Smad3 binds scleraxis and mohawk and regulates tendon matrix organization

Smad3 Binds Scleraxis and Mohawk and Regulates Tendon MatrixOrganization

Ellora Berthet,1 Carol Chen,1 Kristin Butcher,1 Richard A. Schneider,1,2 Tamara Alliston,1,2,3 Mohana Amirtharajah1

1Department of Orthopaedic Surgery, University of California, San Francisco, 1500 Owens St., San Francisco, CA, 94158, 2Eli and Edythe BroadCenter of Regeneration Medicine and Stem Cell Research, University of California, San Francisco, San Francisco, CA, 3Department ofBioengineering and Therapeutic Sciences, University of California, San Francisco, San Francisco, CA

Received 8 August 2012; accepted 10 April 2013

Published online 7 May 2013 in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/jor.22382

ABSTRACT: TGFb plays a critical role in tendon formation and healing. While its downstream effector Smad3 has been implicated inthe healing process, little is known about the role of Smad3 in normal tendon development or tenocyte gene expression. Using micedeficient in Smad3 (Smad3�/�), we show that Smad3 ablation disrupts tendon architecture and has a dramatic impact on normal geneand protein expression during development as well as in mature tendon. In developing and adult tendon, loss of Smad3 results inreduced protein expression of the matrix components Collagen 1 and Tenascin-C. Additionally, when compared to wild type, tendonfrom adult Smad3�/� mice shows a down regulation of key tendon marker genes. Finally, we have established that Smad3 has theability to physically interact with the critical transcriptional regulators Scleraxis and Mohawk. Together these results indicate acentral role for Smad3 in normal tendon formation and in the maintenance of mature tendon. � 2013 Orthopaedic Research Society.Published by Wiley Periodicals, Inc. J Orthop Res 31:1475–1483, 2013.

Keywords: Tendon; Smad3; Scleraxis; Mohawk; TGFb

Flexor tendon injuries cause significant morbidity in theworking-age population with patients rarely regainingtheir pre-injury range of motion. Surgical repair oftendon injuries initially results in the formation of scartissue rather than recreating true tendon tissue, leavingpatients prone to both re-rupture and excessive scarformation that can impede motion. Elucidation of thepathways involved in normal tendon development maylead to the identification of factors that enhance surgicalreconstruction of damaged tendons and allow us torecreate the normal biomechanical properties of tendon.Unfortunately, relatively little is known about themechanisms directing the development of embryonictendon progenitors into mature tendons. In addition,much remains to be learned about the factors thatcontrol tenocyte gene expression and maintain theunique matrix material properties of tendon.

The transforming growth factor-b (TGFb) family ofgrowth factors plays an essential role in tendon biolo-gy.1 TGFb has been implicated in tendon developmentas well as in the pathogenesis of tendon injury andhealing.2 Prior studies have revealed that simulta-neous ablation of both TGFb2 and TGFb3 signaling inmouse embryos prevents development of almost alltendons and ligaments.3 Similar results were observedfollowing inactivation of the TGFb type II receptor.3

TGFb also induces key markers of tenocyte differentia-tion including Scleraxis (Scx) and type I Collagen invivo.3,4 In addition, TGFb regulates the expression ofmatrix metalloproteinases, which are essential for

tendon function and maturation.4–7 Other studies haverevealed a key role for myostatin (GDF-8), anothermember of the TGFb family, in tendon development inmice.8 Finally, in a rabbit model of flexor tendoninjury, TGFb was found to be upregulated at the site ofinjury, implicating the growth factor in the processesof injury–repair and scar formation.2 Similar effectswere found in a mouse model of tendon repair.9

While the importance of TGFb signaling duringtendon development and in tendon injury and repair isapparent, the precise pathway by which TGFb exertsits effects remains unknown. TGFb acts by binding toa complex of type I and type II transmembrane serine/threonine kinases. The receptor/ligand complex thenphosphorylates several intracellular effectors that reg-ulate cell differentiation. Although both TGFb andmyostatin induce the phosphorylation of commondownstream effectors, including Smad2, Smad3, Erk1/2, and p38 MAP kinases, which of these effectors playa role in the observed effects of TGFb signaling intendon remains to be determined.

In other tissues derived from the mesenchymallineage, TGFb acts via Smad3 to regulate the functionof key lineage-specific transcription factors. For exam-ple, in bone, Smad3 represses the transcriptionalactivity of the osteogenic transcription factor Runx2,which in turn regulates osteoblast differentiation andbone matrix material properties.9 In cartilage, Smad3modulates the activity of the critical chondrogenictranscription factor Sox9, while in muscle it regulatesthe myogenic transcription factor MyoD.10,11 In thisway, TGFb-activated Smad3 is a major regulator of celldifferentiation in tissues of the musculoskeletal systemincluding bone, cartilage, and muscle.9–12 Smad3 mayplay a similar role in the TGFb-dependent regulation oftenocyte differentiation.

