PDF (1424 KB)

Arg-Gly-Asp–Containing Domains of Fibrillins-1 and -2Distinctly Regulate Lung Fibroblast Migration

Stephen E. McGowan1, Amey J. Holmes1, Robert P. Mecham2, and Timothy M. Ritty3

1Department of Veterans Affairs Research Service and University of Iowa Carver College of Medicine, Iowa City, Iowa; 2Department of Cell Biologyand Physiology, Washington University School of Medicine, St. Louis, Missouri; and 3Department of Orthopaedics and Rehabilitation, Pennsylvania

State University School of Medicine, Hershey, Pennsylvania

Development of the extracellular matrix is a critical feature ofalveolar formation and actively involves pulmonary interstitial fibro-blasts. The elastic fiber network is an interconnected system of load-bearing fibers that also influences the behavior of adjacent cells,particularly the interstitial lung fibroblasts (LF). We hypothesizedthat discrete domains of fibrillins-1 and -2 interact with LF integrinsand direct their migration in the presenceof platelet-derived growthfactor (PDGF)-A. Surfaces coated with recombinant peptides lackingor including an arginine-glycine-aspartic acid (RGD)motif were usedto study LF migration across porous filters and on protein-coatedglass. Exon 24 of fibrillin-2 (Fib2 24), which encodes for an RGD-containing transforming growth factor-b–binding (TB) domain,stimulated migration with greater directional persistence and moreeffectively stimulated trans-filter migration at low concentrations.Exons 36–44 of fibrillin-1 (Fib1 36–44), which include epidermalgrowth factor–like domains and an RGD-containing TB domain,induce more lamlellipodia and more widespread remodeling of theleading edge, resulting in greater migration velocity than did Fib224. Distinct structural features in regions that surround the RGDmotifs may differentially regulate how the PDGF receptor-a pro-motes integrin distribution and actin filament remodeling at thecell’s leading edge. Understanding how fibrillins regulate LF migra-tion may help elucidate how the elastic fiber system could berestored as an interconnected unit, which fails to occur in emphy-sematous lungs.

Keywords: alveolus; fibrillin; elastin; integrin; cell migration

In mammalian species whose young are not required to walkimmediately after birth (humans and rats as opposed to sheep, forexample), most of the pulmonary alveoli develop after birth (1).Mammalian alveolar formation occurs in a stereotypic fashionwhereby simplified air sacs, which reside at the ends of the smallconducting airways at birth, gain complexity by progressivelydividing into smaller segments (alveoli) (2). The growing alveolarsepta contain a central structural core of fibroblasts and extra-cellular matrix, which is covered by the more superficial gas-exchange surface. Fibers in the central core are primarily producedby alveolar fibroblasts and maintain the cellular integrity of theseptum during the distortion that accompanies phasic respiration(3). The factors that regulate the initiation, localization, andtermination of the synthesis of these fibrillar proteins remainincompletely understood.

Mice that are platelet-derived growth factor (PDGF)-A–nulllack fibroblasts, which normally contain a-smooth muscle actin(aSMA) and synthesize elastin, and exhibit deficient secondaryseptation (4, 5). Septation may fail because fibroblasts, whichare present in primary septa, do not migrate during secondaryseptal elongation. Cell migration is augmented by interactionsbetween PDGFs and integrins. Binding of PDGF-AA or -BB totheir plasma membrane tyrosine kinase receptors initiates anintracellular cascade, causing phosphorylation of focal adhesionkinase, and an increase in the migration of fibroblasts (6).PDGFs promote actin remodeling and dorsal wave formation,where new lamellipodia form at the leading edge (7). Moredirect effects of PDGFs on integrins include (1) direct associ-ation of PDGF-receptor-b with aVb3, but not b1 integrins; (2)localization of integrins at the leading edge through recruitmentand activation of the small GTPase, Rac; and (3) promotion ofintegrin recycling via the early endosomes to the leading edgethrough another GTPase, Rab4A (8–10). Because PDGF-A isrequired for alveolar formation, it is important to understandhow PDGF modifies interactions between cellular integrins andextracellular proteins during cell migration. For example, type Icollagen is bound by the b1 portion of a1b1, a2b1, a10b1, anda11b1 integrins and promotes the migration of smooth musclecells and fibroblasts (11).

