The impact of Smad3 loss of function on scarring and adhesion formation during tendon healing

Abstract

Studies were performed evaluating the role of Smad3, a transcription factor mediating canonical TGF-β signaling, on scarring and adhesion formation using an established flexor digitorum longus (FDL) tendon repair model. In unoperated animals the metatarsophalangeal (MTP) range of motion (ROM) was similar in Smad3^−/− and wild type (WT) mice while the basal tensile strength of Smad3^−/− tendons was significantly (39%) lower than in WT controls. At 14 and 21 days following repair Smad3^−/− MTP ROM reached approximately 50% of the level of the basal level and was twice that observed in WT tendon repairs, consistent with reduced adhesion formation. Smad3^−/− and WT maximal tensile repair strength on post-operative day 14 was similar. However, Smad3^−/− tendon repairs maximal tensile strength on day 21 was 42% lower than observed in matched WT mice, mimicking the relative decrease in strength observed in Smad3^−/− FDL tendons under basal conditions. Histology showed reduced "healing callus" in Smad3^−/− tendons while quantitative PCR, in situ hybridization, and immunohistochemistry showed decreased col3a1 and col1a1 and increased MMP9 gene and protein expression in repaired Smad3^−/− tendons. Thus, Smad3^−/− mice have reduced collagen and increased MMP9 gene and protein expression and decreased scarring following tendon FDL tendon repair.

Introduction

Flexor tendon injuries continue to be a difficult problem for hand surgeons. There has been abundant research in the clinical and biomechanical arenas to identify the ideal suture material, surgical technique, timing of repair, and postoperative rehabilitation regimen (1-5). Despite these efforts, scaring and adhesion formation between tendons and the surrounding structures remain a major complication, often necessitating challenging secondary tenolysis procedures. These continued challenges have led researchers to redirect their focus to the cellular and molecular modulators of healing to try to better understand tendon healing and adhesion formation (6-8). In recent years, the effect of growth factors and cytokines on tendon healing has become an important area of research and TGF-β has emerged as a growth factor of significant interest. In vitro studies have shown TGF-β1 to be a potent fibrotic agent (9). In tendon cell culture, TGF-β1 was shown to affect collagen formation (10). Furthermore, several inhibitors of TGF-β1 have been shown to improve range of motion at a single time point in in vivo tendon studies (11; 12).

Despite these studies, little is known about the intracellular mechanism or downstream signals by which TGF-β1 modulates these effects in healing tendons. Recent studies have shown that Smad proteins act as critical transcription factors for TGF-β (13; 14). Three groups of Smad proteins—receptor activated Smads, common mediator Smads and inhibitory Smads—exist (15). Smad3 is a receptor activated Smad that is phosphorylated in response to TGF-β signaling through the TGF-beta type I and TGF-beta type II transmembrane receptors (16). Once activated, Smad3 heterodimerizes with Smad4—a common mediator Smad—and translocates to the nucleus where Smad3 is thought to modulate transcription of genes involved in cell growth (17), inflammatory response (18), and extracellular matrix formation (19).

Thus, since TGF-β signals through the phosphorylation of the intracellular protein Smad3, a plausible approach to abrogate TGF-β’s pro-adhesion role would be to interrupt Smad3 signaling. To test this hypothesis, healing of flexor tendons in wildtype (WT) and Smad3 knockout (Smad3^−/−) mice were compared in terms of functional properties including tendon gliding, the metatarsophalangeal (MTP) joint flexion, and tensile biomechanical properties (using previously described techniques) (20-22). The biological aspects of the healing process were assessed using histological analysis, PCR and in situ hybridization to analyze temporal and spatial gene activity and immunohistochemistry to analyze spatial activation of protein in and around the repair.