TGFb-activated Smad3 has already been shown toinduce mRNA expression of the tenocyte lineage-specific

Grant sponsor: NIH; Grant numbers: NIDCR R01 DE019284 TA,NIDCR R01 DE016402 RS, NIDCR COHORT T32 CC;Grant sponsor: American Society for Surgery of the Hand;Grant sponsor: University of California.Correspondence to: Mohana Amirtharajah (T: 415-514-6065;F: 415-514-6144; E-mail: [email protected])

# 2013 Orthopaedic Research Society. Published by Wiley Periodicals, Inc.

JOURNAL OF ORTHOPAEDIC RESEARCH SEPTEMBER 2013 1475

transcription factor Scx in avian mesenchymal cul-tures.13 However, the ability of Smad3 to regulate Scxactivity and thereby control tendon cell differentiationremains unknown. A second tenocyte lineage-specifictranscription factor, Mohawk (Mkx), has recently beenidentified and found to be necessary for the normaldevelopment of tendons during embryogenesis as wellas for tendon maturation.14 The ability of Smad3 toregulate Mkx expression has not yet been determined.

Smad3 signaling has been shown to play a role intendon healing as well as in the maintenance oftendon mechanical properties.15 Although deletion ofSmad3 decreases adhesion formation in repaired ten-dons, tendons in these same Smad3-deficient miceexhibit diminished tensile strength and abnormalcollagen deposition following injury.15 The mecha-nisms by which Smad3 asserts these effects in normaland injured tendon remain unknown.

Given that Smad3 plays a key role in the differenti-ation of other musculoskeletal tissues as well as intendon healing, we hypothesized that Smad3 is a keyregulator of tendon formation and of tenocyte geneexpression. We used a combination of in vitro and invivo approaches to determine the expression andfunction of Smad3 during embryonic development andthe effect of Smad3 ablation on extracellular matrixproduction and tenocyte gene expression. Further-more, we investigated the ability of Smad3 to bind thetendon transcription factors Scx and Mkx.

METHODSMiceAll animal procedures were performed in accordance withUniversity of California, San Francisco Institutional Careand Use Committee-approved protocols. Smad3-deficient(Smad3�/�) mice were obtained from Dr. X.-F. Wang andwere maintained on a C57Bl/6 background along with theircorresponding wild type controls.16 SBE-Luciferase miceexpress a luciferase reporter gene under control of 12 copiesof a Smad2/3 binding sequence from the Smad7 promoterand were maintained on an FVB/n background.17 Maturetendon was harvested from 5- to 8-week-old male mice, andembryonic tissues were analyzed at embryonic day (E) 16.5.

Histology, Immunohistochemistry, and In Situ HybridizationAchilles and flexor digitorum profundus (FDP) tendons wereharvested from mature male mice. Embryos were harvested atE16.5 and forelimbs were disarticulated for further processing.Harvested tissues were fixed in 4% paraformaldehyde at 4˚Covernight and prepared for sectioning. Achilles tendons weredehydrated and embedded in paraffin while FDP tendons wereembedded in OCT for frozen sectioning. E16.5 forelimbs wereembedded in paraffin as well as OCT. FDP tendons were usedfor immunohistochemistry experiments because their morphol-ogy was better maintained during frozen sectioning, whileAchilles tendons were used for paraffin sectioning.

To analyze tissue morphology, longitudinal sections ofFDP tendon were stained with hematoxylin and eosin.Frozen sections of FDP tendons and E16.5 forelimbs wereanalyzed by immunohistochemistry using antibodies forSmad3 (Abcam, ab63577), Collagen 1 (Abcam, ab292), Tenas-

cin-C (gift of H. Erickson, Duke University) and Luciferase(Abcam, ab21176). Sections were rinsed, dehydrated, andmounted using standard procedures.

In situ hybridization was performed on paraffin sectionsof adult Achilles tendon and E16.5 forelimbs as described.1835S-labeled antisense riboprobes were generated to mouseCol1a1 (gift of E. Vuorio, University of Finland), Mkx (gift ofR. Jiang, Cincinnati Children’s Hospital Medical Center),and Scx and Tenomodulin (Tnmd) (gifts of R. Schweitzer,Shriner’s Hospital for Children).19–21 Sections were counter-stained with Hoechst nuclear dye (Sigma-Aldrich, St. Louis,MO). Hybridization signals were detected using darkfieldillumination and the nuclear stain detected with epifluores-cence. All figures show representative images of N � 3 mice.

