IMP ELASTIC FIBRE SYSTEM

New Insights into Elastic Fiber Assembly

Jessica E. Wagenseil* and Robert P. Mecham

Elastic fibers provide recoil to tissues that undergo repeated stretch,such as the large arteries and lung. These large extracellular matrix(ECM) structures contain numerous components, and our understandingof elastic fiber assembly is changing as we learn more about the variousmolecules associated with the assembly process. The main componentsof elastic fibers are elastin and microfibrils. Elastin makes up the bulk ofthe mature fiber and is encoded by a single gene. Microfibrils consistmainly of fibrillin, but also contain or associate with proteins such asmicrofibril associated glycoproteins (MAGPs), fibulins, and EMILIN-1.Microfibrils were thought to facilitate alignment of elastin monomersprior to cross-linking by lysyl oxidase (LOX). We now know that theirrole, as well as the overall assembly process, is more complex. Elasticfiber formation involves elaborate spatial and temporal regulation of allof the involved proteins and is difficult to recapitulate in adult tissues.This report summarizes the known interactions between elastin and themicrofibrillar proteins and their role in elastic fiber assembly based on invitro studies and evidence from knockout mice. We also propose amodel of elastic fiber assembly based on the current data that incorpo-rates interactions between elastin, LOXs, fibulins and the microfibril, aswell as the pivotal role played by cells in structuring the final functionalfiber. Birth Defects Research (Part C) 81:229–240, 2007. VC 2008Wiley-Liss, Inc.

INTRODUCTIONElastic fibers provide elastic recoilto tissues such as the largearteries, lung, and skin. Theystore energy during the cardiacand respiratory cycles. The maincomponent of elastic fibers is theprotein elastin. During earlystages of fiber development, elas-tin deposits are usually associatedwith microfibrils, but it is notknown exactly how microfibrilsand elastin interact during elasticfiber assembly. Elastic fibers areassembled primarily during earlydevelopment and adult tissues arenot capable of properly recapitu-

lating the process. Damaged elas-tic fibers in adult tissues are oftenrepaired with improperly organ-ized material that does not func-tion normally. This leads to tissuesthat are too stiff and can result incardiac, cardiovascular, and pul-monary disease. Understandingthe assembly process will help usdesign interventions for properelastic fiber repair in adult tissues.Ultrastructural analysis of devel-

oping elastin-rich tissues showsthat elastin assembly typicallyoccurs in two phases. First, smallmicrofibril-rich elastin globulesappear near the surface of the cell

early in the process. Second,these coalesce to form largerstructures, which can be sheet-like lamellae in blood vessels or fil-aments in skin, lung, and othertissues. As the fiber matures, elas-tin becomes the major component,with a small number of microfibrilsdetectable on the periphery of thestructure.During the past few years, new

insight into elastic fibers and theircomponent proteins has been pro-vided by sometimes surprisingphenotypes in knockout mice.These studies identified severalunexpected proteins that play asignificant role in elastic fiber as-sembly. Gene array studies ofdeveloping mice show the coordi-nated expression patterns of elas-tic fiber proteins and emphasizethe critical spatial and temporaltiming of protein expression. Inaddition, the ability to visualizeextracellular matrix (ECM) assem-bly through video microscopyhighlights the important role thatthe cell plays in structuring theelastic fiber. This report will sum-marize the current information oninteractions between elastin andthe expanding family of microfi-brillar proteins, and the role ofeach protein in elastic fiber assem-bly. We also present a model ofelastic fiber assembly that incor-porates information from in vitroand in vivo studies and highlights

REVIE

W

VC 2008 Wiley-Liss, Inc.

Birth Defects Research (Part C) 81:229–240 (2007)

Jessica E. Wagenseil and Robert P. Mecham are from the Department of Cell Biology and Physiology, Washington UniversitySchool of Medicine, St. Louis, Missouri.

Grant sponsor: American Heart Association Fellowship; Grant number: 0525800Z (to J.E.W.); Grant sponsor: National Institutes ofHealth; Grant number: HL53325; HL71960; HL074138 (to R.P.M.).

*Correspondence to: Jessica E. Wagenseil, Department of Cell Biology and Physiology, CB 8228, Washington University School ofMedicine, 660 S. Euclid, St. Louis, MO 63110. E-mail: [email protected]

The Supplementary Material referred to in this article can be accessed at http://www.interscience.wiley.com/jpages/1542-975X/suppmat.

Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/bdrc.20111

the dynamic role played by the cellin organizing elastin in the ECM.

ELASTIC FIBER

COMPONENTS

Elastic fibers are the largeststructures in the ECM and consistof two morphologically distinctcomponents. The major compo-nent is elastin, which is a cross-linked polymer of the monomericsecreted form of the protein tro-poelastin (Sandberg et al., 1971).The second component is visual-ized as small, 10–15-nm ‘‘microfi-brils’’ that localize to the peripheryof the fiber in adult tissues. Whileelastin arises from a single gene,the composition of microfibrils ismore complex. The major struc-tural element of microfibrils is con-tributed by the fibrillins (Sakaiet al., 1986; Zhang et al., 1994).Numerous other proteins associatewith microfibrils or with elastinitself, including the microfibril as-sociated glycoproteins (MAGPs),fibulins and EMILIN-1 (for a com-plete list see Kielty et al., 2002).Based on ultrastructural studies,microfibrils are thought to providea scaffold that facilitates elastinmolecular alignment and subse-quent cross-linking, which is cata-lyzed by one or more members ofthe lysyl oxidase (LOX) gene family(Csiszar, 2001; Kagan and Li,2003; Lucero and Kagan, 2006).It is still uncertain how each of

the elastic fiber components con-tribute to the formation of a func-tional fiber, but clues can beobtained from in vitro studies withisolated proteins and from knock-out mouse models. A summary ofthe biological properties of themajor proteins known to be asso-ciated with elastic fibers or withelastic fiber assembly is presentedbelow. Particular attention is givento the phenotypes apparent inmice in which the genes havebeen inactivated.

Tropoelastin/Elastin

Tropoelastin is a 60–70-kDa pro-tein composed of alternatinghydrophobic and lysine-containingcross-linking domains (Gray et al.,

1973). Of the �40 lysine residuesin the secreted monomer, all butapproximately five are modified byLOX to form cross-links. This highdegree of cross-linking is responsi-ble for the stability and insolubilityof the protein. Tropoelastin is ca-pable of undergoing coacervationunder physiological conditions in aprocess thought to facilitate self-assembly (Cox et al., 1974; Volpinet al., 1976; Clarke et al., 2006).Coacervation is the result of spe-cific interactions between thehydrophobic domains induced byan increase in temperature (Toon-kool et al., 2001). Studies suggestthat coacervation is an importantprerequisite for cross-linking (Nar-ayanan et al., 1978), but it is notknown at which point in assemblythis occurs. The accepted modelhas been that interactions be-tween tropoelastin and microfibrilsin the extracellular space serve tofacilitate alignment of cross-link-ing domains within tropoelastinprior to cross-linking (Kielty et al.,2002). Alternatively, tropoelastinmay self assemble in the absenceof microfibrils (e.g., through coac-ervation) on the cell surface andthen be transferred as aggregatesto microfibrils. In this case, cross-linking could occur on the cell sur-face within the microaggregate,once the microaggregate is trans-ferred to the microfibril, or, mostlikely, at both steps (see model inFig. 7). The propensity of tropoe-lastin to self-assemble (i.e., coac-ervate) also raises the possibilitythat some of the associated pro-teins thought to promote fiber as-sembly may actually serve to limitassembly so that large elastinaggregates do not form (Kozelet al., 2006).Mice lacking the elastin gene

(eln2/2) die within a few days ofbirth of vascular obstruction dueto smooth muscle cell (SMC) over-proliferation (Li et al., 1998a). Thevessel wall is thickened at thetime of death in these animals andthe SMCs are misarranged. Micehaving one elastin allele (eln1/2),in contrast, live a normal lifespandespite substantial hypertensionand smaller vessels with thinnerwalls and altered mechanical prop-

erties (Wagenseil et al., 2005).Surprisingly, they also have anincreased number of lamellar unitsin the arterial wall (Li et al.,1998b; Faury et al., 2003) (a la-mellar unit is defined as a layer ofelastin with its associated SMCs).Lamellar units are established inearly development during forma-tion of the vessel wall and lamellarnumber is linearly related to walltension in most mammals(Wolinsky and Glagov, 1967). Ourrecent morphologic studies ofdeveloping eln2/2 and 1/2 aortashow that the changes in wallstructure occur in the last fewdays before birth (between em-bryonic day 18 [E18] and post-natal day 0 [P0]) just as the pres-sure and blood flow are signifi-cantly increasing.