MATERIALS AND METHODS

Animals and Tendon Repair Surgery

All animal procedures were approved by the University of Rochester Committee on Animal Research. Mice for this study were operated on from 6-8 weeks of age as previously described (22). This was a homogeneous group of Smad3^−/− animals that included age matched WT controls. Briefly, mice were anesthetized, the flexor digitorum longus (FDL) tendon was transected at the plantar surface of the metatarsal bones and immediately repaired using 8-0 nylon sutures in a modified Kessler pattern (22). The myotendinous junction was released to prevent the generation of active forces across the tendon repair and to protect against disruption during early tendon repair. However, following surgery the mice had unrestricted activity that allowed passive motion of the toes, foot and ankle, and tendon. Limbs were harvested at 14 and 21 days post surgery for non-destructive adhesion testing and biomechanical testing (a minimum of 8 animals per time point were used based on statistical power analysis). Additional samples were harvested on post-operative days 7, 14, and 21 for histological analysis (N=4 repairs per time point), and days 3, 7, 10, and 14 for in situ hybridization and immunohistochemistry (N=4 repairs per time point) and RNA extraction for real-time RT-PCR (N=5 repairs per time point).

Non-Destructive MTP Joint Flexion and Failure Tensile Testing

Limbs were prepared and MTP joint flexion testing was preformed as previously described (20). Briefly, the lower hind limb was rigidly held in place while the proximal FDL tendon was incrementally loaded in the direction of anatomic loading using static weights up to 19 grams. With every increment of the load, digital images were taken to quantify the metatarsalphalageal (MTP) flexion angle relative to the neutral position. MTP joint flexion angles corresponding to each applied load were measured by two independent observers using ImageJ software (http://rsb.info.nih.gov/ij/). The gliding coefficient was determined by fitting the flexion data to a single-phase exponential equation where the MTP flexion angle = β × [1-exp(−m/α)]; where m was the applied load (Prism Graphpad 5.0a for Mac OS X; GraphPad Software, San Diego, CA). This curve fit was constrained to the maximum flexion angle (β) for normal tendons that was previously determined to be 75 degrees for the 19 g applied load (21). Non-linear regression was used to determine the gliding coefficient (α), which was previously demonstrated to correlates inversely with the MTP joint flexion range (21). This corroborates the validity of the Gliding Coefficient as a quantitative measure of the resistance to joint flexion, and is concurrent with published data (23) that demonstrates that gliding resistance significantly correlates with the work of flexion. Immediately following MTP flexion testing, biomechanical testing was performed as previously described (21). In order to preserve the skin and other healing tissues surrounding the repair site, repair sutures were not cut prior to testing. The FDL tendon was mounted in the Instron 8841 DynaMight axial servohydraulic testing system (Instron, Norwood, MA). The sample was then loaded at a rate of 30 mm/min until failure. The ultimate failure force (tensile strength) and the stiffness (the slope of the linear portion of the load-deformation curve) were then determined from the force-deformation curves (21).

Histology

For tendon harvest, the lower hind limb was disarticulated at the knee and the fur was removed down to the ankle without disrupting the skin at the ankle or foot. The foot was fastened to cardboard using 18 gauge needles in order to keep the tendon straight. The tissue was then fixed in 10% neutral buffered formalin for 48 hours. Fixed tissues were washed in PBS and then decalcified for 21 days in 14% EDTA (pH 7.2). The tissues were then processed and embedded in paraffin. Longitudinal 3 μm sections were prepared and stained with hematoxylin and eosin.

RNA Extraction and Real-Time RT-PCR

RNA extraction and real-time RT-PCR were performed as previously described (22). Briefly, FDL tendons were harvested from sacrificed mice and immediately frozen in liquid nitrogen. 5 repaired tendons per time point were pooled and minced using the Ultra Turrax T8 homogenizer (IKA Works, Wilmington, NC). Total RNA was extracted using TRIZOL (Invitrogen Corporation, Carlsbad, CA). Single-stranded cDNA was made using a reverse transcription kit (Invitrogen) and used as a template for real-time PCR with SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA) and gene specific primers in a Rotor-Gene 2000 thermocycler (Corbett Research, Sydney, Australia). The primer sequences are shown in Table 1. Gene expression levels were standardized with the internal control β-actin and normalized to day 3 WT expression levels.

