REVIEW

Intraluminal valves: development, function and diseaseXin Geng1, Boksik Cha1, Md. Riaj Mahamud1,2 and R. Sathish Srinivasan1,2,*

ABSTRACTThe circulatory system consists of the heart, blood vessels andlymphatic vessels, which function in parallel to provide nutrients andremove waste from the body. Vascular function depends on valves,which regulate unidirectional fluid flow against gravitational andpressure gradients. Severe valve disorders can cause mortality andsome are associated with severe morbidity. Although cardiac valvedefects can be treated by valve replacement surgery, no treatment iscurrently available for valve disorders of the veins and lymphatics.Thus, a better understanding of valves, their development and theprogression of valve disease is warranted. In the past decade,molecules that are important for vascular function in humans havebeen identified, with mouse studies also providing new insights intovalve formation and function. Intriguing similarities have recentlyemerged between the different types of valves concerning theirmolecular identity, architecture and development. Shear stressgenerated by fluid flow has also been shown to regulate endothelialcell identity in valves. Here, we review our current understanding ofvalve development with an emphasis on its mechanobiology andsignificance to human health, and highlight unanswered questionsand translational opportunities.

KEY WORDS: Wnt/β-catenin signaling, Calcific aortic valve disease,Lymphatic vasculature, Mechanobiology, Valves

IntroductionThe survival ofmulticellular organisms depends on the ability of theircells to receive nutrients and to dispose of waste. In vertebrates, twointerconnected vascular networks (blood and lymphatic) meet thisbasic requirement. Arteries (excluding the pulmonary artery) in thesystemic circulation carry oxygenated blood from the heart to tissuesand organs, and additionally provide them with glucose and othernutrients derived from the small intestine. These arteries undergo aseries of branching events to generate small-diameter capillariescalled arterioles. Owing to high hydrostatic pressure and low oncoticpressure (see Box 1 for a glossary of terms), arterioles release waterand other small molecules, such as glucose and albumin, into theinterstitial space, which are then taken up by cells. Cells release wastematerials, such as lactic acid and carbonic acid, into the interstitialspace, which are mostly taken up by venules, very small veins thatprogressively fuse to form larger veins that return deoxygenatedblood to the heart (Breslin, 2014; Levick andMichel, 2010;Wiig andSwartz, 2012). Low hydrostatic pressure and high oncotic pressureinside the venules is important for interstitial fluid reabsorption.

Approximately 10% of the interstitial fluid is left behind by thevenules, which is absorbed by lymphatic vessels and returned to thevenous blood circulation. Defects in this fluid transportation systemcan lead to stagnation of blood or interstitial fluid, resulting invascular thrombosis or edema, respectively (Box 1).

In mammals, four main types of valve form to regulate theunidirectional flow of fluid in different organs – cardiac valves (inthe heart), venous valves (VVs; in veins), lymphatic valves (LVs; inthe lymphatic vessels) and lymphovenous valves (LVVs; at the siteswhere lymph is returned to blood circulation) (see below). When thedevelopment or functioning of these valves fail or deteriorate, it canresult in morbidity and death. Thus, new therapeutic approaches toprevent valve deterioration are urgently needed.

Here, we provide an overview of how valves contribute to healthand disease, and review recent findings that highlight the similaritiesthat exist between aortic valves (a type of cardiac valve) andvascular valves (LVs, VVs and LVVs). We discuss the mechanismsby which shear stress regulates these commonalities. We proposethat, by exploring these similarities, we may be able to uncoverthe mechanisms that govern the central nature of valves. Finally, wespeculate on how these findings could provide opportunities todiagnose and treat valve disorders.

Mammalian valves: an overviewIn this section we provide an overview of the main types ofmammalian valves and the diseases that are caused by defects inthese structures.

Cardiac valvesThe mammalian heart has four valves: the aortic and pulmonicvalves (known as the semilunar valves), and the mitral and tricuspidvalves (known as the atrioventricular valves). Defects in any one ofthese valves can have serious consequences (Kim and Ruckdeschel,2016; Nkomo et al., 2006; Shah and Raney, 2008). Calcification is apathological condition that causes stiffening (stenosis; Box 1) of thecardiac valves due to changes in the composition of the extracellularmatrix (ECM). The aortic valve is the most prone to this disorder.Calcific aortic valve disease (CAVD) is the most common valvedisease in the developed world, where an estimated 2- to 4-millionpeople suffer with this disease (Yutzey et al., 2014). Approximately15,000 deaths per year are attributed to CAVD in North America,and this number is expected to increase rapidly owing to the agingpopulation and the lack of prevention strategies (Lindman et al.,2016). Calcified aortic valves cannot open fully during systole orclose fully during diastole (Box 1) (Demer and Tintut, 2008; Yutzeyet al., 2014). Aortic valve stenosis causes the left ventricle to workharder to meet the metabolic demands of the body, resulting in left-ventricular hypertrophy, which increases the risk of heart attack andstroke (Lindman et al., 2016).

Prosthetic valves have significantly reduced the mortality associatedwith cardiac valve disorders. However, patients need long-termtreatment with blood thinners to prevent clot formation on the valves,which are associated with uncontrolled bleeding, which itself could

1Cardiovascular Biology Research Program, Oklahoma Medical ResearchFoundation, Oklahoma City, OK 73104, USA. 2Department of Cell Biology,University of Oklahoma Health Sciences Center, Oklahoma City, OK 73104, USA.

*Author for correspondence (sathish-srinivasan@omrf.org)

R.S.S., 0000-0002-4465-3340

This is an Open Access article distributed under the terms of the Creative Commons AttributionLicense (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.

1273

© 2017. Published by The Company of Biologists Ltd | Disease Models & Mechanisms (2017) 10, 1273-1287 doi:10.1242/dmm.030825

Disea

seModels&Mechan

isms

cause severe morbidity (Dangas et al., 2016; Nishimura et al., 2017).Additionally, prosthetic valves have a limited lifespan and might needreplacement or readjustment with time (Nishimura et al., 2017). Hence,research is ongoing to better understand aortic valve disease and todevelop approaches that will stop/slow its progression.

Venous valvesOwing to low intraluminal pressure (Box 1), blood transport withinthe veins (frequently against gravitational pressure) depends on

VVs, the degeneration of which wreaks havoc on normal vascularphysiology. Defective VVs cause primary chronic venousinsufficiency (Box 1), which leads to elevated venous pressure,edema and pooling of the blood. According to a Scottish study,around 6-10% of the general population, and >20% of those over50 years old, are estimated to have some form of chronic venousinsufficiency (Ruckley et al., 2002; Weber et al., 2016). In the US,chronic venous insufficiency is the seventh leading cause of chronicdebilitating disease (Meissner et al., 2007b;Weber et al., 2016). Theseverity of this disease could be mild (spider veins), moderate(varicose veins) or severe (edema, venous eczema and venousulcers) (Box 1) (Meissner et al., 2007b). The most serious forms ofchronic venous insufficiency can cause necrotic ulcers that mayrequire limb amputations (Tsai et al., 2005).

Owing to their architecture, the downstream side (behind thevalves with respect to the direction of blood flow) of VVs are proneto blood stasis (Brooks et al., 2009). Primary chronic venousinsufficiency causes increased blood pooling within the veins,which results in hypoxia, clotting and endothelial cell inflammation,otherwise known as deep vein thrombosis (DVT). Complications ofDVT include pulmonary embolism (Box 1), due to clots thatdislodge from VVs and migrate to the lungs, and secondary chronicvenous insufficiency, due to inflammation permanently damagingVVs (Meissner et al., 2007a). Every year, an estimated 2-3individuals per 10,000 will develop DVT in the developed world(Fowkes et al., 2003). In those older than 60, this number increasesto nearly 10 in 10,000. Venous malformations, long-distanceflights, sedentary lifestyle and hip or pelvic surgery increase thechances of developing DVT (Kyrle and Eichinger, 2005).

Blood thinners, which can cause bleeding, are commonly used toprevent and treat DVT (Hirsh and Hoak, 1996). Venousinsufficiency in peripheral veins [such as saphenous veins (thetwo main veins in the leg)] is treated by laser ablation of the vein. Incontrast, there is no treatment for insufficiency within central veins[such as the iliac vein (in the pelvis) or inferior vena cava (runningbehind the abdominal cavity to the heart)]. Valve replacementtherapy is currently being explored for central veins and mightbecome available in the future (Weber et al., 2016). VV allografts(the transplantation of valves from living donors or cadavers) areshowing promising results in clinical and pre-clinical trials(Burkhart et al., 1997; Dalsing et al., 1999).

Lymphatic and lymphovenous valvesLymph is collected by lymphatic capillaries and transported viacollecting lymphatic vessels (Davis et al., 2012). LVs within thecollecting lymphatic vessels regulate unidirectional lymph flow.Ultimately, lymph collected from the body returns to the bloodcirculation via four LVVs, which are bilaterally located at thejunction of the jugular and subclavian veins, in the neck (Geng et al.,2016; Srinivasan and Oliver, 2011). Unlike cardiac valves and VVs,the functioning of which can be imaged by color Doppler imaging,it is not currently possible to non-invasively image and quantify LVfunctioning in humans (Mellor et al., 2011). As a result, thesignificance of LV defects to human lymphedema (Box 1) is notfully understood. However, a recently developed, non-invasiveultrasound approach to imaging LVVs might yield quantitativeevidence regarding the importance of LVVs to lymphatic vascularphysiology (Seeger et al., 2009).

Several mouse models that have defective LVs or LVVs exhibitlymph reflux, lymphedema or chylothorax (Box 1) (Bazigou et al.,2009; Geng et al., 2016; Kanady et al., 2011, 2015; Kazenwadelet al., 2015; Kriederman et al., 2003; Martin-Almedina et al., 2016;

Box 1. GlossaryBicuspid aortic valve disease: condition in which people are born withonly two flaps in the aortic valve rather than the usual three; occurs in∼1% of humans.Calcification: a pathological condition of the cardiac valves in which thevalvular interstitial cells acquire bone-cell-like characteristics. The valvesbecome rigid, resulting in their inability to open and close properly.Cardiac cushion: specialized cells within the endocardial tube in thedeveloping heart that generate the valves and septum (the wall dividingthe left and right sides of the heart).Chylothorax: accumulation of lymph in the thoracic cavity (whichencapsulates the heart and lungs) due to defective or damagedlymphatic vessels.Chylous ascites: accumulation of lymph in the abdominal cavity due todefective or damaged lymphatic vessels.Diastole: the phase of the heartbeat during which the heart musclesrelax.Edema: accumulation of water and molecules of small molecular mass,resulting in the swelling of the interstitial tissue.Emberger syndrome: a disease caused by mutations in the zinc-fingertranscription factor GATA2. Symptoms include lymphedema andleukemia.Embolism: blockage of blood vessels by an object. When vessels areblocked by a blood clot, the condition is called thromboembolism.Hydrops fetalis: a condition caused by the accumulation of fluid indeveloping embryos.Hydrostatic pressure: pressure exerted on the walls of blood orlymphatic vessels by the weight of the fluid.Intraluminal pressure: pressure exerted on the walls of blood orlymphatic vessels by the weight of the fluid and the pressure generatedby fluid flow.Laminar shear stress: frictional force experienced by the endothelialcells due to the orderly flow of blood or lymph. Straight portions of bloodor lymphatic vessels experience such a flow pattern.Lymphedema: edema caused by defective lymphatic vascularfunctioning.Lymphedema-distichiasis syndrome: a genetic disorder caused bymutations in the forkhead-domain transcription factor FOXC2.Symptoms include lymphedema, two rows of eyelashes and, rarely,heart defects.Oncotic pressure: pressure caused by a higher protein concentrationwithin blood capillaries than in the extracellular milieu. Higher oncoticpressure pulls water into the capillaries.Oscillatory shear stress: frictional force experienced by the endothelialcells due to the disorderly flow of blood or lymph. Branch points of bloodor lymphatic vessels experience such a flow pattern.Stenosis: an abnormal narrowing of valves due to their inability to opencompletely.Systole: the phase of the heartbeat during which the heart musclescontract.Valve vegetations: abnormal growth of clots in the cardiac valves. Suchvegetations could be caused by bacterial infections, inflammation ormetastatic tumors.Venous insufficiency: stagnation of blood within the veins in the lowerextremities due to the inefficient functioning of venous valves.Venous ulcers: wounds caused by stagnation of blood within the veins.Such wounds normally occur on the lower legs.

1274

REVIEW Disease Models & Mechanisms (2017) 10, 1273-1287 doi:10.1242/dmm.030825

Disea

seModels&Mechan

isms

Munger et al., 2017; Petrova et al., 2004). In one study, the severityof LV defects in the thoracic duct of a mouse model [connexin-37(Cx)37−/−;Cx43+/−; see further details in Vascular valve disorders,below] was shown to correlate with the onset of chylothoraxand postnatal death (Kanady et al., 2011). Additionally,overexpression of vascular endothelial growth factor C (VEGFC)in adipocytes resulted in the incompetence of LVs (likely due tothe dilation of lymphatic vessels) and in chylothorax (Nitschkeet al., 2017). These reports validate the importance of LVs andLVVs in lymphatic vascular physiology. According to currentmodels, defective valves cause lymph to stagnate in the collectingvessels; the increased pressure is transmitted upstream to thecapillaries, inhibiting lymph uptake and exacerbating lymphaticvascular defects, such as lymphedema and chylothorax (Daviset al., 2012).In summary, defects in the cardiac valve, VVs, LVs or LVVs

could cause severe morbidity or mortality. However, the etiology of

valve disorders is not fully understood. In the following section, wediscuss the current understanding that underscores the involvementof genetic factors in the initiation and progression of these diseases.