Fibrillins

Fibrillins are large (�350 kDa),cysteine-rich glycoproteins thatmake up the major structural ele-ment of microfibrils (Sakai et al.,1991). A total of three fibrillins(fib-1, -2, and -3) have been iden-tified (Sakai et al., 1986; Leeet al., 1991a; Corson et al.,2004), although the gene for fibril-lin-3 has been disrupted in rodentsdue to chromosome rearrange-ment (Corson et al., 2004). Theprimary structure of fibrillin isdominated by calcium binding epi-dermal growth factor (EGF)-likedomains that develop a rod-likestructure in the presence of cal-cium (Downing et al., 1996).Fibrillin-1 isolated from humanfibroblasts appears as linear fibrilswith a beaded periodicity (Sakaiet al., 1991). Fibrillins have Arg-Gly-Asp (RGD) sequences thatinteract with integrins (Pfaff et al.,1996; Sakamoto et al., 1996; Baxet al., 2003) and heparin-bindingdomains that interact with cell sur-face heparan sulfate proteogly-cans (Tiedemann et al., 2001;Ritty et al., 2003), suggesting thatfibrillins directly signal cellsthrough these receptors. Further-more, in vivo assembly of fibrillinmay require cell-surface receptorsin a similar manner to fibronectin(Wu et al., 1995). Fibrillins also

230 WAGENSEIL AND MECHAM

Birth Defects Research (Part C) 81:229–240, (2007)

have a major role in binding andsequestering growth factors, suchas TGF-b, into the ECM (Neptuneet al., 2003).While microfibrils are most noted

for their association with elasticfibers, they can also be found with-out elastin in the ciliary zonules ofthe eye and in the circulatory sys-tem of invertebrates where theyserve a mechanical role in thosetissues. The addition of elastin isan evolutionary adaptation in ver-tebrate animals to handle the highpulsatile pressures of a closed cir-culatory system (Faury, 2001).Fibrillin-1 and -2 bind tropoelastinin solid phase binding assays(Trask et al., 2000b).Mutations in the fibrillin-1 gene

lead to Marfan syndrome, whilefibrillin-2 mutations lead to con-genital contractural arachnodac-tyly (Lee et al., 1991b; Park et al.,1998; Chaudhry et al., 2001).Mice lacking the fibrillin-1 gene(fbn12/2) die within two weeks ofbirth from vascular and pulmonarycomplications, including rupturedaortic aneurysms, impairedbreathing and diaphragmatic col-lapse. The elastic fibers in theaorta are abnormally thin andfragmented (Carta et al., 2006).Mice lacking the fibrillin-2 gene(fbn22/2) develop syndactyly,but show no defects in the vascu-lar and pulmonary systems andhave a normal lifespan. The elasticfibers in the aorta look normal(Arteaga-Solis et al., 2001;Chaudhry et al., 2001; Cartaet al., 2006). Mice deficient inboth fibrillins (fbn12/2;fbn22/2)die in utero. At E14.5, there aretraces of elastic fibers betweenSMC layers in fbn12/2:fbn22/2mice, but nothing close to theorganized lamellar units visible inwild-type mice. Half of fbn11/2:fbn22/2 mice die in utero andshow a more severe vascular phe-notype than fbn12/2 mice. Theelastic fibers in the aorta are eventhinner and more fragmented thanfbn12/2 mice. This indicates thatfibrillin-1 may compensate for theloss of fibrillin-2 and that fibrillin-2may be involved in the initial as-sembly of aortic microfibrils (Cartaet al., 2006).

MAGP-1 and MAGP-2

MAGP-1 was the first microfibrillarprotein isolated and completelycharacterized (Gibson et al.,1986). A second member of theMAGP family, MAGP-2, was subse-quently identified (Gibson et al.,1996) (the gene symbols forMAGP-1 and -2 are mfap2 andmfap5, respectively). The MAGPsare small (�20 kDa) glycoproteinsthat localize to the beaded regionof microfibrils (Henderson et al.,1996). MAGP-1 has been localizedto most microfibrils and is thoughtto be the only protein besidesfibrillin that is a constitutive com-ponent of the microfibril.Specific interactions between

MAGP-1 and fibrillin-1 have beenshown by coimmunoprecipitationstudies (Trask et al., 2000a) andrecombinant MAGP-1 binds torecombinant fibrillin-1 and torecombinant tropoelastin (Brown-Augsburger et al., 1994; Jensenet al., 2001) in ligand blottingassays. MAGP-1 has also beenshown to bind to fibrillin-2 in yeasttwo-hybrid, ligand blotting, andsolid phase binding assays (Wer-neck et al., 2004). MAGP-2 bindsto fibrillin-1 and -2 in yeast two-hybrid studies (Penner et al.,2002) and in solid phase bindingassays (Hanssen et al., 2004). Theability of MAGP-1 to bind both tro-poelastin and fibrillin suggestedthat it might play an importantrole in elastic fiber assembly byserving as a bridging moleculebetween its two binding partners.In mice lacking the MAGP-1 gene(mfap22/2), however, elasticfiber assembly is normal and thereare no structural abnormalities inany of the elastin-rich tissues (ourunpublished results). Hence,MAGP-1 is not necessary for nor-mal elastic fiber assembly.The mfap22/2 mice developed

in our laboratory were originally ina mixed background of BlackSwiss (BSw) and 129. The miceshowed phenotypes of variablepenetrance that included in-creased body weight, skeletaldeformations, and bone lesions.When the mice were back-crossedinto the C57Bl/6 background,

most of these phenotypes disap-peared. When back-crossed intothe BSw background, many of theoriginal phenotypes reemerged.These phenotypes include a bleed-ing diathesis in both large andsmall vessels, an increase in bodyfat and muscle mass, bonelesions, delayed wound healing,and altered cell adhesion (Wein-berg JS, Broekelmann TJ, PierceRA, Werneck CC, Segade F, Knut-sen RH, Mecham RP, unpublishedresults). All of the phenotypesshow variable penetrance with theexception of the bleeding diathesis,which is completely penetrant in allbackgrounds, including C57Bl/6.The phenotypes observed in theBSw mfap22/2 mouse are similarto matricellular proteins that bindmatrix proteins, but do not contrib-ute to the structural integrity ofthe ECM, such as thrombospondin(Kyriakides et al., 1998) andsecreted protein acidic and rich incysteine (SPARC) (osteonectin)(Bradshaw et al., 2003).