Table 1.

Forward and reverse primer sequences used for quantitative real-time PCR analysis.

Gene	Primer Direction	Primer Sequence
col1a1	Forward	5-GAGCGGAGAGTACTGGATCG-3′
	Reverse	5′-GCTTCTTTTCCTTGGGTTC-3′
col3a1	Forward	5′-GCCCACAGCCTTCTACAC-3′
	Reverse	5′-CCAGGGTCACCATTTCTC-3′
MMP9	Forward	5′-TGAATCAGCTGGCTTTTGTG-3′
	Reverse	5′-ACCTTCCAGTAGGGGCAACT-3′
Smad2	Forward	5′- TGGAAGTGGCCCATTTAGAG-3′
	Reverse	5′- CTGGTTGACAGACTGAGCCA-3′
BMP2	Forward	5′- GCTTTTCTCGTTTGTGGAGC-3′
	Reverse	5′- GCTTTTCTCGTTTGTGGAGC-3′
BMP4	Forward	5′-TGAGCCTTTCCAGCAAGTTT-3′
	Reverse	5′- CTTCCCGGTCTCAGGTATCA-3′
BMP6	Forward	5′- CTCAGAAGAAGGTTGGCTGG-3′
	Reverse	5′- ACCTCGCTCACCTTGAAGAA-3′
BMP7	Forward	5′- GAAAACAGCAGCAGTGACCA-3′
	Reverse	5′- GGT GGC GTT CAT GTA GGA CT-3′
β-Actin	Forward	5′-AGATGTGGATCAGCAAGCAG-3′
	reverse	5′-GCGCAAGTTAGGTTTTGTCA3′

In situ Hybridization

For in situ hybridization, en bloc resections of the surgery site and surrounding soft tissues including the remaining FDL tendon and the skin on the plantar aspect of the foot were harvested with care not to disrupt the injury site. The harvested tendon and surrounding tissues were fixed in 10% neutral buffered formalin, processed and embedded in paraffin. Five micron sections were cut and RNA in situ hybridization was preformed using probes for col1a1, col3a1 and MMP9. Probes were ³⁵S labeled and hybridized as previously described (22). Slides were then placed in emulsion and stored in desiccant boxes at room temperature for 18 to 24 hours. The desiccant boxes were transferred to 4° Celsius for 4 days (col1a1 and col3a1) or 10 days (MMP9). Developed slides were then counterstained with toluidine blue. Photographs were taken of the developed slides in bright field and dark field using identical settings for all sections of each probe. Dark field signals, converted to a pseudocolor red using Photoshop software (Adobe, San Jose, CA), were overlaid on the bright field images.

Immunohistochemistry (IHC)

For protein localization with IHC, five micron sections were cut from the same blocks used for in situ hybridization. Sections were deparaffinized and rehydrated. For type I and type III collagen, antigen retrieval was performed in a 0.4% pepsin solution for 30 minutes at 37°C. Antigen retrieval for MMP9 was performed using a 0.01 M citrate buffer with a pH of 6.0 in a pressure chamber for 10 minutes at 95°C. The following primary antibodies and dilutions were used: goat polyclonal anti-type I collagen (Santa Cruz Biotechnology, Santa Cruz, CA; 1:50 dilution), rabbit polyclonal anti-type III collagen (GeneTex Inc., San Antonio, TX; 1:30), and polyclonal sheep anti-human MMP9 (Biogenesis Inc., Kingston, NH; 1:100). Slides were developed using biotinylated secondary antibodies, streptavidin-HRP, and Romulin AEC chromagen (Biocare Medical, Concord, CA) and then counterstained with hematoxylin.

Statistical Analysis

Data analysis included a 2-way analysis of variance with Bonferroni post-hoc multiple comparisons (α=0.05) and nonlinear regression were performed using Prism GraphPad 5.0a statistical software.