Genetic contributions to valve disordersCalcific aortic valve diseaseAge, hypertension, kidney disease, high dietary fat, smoking and asedentary lifestyle are implicated in CAVD (Yutzey et al., 2014). Inaddition, polymorphisms in genes that regulate lipid metabolism,inflammation and bone formation have been reported in patientswith CAVD (Guauque-Olarte et al., 2015; Thanassoulis et al., 2013;Yutzey et al., 2014). For example, heterozygous mutations inNOTCH1 cause bicuspid aortic valve disease (Box 1) andindividuals with these mutations are prone to develop CAVD(Garg, 2006; Garg et al., 2005; LaHaye et al., 2014) (Table 1).Notch signaling is thought to antagonize the differentiation of aorticvalve cells into bone-like cells (Acharya et al., 2011).

Table 1. Genes associated with human valve disorders

Effect of gene disruption

Gene Gene function AoV LVV VV LV References

NOTCH1 Receptor forNotchsignaling

BAV and CAVDin humansand mousemodels

ND ND ND in humans; abnormalstructure in mouse models

Garg et al., 2005;Murtomaki et al., 2014

FOXC2 Transcriptionfactor

ND ND in humans; absenceof LVVs in Foxc2−/−

mice and reducednumbers in Foxc2+/−

mice

Valve failure and varicoseveins in humans;absence of VVs incentral veins ofFoxc2−/− mouseembryos

Primary lymphedema inhumans but valve defectsND; absence of LV inFoxc2−/− mice; reducednumber of valves andretrograde lymph flow inFoxc2+/− mice

Bell et al., 2001; Brice et al.,2002; Fang et al., 2000;Geng et al., 2016;Kriederman et al., 2003;Lyons et al., 2017; Melloret al., 2007, 2011; Mungeret al., 2016; Petrova et al.,2004

GATA2 Transcriptionfactor

ND ND in humans; absenceof LVVs in mousemodels

DVT in humans; ND inmice

Primary lymphedema inhumans but valve defectsND; absence of valves inmouse models

Geng et al., 2016;Kazenwadel et al., 2015,2012; Ostergaard et al.,2011; Spinner et al., 2014

CX43 Gap-junctionmolecule

ND ND Fewer VVs in humans;defective VVs in centralveins of Cx43−/−

mouse embryos

Primary lymphedema inhumans but valve defectsND; absence of LVs inCx43−/− mouse embryos;fewer and dysfunctional LVsin adult Lyve1-Cre;Cx43F/F

mice

Brice et al., 2013; Kanadyet al., 2011; Lyons et al.,2017; Munger et al., 2017,2016

CX47 Gap-junctionmolecule

ND ND in humans; normalLVVs in a mousemodel

Fewer VVs and thosepresent have shorterleaflets in humans;absence of peripheralVV in Cx47−/− mice

Primary lymphedema inhumans but valve defectsND; normal LVs in a mousemodel

Ferrell et al., 2010; Lyonset al., 2017; Munger et al.,2016

ITGA9 Cell-ECMinteraction

ND ND in humans; normalLVVs in a mousemodel

ND in humans;morphogenesisdefects in mouseItga9−/− mice

Congenital chylothorax inhumans but valve defectsND; morphogenesis defectsin Itga9−/− mice

Bazigou et al., 2011, 2009;Huang et al., 2000; Maet al., 2008

EPHB4 Receptortyrosinekinase

ND Hydrops fetalis inhumans but valvedefects ND;morphogenesisdefects in mousemodels

ND ND in humans; morphogenesisdefects in mouse models

Martin-Almedina et al.,2016; Zhang et al., 2015

RASA1 Inhibitor of Rassignaling

ND ND ND Primary lymphedema inhumans but valve defectsND; absence of LVs due toapoptosis in mice

Burrows et al., 2013;Lapinski et al., 2017

AoV, aortic valve; BAV, bicuspid aortic valve; CAVD, calcific aortic valve disease; DVT, deep vein thrombosis; LV, lymphatic valve; LVV, lymphovenous valve;ND, not defined; VV, venous valve.

1275

REVIEW Disease Models & Mechanisms (2017) 10, 1273-1287 doi:10.1242/dmm.030825

Disea

seModels&Mechan

isms

Vascular valve disordersVaricose veins and chronic venous insufficiency have a strongfamilial association (Krysa et al., 2012), but only a few genes havebeen linked to these diseases. In contrast, numerous gene mutationsare associated with lymphedema or chylothorax in human patients(Brouillard et al., 2014). We discuss here only those genes that areknown to be critical for vascular valve development in mousemodels (Table 1).

FOXC2Dominant heterozygous mutations in the forkhead familytranscription factor FOXC2 are associated with humanlymphedema, venous insufficiency and varicose veins (Bell et al.,2001; Brice et al., 2002; Fang et al., 2000; Lyons et al., 2017;Melloret al., 2007, 2011). In patients with venous insufficiency andvaricose veins, VVs function abnormally and result in a retrogradeblood flow pattern, as revealed by Doppler imaging (Brice et al.,2002; Mellor et al., 2007). Recently, lymphedema patients withFOXC2mutations were also found to possess fewer VVs, and thosepresent had shorter leaflets (Lyons et al., 2017).Foxc2+/− mice, which are models for lymphedema-distichiasis

syndrome (Box 1), display retrograde lymph flow (Kriederman et al.,2003), possibly because they have incompetent LVs. Foxc2+/−

embryos have 50% fewer LVs in their dorsal skin (Kanady et al.,2015); whether the remaining LVs are functional remains unknown.Interestingly, LV numbers are not reduced in the mesentery ofFoxc2+/− embryos, implying a tissue-specific requirement for Foxc2dosage during LV development (Kanady et al., 2015). Loss of oneFoxc2 allele causes a variable LVVphenotype inmice; whereas someFoxc2+/− embryos have no LVVs and develop severe edema, othershave one LVV at the junction of the jugular and subclavian veins(instead of the normal two LVVs) (Geng et al., 2016).

GATA2Dominant heterozygous mutations in the zinc-finger transcriptionfactor GATA2 are found in individuals with Emberger syndrome(Box 1) (Ostergaard et al., 2011). Approximately 30% of patientscarrying mutations in GATA2 develop lymphedema (n=14 from 8families). Another study reported lymphedema with incompletepenetrance in 3 out of 10 patients withGATA2mutations (Kazenwadelet al., 2012). In another cohort of 57 patients with GATA2 mutations,11% developed lymphedema (Spinner et al., 2014).A total of 25% of patients with GATA2 mutations in the above

cohort developed thrombotic events, such as DVT, pulmonaryembolism, portal vein thrombosis and catheter-related thrombosis(Spinner et al., 2014), possibly because GATA2 deficiency in theendothelium causes coagulopathy. The VVs of these patients werenot analyzed.Mice lacking Gata2 in vascular endothelial cells lack LVVs and

LVs (Geng et al., 2016; Kazenwadel et al., 2015). Whether GATA2is necessary for VV development is currently unknown.

CX43Mutations in the gap-junction molecule CX43 are observed inhuman lymphedema patients (Brice et al., 2013). Cx43−/− mouseembryos have no mesenteric LVs (Kanady et al., 2011). Theconditional deletion of Cx43 in the lymphatic vasculature usingLyve1-Cre results in the formation of fewer and dysfunctional LVsin postnatal and adult mice (Munger et al., 2017).CX43 is also important for VV development. The central VVs are

either absent or defective inCx43−/− embryos (Munger et al., 2016).Whether peripheral VVs require CX43 for their formation is

currently unknown because Cx43−/− mice die soon after birth andperipheral VVs develop postnatally (Bazigou et al., 2011; Mungeret al., 2016; Reaume et al., 1995). Mice with a conditional deletionof Cx43 in endothelial cells, using Tie2-Cre, are viable (Liao et al.,2001; Theis et al., 2001). Whether these mice possess peripheralVVs remains to be investigated.

CX47Mutations in CX47 are observed in human lymphedema patients(Ferrell et al., 2010; Lyons et al., 2017), who have fewer and shorterVVs (Lyons et al., 2017).Cx47−/−mice lack VVs in most peripheralveins, but possess LVVs, LVs and VVs in central veins (such asthose located in the jugular vein) (Munger et al., 2016). Whencompared with Cx43−/− littermates, double-null mutants for Cx47and Cx43 have more severe VV defects in the central veins,indicating that these two connexins may have overlapping roles inVV development. Interestingly, Cx47 expression is downregulatedin the LVs of Lyve1-Cre;Cx43f/f embryos, indicating that Cx47expression might also depend on CX43 (Munger et al., 2016).

ITGA9Mutations in integrin-α9 (ITGA9) were identified in human fetuseswith congenital chylothorax (Ma et al., 2008). Itga9−/− micerecapitulated this phenotype and died soon after birth withchylothorax (Huang et al., 2000). These embryos lack LVs andVVs (Bazigou et al., 2011, 2009), but LVVs appear to developnormally in these mutants (Hess et al., 2014).

EPHB4Autosomal dominant mutations in the receptor tyrosine kinaseephrin type-B receptor 4 (EPHB4) were identified in families with ahistory of lymphatic-related hydrops fetalis (Box 1) (Martin-Almedina et al., 2016). Morphogenesis of LVVs is defective inmice lacking EphB4 (Martin-Almedina et al., 2016). Usingfunction-blocking antibodies, it was reported that EPHB4 isnecessary for the development of LVs (Zhang et al., 2015).

RASA1Loss-of-function mutations in the Ras GTPase RASA1 belong to alarge group of diseases known as the rasopathies, in which theRas signaling pathway is hyperactivated (Brouillard et al., 2014).A subset of patients with rasopathies display lymphedema,chylothorax or chylous ascites with variable penetrance (Box 1)(Brouillard et al., 2014). RASA1 mutations are associated withlymphedema in a subset of carriers (Burrows et al., 2013). Deletionof Rasa1 from adult mice results in the progressive deterioration ofLV function due to apoptotic cell death of valvular endothelial cells(Lapinski et al., 2017).

In summary, mutations in certain genes appear to predisposeindividuals to cardiac and vascular valve disorders (Table 1). A betterunderstanding of their actionmight provide uswith new opportunitiesfor clinical intervention. Therefore, we next discuss the mechanismsof valve development and subsequently the roles of the above-mentioned genes in this developmental process.

The stepwise development of valvesCardiac valvesThe heart undergoes complex morphogenesis during developmentand these changes are intimately connected to the formation of cardiacvalves. In this Review, we provide a simplistic model of heart valvedevelopment in the mouse and we refer readers to more in-depthreviews for additional details (Lin et al., 2012; Person et al., 2005).

1276

REVIEW Disease Models & Mechanisms (2017) 10, 1273-1287 doi:10.1242/dmm.030825

Disea

seModels&Mechan

isms

The development of the atrioventricular valves and semilunarvalves starts at approximately embryonic day (E)9.5 in mouseembryos (Fig. 1A). At this stage, the heart is a simple tubeconsisting of the inner endocardial layer, composed of endothelialcells, and an outer myocardial layer. Signals from the myocardialcells trigger the endothelial cells to differentiate into mesenchymal(migratory) cells that delaminate into the space between theendocardial and myocardial layers and proliferate (Figs 1A,B and 2C).This process is known as the endothelial-to-mesenchymal transition(EndMT). The mesenchymal cells also secrete ECM componentsand the resulting structure is known as the endocardial cushion (alsoknown as the cardiac cushion, CC; Fig. 1, Box 1) (Markwald et al.,1975). Notch, Wnt/β-catenin and TGFβ signaling pathwayspromote EndMT during CC formation (Armstrong and Bischoff,2004). In contrast, VEGF inhibits EndMT in the ‘non-valve’ regionsof the endocardium. In the prospective valve region, thetranscription factor nuclear factor of activated T cells 1 (NFATC1)inhibits the expression of vascular endothelial growth factor(VEGF), thereby promoting EndMT (Chang et al., 2004).NFATC1 also terminates EndMT by inhibiting the expression ofgenes such as Snail1 and Snail2 (Wu et al., 2013).Once CC formation is complete, at E12.5, the mesenchymal cells

and the endocardial cells undergo coordinated morphogenesis toform the valve leaflets (Fig. 1C) (Wirrig and Yutzey, 2014).NFATC1 regulates the recruitment of neural crest cells for theelongation of cardiac valve leaflets (Wu et al., 2013). The semilunarand the tricuspid valves form three leaflets each, whereas the mitralvalve forms two. Once the valve leaflets are formed, themesenchymal cells differentiate into valvular interstitial cells inthree layers (fibrosa, spongiosa and ventricularis) with distincttensile properties (Fig. 2A) (Hinton and Yutzey, 2011; Markwaldet al., 1975). These cell layers cause the cardiac valves to bend andbecome streamlined with respect to blood flow (Figs 1C and 2A).The presence of chordae tendineae (CT) distinguishes theatrioventricular valves from the semilunar valves (Fig. 1D). These

fibrous cords connect the downstream side of the atrioventricularvalves to the papillary muscles on the inner walls of the ventricleand are important for closing the valves during systole (Hinton andYutzey, 2011).