Second Harmonic Generation Imaging and DirectionalityAnalysisWe used a modified “M” microscope (FV1000MPE; Olympus,Center Valley, PA) with an Olympus 20� NA 1.0 waterimmersion objective and custom detection instrumentationbased on designs published by Bullen et al.22 Samples wereexcited with a 10 W MaiTai Ti sapphire laser with awavelength of 830 nm. Second harmonic generation wasdetected using a 495-nm LP dichroic and 455–485-nmbandpass emission filters. Images were collected using aC7950 PMT module (Hamamatsu Photonics, HamamatsuCity, Japan) and using standardized PMT and laser settings.Comparable regions of interest were then analyzed fordirectionality in Fiji image processing software using Fouriercomponents analysis and the dispersion value was obtainedfor N ¼ 3 WT and Smad3�/� tendons. Statistical significancewas determined using an unpaired student’s t-test withp � 0.05 considered significant.

Quantitative Reverse Transcriptase-PCR (qPCR)For analysis of gene expression, tail, flexor, and tibialisanterior tendons were isolated from mature male mice. Thetendon sheath and surrounding soft tissue were carefullyremoved with the aid of loupe magnification. The tendonswere snap frozen and stored in liquid N2. Frozen tendonswere placed into Trizol and homogenized with a douncehomogenizer. After homogenization, RNA was extractedusing the Purelink RNA Mini Kit (Invitrogen, Carlsbad, CA).cDNA was reverse transcribed from the RNA using theiScript cDNA synthesis kit (Bio-Rad, Hercules, CA). Tran-scripts were amplified using the following primers: Smad3(Fwd: 50-ACCAAGTGCATTACCATCC-30; Rev: 50-CAGTAGA-TAACGTGAGGGAGCCC -30), Col1a1 (Fwd: 50-GCATGGC-CAAGAAGACATCC-30; Rev: 50-CCTCGGGTTTCCACGTCTC-30),Col1a2 (Fwd: 50-GTCCCCGAGGCAGAGAT-30; Rev: 50-CCTTTGTCAGAATACTGAGCAGC-30), Tnc (Fwd: 50-AGGC-GATCCCAGCCAGTCAGT-30; Rev: 50-ATGGACGGGGCACC-TCCTGTC-30), Tnmd (Fwd: 50-TGTACTGGATCAATCCCAC-TCT-30; Rev: 50-GCTCATTCTGGTCAATCCCCT-30), Scx (Fwd:50-CCTTCTGCCTCAGCAACCAG-30; Rev: 50-GGTCCAAAGT-GGGGCTCTCCGTGAC-30), and Mkx (Fwd: 50-GACTCCGAG-GCTCTGCCGCAA-30; Rev: 50-CAGGAGTCGCCATCGCTG-CTCA-30). Results were detected based on amplicon bindingof SYBR Green (BioRad) using the CFX96 Real-Time PCRDetection System (BioRad).

Cell Culture and ImmunoprecipitationC3TH10T1/2 cells (ATCC) were cultured and maintained inBasal Medium Eagle supplemented with 10% FBS and 2 mM

1476 BERTHET ET AL.

JOURNAL OF ORTHOPAEDIC RESEARCH SEPTEMBER 2013

L-Glutamine at 37˚C and 5% CO2. Cells were transfected usinglipofectamine with expression vectors for Scx, Mkx, and/orSmad3. Smad3 tagged with either Flag or HA epitopes at theN-terminus were cloned into a pRK5 expression vector. Scxtagged with both HA and Flag tags at the N-terminus wascloned into a pRK5 expression vector. The Mkx–Myc expres-sion vector was a gift of A. Rawls, Arizona State University.23

Cells grew for 18 h following transfection before treatmentwith TGFb (5 ng/ml) for 3 h. Cells were rinsed with PBS andlysed in 50 mM Tris pH 7.5, 150 mM NaCl, and 0.5% TritonX-100 for 15 min at 4˚C. Lysates were collected and brieflysonicated before centrifugation. Complexes were immunopreci-pitated by Flag-M2 affinity gel (Sigma, A2220) before analysisby Western blot. In endogenous co-immunoprecipitationexperiments, Achilles, flexor, and tail tendons were collectedfrom 4-week-old wild type 129/J mice, homogenized in lysisbuffer, and briefly sonicated. Slightly less mature tendon wasselected for immunoprecipitation experiments than for qPCRand immunohistochemistry experiments in order to facilitatetissue homogenization. Complexes were immunoprecipitatedwith Smad3 antibody (Abcam ab28379) before analysis byWestern blot for Scx (Abcam ab58655) and Smad3 (Santa Cruzsc-6202) expression. Images are representative of three inde-pendent experiments.

Statistical Analysis of qPCR DataFold change in gene expression was calculated using the DDCt

method based on the average of technical duplicates.24 Allsamples were normalized to the average delta Ct of the wildtypes in their respective litters. Given the apparent largeramount of variability in our wild type animals versus ourSmad3�/� animals, statistical significance was determinedusing a two-tailed student’s t-test assuming unequal variances.Results with p � 0.05 were considered statistically significant.