RESULTS

Non-Destructive MTP Joint Flexion Testing

In unoperated control tendons there was no significant difference between WT and Smad3^−/−mice in MTP joint range of motion (Fig. 1A). There was also no difference between WT and Smad3^−/−mice in gliding coefficient at day 0 (Fig. 1B). However on post-operative day 14, the MTP joint range of motion was significantly lower for WT (19.9° ± 14.4°) compared to Smad3^−/− at the same time point (40.7° ± 10.9°, p<0.01). The gliding coefficients for WT (511.0 ± 442.8) and Smad3^−/− (20.2 ± 10.4) were also significantly different on post-operative day 14 (p<0.001). The range of motion and gliding coefficient on day 21 for both WT (20.2 ± 17.33 and 248.3 ± 214.4, respectively) and Smad3^−/− (42.5 ± 14.1 and 16.8 ± 8.3) were not significantly different from day 14. However, the differences between WT and Smad3^−/−on post-operative day 21 in range of motion and gliding coefficient were still significant (p<0.001 and p<0.05, respectively; Fig. 1A and B).

Figure 1.

Unoperated control tendons and 14 or 21 day post-repair tendons from WT and Smad3^−/− were evaluated for (A) metatarsalphalangeal (MTP) joint flexion range of motion (ROM) at the maximum applied load of 19 grams, (B) Gliding Coefficient, (C) Maximum tensile force, and (D) tensile stiffeness. Results are shown as mean ± SD. *, **, and *** indicate significant differences of p<0.05, p<0.01, and p<0.001, respectively, between WT and Smad3^−/−.

Biomechanical Testing

The mean maximum tensile force in Smad3^−/− unoperated tendons was 61% (p<0.001) of unoperated WT tendons (7.0 ± 1.1 N and 11.4 ± 2.0 N, respectively; Fig. 1C). On post-operative day 14, both WT and Smad3^−/− repaired tendons required much lower breaking tensile forces than the unoperated tendons, however there was no difference between the two groups. On day 14, WT tendon repairs regained 10.5% of the strength of unoperated WT controls, while Smad3^−/− repaired tendons regained 13.9% of the strength of unoperated Smad3^−/− tendons. There were no significant improvements in repair maximum tensile force between days 14 and day 21 for either genotype. On day 21, the breaking tensile force of Smad3^−/− tendon repairs (0.93 N ± 0.28 N) was 42% of the breaking tensile force of WT tendon repairs (1.6 N ± 0.79 N) (p<0.05; Fig. 1C). However, this difference was not significant when normalized by the basal maximum tensile force of the corresponding unoperated controls (14.1% of control for WT and 13.3% of control for Smad3^−/−, respectively).

There were no significant differences in basal stiffness between WT and Smad3^−/− tendons within unoperated groups or either post-operative time point. The mean stiffness of unoperated WT control tendons was 6.5 ± 1.94 N/mm compared to 5.34 ± 1.02 N/mm for unoperated Smad3^−/− control tendons (Fig. 1D). Furthermore, there were no significant improvements in repair tensile stiffness between days 14 and day 21 for either genotype. On day 14, the mean stiffness for WT repair tendons was 1.2 ± 0.52 N/mm compared to 1.09 ± 0.56 N/mm for Smad3^−/− repairs. On day 21, the mean stiffness for WT repair tendons was 1.77 ± 0.70 N/mm compared to 0.97 ± 0.25 N/mm for Smad3^−/− repair tendons.

Histological Analysis of Healing

On post-operative day 7, minimal fibroblastic granulation accumulation or tendon reorganization was observed in WT or Smad3^−/− mice (Fig. 2). By day 14, WT mice had abundant granulation tissue and disorganized collagen fibers present at the repair site (Fig. 2), while a lesser amount of granulation tissue was observed in Smad3^−/− mice. WT tendon histology specimens at 21 days post-repair showed additional granulation tissue and further disorganization of collagen fibers. Smad3^−/− tendon repairs at day 21 continued to have less granulation tissue and collagen remodeling at the repair site, and more native tendon remained present.

