TGF-βmediates early angiogenesis and latent fibrosis in an ...TGF-βmediates early angiogenesis and...

10
© 2014. Published by The Company of Biologists Ltd | Disease Models & Mechanisms (2014) 7, 987-996 doi:10.1242/dmm.015255 987 ABSTRACT Aortic valve disease (AVD) is characterized by elastic fiber fragmentation (EFF), fibrosis and aberrant angiogenesis. Emilin1 is an elastin-binding glycoprotein that regulates elastogenesis and inhibits TGF-β signaling, but the role of Emilin1 in valve tissue is unknown. We tested the hypothesis that Emilin1 deficiency results in AVD, mediated by non-canonical (MAPK/phosphorylated Erk1 and Erk2) TGF-β dysregulation. Using histology, immunohistochemistry, electron microscopy, quantitative gene expression analysis, immunoblotting and echocardiography, we examined the effects of Emilin1 deficiency (Emilin1 /) in mouse aortic valve tissue. Emilin1 deficiency results in early postnatal cell-matrix defects in aortic valve tissue, including EFF, that progress to latent AVD and premature death. The Emilin1 /aortic valve displays early aberrant provisional angiogenesis and late neovascularization. In addition, Emilin1 /aortic valves are characterized by early valve interstitial cell activation and proliferation and late myofibroblast-like cell activation and fibrosis. Interestingly, canonical TGF-β signaling (phosphorylated Smad2 and Smad3) is upregulated constitutively from birth to senescence, whereas non-canonical TGF-β signaling (phosphorylated Erk1 and Erk2) progressively increases over time. Emilin1 deficiency recapitulates human fibrotic AVD, and advanced disease is mediated by non-canonical (MAPK/phosphorylated Erk1 and Erk2) TGF-β activation. The early manifestation of EFF and aberrant angiogenesis suggests that these processes are crucial intermediate factors involved in disease progression and therefore might provide new therapeutic targets for human AVD. KEY WORDS: Elastic fibers, Extracellular matrix, Aortic valve, Fibrosis, Angiogenesis INTRODUCTION Aortic valve disease (AVD) affects more than 2% of the general population and typically manifests later in life (Nkomo et al., 2006). Therapeutic intervention remains primarily surgical valve replacement, which is associated with limited durability and significant complications (Keane et al., 1993; Hammermeister et al., 2000). AVD is characterized by extracellular matrix (ECM) RESEARCH ARTICLE 1 Division of Cardiology, The Heart Institute, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH, USA. 2 Division of Molecular Cardiovascular Biology, The Heart Institute, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH, USA. 3 Department of Molecular Medicine, University of Padua, 35121 Padua, Italy. *Author for correspondence ([email protected]) This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution and reproduction in any medium provided that the original work is properly attributed. Received 19 December 2013; Accepted 5 June 2014 abnormalities that typically manifest as fibrosis then calcification, resulting in stenosis with or without regurgitation. The prevailing view has been that injury and inflammation result in AVD, but there is increasing evidence that the cell-ECM changes that characterize AVD are mediated by abnormalities in molecular programs that regulate cardiac development (Markwald et al., 2010; Hinton and Yutzey, 2011). Our understanding of the molecular mechanisms underlying AVD pathogenesis, especially disease progression, has been limited, in part, by a lack of animal models that recapitulate the natural history of human AVD (Schoen, 2008; Rajamannan et al., 2011). The mature aortic valve comprises three cusps that are hinged to a crown-shaped annulus within the aortic root. Cusp trilaminar ECM is stratified into fibrosa, spongiosa and ventricularis layers with elastic fibers organized as filaments in the ventricularis layer (Schoen, 1997; Hinton et al., 2006). Valve interstitial cells (VICs) comprise a heterogeneous population of cells that can be classified as quiescent or activated (Rabkin-Aikawa et al., 2004), and some VICs possess characteristics similar to those of smooth muscle cells (SMCs) (Bairati and DeBiasi, 1981; Filip et al., 1986). Activated VICs have been implicated in AVD because they are associated with pathological cell proliferation and maladaptive ECM remodeling (Rabkin-Aikawa et al., 2004). Although healthy adult valve tissue is avascular (Duran and Gunning, 1968), neovascularization has been described in non-rheumatic degenerative AVD and attributed to end- stage inflammatory processes, a secondary result of a wound- healing-like response to injury (Rajamannan et al., 2005; Paruchuri et al., 2006; Li et al., 2011). Little is known about the potential primary role of aberrant angiogenesis in AVD. Elastic fiber fragmentation (EFF) has long been associated with AVD, and the conventional view is that structural degeneration reflects end-stage disease processes (Schoen, 1997). Mouse models with elastic fiber assembly defects often have valve abnormalities (Hanada et al., 2007; Hinton et al., 2010), and EFF is present in pediatric, as well as adult, AVD (Fondard et al., 2005; Hinton et al., 2006; Wirrig et al., 2011), suggesting that EFF reflects elastic fiber assembly defects, in addition to elastase initiated EFF, and therefore might have a role in AVD progression. EFF has been associated with increased elastolytic enzyme activity, which has been identified as a factor contributing to AVD progression in both mouse and human (Fondard et al., 2005; Krishnamurthy et al., 2012). Elastic fibers are made up of elastin (the core protein) and microfibrils (fibrillins and associated proteins), as well as various glycoproteins, such as emilins and fibulins (Kielty et al., 2002; Wagenseil and Mecham, 2009). Emilin1 (elastin microfibril interface-located protein) is an elastin- and fibulin-5-binding protein that is necessary for elastogenesis and inhibits transforming growth factor β (TGF-β) signaling (Zanetti et al., 2004; Zacchigna et al., 2006). In addition, fibroblasts, keratinocytes and lymphatic endothelial cells in Emilin1- deficient mice demonstrate increased proliferation due to absent Emilin1-integrin interactions (Danussi et al., 2011; Danussi et al., TGF-β mediates early angiogenesis and latent fibrosis in an Emilin1-deficient mouse model of aortic valve disease Charu Munjal 1 , Amy M. Opoka 1 , Hanna Osinska 2 , Jeanne F. James 1 , Giorgio M. Bressan 3 and Robert B. Hinton 1, * Disease Models & Mechanisms

Transcript of TGF-βmediates early angiogenesis and latent fibrosis in an ...TGF-βmediates early angiogenesis and...

Page 1: TGF-βmediates early angiogenesis and latent fibrosis in an ...TGF-βmediates early angiogenesis and latent fibrosis in an Emilin1-deficient mouse model of aortic valve disease Charu

© 2014. Published by The Company of Biologists Ltd | Disease Models & Mechanisms (2014) 7, 987-996 doi:10.1242/dmm.015255

987

ABSTRACTAortic valve disease (AVD) is characterized by elastic fiberfragmentation (EFF), fibrosis and aberrant angiogenesis. Emilin1 isan elastin-binding glycoprotein that regulates elastogenesis andinhibits TGF-β signaling, but the role of Emilin1 in valve tissue isunknown. We tested the hypothesis that Emilin1 deficiency results inAVD, mediated by non-canonical (MAPK/phosphorylated Erk1 andErk2) TGF-β dysregulation. Using histology, immunohistochemistry,electron microscopy, quantitative gene expression analysis,immunoblotting and echocardiography, we examined the effects ofEmilin1 deficiency (Emilin1−/−) in mouse aortic valve tissue. Emilin1deficiency results in early postnatal cell-matrix defects in aortic valvetissue, including EFF, that progress to latent AVD and prematuredeath. The Emilin1−/− aortic valve displays early aberrant provisionalangiogenesis and late neovascularization. In addition, Emilin1−/−

aortic valves are characterized by early valve interstitial cell activationand proliferation and late myofibroblast-like cell activation andfibrosis. Interestingly, canonical TGF-β signaling (phosphorylatedSmad2 and Smad3) is upregulated constitutively from birth to senescence, whereas non-canonical TGF-β signaling(phosphorylated Erk1 and Erk2) progressively increases over time.Emilin1 deficiency recapitulates human fibrotic AVD, and advanceddisease is mediated by non-canonical (MAPK/phosphorylated Erk1and Erk2) TGF-β activation. The early manifestation of EFF andaberrant angiogenesis suggests that these processes are crucialintermediate factors involved in disease progression and thereforemight provide new therapeutic targets for human AVD.

KEY WORDS: Elastic fibers, Extracellular matrix, Aortic valve,Fibrosis, Angiogenesis

INTRODUCTIONAortic valve disease (AVD) affects more than 2% of the generalpopulation and typically manifests later in life (Nkomo et al., 2006).Therapeutic intervention remains primarily surgical valvereplacement, which is associated with limited durability andsignificant complications (Keane et al., 1993; Hammermeister et al.,2000). AVD is characterized by extracellular matrix (ECM)

RESEARCH ARTICLE

1Division of Cardiology, The Heart Institute, Cincinnati Children’s Hospital MedicalCenter, Cincinnati, OH, USA. 2Division of Molecular Cardiovascular Biology, TheHeart Institute, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH,USA. 3Department of Molecular Medicine, University of Padua, 35121 Padua,Italy.

*Author for correspondence ([email protected])

This is an Open Access article distributed under the terms of the Creative CommonsAttribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricteduse, distribution and reproduction in any medium provided that the original work is properlyattributed.