At the end of morphogenesis, each leaflet of every cardiac valvehas an upstream side that directly faces the incoming blood and adownstream side that is exposed to the outflowing blood (Figs 1Dand 2A). A single layer of endothelial cells surrounds the CC.Owing to its clinical significance, most of the molecular detailsknown about cardiac valve endothelial cells come from the aorticvalve (Mongkoldhumrongkul et al., 2016), and these details arepertinent to the other valves. Evidence from dogs indicates that theupstream and downstream endothelial cells of the semilunar valvesalign circumferentially along the rim of the valve (Deck, 1986)(Fig. 2B). The author hypothesizes that the perpendicular alignmentof valvular endothelial cells might optimize the endothelial cellresponse to the backflow pressure that forms during diastole.Despite their structural similarity, the endothelial cells on theupstream and downstream sides of the aortic valve have distinctmolecular characteristics (Fig. 2A). For example, endothelial cellson the upstream side of the aortic valves express CX43 and KLF2(Inai et al., 2004; Lee et al., 2006; Simmons et al., 2005), whereasendothelial cells on the downstream side of the aortic valvesexpress TIE1, FGFR2, BMP4, PECAM1, VCAM1 and P-selectin(Porat et al., 2004; Simmons et al., 2005). Ephrin-B2, CX37,β-catenin, PROX1, FOXC2 and GATA2 are also expressed inendothelial cells on the downstream side of the aortic valves (Chaet al., 2016; Cowan et al., 2004; Inai et al., 2004; Kazenwadelet al., 2015; Simmons et al., 2005; Srinivasan and Oliver, 2011).Owing to the expression of pro-inflammatory molecules such asVCAM1 and P-selectin, the endothelial cells on the downstreamside could be permissive to valvular calcification (Simmons et al.,2005). However, they also express genes such as endothelial nitricoxide synthase (eNOS) that could protect valves against thispathology (Simmons et al., 2005).

E9.5 E12.5 E14.5

CT

CC CC

Birth

A Cardiac looping B EndMT C Proliferation D Remodeling

AoV

AVV

Myocardium Cardiac jelly Mesenchymal cellsEndocardiumKey

Fig. 1. Stepwise development of cardiac valves in mice. (A) The heart initially forms as a simple tube with an inner endocardial (blue) and outer myocardial(red) layer. The atrium (‘A’) is yet to separate into the left and right chambers by septation, and the ventricle is a single chamber [the future left (LV) and right (RV)ventricles are depicted]. The outflow tract (OFT) has yet to separate into the pulmonary artery and dorsal aorta. The cardiac cushions (CCs) form at theatrioventricular (right box) and RV-OFT (left box) junctions. Mitral and tricuspid valves develop from the atrioventricular cushion and are collectively known as theatrioventricular valves (AVVs). The aortic (AoV) and pulmonic valves develop from the RV-OFT cushion and are collectively known as semilunar valves. Forsimplicity, the development of an AoV and an AVV is shown in B-D. (B) During endothelial-to-mesenchymal transition (EndMT), the endocardial cells gain amesenchymal-cell-like characteristic and migrate into the CC. These cells are known as valvular interstitial cells (orange cells). (C) Valvular interstitial cellsproliferate and secrete ECM proteins, which results in the bending of primitive valves along the direction of blood flow (arrows). (D) Valves undergo furtherremodeling as they mature. Fibrous chordae tendineae (CT), which distinguish AVVs from semilunar valves, attach the AVVs to the ventricular wall. Panel Awasadapted with permission from Person et al. (2005); panels B-D were adapted with permission from Lin et al. (2012).

1277

REVIEW Disease Models & Mechanisms (2017) 10, 1273-1287 doi:10.1242/dmm.030825

Disea

seModels&Mechan

isms

Lymphovenous valvesLVVs are the first valves to form outside of the heart. In mice, they startforming at E12 (Geng et al., 2016; Srinivasan and Oliver, 2011) at thejunction of the jugular and subclavian veins (Fig. 3A,B). LVVdevelopment begins with the differentiation of two cell types – thelymphatic endothelial cells (LECs) and LVV-forming endothelial cells(LVV-ECs) – that interact to form the LVVs. These cells upregulate the

expression of the homeobox transcription factor PROX1 where thelymph sac (primitive lymphatic vessel) touches the jugulo-subclavianvein junction. Nevertheless, differences exist between these two celltypes. Expression of the receptor tyrosine kinase vascular endothelialgrowth factor receptor 3 (VEGFR3) and CX43 is upregulated by LECs(Geng et al., 2016; Munger et al., 2016). In contrast, LVV-ECsupregulate the expression of FOXC2, GATA2 and CX37 (Geng et al.,

CX37; EFNB2; GATA2; FOXC2; PROX1; TIE1

CX37; ITGA9; GATA2; FOXC2; PROX1; ITGA9*; ITGA5*

CX37; ITGA9; EFNB2; FOXC2; PROX1; ITGA9*; ITGA5*

*

10 µm10 µm

Dow

nstre

am s

ide

of m

atur

e va

lves

Initi

atio

n of

val

vede

velo

pmen

t

AoV LVV LV and VVD G

CX43; KLF2 CX43; PROX1; VEGFR3; ITGA9*; ITGA5*

CX43; CX47; PROX1; ITGA9*; ITGA5*

B E H

C F I

A

CX37CC ; ITGA9; GATA2;FOXC2; PROX1; ITGA9*; ITGA5*

CX37CC ; ITGA9; EFNB2;FOXC2; PROX1;ITGA9*; ITGA5*

D G

CX43; KLF2

CX37CC ; EFNB2; GATA2; FOXC2; PROX1; TIE1

CX43; PROX1; VEGFR3; ITGA9*; ITGA5*

CX43; CX47; PROX1; ITGA9*; ITGA5*

AA

Fig. 2. Structural and molecular features of intraluminal valves. (A) Schematic of a single leaflet of the aortic valve (AoV). Ao represents the aorta, Mrepresents the myocardial layer of the left ventricle and A represents the annulus, which is a fibrous ring that supports the AoV. The small arrows point to the threeinner layers of the valve: ventricularis (V), spongiosa (S) and fibrosa (F). The orange stars represent the ECM-producing activated valvular interstitial cells. Thenucleated cells are the endothelial cells of the AoV. The thick arrow indicates the direction of blood flow from the left ventricle to the aorta. The endothelialcells directly facing the blood flow (upstream side) are in red and their expression profile is presented in the red box. The endothelial cells on the downstream sideof the AoV are in yellow; their expression profile is presented in the light green box. (B) Scanning electron micrograph (SEM) of the downstream side of theAoV of a dog, showing the elongated morphology of the endothelial cells. Image reproduced with permission from Deck (1986). The sample is approximately270 µm in length (left to right). (C) SEM of endothelial-to-mesenchymal transition (EndMT) occurring within a developing aortic valve of rats. Endocardial cells(E, magenta) give rise to ECM-producing valvular interstitial cells (*). Gaps are observed between the endocardial cells (arrow). The picture is a 2600×magnification of the sample. Image reproduced with permission from Markwald et al. (1975). (D) Schematic of a lymphovenous valve (LVV). The arrow indicatesthe direction of lymph flow. Green cells represent the lymphatic endothelial cells (LECs) of the lymph sac, blue cells represent the venous endothelial cells, redcells the LVV-forming endothelial cells (LVV-ECs), yellow cells the specialized LECs on the upstream side of LVVs, and orange cells, mural cells that lie inbetween the upstream and downstream sides of LVVs. The expression profiles of LVV-ECs on the downstream side of LVVs and the LECs on the upstream side ofLVVs are presented in the red and light green boxes, respectively. (E) SEM of the downstream side of a mature LVV from a newborn mouse pup, showing theelongated architecture of LVV-ECs (arrows). The asterisk shows the opening through which lymph is drained from the lymph sac (located behind the plane of thisimage) into the veins. Image reproduced with permission from Geng et al. (2016). (F) SEM of LVV-ECs (magenta) delaminating into the lumen of the embryonicveins in an E12.0mouse embryo. The cells pile on top of each other and form filopodia-like projections. Image reproducedwith permission fromGeng et al. (2016).(G) Schematic of a lymphatic (LV) or a venous (VV) valve. The arrow represents the direction of fluid flow. The endothelial cells of the vessel are represented ingreen, the endothelial cells on the upstream and downstream sides of the valve in red and yellow, respectively. The expression profiles of upstream anddownstream cells are presented in the red and light green boxes, respectively. (H) SEM of a mature VV located at the opening of the external jugular vein in anewborn mouse pup. Notice the elongated valvular endothelial cells along the rim of the valve (arrows). Image reproduced with permission from Geng et al.(2016). (I) SEM of valvular endothelial cells (magenta) from an E14.5 mouse embryo, which migrate in a ‘knitting-like’ manner during VV morphogenesis in thejugular vein. Image reproduced with permission from Geng et al. (2016).

1278

REVIEW Disease Models & Mechanisms (2017) 10, 1273-1287 doi:10.1242/dmm.030825

Disea

seModels&Mechan

isms

2016; Munger et al., 2016). Additionally, integrins (α9 and α5)mediate LVV-EC–ECM and LEC-ECM interaction in the LVVs(Geng et al., 2016; Turner et al., 2014).Immediately after differentiation, the LVV-ECs delaminate from

the vein walls in the luminal orientation (Fig. 2F) (Geng et al.,2016). This process is reminiscent of the EndMT that occurs duringCC formation. However, in contrast to LVV-EC delamination,endocardial cells delaminate in the abluminal direction, away fromthe blood flow during CC formation. LVV-ECs appear to protrudeout of the venous endothelial cell layer during their delamination(Geng et al., 2016) and can withstand the force of blood flow, likelydue to their association with the ECM.The delaminated LVV-ECs dramatically elongate, reaggregate

and align perpendicular to the direction of blood flow (Geng et al.,2016) (Fig. 3C), and concurrently pile up on top of each other andinvaginate into the veins. An opening is created in the middle of this

pile, establishing a connection between the venous and lymphaticvasculatures (Fig. 2E, asterisk) (Geng et al., 2016).

Finally, the valves recruit a fewmural cells into the space betweenthe valvular endothelial cells and the LECs (Fig. 3D) (Geng et al.,2016). At the end of the morphogenetic process, endothelial cells onboth the upstream and downstream sides of LVVs are elongated andaligned perpendicular to the direction of both the venous blood flowand the lymph flow (Fig. 2E) (Geng et al., 2016).

Venous and lymphatic valvesVVs and LVs develop in an identical manner, sharing manysimilarities with LVVs (Fig. 4). In mice, VVs in the central veins( jugular and subclavian veins) start developing at ∼E14.5 and arefully formed at E16.5 (Geng et al., 2016; Munger et al., 2016). VVsin peripheral veins (such as in the saphenous veins) developpostnatally (Bazigou et al., 2011; Munger et al., 2016). LVs of the

E12.0

LS

SCV

LS

SVC

E12.5 E14.5 E16.5

A Differentiation B Delamination

EJV

LS

C Aggregation D Maturation

Fig. 3. Development of mouse lymphovenous valves (LVVs). (A) Schematic of the junction that forms between the lymph sacs (LS; green), the internal jugularvein (IJV), external jugular vein (EJV), subclavian vein (SCV) and superior vena cava (SVC) in an E12.0 mouse embryos. The head and the heart are respectivelylocated anterior and posterior to this location. LVVs (arrowheads) develop at the two sites of contact between the LS and the veins. (B-D) Cross-sectionof the junction between the LS and veins in the developing mouse embryos. (B) LVV-forming endothelial cells (LVV-ECs; red) differentiate at E12.0. Immediatelyafter differentiation, LVV-ECs delaminate into the lumen of the vein (towards the right of this picture). Lymphatic endothelial cells (LECs) in close proximity toLVV-ECs have a distinct molecular profile relative to LECs in the LS (green), and are depicted in yellow. (C) LVV-ECs quickly reaggregate and the entire valvecomplex invaginates into the vein. (D) LVVs undergo further maturation by recruiting mural cells (orange) to the space between LVV-ECs and thespecialized LECs. Pictures were adapted with permission from Geng et al. (2016).

Mesenteric

VV

LV

A Differentiation B Reorientation C Matrix organization D Elongation

Central

Peripheral

E16 E17 E18 P5E17 E18

P4P3 P4P3

E14.5 E15.5 E16.5

P1

Fig. 4. Development of mouse lymphatic valves (LVs) and venous valves (VVs). (A) Top: schematic sagittal section of a VV (red cells) located within thevenous lumen (depicted in blue). Bottom: an LV (red cells) located within a mesenteric lymphatic vessel (depicted in green). VVs of central veins startdeveloping at E14.5 and VVs of peripheral veins start developing at postnatal day (P)1. LVs of the mesentery start developing at E16. Images modified withpermission from Bazigou and Makinen (2013). (B-D) Despite differences in their developmental time points, the morphogenesis of VVs and LVs are similar. Forsimplicity, a schematic of developing VVs is presented. The developmental time points corresponding to the appropriate valves are presented at the topof the pictures. The arrow within the lumen of the vessel represents the direction of fluid flow. (B) The valvular endothelial cells undergo circumferentialreorientation along the rim of the vessels. (C) ECM (yellow) is organized in between the valvular endothelial cell layers. (D) The valve leaflets elongate along thedirection of fluid flow to form mature valves. Images modified with permission from Bazigou et al. (2014).