Figure 2.

Representative histologic sections of WT and Smad3^−/− unoperated flexor digitorum longus (FDL) tendon and FDL repair tendons on post-operative days 7, 14, and 21. Tendon ends (T) and sutures (↑) are indicated. Sections were stained with alcian blue hematoxylin/orange G eosin and representative images are shown at 100x magnification.

Temporal Expression of Genes During Tendon Repair

Since histological analysis suggested decreased granulation tissue and scarring in Smad3^−/− tendon repairs, the expression of a panel of repair and remodeling genes was explored using real-time PCR. In the initial phase of tendon healing, within the first 3 days following injury and repair, col3a1 gene expression levels in Smad3^−/−mice were not statistically different compared to WT specimens (Fig. 3A). By day 7, Smad3^−/− col3a1 gene expression was five-fold less than WT gene expression (p<0.001). Col3a1 gene expression peaked in WT specimens at day 10 and expression was 2.5 times less in Smad3^−/−at this time point (p<0.001). On day 14 in WT tendons, col3a1 gene expression decreased, but was still nearly two-fold higher than gene expression in Smad3^−/− repairs (p<0.001).

Figure 3.

Gene expression of (A) col3a1, (B) col1a1, and (C) MMP9 in WT and Smad3^−/− FDL tendon repair tissue on post-operative days 3, 7, 10 and 14. Total RNA was extracted and from five pooled FDL repair tendons and processed for real-time RT-PCR. Expression was standardized with the internal beta-actin control. Data is presented as the mean fold induction (over WT post-operative day 3 repairs) ± SD. *, **, and *** indicate significant differences of p<0.05, p<0.01, and p<0.001, respectively, between WT and Smad3^−/−.

WT col1a1 gene expression progressively increased over the 14 day time course and was significantly greater than that observed in the tendon repairs in Smad3 mice at each time point. (Fig. 3B). On post-operative day 3, col1a1 gene expression in Smad3^−/− was three times less than expression in WT (p<0.05). By day 7, 10, and 14, Smad3^−/−col1a1 expression was 4.5, 3.3, and 4.3 times less, respectively than in WT (p<0.001).

While collagen gene expression during tendon repair was greater in WT mice compared to Smad3^−/−mice, the expression of MM9, a gene involved in early tendon remodeling during repair, was greater in Smad3^−/− repair tendons (Fig. 3C). On day 7 following tendon repair, MMP9 expression was 1.7 times greater in Smad3^−/− repair tendons than in WT tendon repairs (p<0.01). On day 10, MMMP9 expression peaked for both Smad3^−/−and WT specimens. At this time, MMP9 expression was nearly 2-fold greater for Smad3^−/− than for WT samples (p<0.01). While MMP9 expression declined at 14 days post repair for both Smad3^−/− and WT repair samples, MMP9 expression remained approximately 2-fold higher in Smad3^−/− tendon repairs compared to WT tendon repairs at the same time point (p<0.001).

Additional gene expression experiments were performed in order to determine if any compensatory changes occurred in TGF-β/BMP signaling as a result of the loss of expression of Smad3. Smad2 and BMP2, 4, 6 and 7 gene expressions were investigated (Figure 4). Rather than a compensatory increase, gene deletion of Smad3 resulted in an overall reduction in the expression of Smad2 and the BMPs. Smad2 gene expression was approximately 4.5 fold greater in WT compared to Smad3^−/− tendon repairs at day 3 (p<0.05; Fig.4E). Smad2 gene expression peaked in WT tendon repair samples on day 7 when gene expression in WT specimens was approximately 4 fold greater than in Smad3^−/− specimens (p<0.001). On days 10 and 14 Smad2 gene expression fell to baseline levels in the WT repair tendons and there was no significant differences in Smad2 gene expression between WT and Smad3^−/− tendon repairs.