Received 19 December 2013; Accepted 5 June 2014

abnormalities that typically manifest as fibrosis then calcification,resulting in stenosis with or without regurgitation. The prevailing viewhas been that injury and inflammation result in AVD, but there isincreasing evidence that the cell-ECM changes that characterize AVDare mediated by abnormalities in molecular programs that regulatecardiac development (Markwald et al., 2010; Hinton and Yutzey,2011). Our understanding of the molecular mechanisms underlyingAVD pathogenesis, especially disease progression, has been limited,in part, by a lack of animal models that recapitulate the natural historyof human AVD (Schoen, 2008; Rajamannan et al., 2011).

The mature aortic valve comprises three cusps that are hinged toa crown-shaped annulus within the aortic root. Cusp trilaminar ECMis stratified into fibrosa, spongiosa and ventricularis layers withelastic fibers organized as filaments in the ventricularis layer(Schoen, 1997; Hinton et al., 2006). Valve interstitial cells (VICs)comprise a heterogeneous population of cells that can be classifiedas quiescent or activated (Rabkin-Aikawa et al., 2004), and someVICs possess characteristics similar to those of smooth muscle cells(SMCs) (Bairati and DeBiasi, 1981; Filip et al., 1986). ActivatedVICs have been implicated in AVD because they are associated withpathological cell proliferation and maladaptive ECM remodeling(Rabkin-Aikawa et al., 2004). Although healthy adult valve tissue isavascular (Duran and Gunning, 1968), neovascularization has beendescribed in non-rheumatic degenerative AVD and attributed to end-stage inflammatory processes, a secondary result of a wound-healing-like response to injury (Rajamannan et al., 2005; Paruchuriet al., 2006; Li et al., 2011). Little is known about the potentialprimary role of aberrant angiogenesis in AVD.

Elastic fiber fragmentation (EFF) has long been associated withAVD, and the conventional view is that structural degenerationreflects end-stage disease processes (Schoen, 1997). Mouse modelswith elastic fiber assembly defects often have valve abnormalities(Hanada et al., 2007; Hinton et al., 2010), and EFF is present inpediatric, as well as adult, AVD (Fondard et al., 2005; Hinton et al.,2006; Wirrig et al., 2011), suggesting that EFF reflects elastic fiberassembly defects, in addition to elastase initiated EFF, and thereforemight have a role in AVD progression. EFF has been associated withincreased elastolytic enzyme activity, which has been identified asa factor contributing to AVD progression in both mouse and human(Fondard et al., 2005; Krishnamurthy et al., 2012). Elastic fibers aremade up of elastin (the core protein) and microfibrils (fibrillins andassociated proteins), as well as various glycoproteins, such asemilins and fibulins (Kielty et al., 2002; Wagenseil and Mecham,2009). Emilin1 (elastin microfibril interface-located protein) is anelastin- and fibulin-5-binding protein that is necessary forelastogenesis and inhibits transforming growth factor β (TGF-β)signaling (Zanetti et al., 2004; Zacchigna et al., 2006). In addition,fibroblasts, keratinocytes and lymphatic endothelial cells in Emilin1-deficient mice demonstrate increased proliferation due to absentEmilin1-integrin interactions (Danussi et al., 2011; Danussi et al.,

TGF-β mediates early angiogenesis and latent fibrosis in anEmilin1-deficient mouse model of aortic valve diseaseCharu Munjal1, Amy M. Opoka1, Hanna Osinska2, Jeanne F. James1, Giorgio M. Bressan3 and Robert B. Hinton1,*

Dis

ease

Mod

els

& M

echa

nism

s

Page 2: TGF-βmediates early angiogenesis and latent fibrosis in an ...TGF-βmediates early angiogenesis and latent fibrosis in an Emilin1-deficient mouse model of aortic valve disease Charu

988

2012). Importantly, Emilin1 is expressed in the developing andmature heart valves (Braghetta et al., 2002; Angel et al., 2011;Votteler et al., 2013), and Emilin1 deficiency results in EFF andTGF-β activation in the aorta. The role of Emilin1 in AVDpathogenesis is unknown.

We tested the hypothesis that Emilin1 deficiency results in AVD,which is characterized by early EFF and late fibrosis, mediated bynon-canonical TGF-β signaling [through mitogen-activated proteinkinases (MAPK) and extracellular-signal-regulated kinases 1 and 2(Erk1/2)]. We have demonstrated here that Emilin1−/− aortic valvesexhibit EFF and early aberrant angiogenesis, as well as late fibrosisand stenosis, and that these pathologic findings are due to complexTGF-β dysregulation, including progressive upregulation of non-canonical phosphorylated Erk1/2 signaling. This new model of AVDrecapitulates the natural history of human AVD, thereby providinginsights into mechanisms of disease progression that might identifynew therapeutic targets.

RESULTSEmilin1−/− aortic valve tissue is characterized by early EFFand progressive elastolysisJuvenile Emilin1−/− aortic valve tissue exhibits EFF, which ischaracterized by decreased elastic fiber content and dispersion of

elastic fiber components to all cusp layers, namely the fibrosa,spongiosa and ventricularis, as well as the annulus, indicating faultyelastic fiber assembly (Fig. 1A,B). EFF and delamination worsensover time (Fig. 1B,D,F). Ultra-structural analysis of aged Emilin1−/−

aortic valves confirmed marked EFF, delamination and dispersion(Fig. 1H). MMP-9 was examined because it has been shown to playa crucial role in pathological valve remodeling. MMP-9 expressionwas unaltered in adult Emilin1−/− valves but was increased in agedEmilin1-deficient aortic valves in comparison with age-matchedcontrols (Fig. 1J) (Fondard et al., 2005). This observation wassupported by immunoblot analyses, which demonstrated increasedexpression of active MMP-9, as well as MMP-2, in aged mutantvalves (Fig. 1K). Importantly, elastase expression, which is notexpressed in normal healthy valve, was increased in juvenile andadult Emilin1−/− aortic valve tissue and further increased at the agedstage (Fig. 1M,O,Q), demonstrating a role for elastase andremodeling enzymes in the progression of AVD. In aged Emilin1−/−

aortic valve tissue, elastin mRNA expression was significantlydecreased, whereas that of fibrillin-1 was unchanged (Fig. 1R).Elastin mRNA expression was also decreased in ascending aorta andleft myocardial tissue that had been isolated from Emilin1−/− mice(supplementary material Fig. S2O). Taken together, elastic fiberassembly defects in Emilin1−/− mice predispose aortic valve tissueto progressive elastolysis and worsening EFF.

Aberrant provisional angiogenesis is an early finding thatprogresses to neovascularization in Emilin1−/− aortic valvetissueNeovascularization (overt angiogenesis) was present in Emilin1−/−

aortic valves at the aged stage only and localized to the annulusand proximal cusp, these neovessels also stained positive for CD-31 (Fig. 2B,D). At the earlier adult stage, Emilin1−/− aortic valvesexhibited aberrant provisional angiogenesis, characterized byincreased interstitial VEGF-A and VEGF-R1 (Flt1) expression,which are normally restricted to the valve endothelium, butunchanged VEGF-R2 (Flk1) expression (Fig. 2F,J,N). In addition,circulatory VEGF-A levels were increased significantly in agedEmilin1−/− mice (data not shown). Interestingly, the angiostaticfactors chondromodulin (Chm), endostatin and fibulin-5, which arenormally ubiquitously expressed in valve tissue, were markedlydecreased in aged Emilin1−/− aortic valves (Fig. 2R,T,V),suggesting that aberrant angiogenesis is the result of bothincreased pro-angiogenic factors and decreased angiostatic factors.Quantitative reverse transcription PCR (QRT-PCR) analysis wasperformed in aged aortic valve tissue that showed a significantincrease in both VEGF-A and VEGF-R1, as well as a significantdecrease in endostatin. Overall, these results demonstrate thataberrant angiogenesis is an early disease process that worsens overtime, identifying a crucial role for angiogenesis in AVDprogression.

Dynamic activation of non-canonical TGF-β (phosphorylatedErk1/2) signaling in Emilin1−/− aortic valve tissueTo determine the effects of Emilin1 deficiency on TGF-β signalingin aortic valve tissue, canonical and non-canonical pathways wereexamined. In Emilin1−/− aortic valves at the juvenile stage, levels ofphosphorylated Smad2 and Smad3 (Smad2/3) were increased whencompared with age-matched controls (Fig 3A,B). In Emilin1−/−

aortic valves at both the adult and aged stages, there was asignificant and similar increase in the absolute number and ratio ofphosphorylated Smad2/3-positive nuclei, as well as proteinexpression using immunoblotting (Fig. 3D,F-H). In Emilin1+/+ aortic

RESEARCH ARTICLE Disease Models & Mechanisms (2014) doi:10.1242/dmm.015255

TRANSLATIONAL IMPACTClinical issue Aortic valve disease (AVD) is a major cause of cardiovascular morbidityand affects more than 2% of the general population of the USA. Aorticvalve replacement remains the principal treatment strategy for end-stageAVD; however, this process is associated with considerable complications.Presently, there are no pharmacologic therapies to directly treat AVD. Abetter understanding of the cellular and molecular events that underlieAVD progression is required to develop novel approaches to treatment.Therefore, animal models that mimic the natural history of human AVDare needed to allow researchers to examine the early disease processand test new therapies. Emilin1 is an essential elastogenic protein that ispresent in both developing and mature aortic valve tissue. Previousstudies have shown that Emilin1 is required for normal elastic fiberassembly and inhibits TGF-β signaling in vasculature. However, the roleof Emilin1 dysregulation in AVD pathogenesis is unknown.