RESULTSSmad3 Is Required for Normal Tendon ArchitectureFormationSmad3 has been implicated as a critical downstreameffector of TGFb signaling in tendon development, andpreviously published work has demonstrated reducedtensile strength in tendon from Smad3�/� mice relativeto wild type, implicating Smad3 in the establishment ofnormal tendon mechanical properties.3,15 Given thesefindings, we assessed the requirement for Smad3 innormal tendon formation using an established Smad3�/�

mouse model. Histologic analysis revealed that whiletendons from wild type mice displayed normal matrixorganization (Fig. 1A), tendons from Smad3�/� micedisplayed disorganization of the collagen fibers withcrimping of the fibrils and increased gap formation(Fig. 1B). Furthermore, second harmonic generationimaging of wild type and Smad3�/� tendons highlight-ed the severity of the disruption of collagen fiberalignment in the absence of Smad3 (Fig. 1C and D). Inorder to quantitatively assess the relative organizationof the collagen fibers within these tendons, we ana-lyzed the directionality of the fibers using the image-processing program Fiji and compared the degree ofdispersion. The degree of dispersion is the standarddeviation of the distribution of fiber orientation angles.While the wild type tendon showed a very high degree

of collagen fiber organization with an average disper-sion value of 4.1˚, the Smad3�/� tendons revealed asignificantly larger average dispersion value of 6.2˚(p ¼ 0.03) reflecting their greater collagen fiber disor-ganization (Fig. 1E). This finding confirms that in theabsence of Smad3, the normal organization of collagenfibers in tendon extracellular matrix is disrupted.Together, these results indicate that Smad3 is essen-tial for the formation of normal tendon architecture.

Smad3 Is Required for Normal Gene and ProteinExpression in Adult TendonWe next examined whether the fibrillar disorganiza-tion observed in Smad3�/� tendons was associatedwith abnormal expression of tendon extracellular

Figure 1. Loss of Smad3 disrupts normal tendon architecture.H&E staining of flexor digitorum profundus (FDP) tendon from8-week-old wild type (WT) (A) and Smad3�/� (B) mice. Matrixorganization was assessed by second harmonic generation mi-croscopy of WT (C) and Smad3�/� (D) tendon followed bydirectionality analysis and comparison of the average dispersionangle (E). Images are representative of N � 3 WT and Smad3�/�

littermates (C57Bl/6 background). Scale bar ¼ 20 mm. �Indicatessignificance of p � 0.05.

SMAD3 BINDS SCLERAXIS AND MOHAWK 1477


matrix proteins, such as Collagen 1 (Col1) and Tenas-cin-C (Tnc). Immunohistochemistry of longitudinalsections of FDP tendons revealed significant reduc-tions in the expression of Col1 and Tnc in Smad3�/�

tendons relative to wild type tendons (Fig. 2A),suggesting that Smad3 regulates tendon matrix pro-tein expression.

In order to determine whether this decrease inprotein expression was accompanied by a correspond-ing reduction in gene expression, in situ hybridizationanalyses were performed on adult Achilles tendon,revealing reduced expression of genes encoding thepro-alpha1(I) chain of type 1 collagen (Col1a1) and thecollagen packaging regulator Tenomodulin (Tnmd)(Fig. 2B). In addition, while we could not detect adifference in Scx expression, we did observe a decreasein expression of the tendon transcription factor Mkx inSmad3�/� tendon relative to wild type (Fig. 2B). Inorder to quantify this reduction, we examined geneexpression by qPCR. The expression of Col1a1 inSmad3�/� mice was significantly reduced to 10% ofwild type (p � 0.001), and Col1a2 to 11% of wild type(p � 0.001) (Fig. 3). The expression of Tnc was reducedto 68% of wildtype (p ¼ 0.17) and Tnmd to 23%(p ¼ 0.058), but neither reduction reached statisticalsignificance. Our analysis of tendon-specific transcrip-tion factor expression revealed that Scx expressionwas reduced to 70% (p ¼ 0.070) and Mkx to 48% of

wildype (p ¼ 0.063). Although these differences appearlarge, they did not reach statistical significance, possi-bly due to the high variability in expression in thewild type tendon. The dramatically reduced expressionof genes encoding essential extracellular matrix com-ponents in the absence of Smad3 strongly suggeststhat Smad3 is required for the maintenance of adulttendon architecture.

Smad3 Is Required for Normal Gene and ProteinExpression During Tendon DevelopmentGiven the effects of Smad3 deletion in adult tendon,we investigated the role of Smad3 expression duringtendon development. First we assessed overall Smad3activity during embryonic development using a report-er mouse model. Embryos carrying a luciferase report-er gene driven by a Smad2/3 responsive element wereharvested at E16.5. This developmental stage waschosen because the flexor tendons are largely formedand identifiable at this point.22 Immunohistochemistryagainst luciferase revealed broad Smad2/3 activitythroughout the limb bud including in the FDP tendon(Fig. 4A).