Figure 4.

Gene expression of (A) BMP2, (B) BMP4, (C) BMP6, (D) BMP7 and (E) Smad2 in WT and Smad3^−/− FDL tendon repair tissue on post-operative days 3, 7, 10 and 14. Total RNA was extracted and from five pooled FDL repair tendons and processed for real-time RT-PCR. Expression was standardized with the internal beta-actin control. Data is presented as the mean fold induction (over WT post-operative day 3 repairs) ± SD. *, **, and *** indicate significant differences of p<0.05, p<0.01, and p<0.001, respectively, between WT and Smad3^−/−.

On day 3, BMP2 gene expression was nearly 2-fold greater in WT than in Smad3^−/− repair tendon samples (p<0.01). BMP2 gene expression peaked for both WT and Smad3^−/− repair tendons on day 7 when BMP2 gene expression was approximately 1.5 fold greater in WT than in Smad3^−/− repair tendons (p<0.001). BMP2 gene expression was not significantly different between WT and Smad3^−/− samples on days 10 or 14 (Fig. 4A). For BMP4 gene expression there was no significant difference between WT and Smad3^−/− repair tendons at any time point (Fig. 4B). BMP6 gene expression on day 3 was approximately 25 times greater in WT samples than in Smad3^−/− samples (p<0.001; Fig. 4C). There was no difference in BMP6 gene expression between WT and Smad3^−/− samples on day 7. On day 10, BMP6 gene expression peaked for both WT and Smad3^−/− tendon repair samples. Gene expression was approximately 3 fold greater in WT samples than in Smad3^−/− samples at that time (p<0.001). BMP6 gene expression was approximately 5.4 fold greater in WT samples than in Smad3^−/− samples on day 14 (p<0.01). BMP7 gene expression on day 3 was 125 fold higher in WT repair samples than in Smad3^−/− samples (p<0.001; Fig. 4D). On day 7, BMP7 gene expression was approximately 16 fold greater in the WT than in Smad3^−/− (p<0.05). On days 10 and 14 there were no significant differences in BMP7 gene expression between WT and Smad3^−/− specimens.

Spatial Localization of col3a1, col1a1 and Mmp9 mRNA During Tendon Repair

Various studies have investigated quantitative gene expression in healing tendon (24; 25); however, less is known about spatial localization of gene expression during flexor tendon healing. On post-operative day 3, col3a1 expression could not be detected in the tendon repairs from WT and Smad3^−/− mice (Fig. 5A). By day 7, WT repairs had high levels of col3a1 gene expression that localized to the proliferative fibroblastic granulation tissue in and around the injury site, while only minimal expression was observed in the repair sites in Smad3^−/− samples. No col3a1 expression was observed in the actual tendon tissue in either Smad3^−/− or WT mice. Col3a1 expression was maximal in WT tendons on post-repair day 10 and remained highly elevated at day 14. In contrast, Col3a1 gene expression in Smad3^−/− tendon repairs also had maximal col3a1 expression at day 10, but the levels of expression were much lower than observed in the WT tendon repairs. Similar to col3a1, col1a1 was expressed in granulation tissue surrounding the injury site and was absent in the native tendon tissues. Overall, col1a1expression levels mimicked those of col3a1 in that WT samples exhibited much greater expression at each time point compared to Smad3^−/− samples (Fig. 5B).

Figure 5.

In-situ hybridization of (A) col3a1, (B) col1a1, and (C) MMP9 in FDL repair tendons on post-operative days 3, 7, 10, and 14. Area of the tendon is outlined and tendon ends (T) are indicated. Representative photographs taken at 50x original magnification are shown.