1279

REVIEW Disease Models & Mechanisms (2017) 10, 1273-1287 doi:10.1242/dmm.030825

Disea

seModels&Mechan

isms

skin and mesentery start developing at around E15.5 and E16.5,respectively (Bazigou et al., 2009; Cha et al., 2016; Norrmén et al.,2009).The differentiation of PROX1high; FOXC2high cells is the first step

of VV and LV development (Fig. 4A). Upregulated PROX1expression seems to precede that of FOXC2, at least in the LVs(Sabine et al., 2012). These valvular endothelial cells delaminate andmigrate circumferentially along the rim of the vessels. In the VVs ofcentral veins, the VV-forming endothelial cells crisscross each otherin a ‘knitting’-like process (Fig. 2I) (Geng et al., 2016). This process,called the ‘reorientation’ step because cells undergo a 90o change inorientation, forms a narrow layer of valvular endothelial cells thatencircles the entire circumference of the vessels (Fig. 4B) (Tatin et al.,2013). Next, the collective migration of cells constricts the lumen ofthe vessel while leaving a narrow aperture in the middle (Bazigouet al., 2011; Geng et al., 2016). Subsequently, integrins mediate theorganization of the ECM in the valve region (Fig. 4C) (Bazigou et al.,2009), and cells along the inner edge of the circular shelf elongate inboth directions and touch the vessel wall. This completes theformation of dome-shaped bicuspid valves with two commissures(Figs 2H and 4D). Finally, VVs and LVs recruit a few mural cellsbetween the upstream and downstream endothelial cell layers (Genget al., 2016; Kanady et al., 2011).To summarize, valves develop in a stepwise manner. Defects in

any one of these steps could result in defective valves or in valvesthat are susceptible to degeneration. In the following section, wediscuss disease-associated genes and their mechanisms of actionduring valve development. We also describe the signaling pathwaysthat regulate the expression of these genes. Readers might havealready noted that most of the genes that are necessary for VVdevelopment are also required for LV or LVV development and viceversa. We hope to demonstrate that the role of these genes extendsalso to the aortic valve.

Molecular underpinnings of valve development and diseaseShear stress as a developmental signalIn blood vessels, shear stress is the frictional force experienced bythe endothelial cells due to fluid flow. Depending on the nature offluid flow, the cells could experience either laminar or oscillatoryshear stress (Box 1). In this section we will discuss the dataregarding the relationship between the nature of shear stress, valvedevelopment and valve disease.

Specification of valvular endothelial cellsA high-speed imaging analysis in zebrafish embryos has revealedthat significant backflow occurs at the atrioventricular junction priorto valve development (Vermot et al., 2009). The expression of genessuch as klf2a, bmp4 and notch1b, which are necessary for valveformation, is enriched in this area. Importantly, reducing thereversing flow or shear stress can reduce the expression of klf2a,bmp4 and notch1b, and abolish valve formation. Consistent withthis report, KLF2 has been shown to be necessary for CC formationand for the expression of genes such as Gata4, Tbx5 and Ugdh inmouse embryos (Chiplunkar et al., 2013).Based on these findings, a study by Sabine et al. proposed

oscillatory shear stress as being the signal that specifies LV-ECs(Sabine et al., 2012). Consistent with this model, oscillatory shearstress can enhance the expression of FOXC2, GATA2, CX37,ephrin-B2 and ITGA9 in LECs (Cha et al., 2016; Kazenwadel et al.,2015; Liu et al., 2014; Sabine et al., 2012; Sweet et al., 2015).Together, these data indicate that oscillatory shear stress is the mostupstream regulator of valve specification.

Differentiation of aortic valve endothelial cellsAs mentioned previously, the endothelial cells on the upstream anddownstream sides of aortic valves have distinct molecular profiles.The upstream side of the valve faces the incoming blood;endothelial cells on this surface are exposed to high laminar shearstress. In contrast, the downstream side of the valve is exposed toturbulent and reversing flow and oscillatory shear stress (Fig. 2A). Itis believed that these distinct patterns of mechanical force controlthe identity of the valvular endothelial cells (Chien, 2007). Resultsfrom in vitro flow culture models strongly support this hypothesis.Porcine aortic vascular endothelial cells were exposed to definedshear stress for 24-48 h before analysis (Dekker et al., 2002). Aorticvascular endothelial cells exposed to laminar shear stressupregulated the expression of markers such as KLF2 and CX43that are enriched on the upstream side of the cardiac valves. Incontrast, aortic vascular endothelial cells exposed to oscillatoryshear stress expressed genes such as SELP that are normallyexpressed on the downstream side of the aortic valves.

Therefore, high laminar shear stress and oscillatory shear stressregulate the identities of endothelial cells on the upstream anddownstream sides of aortic valves, respectively. In turn, endothelialcell identity is essential for inhibiting the progression of CAVD.

Conserved polarized expression of shear-stress-responsive genesThe expression of PROX1, FOXC2, GATA2, connexins and ephrin-B2 is conserved between the aortic and vascular valves (Fig. 2A,D,G).Intriguingly, some of these molecules show side-specific expressionpatterns. For example, PROX1 expression is enriched on both theupstream and downstream sides of vascular valves and it is specificallyexpressed on the downstream side of the aortic valve (Bazigou et al.,2011; Geng et al., 2016; Sabine et al., 2012; Srinivasan and Oliver,2011). CX43 is specifically localized to the upstream side of the aorticvalves, and of VVs, LVs and LVVs (Inai et al., 2004; Kanady et al.,2011; Munger et al., 2017, 2016, 2013; Simmons et al., 2005). Incontrast, CX37 expression is enriched on the downstream side ofvalves (Geng et al., 2016; Inai et al., 2004; Kanady et al., 2011;Munger et al., 2017, 2016, 2013; Sabine et al., 2012). AlthoughFOXC2 is expressed in all endothelial cells, it is enriched on thedownstream side of the aortic valve, VVs, LVVs and LVs (De Valet al., 2008; Geng et al., 2016; Kazenwadel et al., 2015;Munger et al.,2016; Sabine et al., 2012; Simmons et al., 2005). Likewise, althoughGATA2 is expressed in all endothelial cells, its expression is enrichedon the downstream side of aortic valves and LVVs (Geng et al., 2016;Kazenwadel et al., 2015; Khandekar et al., 2007). In cardiac valves,the GATA2 expression domain is broader than that of PROX1 andFOXC2 (Kazenwadel et al., 2015). Whether there is any polarizedexpression of GATA2 in LVs and VVs is currently unknown. Finally,ephrin-B2 is expressed on the downstream side of aortic valves andVVs (Bazigou et al., 2011; Cowan et al., 2004). Owing to theirimportant roles in vascular valve development, we speculate thatPROX1, FOXC2, GATA2, CX37 and ephrin-B2 might protect thecalcification-prone downstream side of aortic valves. As mentionedpreviously, the expression of most of these molecules is enhanced byoscillatory shear stress in LECs. In the following paragraphs wediscuss these mechanisms further.

FOXC2FOXC2 was the first molecule shown to be regulated by oscillatoryshear stress in primary human LECs (Sabine et al., 2012), and isnecessary for the expression of genes such as CX37 (Kanady et al.,2011; Sabine et al., 2012). Oscillatory shear stress promotes anincrease in cytoplasmic calcium levels in a CX37-dependent manner

1280

REVIEW Disease Models & Mechanisms (2017) 10, 1273-1287 doi:10.1242/dmm.030825

Disea

seModels&Mechan

isms

(Sabine et al., 2012). The mechanisms of CX37 activity during thisprocess are not fully understood. NFATC1, a FOXC2 cofactor, isdephosphorylated by the calcium-dependent serine/threonine proteinphosphatase calcineurin (Norrmén et al., 2009). DephosphorylatedNFATC1 translocates into the nucleus of oscillatory-shear-stress-exposed LECs and associates with FOXC2 (Sabine et al., 2012).NFATC1 and FOXC2 are expressed on the opposite sides of

venous valves (Munger et al., 2016). Furthermore, NFATC1 isexpressed on both the upstream and downstream sides of cardiacvalves (Wu et al., 2011). Therefore, the relationship betweenFOXC2 and NFATC1 remains to be better elucidated. It is possiblethat these two transcription factors interact during the early stages ofvalve development, as recently reported (Lyons et al., 2017).FOXC2 and its paralog FOXC1 are both strongly expressed on

the downstream side of aortic valves (Simmons et al., 2005).Deletion of both Foxc1 and Foxc2 in mice results in the dysplasia ofCC with increased cell death in neural crest cells (Seo and Kume,2006). However, the role of FOXC2 in CAVD is currently unknown.

GATA2Flow chamber experiments have revealed that oscillatory shearstress activates GATA2 expression in primary human LECs(Kazenwadel et al., 2015; Sweet et al., 2015). In turn, GATA2upregulates the expression of numerous molecules, includingFOXC2 and ITGA9, in oscillatory-shear-stress-exposed LECs.GATA2 also directly activates PROX1 expression, presumably in anoscillatory-shear-stress-independent manner (Kazenwadel et al.,2015). Consistent with these findings, the expression of FOXC2 andPROX1 is downregulated in the valvular endothelial cells of micelacking Gata2 (Geng et al., 2016; Kazenwadel et al., 2015).Whether GATA2 is involved in aortic valve development or

disease is currently unknown. Two Emberger syndrome patientswith mitral valve vegetations (Box 1) and embolic strokes had theirsymptoms attributed to defective endocardial function of GATA2(Spinner et al., 2014). Moreover, single-nucleotide polymorphisms(SNPs) in GATA2 are associated with coronary artery disease andatherosclerosis, diseases that frequently occur together with CAVD(Connelly et al., 2006; Muiya et al., 2014).

PROX1PROX1 is the master regulator of lymphatic vascular developmentand a pioneer regulator of vascular valve development. Analysis ofmouse embryos has revealed that PROX1 upregulation is adefinitive sign that vascular valve development has begun(Bazigou et al., 2009; Norrmén et al., 2009; Srinivasan andOliver, 2011). GATA2 is necessary for PROX1 upregulation inLVV-ECs and LV-ECs (Kazenwadel et al., 2015). Whether GATA2regulates PROX1 expression in VVs remain unknown.In mice, deleting a single Prox1 allele results in severe embryonic

edema (Wigle and Oliver, 1999). An analysis of Prox1+/− mouseembryos just before birth revealed that they lack LVVs, LVs andVVs (Geng et al., 2016; Srinivasan and Oliver, 2011). Oscillatoryshear stress does not regulate PROX1 expression in primary humanLECs (Sabine et al., 2012; Sweet et al., 2015). However, PROX1 isessential for the proper response of LECs to oscillatory shear stress(Sabine et al., 2012). LECs that lack PROX1 dramaticallydownregulate the expression of CX37 and VE-cadherin, and theyalign abnormally with respect to the direction of shear stress. Thus,PROX1 plays an important role in the shear stress response of LECs.The mechanisms that regulate PROX1 expression in the aortic

valves and whether PROX1 plays any role in CAVD are currentlynot known.

ConnexinsConnexins are gap-junction molecules that regulate the transport ofsmall molecules between adjacent cells (Kanady and Simon, 2011).Six connexin molecules oligomerize to form a hemichannel in onecell. The interaction between two hemichannels from adjacent cellsforms a gap-junction channel. Gap-junction channels clustertogether to form gap junctions. Three separate connexins areknown to play important roles in valve development, two of whichare regulated by shear stress.

Connexin 37Cx37−/−mice lack VVs and have reduced numbers of LVs (Kanadyet al., 2011; Munger et al., 2013). Although Cx37−/− embryos haveLVV-ECs, their morphogenesis is defective and they do notinvaginate into the vein (Geng et al., 2016).

CX37 expression is activated by oscillatory shear stress and byFOXC2 in LECs (Kanady et al., 2011; Sabine et al., 2012). CX37promotes the nuclear translocation of the FOXC2 cofactorNFATC1 inLECs exposed to oscillatory shear stress (Sabine et al., 2012). Thislikely establishes a positive feedback loop between CX37 and FOXC2activity. In support of this possibility, a genetic interaction has beenshown to occur between mouse CX37 and FOXC2 (Kanady et al.,2015; Sabine et al., 2012). Foxc2+/− and Cx37+/− mice have normalnumbers of LVs in the mesentery. However, Foxc2+/−;Cx37+/− micehave significantly fewer mature LVs (Sabine et al., 2012).Additionally, whereas Cx37−/− embryos have a reduced number ofLVs compared to wild-type littermates, Foxc2+/−;Cx37−/− embryoshave no LVs (Kanady et al., 2011, 2015).

CX37 is strongly expressed on the downstream side of the aorticvalve (Inai et al., 2004). Whether the progression of CAVD isinfluenced by the loss of CX37 is currently unknown. However,Cx37−/−; ApoE−/− mice are more prone to high-fat-diet-inducedinflammation of aortic endothelial cells and to atherosclerosis (Wonget al., 2006). CAVD and atherosclerosis are often comorbidities inhuman patients. Therefore, it is possible that CX37 plays a protectiverole in CAVD.