To further assess the functional role of Smad3activity during tendon development, the expression ofkey tendon matrix proteins Col1 and Tnc was analyzedin Smad3�/� embryos. As expected, immunohistochem-istry of E16.5 forelimbs revealed an absence of Smad3

Figure 2. Loss of Smad3 results in reduced matrix protein and gene expression in mature tendon. A: Protein expression of Col1 andTnc in WT and Smad3�/� FDP tendons. The specificity of staining is indicated by the IgG control. Scale bar ¼ 100 mm. B: In situhybridization experiments show mRNA expression of Col1a1, Tnmd, Mkx, and Scx in Achilles tendon from adult WT and Smad3�/�

mice. Images are representative of N � 3 littermates (C57Bl/6 background).

1478 BERTHET ET AL.


expression in Smad3�/�mice relative to wild type (Fig.4B). In wild type mice, we observed broad expressionof Col1 and Tnc throughout the limb bud with enrich-ment in the tendon. In contrast, in Smad3�/� limbs,we observed a reduction in Col1 and Tnc expressionacross the limb as well as within the tendon (Fig. 4B).

Next, the expression of genes encoding key tendontranscription factors and extracellular matrix proteinsin the absence of Smad3 was analyzed qualitatively bysection in situ hybridization. These experimentsrevealed a subtle upregulation of mRNA encodingCol1a1 in Smad3�/� mice relative to wild type, incontrast to the reduced levels of Col1 protein expres-sion observed in Smad3�/� tendon (Fig. 4C). WhileTnmd expression appeared slightly elevated in theSmad3�/� tendon relative to wild type, the observeddifference is too subtle to be conclusive given thequalitative nature of the assay. Interestingly, mRNAexpression for the gene encoding the critical tendontranscription factor Scx was increased while expres-sion for the gene Mkx appeared to be decreased inSmad3�/� embryos (Fig. 4C).

Smad3 Binds the Tendon Transcription Factors Scleraxisand MohawkGiven the profound impact of Smad3 ablation on geneexpression in tendon, we sought to understand themechanism by which Smad3 exerts its effects. In othertissues of the musculoskeletal system, includingbone and cartilage, Smad3 has been shown to regulate

differentiation through direct interaction with tissue-specific transcription factors.12,25 We therefore hypoth-esized that Smad3 impacts the expression of keytendon marker genes in a similar manner by bindingto the tendon-specific transcriptional regulators Scxand Mkx. Co-immunoprecipitation experiments wereperformed to assess the capability of Smad3 to physi-cally interact with both Scx and Mkx.

In order to investigate the ability of Smad3 to bindScx in vitro, epitope-tagged versions of these proteinswere overexpressed in 10T1/2 cells. Given the chal-lenges of effectively growing primary tenocytes on thescale necessary for immunoprecipitation experiments,we elected to use an established mouse embryonicfibroblast line. Immunoprecipitation of lysates withanti-Flag antibodies resulted in the co-precipitation ofSmad3 with Flag-tagged Scx (Fig. 5A). The presence ofSmad3 following the immunoprecipitation of Scx indi-cates that these two proteins can physically interact.To determine if Smad3 was similarly capable ofbinding the tendon transcription factor Mkx, Myc-tagged Mkx was overexpressed in 10T1/2 cells alongwith Flag-tagged Smad3. Immunoprecipitation oflysates with anti-Flag antibodies and detection byWestern blot analysis revealed a similar interactionbetween Smad3 and Mkx (Fig. 5B). Neither of thesecomplexes appeared to be affected by the addition ofexogenous TGFb under these conditions (Fig. 5A, B);however, this does not preclude the possibility thatTGFb may regulate complex formation in vivo.

Figure 3. Expression of key tendon genes is dramatically reduced with the loss of Smad3. Gene expression analysis by qPCR ofSmad3, Col1a1, Col1a2, Tnc, Tnmd, Scx, and Mkx inWT and Smad3�/� tendon. Gene expression data for individual mice (WT: N ¼ 7,Smad3�/�: N ¼ 3) is shown alongside the mean for each genotype. �Indicates a significant difference of p � 0.001.



In order to investigate whether the same interac-tion between Smad3 and Scx occurs in vivo, further co-immunoprecipitation experiments were performed onwhole tendon lysates harvested from mature mice.Lysates were precipitated using an antibody againstSmad3, and a co-precipitating Scleraxis band wasobserved by Western blot analysis (Fig. 5C). Thisfinding confirms that Smad3 also binds Scx in vivo.