Elastic fibers are very resilient, have a long biological half-life,and are composed of at least seven different insoluble polymers(12). As one of the earliest proteins that are observed in nascentelastic fibers, fibrillins may serve as scaffolding on which othercomponents such as tropoelastin are deposited, and therebyinfluence the location of elastic fibers in the alveolar septum(3, 12–14). Fibrillins interact with proteins on the cell surface,including integrins, and influence the adhesion to, spreading, andmigration of ligamentum nuchae fibroblasts and mesangial cellsin vitro (14–17). Cellular contacts with fibrillin may also helpassemble a cadre of molecules (including fibulin-5, microfibril-associated glycoproteins, and lysyl oxidase) at the cell surface thatare required for elastic fiber formation (18–20).

Three fibrillin genes have been identified in mammals—fibrillins-1, -2, and -3—and the fibrillin-1 gene product is themost abundant in adults and is synthesized throughout life (21).Fibrillin-2 is maximal during early gestation and its synthesis islargely complete by birth. Fibrillin-3 is expressed in fetal humanbut not mouse or rat lungs (22). Fibrillin-1 is the most abundantfibrillin in pulmonary elastic fibers after birth, is synthesized asa monomer that polymerizes extracellularly, and undergoes aseries of post-translational modifications (23).

CLINICAL RELEVANCE

This research defines interactions between proteins in theelastic fiber and lung fibroblasts that may be involved in thealveolar septation process. The findings foster the devel-opment of new treatments for emphysema.

(Received in original form July 23, 2007 and in final form September 25, 2007)

This research was supported by the Department of Veterans Affairs Research

Service.

Correspondence and requests for reprints should be addressed to Stephen E.

McGowan, M.D., Division of Pulmonary, Critical Care, and Occupational Medicine,

Department of Internal Medicine, C33B GH, University of Iowa Hospital, 200

Hawkins Dr., Iowa City, IA 52242. E-mail: [email protected]

Am J Respir Cell Mol Biol Vol 38. pp 435–445, 2008

Originally Published in Press as DOI: 10.1165/rcmb.2007-0281OC on November 15, 2007

Internet address: www.atsjournals.org

The coding regions of the fibrillin genes are primarily com-posed of epidermal growth factor (EGF)-like repeats, whichencode for calcium-binding motifs and transforming growth fac-tor-b–binding protein-like (TB) repeats that contain 8-cysteinedomains and are involved in disulfide bonding (24). Exons 37of fibrillin-1 and fibrillin-2, both TB regions, each contain anarginine-glycine-aspartic acid (RGD) motif and N-glycosylationsites (25). Exon 24 in fibrillin-2 contains an additional RGD, al-though the adjacent residues differ from those in exon 37 (24, 26).Cellular attachment has been shown to involve interactions be-tween aVb3 or b1integrins and RGD motifs in fibrillins-1 and -2(14, 25). An alternate heparin-binding cell-binding domain in thecarboxy terminal region of fibrillin-1 also mediates attachment(27). This heparin-binding domain mediates interactions betweencells and fibrillin polymers that are required for cellular migra-tion and normal gastrulation in Xenopus embryos (28). We haveexamined how peptides representing particular exons in thefibrillin-1 and fibrillin-2 proteins influence neonatal lung fibro-blast attachment and migration.