Connexin 43CX43 is expressed on the upstream side of all the valves (Fig. 2A,D,G), and laminar shear stress activates its expression in bloodendothelial cells (Butcher and Nerem, 2007; Inai et al., 2004). Incontrast, CX43 expression is downregulated by oscillatory shearstress in both blood endothelial cells and LECs (Butcher and Nerem,2007; Sabine et al., 2012).

When fed a Western diet, Ldlr−/− mice, which lack the low-density-lipoprotein receptor (LDLR) that is necessary for cholesterolendocytosis, develop atherosclerosis (Wong et al., 2003). In contrast,Cx43+/−; Ldlr−/− mice fed a Western diet are resistant toatherosclerosis, indicating that CX43 plays a pro-atherogenic role inmouse blood vessels (Wong et al., 2003). Its downregulation on thedownstream side of the aortic valve could therefore fulfill anatheroprotective role in the vasculature (Mongkoldhumrongkul et al.,2016). As mentioned previously, Tie2-Cre; Cx43f/f mice are viable(Liao et al., 2001; Theis et al., 2001). Whether these mice are betterprotected from CAVD needs to be explored.

Connexin 47Expression analysis has revealed that CX47 widely colocalizes withCX43 on the upstream side of LVs and VVs during their earlydevelopment (Kanady et al., 2011; Munger et al., 2016). At laterstages, CX47 expression continues to colocalizewith CX43, but it isrestricted to fewer cells. CX47 is not expressed in LVVs (Munger

1281

REVIEW Disease Models & Mechanisms (2017) 10, 1273-1287 doi:10.1242/dmm.030825

Disea

seModels&Mechan

isms

et al., 2016). Whether CX47 is regulated by shear stress and whetherit plays any role in aortic valve development or disease is unknown.In summary, connexins are important for valve development.

However, their mechanisms of action in valve formation andfunction are not fully understood. As mentioned previously,CX37 is necessary for oscillatory-shear-stress-induced NFATC1activation. Whether any other signals are regulated by connexinsremains to be investigated. Connexins could also play non-channelroles, such as in regulating cell-cell interaction and the production ofthe vasodilator nitric oxide (Kanady and Simon, 2011; Meens et al.,2015; Pfenniger et al., 2010).

EPHB4 and ephrin-B2The receptor tyrosine kinase EPHB4 was originally identified as avein-specific marker and its membrane-bound ligand ephrin-B2 wasidentified as an artery-specific marker (Gerety et al., 1999). Thesemolecules have mutually exclusive expression patterns and producerepellent signals that promote the segregation of arteries and veins(Gerety et al., 1999). Signals downstream of EPHB4 are known asforward signaling and those downstream of ephrin-B2 as reversesignaling.Expression of ephrin-B2 is enhanced by oscillatory shear stress in

a GATA2-independent manner (Sweet et al., 2015). However, therelationship between ephrin-B2 and shear stress has not beenexplored further. Ephrin-B2 is required for the development of LVsand VVs (Bazigou et al., 2011; Makinen et al., 2005). Loss of thecytoplasmic domain of ephrin-B2 results in thickened cardiacvalves (Cowan et al., 2004). Therefore, ephrin-B2 reverse signalingmight be important to inhibit the progression of CAVD.There is an interesting controversy regarding the role of ephrin-B2

reverse signaling during LV development. Mutating thephosphorylated tyrosine residues to phenylalanine in the cytoplasmictail of ephrin-B2 results in the arrest of LV morphogenesis (Makinenet al., 2005). However, deletion of the entire cytoplasmic tailseemingly does not affect LV morphogenesis (Zhang et al., 2015). Itis proposed that the tyrosine-to-phenylalanine mutations inhibit thecytoplasmic-to-cell-surface translocation of the ligand and precludesEPHB4 forward signaling (Cowan et al., 2004).Antibodies that specifically promote EPHB4 signaling were

recently shown to rescue the LV defects in mice lacking ephrin-B2(Zhang et al., 2015), highlighting this as a potential therapeuticapproach to treating the LVV and LV defects of patients withEPHB4 mutations. Furthermore, if EPHB4 or ephrin-B2 function inCAVD, signal-modulating antibodies could play a role in the fightagainst this devastating disease.

ITGA9ITGA9 expression is enhanced by oscillatory shear stress in aGATA2-dependent manner (Sweet et al., 2015). NOTCH1 is alsonecessary for the expression of ITGA9 in LVs (Murtomaki et al.,2014). ITGA9 can heterodimerize with integrin-β1 (α5β1) andassociate with ECM proteins, such as with the EIIIA-domain-containing fibronectin (Fn-EIIA), SVEP-1 and EMILIN1 (Bazigouet al., 2009; Danussi et al., 2013; Karpanen et al., 2017; Morookaet al., 2017). Indeed, Fn-EIIA−/−mice recapitulate the phenotype ofItga9−/− animals (Bazigou et al., 2009). Svep-1- and Emilin1-nullmice also have valve defects (Danussi et al., 2013; Karpanen et al.,2017; Morooka et al., 2017).In summary, mechanosensitive molecules that are important for

valve development andmorphogenesis are beginning to be identified.In the following section we will discuss the mechanotransductionmechanisms that activate the expression of these molecules.

Mechanotransduction mechanisms during valve developmentWe are beginning to understand the mechanisms that endothelialcells use to sense the various patterns of fluid flow and translatethem into chemical signals. Here, we discuss the mechanosensorymolecules that function in mechanotransduction and in lymphaticvascular maturation or LV development.

Wnt/β-catenin signalingWe recently showed that oscillatory shear stress promotes thestabilization and nuclear translocation of β-catenin in primaryhuman LECs (Cha et al., 2016), and that oscillatory-shear-stress-enhanced FOXC2 expression depends on β-catenin activity. Wnt/β-catenin signaling also enhances PROX1 expression in LECs in anoscillatory-shear-stress-independent manner. β-catenin associateswith the regulatory elements of PROX1 and FOXC2 in primaryhuman LECs (Cha et al., 2016). Consistent with these findings, theconditional deletion of β-catenin from the LECs of mice results inthe loss of LVVs, VVs and LVs (Cha et al., 2016).

Expression of β-catenin is enriched on the downstream sideendothelial cells of aortic valves (Simmons et al., 2005). Consistentwith this report, the deletion of β-catenin in these cells abolishedFOXC2 expression (Cha et al., 2016). These results suggest thatWnt/β-catenin signaling transduces oscillatory shear stress to activateFOXC2 expression in valvular endothelial cells. As mentionedpreviously, GATA2 enhances the expression of PROX1 and FOXC2in valvular endothelial cells (Kazenwadel et al., 2015). As such, itwill be important to test whether GATA2 synergizes with β-catenin toenhance PROX1 and FOXC2 expression.

Syndecan-4In blood endothelial cells, platelet and endothelial cell adhesionmolecule 1 (PECAM-1), VEGFR2 and VE-cadherin form amechanosensory complex that mediates flow response (Tzimaet al., 2005). VEGFR3 is also a component of this mechanosensorycomplex in blood endothelial cells (Coon et al., 2015). WhetherVEGFR2, VEGFR3 and VE-cadherin are necessary for flowresponse in LECs remains unknown. However, PECAM-1 wasrecently shown to act in parallel with syndecan-4 during mouselymphatic vessel and LV morphogenesis (Wang et al., 2016). Thelymphatic vessels of Sdc4−/− mice are hyperproliferative andhyperbranched, and the LV-forming endothelial cells do notreorient properly during the circumferential elongation process.These defects are more severe in Sdc4−/−;Pecam-1−/− embryos.Mechanistically, syndecan-4 knockdown in LECs affects theirability to align correctly with respect to the direction of laminarshear stress. This defect is due to the overexpression of planar cellpolarity (PCP) protein VANGL2 in the cells with reduced syndecan-4. Knockdown of VANGL2 rescues the flow-induced alignment ofLECs with reduced syndecan-4 (Wang et al., 2016).

VANGL2 and CELSR1 are transmembrane proteins that functionin the Wnt/PCP pathway, which coordinates cell polarity across thetissue plane (Devenport, 2014). During LV morphogenesis,VANGL2 and CELSR1 localize to cell junctions, where theyinhibit the accumulation of VE-cadherin to prevent the stabilizationof adherens junctions (Tatin et al., 2013). Thus, syndecan-4 acts as avital link between the laminar shear stress and PCP pathways.Presumably, VANGL2 and CELSR1 destabilize cell junctionsduring LV morphogenesis, whereas syndecan-4 stabilizes themonce the morphogenetic process is complete. How or whethersyndecan-4 mechanistically interacts with PECAM-1 to regulateVANGL2 expression is currently unknown, as is whether syndecan-4 is important for the oscillatory shear stress response.

1282

REVIEW Disease Models & Mechanisms (2017) 10, 1273-1287 doi:10.1242/dmm.030825

Disea

seModels&Mechan

isms

In summary, the Wnt/β-catenin and Wnt/PCP pathways playimportant roles in endothelial cell shear response during valveformation. Recently, the calcium channel ORAI1 was reported to benecessary for laminar-shear-stress-induced inhibition of Notchsignaling and lymphatic vascular growth (Choi et al., 2017).Whether this pathway operates during valve development alsoremains to be investigated.

Limitations of the shear stress modelEndothelial cells from opposite sides of the aortic valve maintaintheir identities in culture even when cultured under identicalconditions for prolonged periods (Simmons et al., 2005). Thissuggests that the side-dependent phenotypes of these cells are likelyto be dictated by developmental and local environmental factors, aswell as by shear stress and hemodynamics.Endothelial cells derived from the aorta (the major artery

originating from the left ventricle) are commonly used to gaininsights about the shear response of aortic valve endothelial cells.One study evaluated the different responses of aortic vascularendothelial cells and aortic valve endothelial cells to flow (Butcheret al., 2004, 2006). The authors determined that cardiac valveendothelial cells align perpendicular to laminar shear stress in theflow chamber, thus recapitulating the in vivo phenotype of thosecells. However, vascular endothelial cells align parallel to laminarshear stress. In addition, at least 10% of genes are differentiallyexpressed between these two cell types in response to laminar shearstress, and the signaling pathways that regulate the endothelial cellresponse to flow are also distinct. The authors noted that caution wasrequired when extrapolating findings from vascular endothelial cellsto valvular endothelial cells (Butcher and Nerem, 2007).The same caution is also warranted when using LECs to describe

LV development. Although LECs exposed to oscillatory shear stressassume a round morphology, LV-ECs are elongated in vivo and are

perpendicularly organized with respect to fluid flow (Geng et al.,2016; Sabine et al., 2012). It will be important to determine themechanisms that regulate the LV-EC identity, which in turn dictatestheir shear response. Additionally, how the oscillatory and laminarshear stress responses are integrated during valve developmentremains to be understood.

In summary, shear stress plays an important role in valvedevelopment. However, approaches to study the response ofvalvular endothelial cells to shear stress need to be further refined.Whenever possible, valvular endothelial cells should be used for cellculture experiments. However, because these cells are difficult toobtain and maintain, novel approaches, such as transdifferentiation of

Box 2. Proteins involved in shear-stress-unrelatedmechanisms that regulate vascular valve developmentAngiopoietin-2: a ligand for the receptor tyrosine kinase TIE2; promotesthe expression of FOXC2 (Dellinger et al., 2008; Morooka et al., 2017).BMP9: signals through ALK1 to promote the expression of neuropilin-1,FOXC2, ephrin-B2 and CX37 (Levet et al., 2013).CDK5: a cyclin-dependent kinase that phosphorylates FOXC2 andpromotes its activity (Liebl et al., 2015).ITGA5: mediates cell-ECM interaction during valve morphogenesis(Turner et al., 2014).Notch signaling: regulates the expression of ITGA9 (Murtomaki et al.,2014).RASA1: inhibits the Ras signaling pathway and prevents apoptosis(Lapinski et al., 2017).Semaphorin-3A: activates the plexin-A1–neuropilin-1 complex duringleft-ventricle morphogenesis (Bouvree et al., 2012; Jurisic et al., 2012).SVEP1: ECM protein that enhances the signals mediated byangiopoietin-2 (Karpanen et al., 2017; Morooka et al., 2017).TIE1: an orphan tyrosine-kinase receptor that promotes the expressionof PROX1 and FOXC2 (Qu et al., 2015).