Fibulins

Fibulins are 50–200 kDa in size(Timpl et al., 2003) and have a se-ries of calcium binding EGF-likedomains followed by a carboxy-terminal fibulin-like domain (Ar-graves et al., 2003). There arefive members of the fibulin family.Fibulin-1 associates with somebasement membranes and withthe amorphous elastin core ofelastic fibers, but not with individ-ual microfibrils (Roark et al.,1995). Fibulin-2 also associateswith basement membranes (Panet al., 1993) and strongly bindstropoelastin (Sasaki et al., 1999).Unlike fibulin-1, fibulin-2 bindsdirectly to fibrillin-1 and has beenshown to localize to the interfacebetween the amorphous elastincore and microfibrils in skin (Rein-hardt et al., 1996). Fibulin-3 doesnot bind to fibrillin-1 (El-Hallouset al., 2007) and has been shownto only weakly interact with tro-poelastin (Kobayashi et al., 2007).Fibulin-3 is expressed in capilla-ries, but not large blood vessels(Giltay et al., 1999) and mostlikely does not have a role in elas-

NEW INSIGHTS INTO ELASTIC FIBER ASSEMBLY 231


tic fiber assembly. Fibulin-4 bindsfibrillin-1 (El-Hallous et al., 2007)and shows moderate affinity fortropoelastin. Fibulin-4 has beenlocalized to microfibrils (Kobayashiet al., 2007). Fibulin-5 colocalizeswith elastic fibers in vitro (Naka-mura et al., 2002) and can beseen at the elastin/microfibrilinterface in the aorta (Yanagisawaet al., 2002). Fibulin-5 binds tro-poelastin and shows weak affinityfor the C- and N-terminus of fibril-lin-1 (Yanagisawa et al., 2002).Fibulin-5 also binds cell surfaceintegrins (Nakamura et al., 2002).The splice D variant of fibulin-1

has been implicated in humansyndactyly (Debeer et al., 2002).Mice deficient in fibulin-1 (fbln12/2)die within 24–48 hr of birth andexhibit spontaneous bleeding asearly as E12.5. Fbln12/2 miceshow defects in the kidney, lungand capillaries, but the heart andlarge elastic arteries look normal.These findings suggest that fibu-lin-1 is not involved in elastic fiberassembly, but plays a role inorganizing basement membraneand microfibrils in angiogenesisand capillary formation. Otherfibulins, such as fibulin-2, mayalso compensate for the loss offibulin-1 in these mice (Kostkaet al., 2001). No known humandiseases have been associatedwith fibulin-2 and a fibulin-2knockout mouse has no obviouselastin phenotype (Chiu, M-L, per-sonal communication). A mutationin fibulin-3 has been implicated ina macular dystrophy (Stone et al.,1999). A fibulin-3 knockout mousehas not yet been described.A mutation in the fibulin-4 gene

has been recently linked to an auto-somal recessive form of cutis laxa.The patient had loose skin, emphy-sema, aortic tortuosity, and anascending aortic aneurysm (Huc-thagowder et al., 2006). Mice lack-ing the fibulin-4 gene (fbln42/2)show the most severe phenotypeof the existing fibulin knockoutmice. Fbln42/2 mice die perina-tally with severe lung and vasculardefects. The aorta is narrowed byE12.5 and tortuous by E15.5. Theelastic fibers contain irregularelastin aggregates with evenly dis-

tributed rod-like filaments visiblewithin the usual amorphous glob-ules. Elastin cross-links are greatlyreduced, but the expression of tro-poelastin and LOX is normal(McLaughlin et al., 2006). Therod-like filaments in the fbn42/2elastic lamellae resemble thoseobserved in the aorta of chicks fedb-aminopropionitrile (BAPN), an ir-reversible LOX inhibitor. BAPNprevents the proper cross-linkingof elastin by LOX. The rod-like fila-ments in the BAPN-fed chickshave been identified as proteogly-cans (Contri et al., 1985). Thesimilar appearance of fbn42/2mouse lamellae and poorly cross-linked chick lamellae suggests thatfibulin-4 facilitates cross-linking ofelastin at some stage in the as-sembly process. Like fibulin-4,mutations in the fibulin-5 genehave also been linked to cutis laxa(Loeys et al., 2002; Markova et al.,2003; Hu et al., 2006). Fibulin-5knockout (fbln52/2) mice showsimilar phenotypes to the fibulin-42/2 mice, but with less severityas they live a normal lifespan.Fbln52/2 mice have loose skin,tortuous aortas, and disruptedelastic fibers with abnormal elastinaggregates (Nakamura et al.,2002; Yanagisawa et al., 2002).

Elastin Microfibril InterfaceLocated Protein (EMILIN)-1

EMILIN-1 is one of four membersof the EMILIN protein family thathave similar structural featuresincluding an EMI domain, an a-hel-ical domain and a gC1q domain(Colombatti et al., 2000). EMILINstands for ‘‘elastin microfibril inter-face located protein.’’ As the namesuggests, EMILIN-1 is found inelastin-rich tissues and is localizedto the interface between amor-phous elastin and microfibrils.Anti-EMILIN-1 antibodies promotethe formation of elastin aggregatesrather than fibers in SMC cultures(Bressan et al., 1993). EMILIN-1binds elastin in solid-phase bindingassays and both elastin and fibu-lin-5 in immunoprecipitation as-says (Zanetti et al., 2004).No known human disease has

been linked to mutations in EMILIN-

1. EMILIN-1 knockout (emilin12/2)mice live a normal lifespan, butshow alterations in the elastic fibersof the skin and aorta. The amor-phous elastin core appears irregu-larly shaped and split and the cellsappear to have lost their connectionsto the elastic fibers. EMILIN-12/2embryonic fibroblasts form elastinfibers in vitro that are thinner andstraighter than control fibroblastsand lack the complete colocalizationof elastin and fibulin-5 observed incontrols. Fibulin-2 and fibrillin-1fibers appear normal (Zanetti et al.,2004). Similar to eln1/2 mice,emilin12/2 mice have significanthigh renin-induced hypertensionand vessels with smaller diametersand thinner walls (Faury et al.,2003; Zacchigna et al., 2006). Theabsence of EMILIN-1 increases TGF-b signaling and, interestingly, thevascular phenotype can be reversedby inactivating one TGF-b allele(emilin-12/2; tgf-b1/2 mice) (Zac-chigna et al., 2006).

LOX

There are five members of the LOXfamily: LOX and four LOX-like pro-teins (LOXL1–4). Each protein con-tains an N-terminal signal peptide,a variable region in the center ofthe molecule, and a C-terminalregion that shows significantsequence similarity within the fam-ily. LOX and LOXL1 are most similarto each other, while LOXL2–4appear to belong to a separate LOXsubfamily (Lucero and Kagan,2006). Enzyme activation is de-pendent on processing from thepro forms by bone specific pro-teases, such as bone morphogenicprotein-1 (BMP-1) (Borel et al.,2001). LOX and LOXL1 cross-linkboth elastin and collagen. In elas-tin, lysine residues are modified toform bi-, tri-, and tetra-functionalcross-links, with desmosine andisodesmosine being unique to elas-tin in vertebrates (Partridge et al.,1963, 1964). LOX can cross-linkrecombinant tropoelastin in vitrowithout the presence of any otherelastic fiber proteins (Bedell-Hoganet al., 1993). LOX localizes to theelastin/microfibril interface in aorta(Kagan et al., 1986) and LOX and



LOXL1 colocalize in most elastic tis-sues (Hayashi et al., 2004). LOXL1specifically localizes to sites of elas-togenesis and interacts with fibu-lin-5 (Liu et al., 2004). The proregion plays a role in directing LOXand LOXL1 to elastic fibers in vitro,and sequence differences in the proregions may account for some ofthe functional differences betweenthe enzymes (Thomassin et al.,2005).As might be expected for an

enzyme so important for collagenand elastin maturation, numerousdisease states have been linked toalterations in LOX activity. Theseinclude copper metabolism or trans-port disorders (Sacks, 2003), as wellas Alzheimer disease (Gilad et al.,2005), amyotrophic lateral sclerosis(ALS) (Li et al., 2004), and evencancer progression (Erler and Giac-cia, 2006). Recently, single nucleo-tide polymorphisms in LOXL havebeen identified as risk factors forglaucoma (Thorleifsson et al.,2007).Mice lacking the LOXgene (lox2/2)

die at the end of gestation or justafter birth with large aortic aneu-rysms and diaphragmatic collapse.The elastic lamellae in the aorta arefragmented and discontinuous andthe vessel wall is thicker with adecreased lumen size. Desmosinecross-links are reduced approxi-mately 60% in the aorta and lungs(Maki et al., 2002; Hornstra et al.,2003). Lox2/2 mice also showimpaired airway development inthe lungs and abnormal collagenand elastin fibers in the skin (Makiet al., 2005). Mice lacking the LOXLgene (loxl12/2) live a normal life-span and do not show the obviousvascular and pulmonary defectsapparent in lox2/2 mice. However,postpartum loxl12/2 mice developpelvic organ prolapse, loose skin,enlarged airspaces and vasculardeformities. Therefore, LOXL1 mayplay a unique role in elastic fiberhomeostasis (Liu et al., 2004).