We hypothesized that microfibrils may influence where alve-olar fibroblasts synthesize and deposit elastin and thereby ensurethat the elastic fiber network is interconnected and most abun-dant at locations of maximal strain during respiration. We alsoformulated a more specific sub-hypothesis: the RGD moieties infibrillins-1 and -2 enhance the attachment and migration of lungfibroblasts (LF) in vitro, in the presence of PDGF. This enhance-ment is associated with specific cellular processes that participatein the remodeling of integrin attachments at the leading edge ofthe cells. We have used recombinant fibrillin peptides that eithercontain or lack the RGD motif and adjacent EGF-like domains tostudy the adhesion and migration of neonatal rat LF. Migrationwas studied using two different approaches: transmigration acrossporous filters with the lower surface coated with fibrillin peptidesor on a planar surface of fibrillin peptides on a glass coverslip.These two approaches are complementary and allowed us tostudy different properties of the LF.

MATERIALS AND METHODS

Isolation of Neonatal LF

LF were isolated from rats at Postnatal Day 8 by digesting the lungswith a mixture or trypsin and collagenase and separating the less dense,lipid-laden fibroblasts using Percoll (29, 30). The cells were grown toconfluence and were then removed from the tissue culture plates using0.2% trypsin, frozen in fetal bovine serum (FBS) containing 10% DMSO,and maintained in the vapors of liquid nitrogen. Aliquots of cells werethawed, cultured in Ham’s F-12 medium containing 5% FBS and usedwithout further sub-cultivation. Migration and adhesion assays wereperformed in MCDB-201 medium on surfaces that were coated withvarious concentrations of fibrillin peptides.

Preparation of Fibrillin Peptides

pQE bacterial plasmids containing various domains of the humanfibrillin 1 (Fib1)- and fibrillin 2 (Fib2)-coding regions have been de-scribed (13, 27). Fib1 36–44 contains exons 36–44 of fibrillin 1, whichincludes an RGD motif in exon 37; Fib 1–30 contains only exon 30 offibrillin 1, which encodes an EGF-like domain and lacks an RGD motif;Fib2 24 contains exon 24 of fibrillin 2, which contains an RGD. To excludethe effect of an adjacent EGF-like domain, we studied Fib2 24 rather thanFib2 37–38, which contains an EGF-like domain in exon 38. The presenceof exon 38 renders Fib2 37–38 more structurally similar to Fib1 36–44 thanis Fib 2 24. The plasmids were propagated in an Escherichia coli strain thatalsocontains the Rep4 plasmid,allowing inducible expression of the recom-binant peptides in the presence of isopropyl b D 1-thiogalactopyranoside(IPTG). The peptides were purified from bacterial cell lysates in thepresence of 8 M urea using Ni-NTA (nickel bound to nitriloacetic acid)agarose. Purity and Mr were confirmed using SDS-PAGE and silverstaining. Fib 23–44 (LEEC) was contained in the pEE14 mammalian

expression vector, which had been stably transfected into CHO-K1 cells,and its expression was driven by the CMV immediate early gene pro-moter (13, 31). Transfected CHO-K1 cells were propagated in Glasgowmodified Eagle Medium without glutamine and containing 100 mM ofL-methionine sulfoximine. Before collection of the conditioned medium,the growth medium was changed to Hy-QCCM5 serum-free medium(InVitrogen Life Technologies, Carlsbad, CA). The recombinant pep-tide was purified from the conditioned medium, after concentration byultrafiltration on a YM-100 membrane (Amicon, Danvers, MA) fol-lowed by elution from a Superose 6HR column (10 mm 3 300 mm;Pharmacia Biotech, Piscataway, NJ) in 100 mM ammonium bicarbonate,pH 7.5 at a flow rate of 30 ml/hour (31). Protein purity was assessed usingSDS-PAGE and Western immunoblotting.

Type I collagen was included in the migration assays for severalreasons: (1) relatively few LF adhered to polycarbonate filters and glassin the absence of collagen, limiting the accuracy of assays that requiredenumeration of cells; (2) LF retracted during haptotaxis on the glasssurface of the perfused migration chamber, preventing an accurateassessment of migration; and (3) PDGF does not increase fibroblastmigration on collagen, which enabled an assessment of how PDGFpromotes the migration on fibrillin (8).