KLF2A

Notch1b

KLF2

UGDHGATA4TBX5

EndMT

Cardiac valvedevelopment

GATA2

PROX1

CTNNB1

CX37

Cn/NFATC1

ECM

CX43

SD

C4

C4

SD

CELSR1

Cell reorientation

VANGL2

PE

CA

M1

PA

M1

EC

A

ITG

A9

ITG

A9

EFN

B2

EFN

B2

EP

HB

4E

PH

B4

Laminar shear stress

FOXC2

Osillatory shear stress

KLF2

Lymphatic valve morphogenesis

Fig. 5. An integrated model for valvedevelopment. A model of the variousmolecules that regulate valvemorphogenesis and their functionalrelationships. Oscillatory shear stressactivates Klf2 expression and itsdownstream target genes during cardiacvalve development in zebrafish and mice.During lymphatic valve development,oscillatory shear stress enhances theexpression of multiple molecules, suchas ITGA9, ephrin-B2 (EFNB2), GATA2,FOXC2, CX37 and β-catenin (CTNNB1).Oscillatory shear stress also antagonizesCX43 expression. β-catenin is upstream ofGATA2 and FOXC2. Shear-stress-activatedGATA2 is upstream of FOXC2 and ITGA9.PROX1 expression is not regulated byoscillatory shear stress. However, bothGATA2 and β-catenin could enhancePROX1 expression in an oscillatory-shear-stress-independent manner. CX37 regulatesNFATC1 activity through the calcineurin (Cn)signaling pathway. Laminar shear stressinhibits the planar cell polarity moleculesVANGL2 and CELSR1 through syndecan-4(SDC4). PECAM1 acts in parallel with SDC4to regulate the reorientation of valve-formingendothelial cells.

1283

REVIEW Disease Models & Mechanisms (2017) 10, 1273-1287 doi:10.1242/dmm.030825

Disea

seModels&Mechan

isms

vascular endothelial cells to valvular endothelial cells, need to bedeveloped. Importantly, the molecular and cellular mechanisms thatare currently unrelated to the shear-stress response should not beignored (see Box 2). Our current understanding of the mechanisms ofvalve development is presented in Fig. 5.

Conclusions and perspectivesWe end this Review with a few final thoughts regarding thechallenges and opportunities for valve research. Owing to largevariations in the penetrance and the time of disease onset, thetreatment strategies may have to be custom designed for the patients.To achieve this goal, we need to understand the reasons for thevariable penetrance and etiology of the disease. We speculate thatsome of the shear-responsive genes could be used as diagnosticmarkers. Next, once the target patients are chosen, the treatmentstrategies may have to be tailored to their specific needs. Target-specific drugs could be the best option for treating the disease thatstarts in utero. Once again, the molecules that we discussed in thisarticle could be important targets for drug development. Surgicalrepair/replacement of valves could be a feasible option for patientswhose disease starts postnatally. Because there are hundreds of LVsand VVs in the human body, identifying the most critical valvescould help in efficiently targeting this treatment. Furthermore, thereare only four LVVs in mammals and several studies (mouse andhuman) have indicated that these valves could be defective inlymphedema. Therefore, we need to explore whether repairing thesevalves could cure or ameliorate lymphedema.Valve replacement surgery is already available for aortic valves.

As discussed above, this approach has its limitations and a betterunderstanding of the biology of aortic valve endothelial cells isneeded for advancing the treatments. Simmons et al. used humangene expression microarrays to identify genes that are differentiallyexpressed between the upstream and downstream sides of porcineheart (Simmons et al., 2005). Since then, both microarray and RNA-seq technologies have improved tremendously. Thus, repeating thisexperiment using advanced tools might reveal additional side-specific genes. Mouse models could demonstrate the significance ofthese genes during aortic valve development and disease. It will beimportant to develop conditional mouse models to specificallydelete genes on the upstream or downstream sides of aortic valveswithout affecting the vascular valves. This will also require thecreation of new Cre lines.In conclusion, in the past 10 years, creative experiments that are

built on models put forward by aortic valve biologists have revealedthat shear forces regulate the expression of genes that are necessaryfor vascular valve development. Intriguingly, these genes are alsoexpressed in the aortic valves, thus revealing a previouslyunanticipated commonality between the aortic valves and vascularvalves.We hope that this proposal will stimulate discussion betweencardiac valve and vascular valve researchers, and accelerate thesearch for cures for valve diseases.

AcknowledgementsWe thank Dr Alex Simon for insightful discussions, and Dr Angela Andersen (LifeScience Editors) and DMM’s developmental editors for scientific editing.

Competing interestsThe authors declare no competing or financial interests.

FundingThis work is supported by National Institutes of Health (NIH)/NHLBI (R01HL131652and R01HL133216), Oklahoma Center for Adult Stem Cell Research (4340) andAmerican Heart Association (15BGIA25710032 to R.S.S., 15POST25080182 toB.C. and 16PRE31190025 to M.R.M.).

ReferencesAcharya, A., Hans, C. P., Koenig, S. N., Nichols, H. A., Galindo, C. L., Garner,

H. R., Merrill, W. H., Hinton, R. B. andGarg, V. (2011). Inhibitory role of Notch1 incalcific aortic valve disease. PLoS ONE 6, e27743.

Armstrong, E. J. andBischoff, J. (2004). Heart valve development: endothelial cellsignaling and differentiation. Circ. Res. 95, 459-470.

Bazigou, E. and Makinen, T. (2013). Flow control in our vessels: vascular valvesmake sure there is no way back. Cell. Mol. Life Sci. 70, 1055-1066.

Bazigou, E., Xie, S., Chen, C., Weston, A., Miura, N., Sorokin, L., Adams, R.,Muro, A. F., Sheppard, D. andMakinen, T. (2009). Integrin-alpha9 is required forfibronectin matrix assembly during lymphatic valve morphogenesis. Dev. Cell 17,175-186.

Bazigou, E., Lyons, O. T. A., Smith, A., Venn, G. E., Cope, C., Brown, N. A. andMakinen, T. (2011). Genes regulating lymphangiogenesis control venous valveformation and maintenance in mice. J. Clin. Invest. 121, 2984-2992.

Bazigou, E., Wilson, J. T. and Moore, J. E., Jr. (2014). Primary and secondarylymphatic valve development: molecular, functional and mechanical insights.Microvasc. Res. 96, 38-45.

Bell, R., Brice, G., Child, A. H., Murday, V. A., Mansour, S., Sandy, C. J., Collin,J. R., Brady, A. F., Callen, D. F., Burnand, K. et al. (2001). Analysis oflymphoedema-distichiasis families for FOXC2 mutations reveals small insertionsand deletions throughout the gene. Hum. Genet. 108, 546-551.

Bouvree, K., Brunet, I., del Toro, R., Gordon, E., Prahst, C., Cristofaro, B.,Mathivet, T., Xu, Y., Soueid, J., Fortuna, V. et al. (2012). Semaphorin3A,Neuropilin-1, and PlexinA1 are required for lymphatic valve formation. Circ. Res.111, 437-445.

Breslin, J. W. (2014). Mechanical forces and lymphatic transport. Microvasc. Res.96, 46-54.

Brice, G., Mansour, S., Bell, R., Collin, J. R., Child, A. H., Brady, A. F., Sarfarazi,M., Burnand, K. G., Jeffery, S., Mortimer, P. et al. (2002). Analysis of thephenotypic abnormalities in lymphoedema-distichiasis syndrome in 74 patientswith FOXC2 mutations or linkage to 16q24. J. Med. Genet. 39, 478-483.

Brice, G., Ostergaard, P., Jeffery, S., Gordon, K., Mortimer, P. S. and Mansour,S. (2013). A novel mutation in GJA1 causing oculodentodigital syndrome andprimary lymphoedema in a three generation family. Clin. Genet. 84, 378-381.

Brooks, E. G., Trotman, W., Wadsworth, M. P., Taatjes, D. J., Evans, M. F.,Ittleman, F. P., Callas, P.W., Esmon, C. T. andBovill, E. G. (2009). Valves of thedeep venous system: an overlooked risk factor. Blood 114, 1276-1279.

Brouillard, P., Boon, L. and Vikkula, M. (2014). Genetics of lymphatic anomalies.J. Clin. Invest. 124, 898-904.

Burkhart, H. M., Fath, S. W., Dalsing, M. C., Sawchuk, A. P., Cikrit, D. F. andLalka, S. G. (1997). Experimental repair of venous valvular insufficiency using acryopreserved venous valve allograft aided by a distal arteriovenous fistula.J. Vasc. Surg. 26, 817-822.

Burrows, P. E., Gonzalez-Garay, M. L., Rasmussen, J. C., Aldrich, M. B.,Guilliod, R., Maus, E. A., Fife, C. E., Kwon, S., Lapinski, P. E., King, P. D. et al.(2013). Lymphatic abnormalities are associated with RASA1 gene mutations inmouse and man. Proc. Natl. Acad. Sci. USA 110, 8621-8626.

Butcher, J. T. and Nerem, R. M. (2007). Valvular endothelial cells and themechanoregulation of valvular pathology. Philos. Trans. R. Soc. Lond. B Biol. Sci.362, 1445-1457.

Butcher, J. T., Penrod, A. M., Garcia, A. J. and Nerem, R. M. (2004). Uniquemorphology and focal adhesion development of valvular endothelial cells in staticand fluid flow environments. Arterioscler. Thromb. Vasc. Biol. 24, 1429-1434.

Butcher, J. T., Tressel, S., Johnson, T., Turner, D., Sorescu, G., Jo, H. andNerem, R. M. (2006). Transcriptional profiles of valvular and vascular endothelialcells reveal phenotypic differences: influence of shear stress. Arterioscler.Thromb. Vasc. Biol. 26, 69-77.

Cha, B., Geng, X., Mahamud, M. R., Fu, J., Mukherjee, A., Kim, Y., Jho, E.-H.,Kim, T. H., Kahn, M. L., Xia, L. et al. (2016). Mechanotransduction activatescanonical Wnt/beta-catenin signaling to promote lymphatic vascular patterningand the development of lymphatic and lymphovenous valves. Genes Dev. 30,1454-1469.

Chang, C.-P., Neilson, J. R., Bayle, J. H., Gestwicki, J. E., Kuo, A., Stankunas,K., Graef, I. A. and Crabtree, G. R. (2004). A field of myocardial-endocardialNFAT signaling underlies heart valve morphogenesis. Cell 118, 649-663.

Chien, S. (2007). Mechanotransduction and endothelial cell homeostasis: thewisdom of the cell. Am. J. Physiol. Heart Circ. Physiol. 292, H1209-H1224.

Chiplunkar, A. R., Lung, T. K., Alhashem, Y., Koppenhaver, B. A., Salloum,F. N., Kukreja, R. C., Haar, J. L. and Lloyd, J. A. (2013). Kruppel-like factor 2 isrequired for normal mouse cardiac development. PLoS ONE 8, e54891.

Choi, D., Park, E., Jung, E., Seong, Y. J., Yoo, J., Lee, E., Hong, M., Lee, S.,Ishida, H., Burford, J. et al. (2017). Laminar flow downregulates Notch activity topromote lymphatic sprouting. J. Clin. Invest. 127, 1225-1240.

Connelly, J. J., Wang, T., Cox, J. E., Haynes, C., Wang, L., Shah, S. H., Crosslin,D. R., Hale, A. B., Nelson, S., Crossman, D. C. et al. (2006). GATA2 isassociated with familial early-onset coronary artery disease. PLoS Genet.2, e139.

Coon, B. G., Baeyens, N., Han, J., Budatha, M., Ross, T. D., Fang, J. S., Yun, S.,Thomas, J.-L. and Schwartz, M. A. (2015). Intramembrane binding of VE-

1284

REVIEW Disease Models & Mechanisms (2017) 10, 1273-1287 doi:10.1242/dmm.030825

Disea

seModels&Mechan

isms

cadherin to VEGFR2 and VEGFR3 assembles the endothelial mechanosensorycomplex. J. Cell Biol. 208, 975-986.

Cowan, C. A., Yokoyama, N., Saxena, A., Chumley, M. J., Silvany, R. E., Baker,L. A., Srivastava, D. and Henkemeyer, M. (2004). Ephrin-B2 reverse signaling isrequired for axon pathfinding and cardiac valve formation but not early vasculardevelopment. Dev. Biol. 271, 263-271.

Dalsing, M. C., Raju, S., Wakefield, T. W. and Taheri, S. (1999). A multicenter,phase I evaluation of cryopreserved venous valve allografts for the treatment ofchronic deep venous insufficiency. J. Vasc. Surg. 30, 854-866.

Dangas, G. D., Weitz, J. I., Giustino, G., Makkar, R. and Mehran, R. (2016).Prosthetic heart valve thrombosis. J. Am. Coll. Cardiol. 68, 2670-2689.

Danussi, C., Del Bel Belluz, L., Pivetta, E., Modica, T. M. E., Muro, A.,Wassermann, B., Doliana, R., Sabatelli, P., Colombatti, A. and Spessotto, P.(2013). EMILIN1/alpha9beta1 integrin interaction is crucial in lymphatic valveformation and maintenance. Mol. Cell. Biol. 33, 4381-4394.

Davis, M. J., Scallan, J. P., Wolpers, J. H., Muthuchamy, M., Gashev, A. A. andZawieja, D. C. (2012). Intrinsic increase in lymphangion muscle contractility inresponse to elevated afterload. Am. J. Physiol. Heart Circ. Physiol. 303,H795-H808.

Deck, J. D. (1986). Endothelial cell orientation on aortic valve leaflets. Cardiovasc.Res. 20, 760-767.

Dekker, R. J., van Soest, S., Fontijn, R. D., Salamanca, S., de Groot, P. G.,VanBavel, E., Pannekoek, H. and Horrevoets, A. J. (2002). Prolonged fluidshear stress induces a distinct set of endothelial cell genes, most specifically lungKruppel-like factor (KLF2). Blood 100, 1689-1698.

