 precedes the appearance of α-SMA-positive myofibroblasts and is essential to mediate TGF-β1-induced α-SMA expression. ¹⁷ Splinting induces ED-A FN expression in granulation tissue, suggesting a possible regulatory function of this protein during stress-induced myofibroblast differentiation. Regulation may occur at the level of FN expression, ⁵⁵ alternative splicing, ^56,57 fibril assembly, ²² or accessibility of functional residues in the ED-A domain of FN ⁵⁸ as it has been demonstrated for plasma FN. ⁵⁹ Integrins of focal adhesions are potential receptors and transducers of these extracellular signals. ⁶⁰

When myofibroblasts were released from mechanical stress by removing the wound splint or by evacuating the granuloma pouch, disappearance of α-SMA preceded the decrease of ED-A FN and TGF-β1 levels. Thus, ED-A FN and TGF-β1 are crucial for the induction of α-SMA expression but are not sufficient to maintain myofibroblast differentiation in the absence of a mechanical stimulus. Taking into account the early formation of F-actin during wound healing and the rapid loss of actin stress fibers after load release, we suggest that the physical integrity of stress fibers ⁶¹ provides an important prerequisite for α-SMA expression and myofibroblast contraction. Mechanical stress induces actin polymerization and stress fiber formation when applied externally ⁶² or by increasing intracellular tension. ^10,63 The ratio between G-actin/F-actin was shown to directly influence β-actin synthesis. ^64,65 However, the role of the G-actin/F-actin ratio for the expression of other actin isoforms, such as α-SMA, has not been evaluated. It is conceivable that mechanical tension induces actin expression without discriminating among isoforms whereas factors such as TGF-β and ED-A FN exert a specific stimulation of α-SMA. ¹⁷ A similar phenomenon has been shown for the stress-dependent expression of α-skeletal actin by cardiomyocytes in the presence of AT-II. ⁴⁵

In conclusion our results show that mechanical tension is a prerequisite for the development and maintenance of myofibroblast differentiation and hence of granulation tissue contraction. Given the reciprocal relationship between fibroblast contractility and the mechanical state of the matrix, the modulation of extracellular and intracellular tension may help to influence wound healing and development of fibrocontractive diseases.

We thank Drs. Victor Koteliansky (Biogen Inc., Cambridge, MA) for providing the soluble TGF-β-RII, Luciano Zardi (National Institute for Cancer Research, Laboratory of Cell Biology, Genoa, Italy) for providing IST-9 antibodies, and Jo de Mey (Department of Pharmacology, University of Maastricht, Maastrich, The Netherlands) for expert advice and teaching; A. Geinoz, M. Bacchetta, P. Henchoz, and S. Coutant-Zimmerli for technical assistance; J. C. Rumbeli and E. Denkinger for photographic work; and Mrs. S. Josseron for secretarial work.