ELASTIC FIBER ASSEMBLY

The proteins necessary for elasticfiber assembly must be regulatedboth spatially and temporally. Thiscomplex regulatory process can be

examined with gene array andmorphologic studies of developingelastic tissues. Elastic fiber assem-bly can also be recapitulated invitro and watched in real timeusing fluorescently-labeled pro-teins or antibodies. Live imagingstudies have highlighted the activerole of the cell in elastic fiber as-sembly as it coordinates anddirects proteins to the appropriatelocations.

Coordinated Gene Expression

Most of the elastic fiber proteinsare expressed from the secondhalf of development (around E14in the mouse) and show a steadyrise through the early postnatalperiod (around P7–14 in mice).This is followed by a decrease inexpression to low levels that per-sist through adulthood (Kelleheret al., 2004). In adult tissue, elas-tic fibers are difficult to repair andelastin often does not polymerizeor form functional three-dimen-sional (3D) fibers (Chrzanowskiet al., 1980; Shifren and Mecham,2006). This is probably becausethe complicated temporal and spa-tial regulation is difficult to reca-pitulate in adult tissues.Our laboratory has recently

completed a gene expressionstudy of developing mouse aortafrom E12 to P180. The data wasgenerated following similar proto-cols to our previous array studies(Kelleher et al., 2004; McLeanet al., 2005), but was completedusing the Affymetrix Mouse 430v2chip (39,000 genes) (Santa Clara,CA) that includes more matrixprobes. All of the elastic fibergenes mentioned above, exceptfor fibrillin-2 (fbn2) and MAGP-1(mfap2), show an increase inexpression from E14 through�P14 followed by a decrease tolow levels where expressionremains steady from P60 to P180.Fbn2 and mfap2, on the otherhand, start out with high expres-sion values and are probablyturned on prior to E14. Expressionof these proteins remains highuntil P0 when they quickly drop tolower levels and oscillate aroundthis value through adulthood. The

fbn2 expression pattern supportsthe hypothesis that fibrillin-2 isinvolved in the initial assembly ofmicrofibrils, but plays a lesser rolein elastic fiber assembly (Cartaet al., 2006). The relative amountsof each protein on the gene arraymay explain some phenotypes inthe knockout mice. For example,lox and loxl1 show almost identi-cal expression patterns, but loxamounts are substantially higherthan loxl1 and the lox2/2 mousehas a much more severe pheno-type than loxl12/2.The coordinated expression pat-

tern of all the proteins, exceptfibrillin-2 and MAGP-1, impliesthat each protein is necessary fornormal elastic fiber assembly. Italso suggests that homologousproteins in the same family maycompensate for each other inknockout mouse models. Forexample, if fibulin-1 is absent,fibulin-2 may be able to compen-sate during development and pro-duce a relatively normal aortabecause the temporal expressionpattern is correct. The severity ofthe phenotype may depend ongene dosage of total fibulins insteadof the presence or absence of spe-cific fibulin proteins. Our laboratoryhas recently shown that mice withdecreasing amounts of elastin (from100 to 30% normal levels) haveincreasingly severe vascular (Hiranoet al., 2007) and pulmonary(Shifren et al., 2007) phenotypes.If elastin amounts fall below 30%,the mice cannot survive postnatally(Hirano et al., 2007).

Cellular Activity

Cells are actively involved inassembling many ECM molecules,including fibronectin (Mao andSchwarzbauer, 2005) and collagen(Birk and Trelstad, 1986; Kadler,2004), but their role in elastic fiberassembly has been overlooked.Recent live imaging studies in ourlaboratory show that cells areactively involved in the organiza-tion and alignment of elastinaggregates into linear structuresand the deposition of elastinaggregates onto preexisting fibers(Kozel et al., 2006; Czirok et al.,



2006). In these studies, cells weretransfected with an expressionconstruct encoding tropoelastinfused with pTimer, which is a fluo-rescent protein that changes fromgreen to red fluorescence overabout eight hours, allowing us todiscriminate between new (green),old (red) and combined new andold (yellow) elastin. In transfectedcells, small green aggregates ofnewly formed elastin are visible onthe cell surface. These aggregatescoalesce into larger structures andremain in contact with the celllong enough for some moleculesto make the green to red transi-tion. The large, yellow aggregatesmove on the cell surface towardpreexisting red fibers where theyare deposited and the cell movesaway. The initial formation ofsmall aggregates is termed‘‘microassembly’’, while the trans-fer of the older, sometimes largeraggregates to existing fibers istermed ‘‘macroassembly’’ (Kozelet al., 2006). Pulse-chase immu-nolabeling of the fibroblast-like ratfetal lung fibroblast (RFL-6) cellsdemonstrates that tropoelastinglobules aggregate in a hierarchi-cal manner, creating progressively

larger fibrillar structures. By ana-lyzing the correlation between celland ECM movements, both theaggregation process and shapingthe aggregates into fibrillar formwas found to be coupled tocell motion. The motion of nonad-jacent cells becomes more coordi-nated as the physical size ofelastin-containing aggregates in-creases, indicating that the forma-tion of elastic fibers involves theconcerted action and motility ofmultiple cells (Czirok et al., 2006).Frames from one of the pTimer-

tropoelastin live imaging studiesare shown in Figure 1. A cell withyellow elastin on its surface can beseen tracking, stretching, andreleasing an extracellular elastinfiber. Another cell that starts outwith green elastin aggregates,gathers the aggregates into alarger structure that turns yellowand is deposited in the ECM, pre-sumably on an unstained preexist-ing fiber. A movie with all the timeframes for Figure 1 is in the onlineSupplemental Material (MovieA{MOVA}; available online athttp://www.interscience.wiley.com/jpages/1542-975X/suppmat). Unlikecellular assembly of collagen and

fibronectin, the nature of the cel-lular interaction with elastin is rel-atively unknown. Possible bindingpartners include elastin-bindingprotein (Hinek et al., 1988; Wrennet al., 1988), integrins (Rodgersand Weiss, 2004), and cell-surfaceglycosaminoglycans (Broekelmannet al., 2005).Additional studies in our labora-

tory have focused on determiningthe interactions between elastinand other elastic fiber proteinsduring in vitro assembly. To deter-mine if LOX is present at the earlystage of elastin assembly, cellswere incubated with BAPN to in-hibit LOX activity and elastin as-sembly was followed using videoimaging. The presence of BAPNdecreased in vitro elastin matrixassembly in a dose dependentmanner (Fig. 2), suggesting thatcross-linking is important for ag-gregate formation on the cell sur-face and that LOX or some othermember of the LOX family is pres-ent at elastin assembly sites. Liveimaging studies have also been in-structive as to the participation ofother elastic fiber proteins in fiberformation. For example, imagingstudies with fluorescently labeled

Figure 1. Assembly of elastin on the cell surface and transfer to extracellular microfibrils. Frames from a live imaging study withrat fetal lung fibroblast (RFL-6) cells transfected with pTimer-tropoelastin. A movie with all of the time frames is in the online Sup-plemental Material (Movie A). Imaging was started 1 day after transiently transfecting confluent cells. Cells were imaged for24.5 hr with 22 min between frames. pTimer-tropoelastin starts out with green fluorescence (new eln) and changes to red afterabout 8 hr (old eln). Yellow staining indicates colocalization of new and old eln. The arrow shows a cell with yellow elastin stainingon its surface that moves toward an existing elastin fiber, tracks along it to the right, stretches it, then follows it to the left beforereleasing the fiber. The arrowhead shows small green cell-associated elastin aggregates that coalesce into a large yellow aggregatethat is released by the cell into the ECM. Additional cell and elastin aggregate and fiber movement can be seen in Movie A.