MMP9 is necessary for tissue catabolism and remodeling during the healing process. On post-repair days 3 and 7 only trace areas of MMP9 gene expression were detectable in WT and Smad3^−/− tendon repair sites. At post-repair day 10, small areas of MMP9 gene expression were present in and around the healing tendons of WT mice. Conversely, in Smad3^−/− mice on day 10, MMP9 was strongly detected in the damaged tendon ends and the surrounding callus (Fig. 5C). On post-repair day 14, MMP9 expression was not detected in WT samples and was reduced but still present in Smad3^−/− tendon repairs.

Spatial Localization of Matrix and Remodeling Proteins During Tendon Repair

Immunohistochemistry was performed to assess whether the changes in RNA expression corresponded with changes in protein expression in Smad3^−/− mice. Consistent with the RNA data, type III collagen immunostaining in Smad3^−/− tendon repairs was decreased compared to WT specimens at all time points and immunostaining peaked for both WT and Smad3^−/− on post-operative day 10 (Fig. 6A). Similarly, type I collagen immunostaining was decreased in Smad3^−/− tendon repairs compared to WT tendon repairs, with maximal expression occurring in both groups 14 days post-repair. In contrast, MMP9 immunostaining was increased in Smad3^−/− compared to WT tendon repairs at all time points (Fig. 6C). Immunostaining was greatest for MMP9 on post-operative day 10 for both WT and Smad3^−/− tendons, consistent with the gene expression data (See Figures 3 and 5).

Figure 6.

Immunohistochemistry of (A) Type III collagen, (B) Type I collagen, and (C) MMP9 in FDL repair tendons on post-operative days 3, 7, 10, and 14. Tendon ends (T), granulation tissue (G) and sutures (↑) are indicated in representative photomicrographs taken at 200x original magnification.

DISCUSSION

TGF-β is associated with three general biological functions. First, it is critical to cell growth (17). Second, it is a powerful immunomodulator (18) and third, it is critical in the formation of extracellular matrix (19). TGF-β has also been shown to regulate extracellular matrix degradation and reorganization by modulating matrix metalloproteinase activity (26). Prior studies have established a potential role for TGF-β signaling in tendon repair by showing that inhibition of TGF-β1 can lead to decreased adhesions following tendon repair (11; 12). We examined the hypothesis that TGF-β signaling, through the transcription factor Smad3, is involved in the regulation of collagen formation and matrix metalloproteinase activity and tissue remodeling during tendon healing. Thus, this paper provides important new information regarding the specific role of the canonical TGF-β signaling pathway in the process of adhesion formation.

In this study, smad3 deficiency during flexor tendon healing led to several key findings. Non-destructive MTP joint flexion testing showed that Smad3^−/− mice had approximately twice the range of motion of WT mice at both 14 and 21 days post-repair. Additionally, the gliding coefficient, a measure of the resistance to gliding due to scar and adhesion formation, was over 12 times greater for WT mice compared to Smad3^−/− mice on day 14 and over 10 times greater at day 21. Together, the range of motion and gliding coefficient provided strong data indicating decreased adhesion formation in flexor tendon repair in Smad3^−/− mice when compared to WT littermates.

Biomechanical testing data showed that the basal tensile strength (maximum tensile force) of unoperated Smad3^−/− tendons was significantly (39%) lower than unoperated WT tendon strength. On day 14, Smad3^−/− and WT repair tendons had similar maximal tensile force. While on day 21, the maximal tensile force was reduced by 42% in Smad3^−/− tendons compared to WT tendon repairs. However, at both 14 and 21 day time points, there was no significant difference between WT and Smad3^−/− tendon repair strength, when normalized to the basal unoperated tendon strength for each genotype. This suggests that the decreased absolute strength of unoperated and day 21 Smad3^−/− healing tendons may be due to developmental differences in the structure and organization of the collagen matrix due to Smad3 deficiency rather than a decrease in adhesion formation. This is supported by prior work showing that GDF5 results in reduced adhesions without a reduction in tensile strength (22). Thus, additional work is required to further understand the role of TGF-β/Smad3 in tendon structure, organization, and strength.