Results In this study, the Emilin1 homozygous knockout mouse (Emilin1–/–) wasidentified as a new model of latent AVD. Histological and molecularanalyses demonstrated that Emilin1 deficiency is associated with theactivation of non-canonical (phosphorylated Erk1 and Erk2) andcanonical (phosphorylated Smad2 and Smad3) signaling in aortic valvetissue. This results in early elastase-mediated elastic fiber fragmentationand aberrant angiogenesis in association with early valve interstitial cell(VIC) activation. Interestingly, there is progressive up-regulation ofphosphorylated Erk1 and Erk2 signaling over time, which results in theactivation of a subset of VICs with myofibroblast-like characteristics andmarked progression of valve pathology that shows severe fibrosis,neovascularization and inflammation.

Implications and future directions Together, these results demonstrate that the Emilin1-deficient mouse is aunique model of latent fibrotic AVD that provides important insights intoearly and intermediate disease mechanisms. These findings establish acentral role for elastic fiber dysregulation in early AVD pathogenesis,suggesting that faulty elastic fiber assembly and consequent elastase-mediated tissue injury contribute to disease initiation and progression. Abetter understanding of these early disease processes could contribute tothe development of novel treatments for AVD. Because the manifestationof AVD is latent in this model, similar to human AVD, preclinical studiesusing the Emilin1-deficient mouse model of AVD are warranted.

Dis

ease

Mod

els

& M

echa

nism

s

Page 3: TGF-βmediates early angiogenesis and latent fibrosis in an ...TGF-βmediates early angiogenesis and latent fibrosis in an Emilin1-deficient mouse model of aortic valve disease Charu

valves, phosphorylated Smad2/3 was increased with advanced agebut significantly decreased when compared with that of Emilin1−/−

valves. There was no change in phosphorylated Smad2/3 inEmilin1−/− myocardium (supplementary material Fig. S2G). TGF-βR1 mRNA expression was increased in Emilin1−/− aortic valvetissue, as determined by using QRT-PCR (Fig. 3I). The amount ofphosphorylated Erk1/2 was unchanged in juvenile Emilin1−/− aorticvalves but was modestly increased in adult Emilin1−/− aortic valves(Fig. 3J-M) and dramatically increased at the aged stage whencompared with the younger mutant (Fig. 3M,O), identifyingprogressive phosphorylated Erk1/2 activation. There was nophosphorylated Erk1/2 expression in Emilin1−/− myocardium (datanot shown). However, aged Emilin1−/− aorta tissue demonstratedincreased phosphorylated Erk1/2 expression that was localized tothe aortic root (supplementary material Fig S2I,N). Circulatoryplasma levels of active TGF-β were significantly increased at theadult and aged stages in Emilin1-deficient mice (Fig. 3R). Takentogether, these findings demonstrate dynamic dysregulation of TGF-β signaling in Emilin1−/− aortic valve tissue, indicating a possiblerole for non-canonical phosphorylated Erk1/2 activation in AVDprogression.

Early VIC activation and proliferation is maladaptive inEmilin1−/− aortic valve tissueIn order to examine the VIC phenotype, embryonic smooth musclemyosin heavy chain (SMemb) and alpha smooth muscle actin

(αSMA) were studied in Emilin1−/− aortic valve tissue. SMembexpression was increased at both the adult and aged stages,indicating VIC activation (Fig. 4B,D). αSMA expression wassimilarly increased at the adult stage but dramatically increased atthe aged stage (Fig. 4F,H). SM22, which is a myofibroblast marker,was not expressed in Emilin1−/− aortic valve tissue at the adult stagebut was abundantly expressed at the aged stage (Fig. 4J,L),identifying a distinct cell activation pattern consistent with latentmyofibroblast-like cell activation. Interestingly, VEGF-A and SM22were colocalized in some VICs (Fig. 4N), primarily in the annulusregion and proximal cusp where neovessels are present,demonstrating that a subset of VICs, rather than endothelial cells,regulate VEGF-A production in the valve interstitium, consistentwith previous observations (Paranya et al., 2001; Paruchuri et al.,2006; Syväranta et al., 2010). The total number of proliferative cellswas quantified by proliferation index analysis using phosphorylatedhistone H3 (HH3) staining. The results showed an increasedproliferative index in Emilin1−/− aortic valve tissue at the juvenileand adult stages that significantly increased by the aged stage (Fig.4O). Ki67 staining demonstrated similar patterns of proliferation bystage and genotype (supplementary material Fig. S1O). Importantly,activated VICs were also positive for the proliferation markerphosphorylated HH3, indicated by colocalization of phosphorylatedHH3 and SMemb (supplementary material Fig. S1L). Overall, theseresults implicate early VIC activation and late myofibroblast-likecell activation in the latent manifestation of AVD.

989

RESEARCH ARTICLE Disease Models & Mechanisms (2014) doi:10.1242/dmm.015255

Fig. 1. Elastic fiber assembly defects inEmilin1−/− aortic valve tissue. Sectionsstained using Hart’s stain show intactelastic fiber filaments localized to theventricularis layer in Emilin1+/+ mice(A,C,E) and EFF in Emilin1−/− aortic valvesthat worsened with time (arrowheads,B,D,F). Ultrastructure analyses in agedEmilin1−/− aortic valves show EFF withdispersion (arrowheads, H) whencompared with intact elastic fiber filamentsin Emilin1+/+ valves (G). MMP-9 and MMP-2 expression increased in the agedaortic valve of Emilin1−/− mice whencompared with that of Emilin1+/+ valvesusing immunohistochemistry (J) andwestern blotting (K). Western blottinganalysis showed increased expression ofboth the inactive proenzyme form (pro)and the active form of MMP-2 and MMP-9in Emilin1–/– aortic valves (K). There wasincreased neutrophil elastase expressionin Emilin1−/− aortic valve tissue, especiallyat the hinge region, at the juvenile, adultand aged stages (arrows, M,O,Q). In agedEmilin1−/− aortic valve tissue, elastinmRNA expression decreased significantlybut there was no change in the amount offibrillin1 mRNA (R). All panels are orientedto show the aortic valve in cross-sectionwith the ventricularis layer situated at thebottom. Mean±s.e.m.; n=12; *P<0.05Emilin1+/+ versus Emilin1−/−. Scale bars:10 μm (A-F); 500 nm (G,H); 50 μm (I-P). F,fibrosa; S, spongiosa; V, ventricularis.

Dis

ease

Mod

els

& M

echa

nism

s

Page 4: TGF-βmediates early angiogenesis and latent fibrosis in an ...TGF-βmediates early angiogenesis and latent fibrosis in an Emilin1-deficient mouse model of aortic valve disease Charu

990

The aged Emilin1−/− aortic valve demonstrates fibrosis andinflammation, but no calcificationFibrillar collagen deposition in aortic valve tissue was increased inaged Emilin1−/− when compared with Emilin1+/+. Specifically, typeI and type III collagen were increased significantly (Fig. 5G). Therewas evidence of mild aortic valve fibrosis at the earlier adult stage(data not shown). To determine the potential role of inflammation inthe manifestation of AVD in Emilin1-deficient mice, we examinedmacrophage and leukocyte markers. Emilin1−/− aortic valvesdisplayed abundant macrophage expression (Mac-3; also known asLamp2) at the aged stage (Fig. 5D), but not earlier stages (data notshown). Interestingly, CD-45, a pan-leukocyte marker, remainedunchanged compared with controls at all stages (data not shown).Ultrastructure analysis showed a qualitative abundance of fibroblast-like cells in the hinge region of aged Emilin1−/− aortic valves (Fig.5F,H). The identity of fibroblast-like cells was further established bythe presence of vimentin, an intermediate cytoskeletal proteinpresent in myofibroblast cells, which was increased in Emilin1−/−

aortic valve tissue (supplementary material Fig. S1B). Interestingly,Emilin1-deficient aortic valves did not exhibit calcification at anystage, as shown by the calcification marker Alizarin Red and Runx-2 (supplementary material Fig. S1F,H). Overall, these findingsidentify a role for Emilin1 deficiency in the latent development ofvalve fibrosis, similar to the natural history of human AVD.

Latent fibrosis was also present in the ascending aorta andmyocardium of Emilin1−/− mice, as evidenced by ultrastructural andmorphometric analysis (supplementary material Fig. S2). Excessivecollagen deposition was largely restricted to the perivasculature

regions of the myocardium and adventitial region of the ascendingaorta (supplementary material Fig. S2D,K). Emilin1−/− myocardiumshowed altered sarcomere morphology that was characterized by anincreased cisternal space due to gaps between the sarcomeres andsarcoplasmic reticulum, and contracted sarcomeres with thinner Z-discs (supplementary material Fig. S2F); in addition, mitochondrialnumber and morphology were unaltered, consistent with anonspecific Ca2+ handling abnormality. Periostin, a profibroticmarker, was increased in the left ventricular myocardium of agedEmilin1−/− (supplementary material Fig. S2G). Collagen fibers in themedia of Emilin1−/− aorta were randomly arranged andheterogeneous in size (supplementary material Fig. S2M). Type I, IIand III collagen were not significantly changed in ascending aortaor myocardium, suggesting that other collagens contribute to thefibrotic response in these tissues. Aged Emilin1−/− mice alsodemonstrated thoracic aortic aneurysm or aortopathy, restricted tothe aortic root region, and three out of 10 (30%) mice demonstratedaortic dissection. Overall, these findings identify a potential role forEmilin1 in regulating ECM in cardiovascular tissues.