antibodies show that elastin fibersare formed at the same location aspreviously existing fibronectin,fibrillin-1, MAGP-1 and fibulin-5fibers. The proteins are temporallyarranged into fibers in that order:fibronectin, fibrillin-1, MAGP-1,fibulin-5, and elastin in a varietyof elastin producing cells. Figure 3shows frames from a live imagingstudy where the cells have beenincubated with antibodies to fibro-nectin and fibrillin-1. At the begin-ning of the imaging period, only fi-

bronectin fibers are apparent. Af-ter 32 hr of imaging and twosubsequent antibody incubations,fibrillin-1 fibers are visible and areforming in the same location aspreviously existing fibronectinfibers. A movie with all of theframes from the live imagingstudy is included in the onlineSupplemental Material (MovieB{MOVB}). Fibronectin is not gen-erally considered an elastic fiberprotein, but an initial fibronectinscaffold is necessary for collagen

type I and type III matrix assem-bly in vitro (Velling et al., 2002)and may be associated with elasticfiber assembly, at least in vitro.LOX binds to fibronectin with asimilar affinity to tropoelastin, butdoes not act on fibronectin enzy-matically. In fibronectin null (fn2/2)fibroblasts, LOX processing fromthe pro form to the active form isreduced; therefore, fibronectinmay serve as a scaffold for enzy-matically active LOX (Fogelgrenet al., 2005). Fn2/2mice die beforeE14.5 with extensive systemicdefects (George et al., 1993).Figure 4 shows frames from a live

imaging study in which the cellshave been incubated with antibod-ies to fibulin-5 and tropoelastin. Atthe start of the imaging period sev-eral thin fibulin-5 fibers andnumerous aggregates are visible.There is one elastin fiber that coloc-alizes with a fibulin-5 fiber andthere are a few elastin aggregates.After 27 hr, both fibulin-5 and elas-tin show numerous colocalizedaggregates that are linearlyarranged in fibers. Not all fibulin-5and elastin aggregates are colocal-ized, but approximately half are,indicating the fibulin-5 may play arole in the microassembly stepwhere elastin aggregates areformed on the cell surface. Duringthe imaging period, the fibers shiftlocation in the imaging plane due toactive rearrangement by the cells.This can be better appreciated inthe complete movie in the onlineSupplemental Material (MovieC{MOVC}).

Figure 2. Inhibition of LOX activity blocks elastin assembly. RFL6 cells untreated or treated with 1 or 2 mM BAPN, an inhibitor ofLOX, for five days after plating at 50% confluence. Cells were fixed with methanol and stained with a polyclonal antibody to mousetropoelastin (green) and Hoechst nuclear stain (blue).

Figure 3. Colocalization of fibronectin and fibrillin during fiber assembly. Frames froma live imaging study of bovine pigmented epithelial (PE) cells stained with a polyclonalantibody to fibronectin (fn, green) and a monoclonal antibody to fibrillin-1 (fbn-1,red). A movie with all of the time frames is in the online Supplemental Material (MovieB). Imaging began one day after the cells were plated at 50% confluency. Cells wereimaged for 68 hr with 40 min between frames. Primary and secondary antibodies wereadded sequentially to the cells for 1 hr each, then rinsed before imaging. The second-ary antibodies alone showed no signal and the secondary antibodies did not cross-react. Antibodies were added and rinsed again 22 and 32 hr after imaging began. Atthe start of the imaging period (not shown), several small fn fibers and a few fbn-1aggregates are visible. By 22 hr, more fn fibers and fbn-1 aggregates can be seenwith fbn-1 aggregates colocalizing with fn fibers. By 32.6 hr, several small fn fibershave combined to make larger fibers and some of the fbn-1 aggregates have formedfibers that colocalize with previously formed fn fibers. In Movie B, the cells can beseen moving, combining and breaking fibers.



Live imaging studies with lightmicroscopy are useful for studyingthe dynamics of fiber assembly,

but do not have sufficient resolu-tion to examine interactionsbetween individual cells and elastic

fibers. Electron microscopy (EM)has the resolution to study cellularinteractions and has brought newunderstanding to the role of cells inthe assembly of other matrix mole-cules, like collagen. EM studieshave shown that collagen fibrillo-genesis is initiated within plasmamembrane recesses, presumablyunder close cellular regulation (Birkand Trelstad, 1986). Similar struc-tures, given the name fibripositors,have been described in ligamentcells (Canty et al., 2004). Our labo-ratory has recently completed anEM study of developing mouseaorta from E14 to P0. The imagesshow the dramatic increase in elas-tin deposition during the last fewdays of development and are con-sistent with the gene array studies(Kelleher et al., 2004). Figure 5 isan EM image of a P0 aorta takennear the center of the vessel wall.Cross-sections of individual colla-gen fibers and a network of microfi-brils can be seen next to the amor-phous elastin lamella. Small glob-ules of elastin are also present atthe cell surface and interspersedthrough the microfibrillar network.There appears to be an elastin ag-gregate being secreted directlyfrom a cell membrane protrusion tojoin a larger aggregate at the cellsurface. This behavior is consistentwith our live imaging studies (Kozelet al., 2006) and is reminiscent ofcollagen fibripositors (Canty et al.,2004).

Model

Figure 6A summarizes the knownbinding interactions between elas-tin and the microfibrillar proteinsdiscussed in this review. Figure 6Bshows the localization of each pro-tein as described in microscopystudies of elastic tissues. Fibulin-1is the only protein, besides elastin,that has been localized to theamorphous core of elastic fibersusing antibodies in EM studies(Roark et al., 1995). Ironically, thefbn12/2 mouse seems to havenormal arteries, but severe capil-lary defects (Kostka et al., 2001).Obviously, assembly of elasticfibers is a complex process that

Figure 4. Distribution of fibulin-5 and tropoelastin in cultured cells. Frames from a liveimaging study of fetal calf ligament (FCL) cells stained with a monoclonal antibody forfibulin-5 (fbln-5, red) and a polyclonal antibody for tropoelastin (eln, green). A moviewith all of the time frames is in the online Supplemental Material (Movie C). Imagingbegan five days after the cells were plated at 50% confluence. Cells were imaged for40 hr with 30 min between frames. The secondary antibodies alone showed no signaland the secondary antibodies did not cross-react. Primary and secondary antibodieswere added sequentially to the cells for 1 hr each, then rinsed before imaging. At thestart of the imaging period, several thin fibers and numerous fbln-5 aggregates arevisible. One eln fiber and several eln aggregates colocalize with the fbln-5 structures.By 27 hr, most of the fbln-5 and eln aggregates are aligned in colocalized fibers. Thiscoordinated reorganization can be followed in Movie C.

Figure 5. Electron micrograph of elastic fiber assembly in the developing mouse aorta.P0 mouse aorta showing secretion and organization of elastin by cells. Collagen fibers(C) can be seen in cross-section near the large amorphous elastin (E) lamella. Microfi-brils (M) are located between the cell and the elastin. Smaller elastin aggregates (e) dec-orate some of the microfibrils. An elastin aggregate appears to be forming within aplasma membrane protrusion (arrow) to join a larger aggregate at the cell surface.



involves interactions betweennumerous proteins and binding,and localization studies, combinedwith mouse knockout data on these

proteins, are not always consistent.Based on the published data andour unpublished studies, however,we have developed a hypothetical

model for elastic fiber assemblythat includes pivotal roles for indi-vidual cells, elastin, LOX and/orLOXL, fibulin-4 and/or -5, andmicrofibrils (Fig. 7).In our model, tropoelastin is

secreted by the cells and organ-ized into cross-linked aggregateson the cell surface (Kozel et al.,2006), possibly through interac-tions with cell surface glycosami-noglycans (Broekelmann et al.,2005) and one or more LOX familymembers. Fibulin-4 and/or -5interact with these aggregates tofacilitate cross-linking (McLaughlinet al., 2006) or to limit the aggre-gate size. The elastin aggregatesremain on the cell surface longenough for new elastin to besecreted from the cell and addedto the aggregates (Kozel et al.,2006). The cell actively collectssmall aggregates on its surfaceinto larger aggregates that arethen transferred to preexistingmicrofibrils. The microfibrils arecomposed primarily of fibrillin-1and/or -2, but probably containother microfibril-associated pro-teins, such as MAGPs. The microfi-brils bind to the cell surfacethrough integrins. Fibulin-4 and -5both bind fibrillin-1 (El-Hallouset al., 2007; Yanagisawa et al.,2002) and may help transfer theelastin aggregates to the microfi-bril. The elastin aggregates on themicrofibril coalesce into largerstructures and are further cross-linked by LOX and/or LOXL to formthe complete and functional elasticfiber. Fibulin-4 and/or -5 mayfacilitate the cross-linking step. Itis important to note that few ofthe proteins in the model canbe detected by antibodies withinthe complete, mature, amorphouselastic fiber. They may be coatedwith elastin so that the epitope isunavailable, they may be physi-cally displaced and moved to theedge of the elastic fiber or theymay be degraded by proteasesduring the final cross-linking step.