DISCUSSIONOur work has elucidated a critical role for Smad3 inthe formation of normal tendon architecture and inthe regulation of tenocyte marker genes. Loss ofSmad3 affected not only the expression of key tenocyteproteins, but also the organization of collagen fibrilswithin the extracellular matrix. The collagen fibers inmature tendons of Smad3�/� mice showed significantlyincreased disorganization when compared to wild type.Furthermore, immunohistochemistry revealed that ex-pression of the tendon matrix proteins Col1 and Tnc

are also severely diminished. A previously publishedstudy investigated the effect of Smad3 deletion onmatrix material properties of tendon and demonstrat-ed a significant reduction in the tensile strength ofSmad3�/� tendon relative to wild type.15 This reportedchange in tensile strength might be explained by theobserved abnormalities in collagen architecture andmatrix protein expression found in our current study.It should be noted that while the defects observed inSmad3�/� tendon are striking and the disruption ofmatrix protein expression significant, the phenotype isnot as severe as in the case of complete ablation ofTGFb signaling, suggesting that in tendon develop-ment Smad3 acts in concert with other downstreameffectors that as of yet remain unknown.

Embryonic immunohistochemistry studies revealthat Smad2/3 activity is widespread throughout thedeveloping limb bud. Interestingly, in situ hybridiza-tion analysis of the developing limb suggested anapparent upregulation of Scx and Col1a1 mRNA in the

Figure 4. Loss of Smad3 disrupts normal matrix protein and gene expression in developing tendon. A: Mice with a Smad2/3responsive element driving a luciferase reporter gene display broad luciferase expression across the limb bud and in the FDP tendon(blue arrow) as analyzed by immunohistochemistry at E16.5. B: Immunohistochemistry of axial sections of the mouse embryonicforelimb at E16.5 show protein expression of Smad3, Col1, and Tnc, in WT and Smad3�/� FDP tendon (blue arrow). The IgG controlindicates the specificity of staining. C: In situ hybridization of axial limb sections of WT and Smad3�/� E16.5 embryos show expressionof Col1a1, Tnmd, Scx, and Mkx mRNA in the FDP tendon (yellow arrow). Images are representative of N ¼ 3 mice.

1480 BERTHET ET AL.


absence of Smad3 in contrast to our adult in situ andqPCR data, which showed reduced expression of thesegenes. The mechanism and significance of this appar-ent discrepancy are unclear. It is possible that themechanisms controlling expression of the tenocytegenes are differentially sensitive to Smad3 in embry-onic and adult tissue. Alternatively, upregulation ofalternate downstream effectors such as Smad2 during

development may be able compensate for the loss ofSmad3.

Furthermore, while our data show that expressionof Col1a1 mRNA in the developing limb is upregulatedin the absence of Smad3, immunohistochemistryreveals that total Collagen1 protein levels are simulta-neously reduced, suggesting a potential post-transcrip-tional regulatory role for Smad3. For example, Smad3is known to regulate the expression and activity ofseveral miRNAs as well as of enzymes responsible formatrix protein modification and secretion in othertissues.9,26 Smad3 may similarly play a post-transcrip-tional regulatory role in the developing limb.

In mature tendon, Smad3 ablation resulted insignificantly diminished expression of Col1a1 andCol1a2 mRNA, which together encode the primarytendon matrix protein Collagen 1. In addition, expres-sion of major tenocyte marker genes including thetendon transcription factors Scx and Mkx were alsosomewhat reduced. The effect of Smad3 deficiency ontranscription factor expression suggests that Smad3may impact the expression of genes encoding matrixcomponents such as Collagen 1 not only in a directfashion but also indirectly via its regulation of theexpression of Scx and Mkx.

Smad3 plays a critical role in the differentiation ofother tissues of the musculoskeletal system. In bone,Smad3 binds the osteogenic transcription factor Runx2to confer TGFb-mediated inhibition of terminal osteo-blast differentiation.25 Similarly, during chondrocytedifferentiation, Smad3 binds Sox9 to promote chondro-genic gene expression.12 Likewise, Smad3 participatesin the regulation of myocyte and adipocyte differentia-tion by interactions with the lineage-specific transcrip-tion factors MyoD and C/EBPb respectively.11,27 Ineach of these tissues, Smad3 exerts its effects on theexpression of lineage-specific genes via its interactionwith tissue-specific transcription factors.

Similarly, we have found that in tendon Smad3 hasthe ability to directly interact with Scx in vivo.Unfortunately due to the limitation of available anti-bodies that recognize Mkx, we were unable to confirma similar interaction between Smad3 and Mkx in vivo.However, our in vitro work reveals that Smad3 canbind Mkx in a similar manner to Scx. In addition,recent work has demonstrated that Smad3 and Scxare able to bind to and synergistically activate theCol1a2 promoter, suggesting that the Smad3–Scxcomplex plays a key functional role in the regulationof extracellular matrix production.28 Together theseresults strongly suggest that Smad3 is able to regulatetendon differentiation via its interaction with tendon-specific transcription factors in a manner similar tothat observed in other tissues of the musculoskeletalsystem.