Adhesion Assay

The adhesion of LF to fibrillin peptides was assessed using Costar 3590,96-well microplates (Corning Life Science, Lowell, MA). Quadrupli-cate wells were coated overnight at 48C with 0, 0.1, 0.4 1, or 4 mg/ml ofthe various peptides in 0.14 M NaCl, 50 mM Tris-HCl, pH 7.4 (TBS),0.02% sodium azide. After removing the unbound peptides, the wellswere washed and the surface was incubated for 1 hour in 1 mg of bovineserum albumin (BSA) in TBS without azide. Fifty thousand LF wereadded to each well and incubated at 378 for 2 hours in 5% CO2 to allowattachment. The nonadherent LF were removed by washing the wellswith warm TBS without azide. The adherent cells were quantified byassaying their DNA contents using Yo-Pro-1 dye (InVitrogen, MolecularProbes, San Diego, CA) (32). The adherent LF were first lysed in 50 ml of0.1% Triton X-100 in H2O for 30 minutes at room temperature. An equalvolume of 4 mM Yo-Pro-1 in 10 mM Tris-HCl pH 7.4, 2 M NaCl, 1 mMEDTA was added and the fluorescence intensity was assessed usinga FluoStar Microplate reader (BMG-Labtech, Durham, NC) at excita-tion and emission wavelengths of 485 and 530 nm, respectively. Astandard curve was generated using calf-thymus DNA. The DNAcontained in known quantities of LF was quantified using calf-thymusDNA to determine the ng of DNA per 1,000 LF. In all cases, the quantityof DNA that was contained in LF that adhered to the microwells thatonly received BSA was subtracted from the quantity of DNA in wells thatwere coated with the fibrillin proteins.

To assess whether aV-integrin is involved in LF adhesion, wellswere coated with Fib1 36–44. Lung fibroblasts were pre-incubated for30 minutes with various concentrations of anti-aV (monoclonal ratanti-CD51, clone RMV-7; eBioscience, San Diego, CA) at 48C and thecell suspensions in the antibody solutions were added to quadruplicatewells. After incubating for 2 hours at 378C, the DNA in adherent LFwas quantified. Three separate experiments were performed.

Transmigration of LF across Polycarbonate Filters

Lung fibroblasts were grown to confluence and then starved overnightby reducing serum concentration to 0.5% in MCDB 201. The AP48chemotaxis chamber and the 8-mm polycarbonate filters were obtainedfrom Neuro Probe (Gaithersburg, MD). Various concentrations, rang-ing from 0.05 to 2 mg/ml, of the fibrillin peptides were placed in thelower chambers. After assembling the apparatus, 2.5 mg of rat tail ten-don collagen (Type 1; Sigma-Aldrich Chemical Co., St. Louis, MO) in0.02 M acetic acid was added to the upper wells and the apparatus waskept at 48C overnight. The next morning, the solution was removedfrom the upper chamber, and the wells were washed three times with0.145 M NaCl, 0.0015 M KH2PO4, 0.0027 M KCl, 0.0086 M Na2HPO4,pH 7.4 (PBS). The top plate of the chemotaxis apparatus was removedwith the filter still attached. The bottom wells were evacuated, washedwith PBS, and refilled with MCDB medium containing 1 mg/ml BSAand 50 ng of PDGF-AA per ml (R&D Systems, Minneapolis, MN).The apparatus was reassembled with the top plate and filter and asuspension containing 2.5 3 104 LF was added to the top chamber (33).