Emilin1 deficiency results in AVD in vivoIn order to evaluate aortic valve function in the Emilin1−/− mouse invivo, we performed echocardiography in adult and aged mice. Theaortic valve peak velocity and corresponding pressure gradient weresignificantly increased in aged Emilin1−/− mice compared with age-matched controls (Table 1), consistent with aortic valve stenosis. Inaddition, 20% demonstrated aortic insufficiency. The aortic root indexwas increased in aged Emilin1−/− mice, consistent with mild aortic

RESEARCH ARTICLE Disease Models & Mechanisms (2014) doi:10.1242/dmm.015255

Fig. 2. Emilin1 deficiency is associated with aberrant provisional angiogenesis and neovascularization. Overt neovascularization is shown in agedEmilin1−/− aortic valve tissue, restricted to the annulus and proximal cusp, as indicated by Movat’s pentachrome staining (arrowheads, B) and CD-31-positivestained neovessels (arrowheads, D). Emilin1+/+ controls are shown in A and C. VEGF-A (E-H) and VEGF-R1 (I-L) expression was increased at adult (F,J) andaged (H,L) stages in the interstitium of Emilin1−/− aortic valves when compared with that of control Emilin1+/+ tissue, which shows normal expression around theendothelial layer (arrowheads, E,G). Levels of endostatin (Q-R, green; blue, nuclei), chondromodulin (Chm) (S-T) and fibulin-5 (U-V) were lower at the agedstage (R,T,V) when compared with Emilin1+/+ controls (arrowheads, Q,S,U). mRNA expression for VEGF-A, VEGF-R1, VEGF-R2 and endostatin (Endos) inaged Emilin1−/− aortic valve tissue relative to Emilin1+/+ control values (normalized to 1) for each gene (W). Scale bars: 50 μm. Mean±s.e.m.; n=10; *P<0.05Emilin1+/+ versus Emilin1−/−.

Dis

ease

Mod

els

& M

echa

nism

s

Page 5: TGF-βmediates early angiogenesis and latent fibrosis in an ...TGF-βmediates early angiogenesis and latent fibrosis in an Emilin1-deficient mouse model of aortic valve disease Charu

root dilation, a finding that is commonly associated with AVD (Hahnet al., 1992). The ascending aorta dimension was normal.Paradoxically, the aortic valve annulus dimension was increased,which together with aortic root dilation is reminiscent of annulo-aorticectasia in human connective tissue disorders. In 75% of agedEmilin1−/− mice, there was significant left ventricular dysfunction, acommon complication of severe AVD. Left ventricular mass and enddiastolic dimension were unchanged in Emilin1−/− mice, consistentwith the absence of a primary cardiomyopathy. Interestingly, 25% ofEmilin1−/− mice die prematurely between 14 and 18 months due tounclear causes when compared with Emilin1+/+ mice, which arereported to have a longevity of 25 months (Yuan et al., 2009). Nosignificant changes in aortic valve and ventricular functions wereobserved in adult mutant mice when compared with age-matchedcontrol (data not shown). Taken together, these findings identify theEmilin1−/− mouse as a model of latent fibrotic AVD.

DISCUSSIONIn the present study, we demonstrated that the Emilin1 deficientmouse is a model of human fibrotic AVD. Importantly this modelrecapitulates the natural history of human AVD allowinginvestigation of early disease processes. In summary, Emilin1−/−

aortic valve tissue demonstrates complex TGF-β dysregulation thatcauses downstream activation of both non-canonical (phosphorylatedErk1/2) and canonical (phosphorylated Smad2/3) TGF-β signaling,and ultimately results in aberrant angiogenesis and mild fibrosis inthe adult stage due to VIC activation and EFF mediated in-part dueto increased expression of elastolytic enzymes. Interestingly, overtime, there is progressive upregulation of phosphorylated Erk1/2signaling that results in activation of a subset of VICs withmyofibroblast-like characteristics, which results in neovascularizationand severe fibrosis at the aged stage (Fig. 6). Taken together, theEmilin1−/− model of AVD provides unique opportunities to identify

991

RESEARCH ARTICLE Disease Models & Mechanisms (2014) doi:10.1242/dmm.015255

Fig. 3. Emilin1 deficiency is associatedwith canonical and non-canonical TGF-β activation in aortic valve tissue.Levels of phosphorylated Smad2/3 (p-Smad2/3) are increased in juvenile(arrowheads, B), adult (arrowheads, D)and aged (arrowheads, F) Emilin1−/−

aortic valves compared those in juvenile(A), adult (C) and aged (E) Emilin1+/+

valves. Increased phosphorylatedSmad2/3 in adult and aged tissue wasconfirmed by immunoblotting (G),quantification of Smad2/3-positive nucleito total nuclei is shown in H. QRT-PCRshowed increased mRNA expression ofTGF-βR1 in aged Emilin1+/+ (white bar)and Emilin1−/− (black bar) aortic valve (I).MAPK/phosphorylated Erk1/2 signalingprogressively increased in Emilin1−/−

aortic valves (arrowheads, K,M,O) whencompared with age-matched controls(J,L,N). Increased phosphorylated Erk1/2activation (p-Erk1/2) was confirmed byimmunoblotting (P) in both adult and agedEmilin1−/− aortic valve tissue; this increasewas significant at the aged stage, asassessed using ImageJ densitometryquantification (Q). Histogram showscirculatory levels of active TGF-β in theplasma samples of Emilin1+/+ andEmilin1−/− mice (R). Scale bars: 50 μm.Mean±s.e.m.; n=14; *P<0.05 Emilin1+/+

versus Emilin1−/−; #P<0.05 adult versusaged.

Dis

ease

Mod

els

& M

echa

nism

s

Page 6: TGF-βmediates early angiogenesis and latent fibrosis in an ...TGF-βmediates early angiogenesis and latent fibrosis in an Emilin1-deficient mouse model of aortic valve disease Charu

992

predictive biomarkers and to test new therapeutic targets that couldrepresent much needed early intervention strategies to preventadvanced AVD and the need for surgery.

Elastic fibers play a crucial role in continuous valve motion bycontributing to the normal recoil process during the cardiac cycle,and EFF and elastase-mediated disease progression are universalfindings of AVD that contribute to valve dysfunction (Vesely,1997; Fondard et al., 2005; Schoen, 2008). Emilin1 deficiency andthe resulting elastic fiber assembly defects have been studied intissue of the aorta, showing that Emilin1 inhibits TGF-β signalingand that Emilin1 deficiency causes TGF-β upregulation (Zanetti etal., 2004; Zacchigna et al., 2006). The findings of the current studyshow that Emilin1−/− aortic valves have intrinsic elastic fiberassembly defects, suggesting that EFF might be the result of elasticfiber assembly defects that trigger further elastase-mediated EFFand a progressive disease process. Several mouse models ofmutated elastic fiber components have valve defects (Ng et al.,2004; Hanada et al., 2007; Hinton et al., 2010; Krishnamurthy etal., 2012), suggesting that elastic fiber assembly defects have acentral role in AVD pathogenesis. Interestingly, in normal aortic

valve tissue, Emilin1 is present in the annulus and other cusplayers, in addition to its presence in the mature elastic fiberfilaments in the ventricularis layer, suggesting that Emilin1 hasfunctional roles additional to those associated with elastic fiberassembly (Angel et al., 2011; Votteler et al., 2013). Overall, thisstudy identifies Emilin1 as an ECM protein that is necessary formature valve structure and function, and the Emilin1−/− mouse asan important model of viable AVD.

Neovascularization is present in the proximal cusp and annulusregions of aged Emilin1−/− aortic valves in proximity to EFF,consistent with findings in human AVD and the idea that cell-ECMhomeostasis is disrupted primarily at the hinge region of the valve(Thubrikar et al., 1986; Hinton et al., 2010; Wirrig et al., 2011). Inthe present study, we have shown a marked imbalance between pro-angiogenic and anti-angiogenic factors that promotes aberrantangiogenesis in the typically avascular valve tissue. Angiogenicsignaling pathways regulate normal embryonic valve development(Combs and Yutzey, 2009), but it is largely unknown whether thesepathways contribute to the maintenance of mature adult valve tissue.Previous studies have shown that mechanisms of AVD recapitulate

RESEARCH ARTICLE Disease Models & Mechanisms (2014) doi:10.1242/dmm.015255

Fig. 4. Emilin1 deficiency is associated with early VIC and late myofibroblast activation in aortic valve tissue. VIC activation is demonstrated byincreased SMemb and αSMA expression at the adult (arrowheads, B,F) and aged stages (arrowheads, D,H) in Emilin1–/–aortic valves when compared withEmilin1+/+ valves (A,C,E,G). Myofibroblast activation is shown by increased SM22 at the aged stage (arrowheads, L) but not the adult stage (J) in Emilin1–/–

aortic valves when compared with Emilin1+/+ valves (I,K). Confocal imaging shows colocalization of SM22 (green) and VEGF-A (red) in the endothelium ofEmilin1+/+ (M) and the interstitium of Emilin1–/– aortic valve tissue (white arrowheads, N). Emilin1–/– aortic valves demonstrate an increase in theproliferation index at all stages when compared with age-matched Emilin1+/+ valves (O). Aged Emilin1–/– aortic valves also showed significantly increasedproliferation when compared with juvenile and adult Emilin1–/– aortic valves. Scale bars: 50 µm (A-L); 10 µm (M,N). Mean±s.e.m.; n=8; *P<0.05 Emilin1+/+

vs Emilin1–/–; #P<0.05 adult Emilin1–/– vs aged Emilin1–/–; ^P<0.05 juvenile Emilin1–/– vs aged Emilin1–/–.