In other mesenchymally derived cells, Smad3 modu-lates gene expression not only through its interactionwith tissue specific transcription factors, but alsothrough the additional recruitment of transcriptional

Figure 5. Smad3 binds Scx and Mkx. 10T1/2 cells were trans-fected with the indicated plasmids (A: Smad3 and/or Scx; B:Smad3 and/or Mkx) and treated with TGFb or vehicle control.Lysates were immunoprecipitated with anti-Flag antibodies.Smad3 binds Scx (A) and Mkx (B) as indicated by the co-precipitating band in the immunoprecipitated (IP) fraction.Whole cell lysate (WCL) shows total protein expression. C:Smad3 binds Scx in tendon in vivo as indicated by a co-precipitating Scx band following immunoprecipitation of wholetendon lysates (WTL) by anti-Smad3.



co-activators such as CREB binding protein duringchondrogenesis or co-repressors such as class IIahistone deacetylases during osteogenesis.12,29 Whetherthe Smad3–Scx and Smad3–Mkx interactions we haveidentified exist within a larger complex or are capableof recruiting additional co-factors presents an interest-ing avenue for further investigation.

While previous studies have suggested that Smad3activity increases the formation of pathologic adhe-sions after tendon injury, our findings revealan activerole for Smad3 in normal tendon development and inmaintenance of tendon architecture.15 TGFb similarlyplays a role in both pathologic adhesion formation andin normal tendon development. Therefore, TGFb mayexert its crucial regulation of both tendon differentia-tion and adhesion formation via a Smad3-dependentpathway.

The formation of flexor tendon adhesions after tendoninjury and repair may be induced by the expression offactors from the tissues of the surrounding tendonsheath rather than from within the tendon itself. Thus,it is possible for Smad3 activity to induce scar andadhesion formation when expressed by the tissues of thesurrounding tendon sheath while promoting tendonhealing when expressed within the tendon itself. Ourstudy adds to the growing body of evidence indicatingthat the same pathways regulate both tendon healingand tendon adhesion formation. Further studies differ-entiating the effects of Smad3 expression in cells of thesurrounding tendon sheath from those of the tenocytesthemselves may provide some clue as to how thispathway may be modulated in order to prevent adhesionformation while promoting tendon healing.

ACKNOWLEDGMENTSWe thank Melissa Ehlers, Brian Black, and Alan Rawls forexpression plasmids, Rulang Jiang and Ronen Schweitzer forin situ probe plasmids, Ashton Mortazavi for assistance withhistology, and Christopher DuFort for technical assistance.This work was supported in part by research grants from theNIH (NIDCR R01 DE019284 TA, NIDCR R01 DE016402 RS,NIDCR COHORT T32 CC), the American Society for Surgeryof the Hand (M.A.), and the Resource Allocation Program atthe University of California, San Francisco (M.A.).

REFERENCES1. Derynck R, Schneider RA, Piek E, et al. 2008. TGF-b family

signaling in mesenchymal differentiation. In: Derynck R,Miyazono K, editors. The TGF-b family. Woodbury, NY: ColdSpring Harbor Press, p. 613–655.

2. Chang J, Most D, Stelnicki E, et al. 1997. Gene expression oftransforming growth factor beta-1 in rabbit zone II flexortendon wound healing: evidence for dual mechanism ofrepair. Plast Reconstr Surg 100:937–944.

3. Pryce BA, Watson SS, Murchison ND, et al. 2009. Recruit-ment and maintenance of tendon progenitors by TGF{beta}signaling are essential for tendon formation. Development136:1351–1361.

4. Chan KM, Fu SC, Wong YP, et al. 2008. Expression oftransforming growth factor beta isoforms and their roles intendon healing. Wound Repair Regen 16:399–407.

5. Harris SE, Bonewald LF, Harris MA, et al. 1994. Effects oftransforming growth factor beta on bone nodule formationand expression of bone morphogenetic protein 2, osteocalcin,osteopontin, alkaline phosphatase, and type I collagenmRNA in long-term cultures of fetal rat calvarial osteo-blasts. J Bone Miner Res 9:855–863.

6. Rydziel S, Varghese S, Canalis E. 1997. Transforminggrowth factor beta1 inhibits collagenase 3 expression bytranscriptional and post-transcriptional mechanisms in oste-oblast cultures. J Cell Physiol 170:145–152.

7. Jung J, Wang P, Zhang G, et al. 2009. Collagen fibril growthduring chicken tendon development: matrix metalloprotei-nase-2 and its activation. Cell Tissue Res 336:79–89.

8. Mendias CL, Bakhurin KI, Faulkner JA. 2008. Tendons ofmyostatin-deficient mice are small, brittle, and hypocellular.Proc Natl Acad Sci U S A 105:388–393.