Dellinger, M., Hunter, R., Bernas, M., Gale, N., Yancopoulos, G., Erickson, R.and Witte, M. (2008). Defective remodeling and maturation of the lymphaticvasculature in Angiopoietin-2 deficient mice. Dev. Biol. 319, 309-320.

Demer, L. L. and Tintut, Y. (2008). Vascular calcification: pathobiology of amultifaceted disease. Circulation 117, 2938-2948.

De Val, S., Chi, N. C., Meadows, S. M., Minovitsky, S., Anderson, J. P., Harris,I. S., Ehlers, M. L., Agarwal, P., Visel, A., Xu, S.-M. et al. (2008). Combinatorialregulation of endothelial gene expression by ets and forkhead transcriptionfactors. Cell 135, 1053-1064.

Devenport, D. (2014). The cell biology of planar cell polarity. J. Cell Biol. 207,171-179.

Fang, J., Dagenais, S. L., Erickson, R. P., Arlt, M. F., Glynn, M. W., Gorski, J. L.,Seaver, L. H. and Glover, T. W. (2000). Mutations in FOXC2 (MFH-1), a forkheadfamily transcription factor, are responsible for the hereditary lymphedema-distichiasis syndrome. Am. J. Hum. Genet. 67, 1382-1388.

Ferrell, R. E., Baty, C. J., Kimak, M. A., Karlsson, J. M., Lawrence, E. C., Franke-Snyder, M., Meriney, S. D., Feingold, E. and Finegold, D. N. (2010). GJC2missense mutations cause human lymphedema. Am. J. Hum. Genet. 86,943-948.

Fowkes, F. J., Price, J. F. and Fowkes, F. G. R. (2003). Incidence of diagnoseddeep vein thrombosis in the general population: systematic review. Eur. J. Vasc.Endovasc Surg. 25, 1-5.

Garg, V. (2006). Molecular genetics of aortic valve disease. Curr. Opin. Cardiol. 21,180-184.

Garg, V., Muth, A. N., Ransom, J. F., Schluterman, M. K., Barnes, R., King, I. N.,Grossfeld, P. D. and Srivastava, D. (2005). Mutations in NOTCH1 cause aorticvalve disease. Nature 437, 270-274.

Geng, X., Cha, B., Mahamud, M. R., Lim, K.-C., Silasi-Mansat, R., Uddin,M. K. M., Miura, N., Xia, L., Simon, A. M., Engel, J. D. et al. (2016). Multiplemouse models of primary lymphedema exhibit distinct defects in lymphovenousvalve development. Dev. Biol. 409, 218-233.

Gerety, S. S., Wang, H. U., Chen, Z.-F. and Anderson, D. J. (1999). Symmetricalmutant phenotypes of the receptor EphB4 and its specific transmembrane ligandephrin-B2 in cardiovascular development. Mol. Cell 4, 403-414.

Guauque-Olarte, S., Messika-Zeitoun, D., Droit, A., Lamontagne, M., Tremblay-Marchand, J., Lavoie-Charland, E., Gaudreault, N., Arsenault, B. J., Dube,M. P., Tardif, J.-C. et al. (2015). Calcium signaling pathway genes RUNX2 andCACNA1C are associated with calcific aortic valve disease. Circ CardiovascGenet 8, 812-822.

Hess, P. R., Rawnsley, D. R., Jakus, Z., Yang, Y., Sweet, D. T., Fu, J., Herzog, B.,Lu, M. M., Nieswandt, B., Oliver, G. et al. (2014). Platelets mediatelymphovenous hemostasis to maintain blood-lymphatic separation throughoutlife. J. Clin. Invest. 124, 273-284.

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

Hirsh, J. andHoak, J. (1996). Management of deep vein thrombosis and pulmonaryembolism. A statement for healthcare professionals. Council on Thrombosis (inconsultation with the Council on Cardiovascular Radiology), American HeartAssociation. Circulation 93, 2212-2245.

Huang, X. Z., Wu, J. F., Ferrando, R., Lee, J. H., Wang, Y. L., Farese, R. V., Jr. andSheppard, D. (2000). Fatal bilateral chylothorax in mice lacking the integrinalpha9beta1. Mol. Cell. Biol. 20, 5208-5215.

Inai, T., Mancuso, M. R., McDonald, D. M., Kobayashi, J., Nakamura, K. andShibata, Y. (2004). Shear stress-induced upregulation of connexin 43 expression

in endothelial cells on upstream surfaces of rat cardiac valves. Histochem. CellBiol. 122, 477-483.

Jurisic, G., Maby-El Hajjami, H., Karaman, S., Ochsenbein, A. M., Alitalo, A.,Siddiqui, S. S., Ochoa Pereira, C., Petrova, T. V. and Detmar, M. (2012). Anunexpected role of semaphorin3a-neuropilin-1 signaling in lymphatic vesselmaturation and valve formation. Circ. Res. 111, 426-436.

Kanady, J. D. and Simon, A. M. (2011). Lymphatic communication: connexinjunction, what’s your function? Lymphology 44, 95-102.

Kanady, J. D., Dellinger, M. T., Munger, S. J., Witte, M. H. and Simon, A. M.(2011). Connexin37 and Connexin43 deficiencies in mice disrupt lymphatic valvedevelopment and result in lymphatic disorders including lymphedema andchylothorax. Dev. Biol. 354, 253-266.

Kanady, J. D., Munger, S. J., Witte, M. H. and Simon, A. M. (2015). CombiningFoxc2 and Connexin37 deletions in mice leads to severe defects in lymphaticvascular growth and remodeling. Dev. Biol. 405, 33-46.

Karpanen, T., Padberg, Y., van de Pavert, S. A., Dierkes, C., Morooka, N.,Peterson-Maduro, J., van de Hoek, G., Adrian, M., Mochizuki, N., Sekiguchi,K. et al. (2017). An evolutionarily conserved role for polydom/Svep1 duringlymphatic vessel formation. Circ. Res. 120, 1263-1275.

Kazenwadel, J., Secker, G. A., Liu, Y. J., Rosenfeld, J. A., Wildin, R. S., Cuellar-Rodriguez, J., Hsu, A. P., Dyack, S., Fernandez, C. V., Chong, C.-E. et al.(2012). Loss-of-function germline GATA2 mutations in patients with MDS/AML orMonoMAC syndrome and primary lymphedema reveal a key role for GATA2 in thelymphatic vasculature. Blood 119, 1283-1291.

Kazenwadel, J., Betterman, K. L., Chong, C.-E., Stokes, P. H., Lee, Y. K., Secker,G. A., Agalarov, Y., Demir, C. S., Lawrence, D. M., Sutton, D. L. et al. (2015).GATA2 is required for lymphatic vessel valve development and maintenance.J. Clin. Invest. 125, 2979-2994.

Khandekar, M., Brandt, W., Zhou, Y., Dagenais, S., Glover, T. W., Suzuki, N.,Shimizu, R., Yamamoto, M., Lim, K.-C. and Engel, J. D. (2007). A Gata2 intronicenhancer confers its pan-endothelia-specific regulation. Development 134,1703-1712.

Kim, Y. Y. and Ruckdeschel, E. (2016). Approach to residual pulmonary valvedysfunction in adults with repaired tetralogy of Fallot. Heart 102, 1520-1526.

Kriederman, B. M., Myloyde, T. L., Witte, M. H., Dagenais, S. L., Witte, C. L.,Rennels, M., Bernas, M. J., Lynch, M. T., Erickson, R. P., Caulder, M. S. et al.(2003). FOXC2 haploinsufficient mice are amodel for human autosomal dominantlymphedema-distichiasis syndrome. Hum. Mol. Genet. 12, 1179-1185.

Krysa, J., Jones, G. T. and van Rij, A. M. (2012). Evidence for a genetic role invaricose veins and chronic venous insufficiency. Phlebology 27, 329-335.

Kyrle, P. A. and Eichinger, S. (2005). Deep vein thrombosis. Lancet 365,1163-1174.

LaHaye, S., Lincoln, J. and Garg, V. (2014). Genetics of valvular heart disease.Curr. Cardiol. Rep. 16, 487.

Lapinski, P. E., Lubeck, B. A., Chen, D., Doosti, A., Zawieja, S. D., Davis, M. J.and King, P. D. (2017). RASA1 regulates the function of lymphatic vessel valvesin mice. J. Clin. Invest. 127, 2569-2585.

Lee, J. S., Yu, Q., Shin, J. T., Sebzda, E., Bertozzi, C., Chen, M., Mericko, P.,Stadtfeld, M., Zhou, D., Cheng, L. et al. (2006). Klf2 is an essential regulator ofvascular hemodynamic forces in vivo. Dev. Cell 11, 845-857.

Levet, S., Ciais, D., Merdzhanova, G., Mallet, C., Zimmers, T. A., Lee, S.-J.,Navarro, F. P., Texier, I., Feige, J.-J., Bailly, S. et al. (2013). Bonemorphogenetic protein 9 (BMP9) controls lymphatic vessel maturation andvalve formation. Blood 122, 598-607.

Levick, J. R. and Michel, C. C. (2010). Microvascular fluid exchange and therevised Starling principle. Cardiovasc. Res. 87, 198-210.

Liao, Y., Day, K. H., Damon, D. N. and Duling, B. R. (2001). Endothelial cell-specific knockout of connexin 43 causes hypotension and bradycardia in mice.Proc. Natl. Acad. Sci. USA 98, 9989-9994.

Liebl, J., Zhang, S., Moser, M., Agalarov, Y., Demir, C. S., Hager, B., Bibb, J. A.,Adams, R. H., Kiefer, F., Miura, N. et al. (2015). Cdk5 controls lymphatic vesseldevelopment and function by phosphorylation of Foxc2. Nat. Commun. 6, 7274.

Lin, C.-J., Lin, C.-Y., Chen, C.-H., Zhou, B. and Chang, C.-P. (2012). Partitioningthe heart: mechanisms of cardiac septation and valve development.Development139, 3277-3299.

Lindman, B. R., Clavel, M.-A., Mathieu, P., Iung, B., Lancellotti, P., Otto, C. M.and Pibarot, P. (2016). Calcific aortic stenosis. Nat Rev Dis Primers 2, 16006.

Liu, X., Pasula, S., Song, H., Tessneer, K. L., Dong, Y., Hahn, S., Yago, T.,Brophy,M. L., Chang, B., Cai, X. et al. (2014). Temporal and spatial regulation ofepsin abundance and VEGFR3 signaling are required for lymphatic valveformation and function. Sci. Signal. 7, ra97.

Lyons, O., Saha, P., Seet, C., Kuchta, A., Arnold, A., Grover, S., Rashbrook, V.,Sabine, A., Vizcay-Barrena, G., Patel, A. et al. (2017). Human venous valvedisease caused by mutations in FOXC2 and GJC2. J. Exp. Med. 214, 2437.

Ma, G.-C., Liu, C.-S., Chang, S.-P., Yeh, K.-T., Ke, Y.-Y., Chen, T.-H., Wang,B. B.-T., Kuo, S.-J., Shih, J.-C. andChen,M. (2008). A recurrent ITGA9missensemutation in human fetuses with severe chylothorax: possible correlation with poorresponse to fetal therapy. Prenat. Diagn. 28, 1057-1063.

1285

REVIEW Disease Models & Mechanisms (2017) 10, 1273-1287 doi:10.1242/dmm.030825

Disea

seModels&Mechan

isms

Makinen, T., Adams, R. H., Bailey, J., Lu, Q., Ziemiecki, A., Alitalo, K., Klein, R.and Wilkinson, G. A. (2005). PDZ interaction site in ephrinB2 is required for theremodeling of lymphatic vasculature. Genes Dev. 19, 397-410.

Markwald, R. R., Fitzharris, T. P. and Smith, W. N. (1975). Sturctural analysis ofendocardial cytodifferentiation. Dev. Biol. 42, 160-180.

Martin-Almedina, S., Martinez-Corral, I., Holdhus, R., Vicente, A., Fotiou, E.,Lin, S., Petersen, K., Simpson, M. A., Hoischen, A., Gilissen, C. et al. (2016).EPHB4 kinase-inactivating mutations cause autosomal dominant lymphatic-related hydrops fetalis. J. Clin. Invest. 126, 3080-3088.

Meens, M. J., Alonso, F., Le Gal, L., Kwak, B. R. and Haefliger, J.-A. (2015).Endothelial Connexin37 and Connexin40 participate in basal but not agonist-induced NO release. Cell Commun Signal 13, 34.

Meissner, M. H., Eklof, B., Smith, P. C., Dalsing, M. C., DePalma, R. G.,Gloviczki, P., Moneta, G., Neglen, P., O’Donnell, T., Partsch, H. et al. (2007a).Secondary chronic venous disorders. J. Vasc. Surg. 46 Suppl. S, 68S-83S.

Meissner, M. H., Gloviczki, P., Bergan, J., Kistner, R. L., Morrison, N., Pannier,F., Pappas, P. J., Rabe, E., Raju, S. and Villavicencio, J. L. (2007b). Primarychronic venous disorders. J. Vasc. Surg. 46 Suppl. S, 54S-67S.

Mellor, R. H., Brice, G., Stanton, A. W. B., French, J., Smith, A., Jeffery, S.,Levick, J. R., Burnand, K. G., Mortimer, P. S. and Lymphoedema Research,C. (2007). Mutations in FOXC2 are strongly associated with primary valve failure inveins of the lower limb. Circulation 115, 1912-1920.