CONCLUSIONS

We have summarized the currentdata on elastin and elastic fiberproteins and their role in elastic

Figure 6. Binding interactions and spatial localization of elastic fiber proteins.A: Known binding interactions between elastin and the microfibrillar proteins.B: Observed localization of elastin and the microfibrillar proteins with relationship tothe amorphous elastin core and the associated microfibrils.

Figure 7. Model of elastic fiber assembly. (1) Tropoelastin is transported to assemblysites on the plasma membrane where it is organized into small aggregates that arecross-linked by a LOX. Cell surface receptors/binding proteins such as heparan sulfateproteoglycans at these sites may serve to assist with the initial assembly steps. Inter-action with Fibulin-4 and/or -5 may facilitate cross-linking or possibly help limit thesize of the aggregates. (2) The aggregates remain on the cell surface while newlysecreted elastin is added to increase the size. (3) The aggregates are then transferredto extracellular microfibrils, which interact with the cell through integrins. Fibulin-4and/or -5, or another microfibril-associated protein, may assist the transfer of elastinaggregates to the microfibril. (4) Elastin aggregates on the microfibril coalesce intolarger structures. Fibulin-4 and/or -5 may facilitate this process. (5) The elastin aggre-gates are further cross-linked by LOX to form the complete elastic fiber.



fiber assembly. We have also pre-sented a working model thatassimilates much of the current in-formation obtained through bothin vivo and in vitro methods. Thenext step in understanding elasticfiber formation will be identifyinghow key elastic fiber moleculesinteract at the cell surface, howthese molecules are sorted withinthe cell and transported to sites ofassembly, how the cell regulatesthe overall assembly process, andhow multiple cells cooperateto structure a functional fiber.The answers to these questionswill require new experimental ap-proaches that combine the powerof dynamic imaging with specificantibodies, novel protein con-structs, and knockout cell lines.

ACKNOWLEDGMENTS

We thank Sean McLean, BrighamMecham, and Benjamin Scruggsfor assistance with the gene arraydata, and Beth Kozel, BrendaRongish, and Andras Czirok for thep-Timer-tropoelastin and otherlive imaging studies. We alsothank Russel Knutsen and MarilynLevy for assistance with EM.

REFERENCES

Argraves WS, Greene LM, Cooley MA,Gallagher WM. 2003. Fibulins: physi-ological and disease perspectives.EMBO Rep 4:1127–1131.

Arteaga-Solis E, Gayraud B, Lee SY,et al. 2001. Regulation of limb pat-terning by extracellular microfibrils. JCell Biol 154:275–281.

Bax DV, Bernard SE, Lomas A, et al.2003. Cell adhesion to fibrillin-1 mol-ecules and microfibrils is mediatedby alpha 5 beta 1 and alpha v beta 3integrins. J Biol Chem 278:34605–34616.

Bedell-Hogan D, Trackman P, AbramsW, et al. 1993. Oxidation, cross-link-ing, and insolubilization of recombi-nant tropoelastin by purified lysyloxidase. J Biol Chem 268:10345–10350.

Birk DE, Trelstad RL. 1986. Extracellu-lar compartments in tendon morpho-genesis: collagen fibril, bundle, andmacroaggregate formation. J CellBiol 103:231–240.

Borel A, Eichenberger D, Farjanel J,et al. 2001. Lysyl oxidase-like proteinfrom bovine aorta. Isolation andmaturation to an active form by bonemorphogenetic protein-1. J BiolChem 276:48944–48949.

Bradshaw AD, Graves DC, Motamed K,Sage EH. 2003. SPARC-null mice ex-hibit increased adiposity without sig-nificant differences in overall bodyweight. Proc Natl Acad Sci USA100:6045–6050.

Bressan GM, Daga-Gordini D, Colom-batti A, et al. 1993. Emilin, a compo-nent of elastic fibers preferentiallylocated at the elastin-microfibrilsinterface. J Cell Biol 121:201–212.

Broekelmann TJ, Kozel BA, Ishibashi H,et al. 2005. Tropoelastin interactswith cell-surface glycosaminoglycansvia its COOH-terminal domain. J BiolChem 280:40939–40947.

Brown-Augsburger P, Broekelmann T,Mecham L, et al. 1994. Microfibril-associated glycoprotein (MAGP)binds to the carboxy-terminal do-main of tropoelastin and is a sub-strate for transglutaminase. J BiolChem 269:28443–28449.

Canty EG, Lu Y, Meadows RS, et al.2004. Coalignment of plasma mem-brane channels and protrusions(fibripositors) specifies the parallel-ism of tendon. J Cell Biol 165:553–563.

Carta L, Pereira L, Arteaga-Solis E,et al. 2006. Fibrillins 1 and 2 performpartially overlapping functions duringaortic development. J Biol Chem281:8016–8023.

Chaudhry SS, Gazzard J, Baldock C,et al. 2001. Mutation of the geneencoding fibrillin-2 results in syndac-tyly in mice. Hum Mol Genet 10:835–843.

Chrzanowski P, Keller S, Cerreta J,et al. 1980. Elastin content of normaland emphysematous lung paren-chyma. Am J Med 69:351–359.

Clarke AW, Arnspang EC, Mithieux SM,et al. 2006. Tropoelastin massivelyassociates during coacervation toform quantized protein spheres. Bio-chemistry 45:9989–9996.

Colombatti A, Doliana R, Bot S,et al.2000. The EMILIN protein family.Matrix Biol 19:289–301.

Contri MB, Fornieri CC, Ronchetti IP.1985. Elastin-proteoglycans associa-tion revealed by cytochemical meth-ods. Connect Tissue Res 13:237–249.

Corson GM, Charbonneau NL, KeeneDR, Sakai LY. 2004. Differentialexpression of fibrillin-3 adds tomicrofibril variety in human andavian, but not rodent, connective tis-sues. Genomics 83:461–472.

Cox BA, Starcher BC, Urry DW. 1974.Coacervation of tropoelastin resultsin fiber formation. J Biol Chem249:997–998.

Csiszar K. 2001. Lysyl oxidases: anovel multifunctional amine oxidasefamily. Prog Nucleic Acid Res Mol Biol70:1–32.

Czirok A, Zach J, Kozel BA, et al. 2006.Elastic fiber macro-assembly is ahierarchical, cell motion-mediatedprocess. J Cell Phys 207:97–106.

Debeer P, Schoenmakers EF, Twal WO,et al. 2002. The fibulin-1 gene(FBLN1) is disrupted in a t(12;22)associated with a complex type ofsynpolydactyly. J Med Genet 39:98–104.

Downing AK, Knott V, Werner JM, et al.1996. Solution structure of a pair ofcalcium-binding epidermal growthfactor-like domains: implicationsfor the Marfan syndrome andother genetic disorders. Cell 85:597–605.

El-Hallous E, Sasaki T, Hubmacher D,et al. 2007. Fibrillin-1 interactionswith fibulins depend on the firsthybrid domain and provide an adap-tor function to tropoelastin. J BiolChem 282:8935–8946.

Erler JT, Giaccia AJ. 2006. Lysyl oxi-dase mediates hypoxic control of me-tastasis. Cancer Res 66:10238–10241.

Faury G. 2001. Function-structurerelationship of elastic arteries in evo-lution: from microfibrils to elastinand elastic fibres. Pathol Biol (Paris)49:310–325.

Faury G, Pezet M, Knutsen RH, et al.2003. Developmental adaptation ofthe mouse cardiovascular system toelastin haploinsufficiency. J ClinInvest 112:1419–1428.

Fogelgren B, Polgar N, Szauter KM,et al. 2005. Cellular fibronectin bindsto lysyl oxidase with high affinity andis critical for its proteolytic activation.J Biol Chem 280:24690–24697.