Dis

ease

Mod

els

& M

echa

nism

s

Page 7: TGF-βmediates early angiogenesis and latent fibrosis in an ...TGF-βmediates early angiogenesis and latent fibrosis in an Emilin1-deficient mouse model of aortic valve disease Charu

developmental programs, suggesting that the dysregulation ofspecific structural proteins and signaling pathways incite disease andcontribute to faulty homeostasis over time and ultimately valvedysfunction later in life (Markwald et al., 2010; Hinton and Yutzey,2011; Mahler and Butcher, 2011). Elastic fiber fragments, ordegradation products, might have pro-angiogenic, pro-proliferative,pro-calcific or pro-inflammatory properties (Perrotta et al., 2011;Pivetta et al., 2014; Parks and Mecham, 2011), in addition to well-described increases in elastase activity (Robinet et al., 2005),suggesting that different elastic fiber fragments have differentmaladaptive effects. In this context, Emilin1-deficiency-related EFFresults in aberrant angiogenesis and fibrosis, but not calcification.The identification of specific mechanisms that regulate differenttypes of disease progression might provide new potential therapeutictargets that prevent AVD progression.

The role of canonical TGF-β signaling (phosphorylated Smad2/3)during valve development and disease has been well established(Camenisch et al., 2002; Walker et al., 2004), but little is knownabout the role of non-canonical TGF-β signaling (phosphorylatedErk1/2). Previous in vitro studies have shown that TGF-β1 signalingactivates canonical Smad2/3 and non-canonical MAPK pathways,resulting in VIC activation and fibrosis, suggesting a potential rolefor phosphorylated Erk1/2 in AVD (Gu and Masters, 2009;Hutcheson et al., 2012). Interestingly, we showed that non-canonicalsignaling was upregulated and progressively increased over time,indicating the presence of a liability threshold for myofibroblastactivation (Fig. 6). We showed that VIC activation occurs early inthe disease process and that myofibroblast-like cell activationoccurred later in the disease process, associated temporally with thelatent findings of marked fibrosis, proliferation and inflammation.Activated myofibroblasts release angiogenic factors in the valveinterstitium, which is typically avascular, providing a putativemechanism for the advancement of late neovascularization fromearly provisional aberrant angiogenesis (Paruchuri et al., 2006).Consistently, our findings showed colocalization of activatedmyofibroblasts with VEGF-A-positive cells. In addition, there wasa progressive increase in the proliferation index in Emilin1−/− aorticvalve tissue, suggesting discrete contributions from each type ofVIC activation. Constitutive phosphorylated Smad2/3 upregulationmight result in early VIC activation, increased elastolytic activity,proliferation and provisional angiogenesis, whereas progressivephosphorylated Erk1/2 upregulation (or a cumulative threshold ofboth canonical and non-canonical TGF-β signaling) might result inlate myofibroblast activation that results in fibrosis andinflammation, consistent with previous reports (Hutcheson et al.,2013). Previous reports have shown robust activation ofphosphorylated Erk1/2 and proliferation in Emilin1−/− fibroblastcells that is mediated through a PTEN-dependent pathway (Danussiet al., 2011; Danussi et al., 2012). However, in the current study,PTEN levels were unaltered despite Erk1/2 activation and increasedproliferation in Emilin1−/− valves, consistent with a role for activatednon-canonical TGF-β signaling in aortic valve tissue. A limitednumber of apoptotic cells were identified as being restricted to thehinge region of aged Emilin1−/− aortic valves (supplementarymaterial Fig. S1M), possibly due to increased phosphorylatedErk1/2 signaling and high mechanical strain, suggesting that thebiomechanical properties of the Emilin1-deficient valves warrant

993

RESEARCH ARTICLE Disease Models & Mechanisms (2014) doi:10.1242/dmm.015255

Fig. 5. Aged Emilin1−/− aortic valve tissue demonstrates fibrosis andinflammation. Increased collagen deposition (blue staining using Masson’strichrome) is seen in the Emilin1−/− aortic valve annulus (arrowheads, B)when compared with age-matched Emilin1+/+ valves (A). Macrophageinfiltration, as evidenced by Mac-3-positive cells (arrowheads, D), is presentin aged Emilin1−/− valves only and not Emilin1+/+ controls (C). Electronmicroscopy (EM) of Emilin1−/− aortic valves showed fibroblast-like cells in allcusp layers (arrowheads, F,H) along with collagen accumulation. RelativemRNA expression of type I, II and III collagen in the aortic valve tissue ofaged Emilin1−/− mice to that of Emilin1+/+ mice (I). Scale bars: 50 μm (A-D);500 nm (E,F); 100 nm (G,H). Mean±s.e.m.; n=14; *P<0.05 Emilin1+/+ versusEmilin1−/−.

Table 1. Echocardiography in aged Emilin1−/− mice

Emilin1+/+ (n=11) Emilin1−/− (n=12)

DimensionsAnnulus (mm) 1.34±0.03 1.58±0.04*Root (mm) 1.76±0.08 1.92±0.06STJ (mm) 1.61±0.03 1.56±0.03AAO (mm) 1.66±0.03 1.60±0.03Root:AAO index 1.08±0.04 1.20±0.03*

Valve functionPeak velocity (mm/s) 380±70 526±66*Peak gradient (mm Hg) 0.60±0.09 1.20±0.06*

Ventricular functionEjection fraction (%) 54.5±1.5 39.5±3.5*Fractional shortening (%) 28.0±0.8 19.5±1.9*LVEDD (mm) 4.60±0.07 4.80±0.13LV mass (mg) 96.4±4.40 97.4±6.60Heart rate (bpm) 322±33 383±16

AAO, ascending aorta; LV, left ventricular; LVEDD, left ventricular enddiastolic diameter; STJ, sino-tubular junction; *P<0.05, n=10. Results areshown as mean±s.e.m.

Dis

ease

Mod

els

& M

echa

nism

s

Page 8: TGF-βmediates early angiogenesis and latent fibrosis in an ...TGF-βmediates early angiogenesis and latent fibrosis in an Emilin1-deficient mouse model of aortic valve disease Charu

994

further investigation. This implicates non-canonical MAPK/phosphorylated Erk1/2 activation as a mechanism underlying theprogression of AVD and the manifestation of advanced disease,controlling for the important adverse effects of aging.

Although non-resident cells have been described in the valveinterstitium of wild-type mice (Hajdu et al., 2011), the inflammationobserved in aged Emilin1−/− aortic valves might be due to otherprocesses. For example, inflammation might be the direct result ofsystemic TGF-β upregulation, given the role of TGF-β signalingin inflammation. Alternatively, in light of the abnormalitiesappreciated in the adventitia layer of the aorta, inflammation mightbe the result of the migration of non-resident cells from theadventitia layer into the valve annulus and proximal cusp,consistent with the localization of neovascularization. Anotheralternative is direct infiltration due to a discontinuous endotheliallayer. This results in elastases in the valve interstitium and migrationof circulating hematopoietic (or non-resident) cells from either thenewly formed valve neovasculature, which is not present until theaged stage, or the disrupted endothelium. There were areas ofendothelial disruption around the edges of the Emilin1−/− aorticvalve (data not shown), which is consistent with previous findings(Zanetti et al., 2004; Danussi et al., 2012). Previous evidencesuggests that activation of the phosphorylated Erk1/2 pathway is acrucial modifier in elastase-mediated diseases (Preston et al., 2002;Ghosh et al., 2012). Taken together, complex TGF-β dysregulationmediates early and late pathologic findings in Emilin1−/− aorticvalves, and phosphorylated Erk1/2 activation might play a moreimportant role than previously appreciated in AVD pathogenesis.