9. Balooch G, Balooch M, Nalla RK, et al. 2005. TGF-betaregulates the mechanical properties and composition of bonematrix. Proc Natl Acad Sci U S A 102:18813–18818.

10. Furumatsu T, Tsuda M, Yoshida K, et al. 2005. Sox9 andp300 cooperatively regulate chromatin-mediated transcrip-tion. J Biol Chem 280:35203–35208.

11. Liu D, Black BL, Derynck R. 2001. TGF-beta inhibits muscledifferentiation through functional repression of myogenictranscription factors by Smad3. Genes Dev 15:2950–2966.

12. Furumatsu T, Tsuda M, Taniguchi N, et al. 2005. Smad3induces chondrogenesis through the activation of SOX9 viaCREB-binding protein/p300 recruitment. J Biol Chem280:8343–8350.

13. Lorda-Diez C, Montero J, Martinez-Cue C, et al. 2009.Transforming growth factors beta coordinate cartilage andtendon differentiation in the developing limb mesenchyme. JBiol Chem

14. Liu W, Watson S, Lan Y, et al. 2010. The atypicalhomeodomain transcription factor Mohawk controls tendonmorphogenesis. Mol Cell Biol 30:4797–4807.

15. Katzel EB, Wolenski M, Loiselle AE, et al. 2011. Impact ofSmad3 loss of function on scarring and adhesion formationduring tendon healing. J Orthop Res 29:684–693.

16. Borton AJ, Frederick JP, Datto MB, et al. 2001. The loss ofSmad3 results in a lower rate of bone formation andosteopenia through dysregulation of osteoblast differentia-tion and apoptosis. J Bone Miner Res 16:1754–1764.

17. Lin AH, Luo J, Mondshein LH, et al. 2005. Global analysisof Smad2/3-dependent TGF-beta signaling in living micereveals prominent tissue-specific responses to injury. JImmunol 175:547–554.

18. Albrecht UEG. 1997. Visualization of gene expression pat-terns by in situ hybridization. In: Gaston GP, editor.Molecular and Cellular Methods in Developmental Toxicolo-gy. Boca Raton, FL: CRC Press, p. 23–48.

19. Metsaranta M, Toman D, De Crombrugghe B, et al. 1991.Specific hybridization probes for mouse type I, II, III and IXcollagen mRNAs. Biochim Biophys Acta 1089:241–243.

20. Liu W, Watson SS, Lan Y, et al. 2010. The atypicalhomeodomain transcription factor Mohawk controls tendonmorphogenesis. Mol Cell Biol 30:4797–4807.

21. Murchison ND, Price BA, Conner DA, et al. 2007. Regulationof tendon differentiation by scleraxis distinguishes force-transmitting tendons from muscle-anchoring tendons. Devel-opment 134:2697–2708.

22. Bullen A, Friedman RS, Krummel MF. 2009. Two-photonimaging of the immune system: a custom technology plat-form for high-speed, multicolor tissue imaging of immuneresponses. Curr Top Microbiol Immunol 334:1–29.

23. Anderson DM, Beres BJ, Wilson-Rawls J, et al. 2009. Thehomeobox gene Mohawk represses transcription by recruit-

1482 BERTHET ET AL.


ing the sin3A/HDAC co-repressor complex. Dev Dyn 238:572–580.

24. Livak KJ, Schmittgen TD. 2001. Analysis of relative geneexpression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) method. Methods 25:402–408.

25. Alliston T, Choy L, Ducy P, et al. 2001. TGF-beta-inducedrepression of CBFA1 by Smad3 decreases cbfa1 and osteo-calcin expression and inhibits osteoblast differentiation.EMBO J 20:2254–2272.

26. Qin W, Chung A, Huang X, et al. 2011. TGF-b/Smad3signaling promotes renal fibrosis by inhibiting miR-39. J AmSoc Nephrol 22:1462–1474.

27. Choy L, Derynck R. 2003. Transforming growth factor-Binhibits adipocyte differentiation by Smad3 interacting withCCAAT/enhancer-binding protein (C/EBP) and repressingC/EBP transactivation function. J Biol Chem 278:9609–9619.

28. Bagchi RA, Czubryt MP. 2012. Synergistic roles of scleraxisand Smads in the regulation of collagen 1alpha2 geneexpression. Biochim Biophys Acta 1823:1936–1944.

29. Kang JS, Alliston T, Delston R, et al. 2005. Repression ofRunx2 function by TGF-beta through recruitment of class IIhistone deacetylases by Smad3. EMBO J 24:2543–2555.



Smad3 binds scleraxis and mohawk and regulates tendon matrix organization

Documents

Transcript of Smad3 binds scleraxis and mohawk and regulates tendon matrix organization