George EL, Georges-Labouesse EN,Patel-King RS, et al. 1993. Defects inmesoderm, neural tube and vasculardevelopment in mouse embryos lack-ing fibronectin. Development 119:1079–1091.

Gibson MA, Hughes JL, Fanning JC,Cleary EG. 1986. The major antigenof elastin-associated microfibrils is a31-kDa glycoprotein. J Biol Chem261:11429–11436.

Gibson MA, Hatzinikolas G, Kumarati-lake JS, et al. 1996. Further charac-terization of proteins associated withelastic fiber microfibrils including themolecular cloning of MAGP-2 (MP25).J Biol Chem 271:1096–1103.

Gilad GM, Kagan HM, Gilad VH. 2005.Evidence for increased lysyl oxidase,the extracellular matrix-formingenzyme, in Alzheimer’s diseasebrain. Neurosci Lett 376:210–214.

Giltay R, Timpl R, Kostka G. 1999.Sequence, recombinant expressionand tissue localization of two novelextracellular matrix proteins, fibulin-3and fibulin-4.Matrix Biol 18:469–480.

Gray WR, Sandberg LB, Foster JA.1973. Molecular model for elastinstructure and function. Nature246:461–466.

Hanssen E, Hew FH, Moore E, GibsonMA. 2004. MAGP-2 has multiple bind-ing regions on fibrillins and has cova-lent periodic association with fibrillin-containing microfibrils. J Biol Chem279:29185–29194.



Hayashi K, Fong KS, Mercier F, et al.2004. Comparative immunocyto-chemical localization of lysyl oxidase(LOX) and the lysyl oxidase-like(LOXL) proteins: changes in theexpression of LOXL during develop-ment and growth of mouse tissues. JMol Histol 35:845–855.

Henderson M, Polewski R, Fanning JC,Gibson MA. 1996. Microfibril-associ-ated glycoprotein-1 (MAGP-1) is spe-cifically located on the beads of thebeaded-filament structure of fibrillin-containing microfibrils as visualizedby the rotary shadowing technique.J Histochem Cytochem 44:1389–1397.

Hinek A, Wrenn DS, Mecham RP, Bar-ondes SH. 1988. The elastin recep-tor: a galactoside-binding protein.Science 239:1539–1541.

Hirano E, Knutsen RH, Sugitani H,et al. 2007. Functional rescue ofelastin insufficiency in mice by thehuman elastin gene. implications formouse models of human disease.Circ Res 101:523–531.

Hornstra IK, Birge S, Starcher B, et al.2003. Lysyl oxidase is required forvascular and diaphragmatic develop-ment in mice. J Biol Chem278:14387–14393.

Hu Q, Loeys BL, Coucke PJ, et al.2006. Fibulin-5 mutations: mecha-nisms of impaired elastic fiber forma-tion in recessive cutis laxa. Hum MolGenet 15:3379–3386.

Hucthagowder V, Sausgruber N, KimKH, et al. 2006. Fibulin-4: a novelgene for an autosomal recessivecutis laxa syndrome. Am J HumGenet 78:1075–1080.

Jensen SA, Reinhardt DP, Gibson MA,Weiss AS. 2001. Protein interactionstudies of MAGP-1 with tropoelastinand fibrillin-1. J Biol Chem 276:39661–39666.

Kadler K. 2004. Matrix loading: assem-bly of extracellular matrix collagenfibrils during embryogenesis. BirthDefects Res C Embryo Today 72:1–11.

Kagan HM, Vaccaro CA, Bronson RE,et al. 1986. Ultrastructural immuno-localization of lysyl oxidase in vascu-lar connective tissue. J Cell Biol103:1121–1128.

Kagan HM, Li W. 2003. Lysyl oxidase:properties, specificity, and biologicalroles inside and outside of the cell.J Cell Biochem 88:660–672.

Kelleher CM, McLean SE, Mecham RP.2004. Vascular extracellular matrixand aortic development. Curr TopDev Biol 62:153–188.

Kielty CM, Sherratt MJ, ShuttleworthCA. 2002. Elastic fibres. J Cell Sci115(Pt 14):2817–2828.

Kobayashi N, Kostka G, Garbe JH,et al. 2007. A comparative analysisof the fibulin protein family. Bio-chemical characterization, bindinginteractions, and tissue localization. JBiol Chem 282:11805–11816.

Kostka G, Giltay R, Bloch W, et al.2001. Perinatal lethality and endo-thelial cell abnormalities in severalvessel compartments of fibulin-1-de-ficient mice. Mol Cell Biol 21:7025–7034.

Kozel BA, Rongish BJ, Czirok A, et al.2006. Elastic fiber formation: adynamic view of extracellular matrixassembly using timer reporters. JCell Phys 207:87–96.

Kyriakides TR, Zhu YH, Smith LT, et al.1998. Mice that lack thrombospondin2 display connective tissue abnor-malities that are associated with dis-ordered collagen fibrillogenesis, anincreased vascular density, and ableeding diathesis. J Cell Biol 140:419–430.

Lee B, Godfrey M, Vitale E, et al.1991a. Linkage of Marfan syndromeand a phenotypically related disorderto two different fibrillin genes. Nature352:330–334.

Lee RM, Berecek KH, Tsoporis J, et al.1991b. Prevention of hypertensionand vascular changes by captopriltreatment. Hypertension 17:141–150.

Li DY, Brooke B, Davis EC, et al.1998a. Elastin is an essential deter-minant of arterial morphogenesis.Nature 393:276–280.

Li DY, Faury G, Taylor DG, et al.1998b. Novel arterial pathology inmice and humans hemizygous forelastin. J Clin Invest 102:1783–1787.

Li PA, He Q, Cao T, et al. 2004. Up-regulation and altered distributionof lysyl oxidase in the central nerv-ous system of mutant SOD1 trans-genic mouse model of amyotrophiclateral sclerosis. Brain Res 120:115–122.

Liu X, Zhao Y, Gao J, et al. 2004. Elas-tic fiber homeostasis requires lysyloxidase-like 1 protein. Nature genet-ics 36:178–182.

Loeys B, Van Maldergem L, Mortier G,et al. 2002. Homozygosity for a mis-sense mutation in fibulin-5 (FBLN5)results in a severe form of cutis laxa.Hum Mol Genet 11:2113–2118.

Lucero HA, Kagan HM. 2006. Lysyl oxi-dase: an oxidative enzyme and effec-tor of cell function. Cell Mol Life Sci63(19–20):2304–2316.

Maki JM, Rasanen J, Tikkanen H, Sor-munen R, Makikallio K, Kivirikko KI,Soininen R. 2002. Inactivation of thelysyl oxidase gene Lox leads to aorticaneurysms, cardiovascular dysfunc-tion, and perinatal death in mice. Cir-culation 106:2503–2509.

Maki JM, Sormunen R, Lippo S, et al.2005. Lysyl oxidase is essential fornormal development and function ofthe respiratory system and for theintegrity of elastic and collagen fibersin various tissues. Am J Pathol167:927–936.

Mao Y, Schwarzbauer JE. 2005. Fibro-nectin fibrillogenesis, a cell-mediated

matrix assembly process. Matrix Biol24:389–399.

Markova D, Zou Y, Ringpfeil F, et al.2003. Genetic heterogeneity of cutislaxa: a heterozygous tandem duplica-tion within the fibulin-5 (FBLN5)gene. Am J Hum Genet 72:998–1004.

McLaughlin PJ, Chen Q, Horiguchi M,et al. 2006. Targeted disruption offibulin-4 abolishes elastogenesis andcauses perinatal lethality in mice.Mol Cell Biol 26:1700–1709.

McLean SE, Mecham BH, Kelleher CM,et al. 2005. Extracellular matrix geneexpression in the developing mouseaorta. In: Miner JH, editor. Extracel-lular matrices and development. Am-sterdam: Elsevier. pp.82–128.

Nakamura T, Lozano PR, Ikeda Y, et al.2002. Fibulin-5/DANCE is essentialfor elastogenesis in vivo. Nature415:171–175.