The findings of the current study identify the Emilin1−/− mouseas a model of latent fibrotic AVD. These findings have substantialclinical implications that emphasize the importance of translationalresearch. AVD progression is a poorly understood process that needsto be defined on a molecular basis in order to test new clinicaltherapies (Rajamannan et al., 2011). Presently, there are nopharmacological treatment strategies for early-stage AVD; therefore,the elucidation of the natural history of the disease will facilitatepreclinical studies aiming to test early intervention strategies,including pharmacological therapies attempting to prevent advancedAVD, potentially encompassing elastase, phosphorylated Erk1/2 orangiogenesis inhibitors. In addition, this model might provideinsights into common processes underlying aortopathy associatedwith AVD. This clinical association is well described and there issome genetic overlap that establishes shared etiologic factors,including TGF-β dysregulation (Hinton, 2012), but this is largelyundefined. Mutations in human EMILIN1 have been associated withcardiovascular disease (Shen et al., 2009; Liu and Xi, 2012), but notspecifically AVD or aortopathy, and therefore could represent animportant candidate gene and/or genetic modifier. Overall, theemilin family of glycoproteins might represent important factors incardiovascular diseases.

MATERIALS AND METHODSAnimalsEmilin1 deficient (Emilin1−/−) and wild-type (Emilin1+/+) littermates werestudied at juvenile (8-10 days), adult (4-6 months) and aged (12-14 months)stages. Mice were maintained on a C57Bl6 genetic background, andgenotyping was performed as described previously (Zanetti et al., 2004). Allprotocols were approved by the Institutional Animal Care and UseCommittee at Cincinnati Children’s Hospital Medical Center.

Histology and immunohistochemistryHearts were isolated and processed as described previously (Hinton et al.,2010). Movat’s pentachrome, Masson trichrome, Hart’s and Alizarin-Redstains were used. Immunohistochemistry was performed to assess markers ofVIC activation, myofibroblast differentiation, angiogenesis, TGF-β signaling,elastolysis and proliferation as described in Table 2. Antibodies were obtainedfrom Cell Signaling (Boston, MA, USA); Abcam (Cambridge, MA, USA);Santa Cruz Biotechnology (Dallas, TX, USA); Sigma-Aldrich (St Louis, MO,USA); Invitrogen (Grand Island, NY, USA) and Millipore (Billerica, MA,USA) with catalog information and other details given in Table 2. A universalstreptavidin-biotin and diaminobenzidine detection system (Pierce) was usedfor colorimetric detection. Secondary antibodies used for immunofluorescencewere goat against mouse IgG Alexa Fluor 568 (Invitrogen, A11004), goatagainst rabbit IgG Alexa Fluor 568 (Invitrogen, A11011), and goat againstmouse Alexa Fluor 488 (Invitrogen, A11001). Immunofluorescence wasimaged using a Nikon A1-R confocal microscope. Cell proliferation wasstudied in histological sections of mouse heart, which were stained using anantibody against phosphorylated histone H3 (phosphorylated HH3) asdescribed previously (Hinton et al., 2006). The proliferation index wascalculated by determining the ratio of positively stained nuclei to the totalnumber of nuclei in the hinge region of the valve. The phosphorylatedSmad2/3 nuclei index was calculated in the same way. Images were analyzedusing NIS-Elements software (Nikon). A total of four to six animals werestudied per genotype per stage.

Electron microscopyUltrastructure analysis was performed using a Hitachi 7600 transmissionelectron microscope (Hitachi Technologies). Aortic valve, ascending aortaand left ventricular myocardium were processed and analyzed as describedpreviously (Hinton et al., 2010). In order to visualize collagen and elasticfibers, sections were counter stained with aqueous solutions of 5% tannicacid followed by 1% uranyl acetate, and then counterstained with leadcitrate.

RESEARCH ARTICLE Disease Models & Mechanisms (2014) doi:10.1242/dmm.015255

Fig. 6. Hypothetical model of complex TGF-β signaling dysregulation inEmilin1-deficient aortic valve tissue. Emilin1 deficiency results in inherentelastic fiber assembly defects and activation of TGF-β signaling. This resultsin downstream activation of both non-canonical (phosphorylated Erk1/2; p-Erk1/2) and canonical (phosphorylated Smad2/3; p-Smad2/3) signaling inEmilin1−/− aortic valve tissue, resulting in early mild fibrosis and aberrantangiogenesis due to early VIC activation and elastic fiber fragmentation(EFF) mediated, in-part, by increased expression of proteolytic enzymes,such as elastases, MMP-2 and MMP-9. Over time (red), there is aprogressive upregulation of phosphorylated Erk1/2 signaling, which results inmyofibroblast activation, neovascularization and severe fibrosis at the agedstage.

Dis

ease

Mod

els

& M

echa

nism

s

Page 9: TGF-βmediates early angiogenesis and latent fibrosis in an ...TGF-βmediates early angiogenesis and latent fibrosis in an Emilin1-deficient mouse model of aortic valve disease Charu

QRT-PCRAortic valves were dissected from adult and aged mouse heart fromEmilin1−/− or Emilin1+/+ mice. Aortic valve tissue from three mice waspooled and RNA was isolated using the standard Trizol method. cDNA wasgenerated using ~500 ng RNA followed by QRT-PCR and amplified by PCRusing gene specific primers (supplementary material Table S1) (Hinton etal., 2010). Ct values were obtained using Bio-Rad software. The ΔΔCtmethod was used to represent the mRNA fold change. Three experimentswere performed for each stage and genotype.

ImmunoblottingAnalyses were performed on aortic valves isolated from five stage- andgenotype-matched mice, as described previously (Givvimani et al., 2012).A bicinchoninic acid protein assay kit (Thermo Scientific, Rockford, IL,USA) was used to estimate total protein, and 30 μg of protein lysate wasloaded onto 8-12% SDS-PAGE gels and then transferred onto nitrocellulosemembranes. The membranes were blocked with 3% non-fat dry milk in TBSTween 20 and incubated with primary antibodies against MMP-2 (1:1000),MMP-9 (1:1000), phosphorylated Smad2/3 (1:300) and phosphorylatedErk1/2 (1:1000) overnight at 4°C. Immunoblots were probed withhorseradish-peroxidase-conjugated secondary antibodies for 1 hour at roomtemperature and developed using chemiluminescence (AmershamBiosciences). After applying a stripping buffer (Thermo Fisher), the blotswere re-probed with antibodies against GAPDH (Santa Cruz Biotechnology)or total Erk1/2 (Cell Signaling). Signal intensity was quantified usingNational Institutes of Health ImageJ software. The arbitrary pixel densitiesof each protein were normalized to GAPDH or total Erk1/2.

EchocardiographyCross-sectional, two-dimensional and color Doppler transthoracicechocardiography was used to assess cardiac structure and function in vivo.Mice were studied at the aged stage. Aortic valve structure and functionwere quantified using a mouse valve protocol that we developed inaccordance with consensus guidelines (Cheitlin et al., 2003; Hinton et al.,2008; Baumgartner et al., 2009). A Visual Sonics Vivo 2100 ImagingSystem (Toronto) with a 30 MHz transducer was used. Mice wereanesthetized with 1-2% iso-fluorane with continuous temperature and heartrate monitoring (Hinton et al., 2008; Hinton et al., 2010). The aortic rootdimension was indexed by calculating the ratio of the aortic root to theascending aorta as described previously (Kemna et al., 2009).

Statistical analysisDescriptive statistics were reported as the mean±s.e.m. Student’s t-test wasused to compare groups using Graph Pad Prism software. A P-value of<0.05 was considered significant.

AcknowledgementsWe thank Osniel Gonzales-Ramos, Varun Krishnamurthy and the Cardiac ImagingCore Research Laboratory for their assistance.

Competing interestsThe authors declare no competing financial interests.

Author contributionsC.M. and R.B.H. developed the concepts and approaches, analyzed data andwrote the manuscript. C.M., A.O., H.O., J.J. performed experiments and analyzeddata. G.B. provided the mouse model studied and analyzed the data. All authorsapproved the final manuscript.

FundingThis work was supported by the Cincinnati Children’s Research Foundation(R.B.H.); the National Institutes of Health (NIH) [HL085122]; and an InstitutionalClinical and Translational Science Award [NIH/NCRR 8UL1TR000077].

Supplementary materialSupplementary material available online athttp://dmm.biologists.org/lookup/suppl/doi:10.1242/dmm.015255/-/DC1

ReferencesAngel, P. M., Nusinow, D., Brown, C. B., Violette, K., Barnett, J. V., Zhang, B.,

Baldwin, H. S. and Caprioli, R. M. (2011). Networked-based characterization ofextracellular matrix proteins from adult mouse pulmonary and aortic valves. J.Proteome Res. 10, 812-823.

Bairati, A. and DeBiasi, S. (1981). Presence of a smooth muscle system in aorticvalve leaflets. Anat. Embryol. (Berl.) 161, 329-340.

Baumgartner, H., Hung, J., Bermejo, J., Chambers, J. B., Evangelista, A., Griffin,B. P., Iung, B., Otto, C. M., Pellikka, P. A., Quiñones, M.; EAE/ASE (2009).Echocardiographic assessment of valve stenosis: EAE/ASE recommendations forclinical practice. Eur. J. Echocardiogr. 10, 1-25.

Braghetta, P., Ferrari, A., de Gemmis, P., Zanetti, M., Volpin, D., Bonaldo, P. andBressan, G. M. (2002). Expression of the EMILIN-1 gene during mousedevelopment. Matrix Biol. 21, 603-609.

Camenisch, T. D., Molin, D. G., Person, A., Runyan, R. B., Gittenberger-de Groot,A. C., McDonald, J. A. and Klewer, S. E. (2002). Temporal and distinct TGFbetaligand requirements during mouse and avian endocardial cushion morphogenesis.Dev. Biol. 248, 170-181.