436 AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL 38 2008

The apparatus was incubated at 378C in a humidified atmospherecontaining 5% CO2 for 4 hours. The apparatus was disassembled andthe filter was carefully inverted and floated in PBS. A weight wasapplied to one edge of the filter and a clamp to the opposite edge. Thetop side of the filter was scraped over a rubber squeegee, as directed in theNeuro Probe instruction manual. The filter was fixed in methanol at roomtemperature for 20 minutes. After washing the filter it was stained for5 minutes with Diff Quik stain (IMEB Corp., San Marcos, CA). Afterwashing and dehydration by gradually increasing the ethanol concentra-tion in the bath from 70% to 100%, the filters were allowed to air dry ona glass slide. Immersion oil was applied to adhere the filter to glass slide.The slides were evaluated using an Olympus BX40 microscope (Olym-pus, Center Valley, PA) using a 3100 oil objective. The regions corre-sponding to the wells were located and the focus was adjusted to identifythe top and bottom sides of the filter. The plane of focus was adjusted tothe bottom of the filter, and all of the cells in the region corresponding tothe well were counted. The focus was adjusted up and down to ensure thatonly cells on the bottom of the filter were counted. The data wereexpressed as cells migrated per 20 oil fields.

Preparation of Coverslips to Study Haptotaxis of

LF on Fibrillin

To prepare the coverslips to receive the fibrillin coating, the coverslip wasfirst coated with a solution containing nickel, which was accomplished asfollows (34, 35). The coverslips were first cleaned by successive exposuresto acetone, ethanol, a solution of 5% ammonium hydroxide and 3%H2O2 in deionized H2O, and then air dried. Next, the coverslips wereexposed to 20% 3-aminopropyl triethoxysilane in 100% ethanol (vol/vol)overnight, rinsed twice with 95% ethanol, and then autoclaved at 1208Cfor 40 minutes. The coverslips were then coated with a 0.1% solution ofpoly-octadecene-maleic anhydride (POMA) in tetrahydrofuran usinga spin coater (Specialty Coating Systems, Indianapolis, IN). The cover-slips were autoclaved at 1208C for 20 min and stored at 48C individuallyin wells of a 12-well plate in a sterile environment. Next they were incu-bated overnight at room temperature in 10 mM of Na-Na-bis carbox-ymethyl-lysine in 0.1 M sodium phosphate (pH 8.0). The coverslips werewashed with 50 mM Tris-HCl pH 7.5, 150 mM NaCl, 0.05% Tween 20 andthen blocked for 2 hours with PBS containing 2% BSA in 50 mM Tris-HClpH 7.5, 150 mM NaCl, 0.05% Tween 20. At room temperature, the cov-erslips were then successively washed with (1) 50 mM Tris HCl, pH 7.5; (2)sterile deionized H2O; (3) 100 mM EDTA, and once again in steriledeionized H2O. The coverslips were incubated for 20 minutes with 20 mMNiSO4, washed with deionized water, followed by 50 mM Tris-HCl pH 7.5,150 mM NaCl, and finally highly purified water. The surface-bound nickelwas allowed to bind the 6-histidine–tagged Fib1 36–44 or Fib2 24 peptidesfor 6 hours at room temperature and then the coverslips were incubatedovernight with PBS containing either 10 or 12.5 mg of collagen per ml.Equivalent molar quantities of RGD were presented when Fib1 36–44 andFib2 24 were present at 10 and 4 mg per ml, respectively. The coatingconcentration of Fib1 30 was 10 mg per ml.

Migration and Time-Lapse Photography of LF Haptotaxis

Migration was evaluated using a POCmini chamber (HemoGenex,Colorado Springs, CO) in the closed chamber perfusion format. The top,22-mm cover glass had been coated with the fibrillin peptides, whereasthe bottom coverslip provided a transparent floor for the chamber.Before the migration assay, LF were maintained in Ham’s F-12 mediumcontaining 5% FBS. The medium was changed to MCDB-201 containing5% by volume of Serum-Plus (SAFC Biosciences, Lenexa, KS) to yieldan equivalent concentration of 1% serum. Twenty-five thousand (6.6 3