Narayanan AS, Page RC, Kuzan F,Cooper CG. 1978. Elastin cross-link-ing in vitro. Studies on factors influ-encing the formation of desmosinesby lysyl oxidase action on tropoelas-tin. Biochem J 173:857–862.

Neptune ER, Frischmeyer PA, ArkingDE, et al. 2003. Dysregulation ofTGF-beta activation contributes topathogenesis in Marfan syndrome.Nat Genet 33:407–411.

Pan TC, Sasaki T, Zhang RZ, et al.1993. Structure and expression offibulin-2, a novel extracellular matrixprotein with multiple EGF-likerepeats and consensus motifs for cal-cium binding. J Cell Biol 123:1269–1277.

Park ES, Putnam EA, Chitayat D, et al.1998. Clustering of FBN2 mutationsin patients with congenital contractu-ral arachnodactyly indicates an im-portant role of the domains encodedby exons 24 through 34 duringhuman development. Am J MedGenet 78:350–355.

Partridge SM, Elsden DF, Thomas J.1963. Constitution of the cross-link-ages in elastin. Nature 197:1297–1298.

Partridge SM, Elsden DF, Thomas J,et al. 1964. Biosynthesis of the des-mosine and isodesmosine cross-bridges in elastin. Biochem J93:30C–33C.

Penner AS, Rock MJ, Kielty CM, ShipleyJM. 2002. Microfibril-associated gly-coprotein-2 interacts with fibrillin-1and fibrillin-2 suggesting a role forMAGP-2 in elastic fiber assembly. JBiol Chem 277:35044–35049.

Pfaff M, Reinhardt DP, Sakai LY, TimplR. 1996. Cell adhesion and integrinbinding to recombinant human fibril-lin-1. FEBS Lett 384:247–250.

Reinhardt DP, Sasaki T, Dzamba BJ,et al. 1996. Fibrillin-1 and fibulin-2interact and are colocalized in sometissues. J Biol Chem 271:19489–19496.

Ritty TM, Broekelmann T, Werneck CC,Mecham RP. 2003. Fibrillin-1 and -2



contain heparin-binding sites impor-tant for matrix deposition and thatsupport cell attachment. Biochem J375:425–432.

Roark EF, Keene DR, Haudenschild CC,et al. 1995. The association ofhuman fibulin-1 with elastic fibers:an immunohistological, ultrastruc-tural, and RNA study. J HistochemCytochem 43:401–411.

Rodgers UR, Weiss AS. 2004. Integrinalpha v beta 3 binds a unique non-RGD site near the C-terminus ofhuman tropoelastin. Biochimie86:173–178.

Sacks MS. 2003. Incorporation ofexperimentally-derived fiber orienta-tion into a structural constitutivemodel for planar collagenous tissues.J Biomech Eng 125:280–287.

Sakai LY, Keene DR, Engvall E. 1986.Fibrillin, a new 350-kD glycoprotein,is a component of extracellularmicrofibrils. J Cell Biol 103(Pt1):2499–2509.

Sakai LY, Keene DR, Glanville RW,Bachinger HP. 1991. Purification andpartial characterization of fibrillin, acysteine-rich structural component ofconnective tissue microfibrils. J BiolChem 266:14763–14770.

Sakamoto H, Broekelmann T, ChereshDA, et al. 1996. Cell-type specific rec-ognition of RGD- and non-RGD-con-taining cell binding domains in fibril-lin-1. J Biol Chem 271:4916–4922.

Sandberg LB, Weissman N, Gray WR.1971. Structural features of tropoe-lastin related to the sites of cross-links in aortic elastin. Biochemistry10:52–58.

Sasaki T, Gohring W, Miosge N, et al.1999. Tropoelastin binding to fibu-lins, nidogen-2 and other extracellu-lar matrix proteins. FEBS lett460:280–284.

Shifren A, Mecham RP. 2006. Thestumbling block in lung repair of em-physema: elastic fiber assembly.Proc Am Thorac Soc 3:428–433.

Shifren A, Durmowicz AG, Knutsen RH,et al. 2007. Elastin protein levels are

a vital modifier affecting normal lungdevelopment and susceptibility toemphysema. Am J Physiol Lung CellMol Physiol 292:L778–L787.

Stone EM, Lotery AJ, Munier FL, et al.1999. A single EFEMP1 mutationassociated with both Malattia Leventi-nese and Doyne honeycomb retinaldystrophy. Nat Genet 22:199–202.

Thomassin L, Werneck CC, Broekel-mann TJ, et al. 2005. The Pro-regions of lysyl oxidase and lysyl oxi-dase-like 1 are required for deposi-tion onto elastic fibers. J Biol Chem280:42848–42855.

Thorleifsson G, Magnusson KP, SulemP, et al. 2007. Common sequencevariants in the LOXL1 gene confersusceptibility to exfoliation glau-coma. Science 317:1397–1400.

Tiedemann K, Batge B, Muller PK,Reinhardt DP. 2001. Interactions offibrillin-1 with heparin/heparan sul-fate, implications for microfibrillar as-sembly. J Biol Chem 276:36035–36042.

Timpl R, Sasaki T, Kostka G, Chu ML.2003. Fibulins: a versatile family ofextracellular matrix proteins. NatRev 4:479–489.

Toonkool P, Jensen SA, Maxwell AL,Weiss AS. 2001. Hydrophobic do-mains of human tropoelastin interactin a context-dependent manner. JBiol Chem 276:44575–44580.

Trask BC, Trask TM, Broekelmann T,Mecham RP. 2000a. The microfibrillarproteins MAGP-1 and fibrillin-1 forma ternary complex with the chondroi-tin sulfate proteoglycan decorin. MolBiol Cell 11:1499–1507.

Trask TM, Trask BC, Ritty TM, et al.2000b. Interaction of tropoelastinwith the amino-terminal domains offibrillin-1 and fibrillin-2 suggests arole for the fibrillins in elastic fiberassembly. J Biol Chem 275:24400–24406.

Velling T, Risteli J, Wennerberg K,et al. 2002. Polymerization of type Iand III collagens is dependent on fi-bronectin and enhanced by integrins

alpha 11beta 1 and alpha 2beta 1. JBiol Chem 277:37377–37381.

Volpin D, Pasquali-Ronchetti I, UrryDW, Gotte L. 1976. Banded fibers inhigh temperature coacervates ofelastin peptides. J Biol Chem251:6871–6873.

Wagenseil JE, Nerurkar NL, KnutsenRH, et al. 2005. Effects of elastinhaploinsufficiency on the mechanicalbehavior of mouse arteries. Am JPhysiol 289:H1209–H1217.

Werneck CC, Trask BC, BroekelmannTJ, et al. 2004. Identification of amajor microfibril-associated glyco-protein-1-binding domain in fibrillin-2. J Biol Chem 279:23045–23051.

Wolinsky H, Glagov S. 1967. A lamellarunit of aortic medial structure andfunction in mammals. Circ Res20:99–111.

Wrenn DS, Hinek A, Mecham RP. 1988.Kinetics of receptor-mediated bindingof tropoelastin to ligamentfibroblasts. J Biol Chem 263:2280–2284.

Wu C, Keivens VM, O’Toole TE, et al.1995. Integrin activation and cytos-keletal interaction are essential forthe assembly of a fibronectin matrix.Cell 83:715–724.

Yanagisawa H, Davis EC, Starcher BC,et al. 2002. Fibulin-5 is an elastin-binding protein essential for elasticfibre development in vivo. Nature415:168–171.

Zacchigna L, Vecchione C, Notte A,et al. 2006. Emilin1 links TGF-betamaturation to blood pressure homeo-stasis. Cell 124:929–942.

Zanetti M, Braghetta P, Sabatelli P,et al. 2004. EMILIN-1 deficiencyinduces elastogenesis and vascularcell defects. Mol Cell Biol 24:638–650.

Zhang H, Apfelroth SD, Hu W, et al.1994. Structure and expression offibrillin-2, a novel microfibrillarcomponent preferentially located inelastic matrices. J Cell Biol 124:855–863.



IMP ELASTIC FIBRE SYSTEM

Documents

Transcript of IMP ELASTIC FIBRE SYSTEM