Cheitlin, M. D., Armstrong, W. F., Aurigemma, G. P., Beller, G. A., Bierman, F. Z.,Davis, J. L., Douglas, P. S., Faxon, D. P., Gillam, L. D., Kimball, T. R. et al.;American College of Cardiology; American Heart Association; AmericanSociety of Echocardiography (2003). ACC/AHA/ASE 2003 guideline update forthe clinical application of echocardiography: summary article: a report of theAmerican College of Cardiology/American Heart Association Task Force on PracticeGuidelines (ACC/AHA/ASE Committee to Update the 1997 Guidelines for theClinical Application of Echocardiography). Circulation 108, 1146-1162.

Combs, M. D. and Yutzey, K. E. (2009). VEGF and RANKL regulation of NFATc1 inheart valve development. Circ. Res. 105, 565-574.

Danussi, C., Petrucco, A., Wassermann, B., Pivetta, E., Modica, T. M., Del BelBelluz, L., Colombatti, A. and Spessotto, P. (2011). EMILIN1-α4/α9 integrin

995

RESEARCH ARTICLE Disease Models & Mechanisms (2014) doi:10.1242/dmm.015255

Table 2. Specification of the primary antibodies used for immunohistochemistryProtein Marker Host Dilution Source Catalogue number Antigen retrieval

αSMA VIC activation Mouse monoclonal 1:250 Sigma-Aldrich ab5694 Hot citrateSMemb VIC activation Mouse monoclonal 1:250 Abcam ab684 Hot citrateSM22 VIC activation Goat polyclonal 1:250 Abcam ab1035 Hot citrateN elastase Elastase Goat polyclonal 1:50 Santa Cruz Biotech sc9518 Hot citrateMMP-2 Elastase Rabbit polyclonal 1:100 Abcam ab7298 Hot citrateMMP-9 Elastase Rabbit polyclonal 1:200 Abcam ab38898 Hot citrateCD-31 Endothelial cell Rabbit polyclonal 1:50 Abcam ab28364 Hot citrateVEGF-R1 Pro-angiogenesis Rabbit polyclonal 1:100 Santa Cruz Biotech sc316 Hot citrateVEGF-R2 Pro-angiogenesis Rabbit polyclonal 1:100 Cell Signaling 55B11 Proteinase KVEGF-A Pro-angiogenesis Rabbit polyclonal 1:100 Santa Cruz Biotech sc152 Hot citrateChondromodulin Anti-angiogenesis Rabbit polyclonal 1:75 Lifespan Bioscience LS-C80668 Hot citrateEndostatin Anti-angiogenesis Rabbit polyclonal 1:100 Abcam ab85661 Hot citrateFibulin-5 Anti-angiogenesis Mouse monoclonal 1:100 Abcam ab66339 Hot citratePhosphorylated Smad TGF-β Rabbit polyclonal 1:100 Cell Signaling 3101 Hot citratePhosphorylated Erk1/2 MAPK Rabbit polyclonal 1:100 Invitrogen 700012 EDTAPhosphorylated HH3 Proliferation Rabbit polyclonal 1:300 Millipore 05-806 Proteinase KKi67 Proliferation Rabbit polyclonal 1:500 Abcam ab15580 Proteinase KCaspase 3 Apoptosis Rabbit polyclonal 1:100 Cell Signaling 9661 Hot citrateMac-3 Macrophage marker Rabbit 1:50 BD Pharmingen 550292 None

Dis

ease

Mod

els

& M

echa

nism

s

Page 10: TGF-βmediates early angiogenesis and latent fibrosis in an ...TGF-βmediates early angiogenesis and latent fibrosis in an Emilin1-deficient mouse model of aortic valve disease Charu

996

interaction inhibits dermal fibroblast and keratinocyte proliferation. J. Cell Biol. 195,131-145.

Danussi, C., Petrucco, A., Wassermann, B., Modica, T. M., Pivetta, E., Del BelBelluz, L., Colombatti, A. and Spessotto, P. (2012). An EMILIN1-negativemicroenvironment promotes tumor cell proliferation and lymph node invasion.Cancer Prev. Res. (Phila.) 5, 1131-1143.

Duran, C. M. and Gunning, A. J. (1968). The vascularization of the heart valves: acomparative study. Cardiovasc. Res. 2, 290-296.

Filip, D. A., Radu, A. and Simionescu, M. (1986). Interstitial cells of the heart valvespossess characteristics similar to smooth muscle cells. Circ. Res. 59, 310-320.

Fondard, O., Detaint, D., Iung, B., Choqueux, C., Adle-Biassette, H., Jarraya, M.,Hvass, U., Couetil, J. P., Henin, D., Michel, J. B. et al. (2005). Extracellular matrixremodelling in human aortic valve disease: the role of matrix metalloproteinases andtheir tissue inhibitors. Eur. Heart J. 26, 1333-1341.

Ghosh, A., DiMusto, P. D., Ehrlichman, L. K., Sadiq, O., McEvoy, B., Futchko, J.S., Henke, P. K., Eliason, J. L. and Upchurch, G. R., Jr (2012). The role ofextracellular signal-related kinase during abdominal aortic aneurysm formation. J.Am. Coll. Surg. 215, 668-680 e661.

Givvimani, S., Munjal, C., Narayanan, N., Aqil, F., Tyagi, G., Metreveli, N. andTyagi, S. C. (2012). Hyperhomocysteinemia decreases intestinal motility leading toconstipation. Am. J. Physiol. 303, G281-G290.

Gu, X. and Masters, K. S. (2009). Role of the MAPK/ERK pathway in valvularinterstitial cell calcification. Am. J. Physiol. 296, H1748-H1757.

Hahn, R. T., Roman, M. J., Mogtader, A. H. and Devereux, R. B. (1992). Associationof aortic dilation with regurgitant, stenotic and functionally normal bicuspid aorticvalves. J. Am. Coll. Cardiol. 19, 283-288.

Hajdu, Z., Romeo, S. J., Fleming, P. A., Markwald, R. R., Visconti, R. P. and Drake,C. J. (2011). Recruitment of bone marrow-derived valve interstitial cells is a normalhomeostatic process. J. Mol. Cell. Cardiol. 51, 955-965.

Hammermeister, K., Sethi, G. K., Henderson, W. G., Grover, F. L., Oprian, C. andRahimtoola, S. H. (2000). Outcomes 15 years after valve replacement with amechanical versus a bioprosthetic valve: final report of the Veterans Affairsrandomized trial. J. Am. Coll. Cardiol. 36, 1152-1158.

Hanada, K., Vermeij, M., Garinis, G. A., de Waard, M. C., Kunen, M. G., Myers, L.,Maas, A., Duncker, D. J., Meijers, C., Dietz, H. C. et al. (2007). Perturbations ofvascular homeostasis and aortic valve abnormalities in fibulin-4 deficient mice. Circ.Res. 100, 738-746.

Hinton, R. B. (2012). Bicuspid aortic valve and thoracic aortic aneurysm: three patientpopulations, two disease phenotypes, and one shared genotype. Cardiol. Res. Pract.2012, 926975.

Hinton, R. B. and Yutzey, K. E. (2011). Heart valve structure and function indevelopment and disease. Annu. Rev. Physiol. 73, 29-46.

Hinton, R. B., Jr, Lincoln, J., Deutsch, G. H., Osinska, H., Manning, P. B., Benson,D. W. and Yutzey, K. E. (2006). Extracellular matrix remodeling and organization indeveloping and diseased aortic valves. Circ. Res. 98, 1431-1438.

Hinton, R. B., Jr, Alfieri, C. M., Witt, S. A., Glascock, B. J., Khoury, P. R., Benson,D. W. and Yutzey, K. E. (2008). Mouse heart valve structure and function:echocardiographic and morphometric analyses from the fetus through the agedadult. Am. J. Physiol. 294, H2480-H2488.

Hinton, R. B., Adelman-Brown, J., Witt, S., Krishnamurthy, V. K., Osinska, H.,Sakthivel, B., James, J. F., Li, D. Y., Narmoneva, D. A., Mecham, R. P. et al.(2010). Elastin haploinsufficiency results in progressive aortic valve malformationand latent valve disease in a mouse model. Circ. Res. 107, 549-557.

Hutcheson, J. D., Ryzhova, L. M., Setola, V. and Merryman, W. D. (2012). 5-HT(2B)antagonism arrests non-canonical TGF-β1-induced valvular myofibroblastdifferentiation. J. Mol. Cell. Cardiol. 53, 707-714.

Hutcheson, J. D., Chen, J., Sewell-Loftin, M. K., Ryzhova, L. M., Fisher, C. I., Su,Y. R. and Merryman, W. D. (2013). Cadherin-11 regulates cell-cell tensionnecessary for calcific nodule formation by valvular myofibroblasts. Arterioscler.Thromb. Vasc. Biol. 33, 114-120.

Keane, J. F., Driscoll, D. J., Gersony, W. M., Hayes, C. J., Kidd, L., O’Fallon, W. M.,Pieroni, D. R., Wolfe, R. R. and Weidman, W. H. (1993). Second natural historystudy of congenital heart defects. Results of treatment of patients with aortic valvarstenosis. Circulation 87 Suppl., I16-I27.