During early tendon healing, epitenon-derived cells have been shown to form granulation tissue and collagens which contribute to the "healing callus" (22). Histological analysis of the tendon repair sites in WT and Smad3^−/− mice suggested structural differences in these "healing calluses," with the repair sites of Smad3^−/− mice forming less granulation tissue and ultimately smaller calluses. TGF–β1 plays an important yet not fully understood role in this proliferative phase of tendon healing and during this early tendon healing period the granulation tissue and early scarring that develops at the repair site are characterized by type III collagen fibers (27)(28). In the tendon repairs of Smad3^−/− mice both quantitative PCR and in situ showed marked reduction in col3a1 gene expression at the repair site, while immunohistochemistry analysis showed reduced protein expression in Smad3^−/− mice. This suggests that Smad3 stimulates anabolic tendon healing which could explain the reduction in adhesion formation and repair strength seen in Smad3^−/− mice.

The proliferative phase of tendon healing is followed by a period of remodeling or maturation. During this stage of early remodeling, type III collagen fibers transition into stronger and stiffer type I collagen fibers. In this study, Smad3^−/− mice had reduced col1a1 gene expression. In situ hybridization localized the expression of col1a1 and col3a1 to the "callus" or reparative tissues adjacent to the native tendon tissue, and immunohistochemistry confirmed a parallel decrease in protein expression. This suggests a decrease in matrix deposition during the remodeling phase of tendon repair and an associated between decreased type I collagen production and reduced adhesions that may account for the reduction in maximal tensile strength observed in the healing tendons at day 21 in Smad3^−/− mice.

Matrix metalloproteinases are a group of zinc dependent endopeptides capable of degrading the extracellular matrix (29). MMP activity is critical during the remodeling phase of tendon repair (25). MMP9 is a gelatinase and is predominantly involved in breaking down damaged collagen (30; 31). Tendon repairs in Smad3^−/− mice had a significant increase in MMP9 expression, with gene and protein expressions localized to the areas of granulation tissue and tendon ends adjacent to the injury. The increased MMP9 production throughout early tendon repair further suggests that the reduction in adhesion formation seen with Smad3 deficiency may be due to enhanced remodeling of the scar tissue.

Smad3 is essential for normal TGF-β signaling and while its deficiency is clearly associated with the reduction in matrix deposition, our findings also showed that Smad3 deletion resulted in further perturbation of the TGF-β and the BMP signaling pathways. Tendon healing in Smad3 mice was associated with reduction in Smad2 gene expression as well as in the gene expressions of BMP2, BMP6, and BMP7. Similar to TGF-β, the BMPs stimulate the expression of genes and signals that accelerate tissue repair (20-22). Thus, rather than a compensatory increase in other pathways or signals involved in the reparative process, the deletion of Smad3 resulted in further reduction in anabolic signals and pathways. These findings confirm a central regulatory role for TGF-β/Smad3 during tendon repair.

The tendon repair model used in these studies is designed to evaluate intrasynovial tendon healing and the model is not an exact representation or recapitulation of the anatomy of the zone II human flexor tendon. However, despite this deficiency, the model provides insights regarding the cellular and the molecular events involved in healing and adhesion formation. The method includes a release of the tendon at the muscle tendon junction in the leg. This release prevents the generation of active muscle forces across the repair site during the early phase of repair. However, passive motion of the toes, foot, and ankle joints is unrestricted so the tendon undergoes passive motion and in this manner also mimics the clinical scenario whereby initially resistance exercises are limited and passive motion is instituted following tendon repair (32).

The current study provides novel insights regarding the role of TGF-β/Smad3 in tendon repair and adhesions. Smad3 regulates matrix deposition and remodeling and, along with TGF-β it is a potentially important molecular target for the regulation of scar and adhesion formation. This work also further establishes the complexity of the tendon healing process and the challenge of recapitulating normal tendon structure and function following the injury and repair process. Further studies are required to define and further clarify the role of TGF-b/Smad3 and the part played by other genetic factors, environmental influences, and the aging process.