Mellor, R. H., Tate, N., Stanton, A. W. B., Hubert, C., Makinen, T., Smith, A.,Burnand, K. G., Jeffery, S., Levick, J. R. and Mortimer, P. S. (2011). Mutationsin FOXC2 in humans (lymphoedema distichiasis syndrome) cause lymphaticdysfunction on dependency. J. Vasc. Res. 48, 397-407.

Mongkoldhumrongkul, N., Yacoub, M. H. and Chester, A. H. (2016). Valveendothelial cells - not just any old endothelial cells. Curr. Vasc. Pharmacol. 14,146-154.

Morooka, N., Futaki, S., Sato-Nishiuchi, R., Nishino, M., Totani, Y., Shimono, C.,Nakano, I., Nakajima, H., Mochizuki, N. and Sekiguchi, K. (2017). Polydom isan extracellular matrix protein involved in lymphatic vessel remodeling. Circ. Res.120, 1276-1288.

Muiya, N. P., Wakil, S., Al-Najai, M., Tahir, A. I., Baz, B., Andres, E., Al-Boudari,O., Al-Tassan, N., Al-Shahid, M., Meyer, B. F. et al. (2014). A study of the role ofGATA2 gene polymorphism in coronary artery disease risk traits. Gene 544,152-158.

Munger, S. J., Kanady, J. D. and Simon, A. M. (2013). Absence of venous valvesin mice lacking Connexin37. Dev. Biol. 373, 338-348.

Munger, S. J., Geng, X., Srinivasan, R. S., Witte, M. H., Paul, D. L. and Simon,A. M. (2016). Segregated Foxc2, NFATc1 and Connexin expression at normaldeveloping venous valves, and Connexin-specific differences in the valvephenotypes of Cx37, Cx43, and Cx47 knockout mice. Dev. Biol. 412, 173-190.

Munger, S. J., Davis, M. J. and Simon, A. M. (2017). Defective lymphatic valvedevelopment and chylothorax in mice with a lymphatic-specific deletion ofConnexin43. Dev. Biol. 421, 204-218.

Murtomaki, A., Uh, M. K., Kitajewski, C., Zhao, J., Nagasaki, T., Shawber, C. J.and Kitajewski, J. (2014). Notch signaling functions in lymphatic valve formation.Development 141, 2446-2451.

Nishimura, R. A., Otto, C. M., Bonow, R. O., Carabello, B. A., Erwin, J. P., III,Fleisher, L. A., Jneid, H., Mack, M. J., McLeod, C. J., O’Gara, P. T. et al. (2017).2017 AHA/ACC focused update of the 2014 AHA/ACC guideline for themanagement of patients with valvular heart disease: a report of the americancollege of cardiology/american heart association task force on clinical practiceguidelines. Circulation 135, e1159-e1195.

Nitschke, M., Bell, A., Karaman, S., Amouzgar, M., Rutkowski, J. M., Scherer,P. E., Alitalo, K. and McDonald, D. M. (2017). Retrograde lymph flow leads tochylothorax in transgenic mice with lymphatic malformations. Am. J. Pathol. 187,1984-1997.

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-based study. Lancet 368, 1005-1011.

Norrmen, C., Ivanov, K. I., Cheng, J., Zangger, N., Delorenzi, M., Jaquet, M.,Miura, N., Puolakkainen, P., Horsley, V., Hu, J. et al. (2009). FOXC2 controlsformation and maturation of lymphatic collecting vessels through cooperation withNFATc1. J. Cell Biol. 185, 439-457.

Ostergaard, P., Simpson, M. A., Connell, F. C., Steward, C. G., Brice, G.,Woollard, W. J., Dafou, D., Kilo, T., Smithson, S., Lunt, P. et al. (2011).Mutations in GATA2 cause primary lymphedema associated with a predispositionto acute myeloid leukemia (Emberger syndrome). Nat. Genet. 43, 929-931.

Person, A. D., Klewer, S. E. and Runyan, R. B. (2005). Cell biology of cardiaccushion development. Int. Rev. Cytol. 243, 287-335.

Petrova, T. V., Karpanen, T., Norrmen, C., Mellor, R., Tamakoshi, T., Finegold,D., Ferrell, R., Kerjaschki, D., Mortimer, P., Yla-Herttuala, S. et al. (2004).Defective valves and abnormal mural cell recruitment underlie lymphatic vascularfailure in lymphedema distichiasis. Nat. Med. 10, 974-981.

Pfenniger, A., Derouette, J. P., Verma, V., Lin, X., Foglia, B., Coombs, W., Roth,I., Satta, N., Dunoyer-Geindre, S., Sorgen, P. et al. (2010). Gap junction proteinCx37 interacts with endothelial nitric oxide synthase in endothelial cells.Arterioscler. Thromb. Vasc. Biol. 30, 827-834.

Porat, R. M., Grunewald, M., Globerman, A., Itin, A., Barshtein, G., Alhonen, L.,Alitalo, K. and Keshet, E. (2004). Specific induction of tie1 promoter by disturbedflow in atherosclerosis-prone vascular niches and flow-obstructing pathologies.Circ. Res. 94, 394-401.

Qu, X., Zhou, B. and Scott Baldwin, H. (2015). Tie1 is required for lymphatic valveand collecting vessel development. Dev. Biol. 399, 117-128.

Reaume, A. G., de Sousa, P. A., Kulkarni, S., Langille, B. L., Zhu, D., Davies,T. C., Juneja, S. C., Kidder, G. M. and Rossant, J. (1995). Cardiac malformationin neonatal mice lacking connexin43. Science 267, 1831-1834.

Ruckley, C. V., Evans, C. J., Allan, P. L., Lee, A. J. and Fowkes, F. G. (2002).Chronic venous insufficiency: clinical and duplex correlations. The Edinburgh VeinStudy of venous disorders in the general population. J. Vasc. Surg. 36, 520-525.

Sabine, A., Agalarov, Y., Maby-El Hajjami, H., Jaquet, M., Hagerling, R.,Pollmann, C., Bebber, D., Pfenniger, A., Miura, N., Dormond, O. et al. (2012).Mechanotransduction, PROX1, and FOXC2 cooperate to control connexin37 andcalcineurin during lymphatic-valve formation. Dev. Cell 22, 430-445.

Seeger, M., Bewig, B., Gunther, R., Schafmayer, C., Vollnberg, B., Rubin, D.,Hoell, C., Schreiber, S., Folsch, U. R. and Hampe, J. (2009). Terminal part ofthoracic duct: high-resolution US imaging. Radiology 252, 897-904.

Seo, S. and Kume, T. (2006). Forkhead transcription factors, Foxc1 and Foxc2, arerequired for the morphogenesis of the cardiac outflow tract. Dev. Biol. 296,421-436.

Shah, P. M. and Raney, A. A. (2008). Tricuspid valve disease. Curr. Probl. Cardiol.33, 47-84.

Simmons, C. A., Grant, G. R., Manduchi, E. and Davies, P. F. (2005). Spatialheterogeneity of endothelial phenotypes correlates with side-specific vulnerabilityto calcification in normal porcine aortic valves. Circ. Res. 96, 792-799.

Spinner, M. A., Sanchez, L. A., Hsu, A. P., Shaw, P. A., Zerbe, C. S., Calvo, K. R.,Arthur, D. C., Gu, W., Gould, C. M., Brewer, C. C. et al. (2014). GATA2deficiency: a protean disorder of hematopoiesis, lymphatics, and immunity. Blood123, 809-821.

Srinivasan, R. S. and Oliver, G. (2011). Prox1 dosage controls the number oflymphatic endothelial cell progenitors and the formation of the lymphovenousvalves. Genes Dev. 25, 2187-2197.

Sweet, D. T., Jimenez, J. M., Chang, J., Hess, P. R., Mericko-Ishizuka, P., Fu, J.,Xia, L., Davies, P. F. and Kahn, M. L. (2015). Lymph flow regulates collectinglymphatic vessel maturation in vivo. J. Clin. Invest. 125, 2995-3007.

Tatin, F., Taddei, A., Weston, A., Fuchs, E., Devenport, D., Tissir, F. andMakinen, T. (2013). Planar cell polarity protein Celsr1 regulates endothelialadherens junctions and directed cell rearrangements during valvemorphogenesis. Dev. Cell 26, 31-44.

Thanassoulis, G., Campbell, C. Y., Owens, D. S., Smith, J. G., Smith, A. V.,Peloso, G. M., Kerr, K. F., Pechlivanis, S., Budoff, M. J., Harris, T. B. et al.(2013). Genetic associations with valvular calcification and aortic stenosis.N. Engl. J. Med. 368, 503-512.

Theis, M., deWit, C., Schlaeger, T.M., Eckardt, D., Kruger, O., Doring, B., Risau,W.,Deutsch, U., Pohl, U. andWillecke, K. (2001). Endothelium-specific replacement ofthe connexin43 coding region by a lacZ reporter gene. Genesis 29, 1-13.

Tsai, S., Dubovoy, A., Wainess, R., Upchurch, G. R., Jr, Wakefield, T. W. andHenke, P. K. (2005). Severe chronic venous insufficiency: magnitude of theproblem and consequences. Ann. Vasc. Surg. 19, 705-711.

Turner, C. J., Badu-Nkansah, K., Crowley, D., van der Flier, A. and Hynes, R. O.(2014). Integrin-alpha5beta1 is not required for mural cell functions duringdevelopment of blood vessels but is required for lymphatic-blood vesselseparation and lymphovenous valve formation. Dev. Biol. 392, 381-392.

Tzima, E., Irani-Tehrani, M., Kiosses, W. B., Dejana, E., Schultz, D. A.,Engelhardt, B., Cao, G., DeLisser, H. and Schwartz, M. A. (2005). Amechanosensory complex that mediates the endothelial cell response to fluidshear stress. Nature 437, 426-431.

Vermot, J., Forouhar, A. S., Liebling, M., Wu, D., Plummer, D., Gharib, M. andFraser, S. E. (2009). Reversing blood flows act through klf2a to ensure normalvalvulogenesis in the developing heart. PLoS Biol. 7, e1000246.

Wang, Y., Baeyens, N., Corti, F., Tanaka, K., Fang, J. S., Zhang, J., Jin, Y., Coon,B., Hirschi, K. K., Schwartz, M. A. et al. (2016). Syndecan 4 controls lymphaticvasculature remodeling during mouse embryonic development. Development143, 4441-4451.

Weber, B., Hafner, J., Willenberg, T. and Hoerstrup, S. P. (2016). Bioengineeredvalves for the venous circulation. Expert Rev. Med. Devices 13, 1005-1011.

Wigle, J. T. and Oliver, G. (1999). Prox1 function is required for the development ofthe murine lymphatic system. Cell 98, 769-778.

Wiig, H. and Swartz, M. A. (2012). Interstitial fluid and lymph formation andtransport: physiological regulation and roles in inflammation and cancer. Physiol.Rev. 92, 1005-1060.

Wirrig, E. E. and Yutzey, K. E. (2014). Conserved transcriptional regulatorymechanisms in aortic valve development and disease. Arterioscler. Thromb.Vasc. Biol. 34, 737-741.

Wong, C.W., Burger, F., Pelli, G., Mach, F. and Kwak, B. R. (2003). Dual benefit ofreduced Cx43 on atherosclerosis in LDL receptor-deficient mice. Cell Commun.Adhes. 10, 395-400.

1286

REVIEW Disease Models & Mechanisms (2017) 10, 1273-1287 doi:10.1242/dmm.030825

Disea

seModels&Mechan

isms

Wong,C.W.,Christen,T., Roth, I., Chadjichristos,C.E.,Derouette, J.-P., Foglia,B.F.,Chanson, M., Goodenough, D. A. and Kwak, B. R. (2006). Connexin37 protectsagainst atherosclerosis by regulating monocyte adhesion. Nat. Med. 12, 950-954.

Wu, B., Wang, Y., Lui, W., Langworthy, M., Tompkins, K. L., Hatzopoulos, A. K.,Baldwin, H. S. and Zhou, B. (2011). Nfatc1 coordinates valve endocardial celllineage development required for heart valve formation. Circ. Res. 109, 183-192.

Wu,B., Baldwin, H. S. andZhou, B. (2013). Nfatc1 directs theendocardial progenitorcells to make heart valve primordium. Trends Cardiovasc. Med. 23, 294-300.

Yutzey, K. E., Demer, L. L., Body, S. C., Huggins, G. S., Towler, D. A., Giachelli,C. M., Hofmann-Bowman,M. A., Mortlock, D. P., Rogers, M. B., Sadeghi, M. M.et al. (2014). Calcific aortic valve disease: a consensus summary from the allianceof investigators on calcific aortic valve disease. Arterioscler. Thromb. Vasc. Biol.34, 2387-2393.

Zhang, G., Brady, J., Liang, W.-C., Wu, Y., Henkemeyer, M. and Yan, M. (2015).EphB4 forward signalling regulates lymphatic valve development. Nat. Commun.6, 6625.

1287

REVIEW Disease Models & Mechanisms (2017) 10, 1273-1287 doi:10.1242/dmm.030825

Disea

seModels&Mechan

isms