Kemna, M. S., Murphy, D. J. and Silverman, N. H. (2009). Screening for aortic rootdilation in marfan syndrome using the ratio of the aortic root to descending aorticdiameters in children. J. Am. Soc. Echocardiogr. 22, 1109-1113.

Kielty, C. M., Sherratt, M. J. and Shuttleworth, C. A. (2002). Elastic fibres. J. CellSci. 115, 2817-2828.

Krishnamurthy, V. K., Opoka, A. M., Kern, C. B., Guilak, F., Narmoneva, D. A. andHinton, R. B. (2012). Maladaptive matrix remodeling and regional biomechanicaldysfunction in a mouse model of aortic valve disease. Matrix Biol. 31, 197-205.

Li, C., Xu, S. and Gotlieb, A. I. (2011). The response to valve injury. A paradigm tounderstand the pathogenesis of heart valve disease. Cardiovasc. Pathol. 20, 183-190.

Liu, C. and Xi, B. (2012). Pooled analyses of the associations of polymorphisms in theGRK4 and EMILIN1 genes with hypertension risk. Int. J. Med. Sci. 9, 274-279.

Mahler, G. J. and Butcher, J. T. (2011). Inflammatory regulation of valvularremodeling: the good(?), the bad, and the ugly. Int. J. Inflam. 2011, 721419.

Markwald, R. R., Norris, R. A., Moreno-Rodriguez, R. and Levine, R. A. (2010).Developmental basis of adult cardiovascular diseases: valvular heart diseases. Ann.N. Y. Acad. Sci. 1188, 177-183.

Ng, C. M., Cheng, A., Myers, L. A., Martinez-Murillo, F., Jie, C., Bedja, D.,Gabrielson, K. L., Hausladen, J. M., Mecham, R. P., Judge, D. P. et al. (2004).TGF-beta-dependent pathogenesis of mitral valve prolapse in a mouse model ofMarfan syndrome. J. Clin. Invest. 114, 1586-1592.

Nkomo, V. T., Gardin, J. M., Skelton, T. N., Gottdiener, J. S., Scott, C. G. andEnriquez-Sarano, M. (2006). Burden of valvular heart diseases: a population-basedstudy. Lancet 368, 1005-1011.

Paranya, G., Vineberg, S., Dvorin, E., Kaushal, S., Roth, S. J., Rabkin, E., Schoen,F. J. and Bischoff, J. (2001). Aortic valve endothelial cells undergo transforminggrowth factor-beta-mediated and non-transforming growth factor-beta-mediatedtransdifferentiation in vitro. Am. J. Pathol. 159, 1335-1343.

Parks, W. and Mecham, R. (2011). Extracellular Matrix Degradation. Berlin;Heidelberg: Springer.

Paruchuri, S., Yang, J. H., Aikawa, E., Melero-Martin, J. M., Khan, Z. A.,Loukogeorgakis, S., Schoen, F. J. and Bischoff, J. (2006). Human pulmonaryvalve progenitor cells exhibit endothelial/mesenchymal plasticity in response tovascular endothelial growth factor-A and transforming growth factor-beta2. Circ. Res.99, 861-869.

Perrotta, I., Russo, E., Camastra, C., Filice, G., Di Mizio, G., Colosimo, F., Ricci, P.,Tripepi, S., Amorosi, A., Triumbari, F. et al. (2011). New evidence for a critical roleof elastin in calcification of native heart valves: immunohistochemical andultrastructural study with literature review. Histopathology 59, 504-513.

Pivetta, E., Danussi, C., Wassermann, B., Modica, T. M., Del Bel Belluz, L.,Canzonieri, V., Colombatti, A. and Spessotto, P. (2014). Neutrophil elastase-dependent cleavage compromises the tumor suppressor role of EMILIN1. MatrixBiol. 34, 22-32.

Preston, G. A., Zarella, C. S., Pendergraft, W. F., III, Rudolph, E. H., Yang, J. J.,Sekura, S. B., Jennette, J. C. and Falk, R. J. (2002). Novel effects of neutrophil-derived proteinase 3 and elastase on the vascular endothelium involve in vivocleavage of NF-kappaB and proapoptotic changes in JNK, ERK, and p38 MAPKsignaling pathways. J. Am. Soc. Nephrol.13, 2840-2849.

Rabkin-Aikawa, E., Farber, M., Aikawa, M. and Schoen, F. J. (2004). Dynamic andreversible changes of interstitial cell phenotype during remodeling of cardiac valves.J. Heart Valve Dis. 13, 841-847.

Rajamannan, N. M., Nealis, T. B., Subramaniam, M., Pandya, S., Stock, S. R.,Ignatiev, C. I., Sebo, T. J., Rosengart, T. K., Edwards, W. D., McCarthy, P. M. etal. (2005). Calcified rheumatic valve neoangiogenesis is associated with vascularendothelial growth factor expression and osteoblast-like bone formation. Circulation111, 3296-3301.

Rajamannan, N. M., Evans, F. J., Aikawa, E., Grande-Allen, K. J., Demer, L. L.,Heistad, D. D., Simmons, C. A., Masters, K. S., Mathieu, P., O’Brien, K. D. et al.(2011). Calcific aortic valve disease: not simply a degenerative process: A reviewand agenda for research from the National Heart and Lung and Blood Institute AorticStenosis Working Group. Executive summary: Calcific aortic valve disease-2011update. Circulation 124, 1783-1791.

Robinet, A., Fahem, A., Cauchard, J. H., Huet, E., Vincent, L., Lorimier, S.,Antonicelli, F., Soria, C., Crepin, M., Hornebeck, W. et al. (2005). Elastin-derivedpeptides enhance angiogenesis by promoting endothelial cell migration andtubulogenesis through upregulation of MT1-MMP. J. Cell Sci. 118, 343-356.

Schoen, F. J. (1997). Aortic valve structure-function correlations: role of elastic fibersno longer a stretch of the imagination. J. Heart Valve Dis. 6, 1-6.

Schoen, F. J. (2008). Evolving concepts of cardiac valve dynamics: the continuum ofdevelopment, functional structure, pathobiology, and tissue engineering. Circulation118, 1864-1880.

Shen, C., Lu, X., Li, Y., Zhao, Q., Liu, X., Hou, L., Wang, L., Chen, S., Huang, J. andGu, D. (2009). Emilin1 gene and essential hypertension: a two-stage associationstudy in northern Han Chinese population. BMC Med. Genet. 10, 118.

Syväranta, S., Helske, S., Laine, M., Lappalainen, J., Kupari, M., Mäyränpää, M. I.,Lindstedt, K. A. and Kovanen, P. T. (2010). Vascular endothelial growth factor-secreting mast cells and myofibroblasts: a novel self-perpetuating angiogenicpathway in aortic valve stenosis. Arterioscler. Thromb. Vasc. Biol. 30, 1220-1227.

Thubrikar, M. J., Aouad, J. and Nolan, S. P. (1986). Patterns of calcific deposits inoperatively excised stenotic or purely regurgitant aortic valves and their relation tomechanical stress. Am. J. Cardiol. 58, 304-308.

Vesely, I. (1997). The role of elastin in aortic valve mechanics. J. Biomech. 31, 115-123.

Votteler, M., Berrio, D. A., Horke, A., Sabatier, L., Reinhardt, D. P., Nsair, A.,Aikawa, E. and Schenke-Layland, K. (2013). Elastogenesis at the onset of humancardiac valve development. Development 140, 2345-2353.

Wagenseil, J. E. and Mecham, R. P. (2009). Vascular extracellular matrix and arterialmechanics. Physiol. Rev. 89, 957-989.

Walker, G. A., Masters, K. S., Shah, D. N., Anseth, K. S. and Leinwand, L. A.(2004). Valvular myofibroblast activation by transforming growth factor-beta:implications for pathological extracellular matrix remodeling in heart valve disease.Circ. Res. 95, 253-260.

Wirrig, E. E., Hinton, R. B. and Yutzey, K. E. (2011). Differential expression ofcartilage and bone-related proteins in pediatric and adult diseased aortic valves. J.Mol. Cell. Cardiol. 50, 561-569.

Yuan, R., Tsaih, S. W., Petkova, S. B., Marin de Evsikova, C., Xing, S., Marion, M.A., Bogue, M. A., Mills, K. D., Peters, L. L., Bult, C. J. et al. (2009). Aging in inbredstrains of mice: study design and interim report on median lifespans and circulatingIGF1 levels. Aging Cell 8, 277-287.

Zacchigna, L., Vecchione, C., Notte, A., Cordenonsi, M., Dupont, S., Maretto, S.,Cifelli, G., Ferrari, A., Maffei, A., Fabbro, C. et al. (2006). Emilin1 links TGF-betamaturation to blood pressure homeostasis. Cell 124, 929-942.

Zanetti, M., Braghetta, P., Sabatelli, P., Mura, I., Doliana, R., Colombatti, A., Volpin,D., Bonaldo, P. and Bressan, G. M. (2004). EMILIN-1 deficiency induceselastogenesis and vascular cell defects. Mol. Cell. Biol. 24, 638-650.

RESEARCH ARTICLE Disease Models & Mechanisms (2014) doi:10.1242/dmm.015255

Dis

ease

Mod

els

& M

echa

nism

s