103/cm2) LF were allowed to adhere to the coated top coverslip for 6hours and the POCmini apparatus was assembled. The chamber wasperfused with MCDB-201 medium containing 5% Serum-Plus at 0.5 mlper hour while the chamber was maintained at 358C using a TC-324temperature controller and heater (Warner Instruments, Hamden, CT).We used MCDB-201 medium containing 2 mg/ml human serum albuminduring the adherence and migration, but consistently observed disrup-tion of focal contacts and contraction of many of the LF. We could notreliably assess migration under these conditions. The POCmini chamberwas fixed to a motorized mechanical stage (Ludl Electronics, Hawthorne,NY) on an Olympus BX40 microscope and viewed using a phase-contrastcondenser and an Olympus LUC PLAN FL 320 objective (N.A. 0.45)

with a long working distance. Images were acquired every 15 minutesfrom five different fields, captured by an Optronix MicroFire charged-coupled device (CCD) camera, and saved in a grayscale TIF format. TheOptronix Picture Frame software was used to prepare image sequencesfor each field, which were then analyzed using IP-Lab for Windows (BD-Scanalytics, Rockville, MD). An obvious nuclear landmark (usuallya nucleolus) was chosen and its path was marked at hourly intervals overthe 18-hour course of the experiment. The x- and y-coordinates of thisstructure were used to calculate the root mean squared distance moved ateach hourly interval (36). These data were used to determine theinstantaneous velocity, total distance moved, and to plot the path ofthe cells. The accumulated distances were assessed using the generalizedleast-square method of Levenberg-Marquardt to calculate root meansquared speed (S) and persistence (P) for each cell that moved and stayedwithin the microscopic field during the entire 18 hours. The data for D2

versus time in hours were fit to the equation: D2 5 S2P2(T/P-1 1 e2T/P).D2 is the square of the distance moved in the preceding 1 hour (squareddisplacement in mm2); P is the persistence time in hours, T is time inhours, and S is speed or velocity in mm per hour. Directional persistence isa measure of the persistence of motion at a similar velocity and direction,and is based on a diffusion model that predicts the probability that eitherspeed or direction will change with time (37). This calculation wasperformed using nonlinear regression and Prism4 (Graph Pad, SanDiego, CA). The data for individual cells were combined and the meanand SEM were calculated.

Cellular shape was compared after LF had been allowed to migratefor 6 hours on collagen, Fib1 36–44, or Fib2 24. Only cells that did notdivide during a 12-hour period surrounding this time point were analyzed.The lamellipodia at the leading edge were counted, the perimeters of thecells were traced using IP-Lab software, and the perimeter and area werecalculated in mm and mm2, respectively. The ratio of the perimeter to thearea (P/A) was calculated as an index of the complexity of the cell shape(P/A should be greater in cells with more protrusions).

The areas occupied by the leading edges of LF were assessed duringmigration on only collagen or on collagen plus either Fib1 36–44 orFib2 24. The perimeters of the cells were traced at hourly intervalsduring a 6-hour period of the time-lapse recording when the LF werethe most motile. The areas circumscribed by the perimeters, the lengthsof the major axes of the cells, and the centroids were calculated using IPLab. By tracing the cells at hourly intervals, we were able to determinethe area of protrusion of the leading edge (% positive flow) 5 (area attime n 2 area at time n 2 1 / area at time n) 3 100 (38). The % negativeflow was calculated using the areal difference between the retractions ofthe trailing edges of same tracings. By comparing the distance betweenthe centroids of the protrusion and retraction at a particular 1-hourinterval to the major axis of the cell at time n, we calculated the polarity(distance between centroids / length along the major axis of the cell) (39).Twelve cells were analyzed for each condition, and the mean % positiveflow and polarity over the entire 6-hour period were calculated for eachcell. Selection bias was minimized by progressing through the sequencesof the fields that were photographed in a systematic way and identifyingall of the cells in field 1 that could be analyzed, performing the analyses,and then moving onto field 2. Four cells were analyzed in each of threeseparate experiments for each coating condition.

Immunohistochemistry

The association of aVb3 integrins with the RGD motifs in the fibrillinpeptides was examined by immunolocalization of the 6-His portion ofthe peptides and the aV chain (CD51) of the integrin. Paxillin was alsovisualized to localize focal contacts. The effects of the various fibrillinpeptides and PDGF-A on the intracellular distribution of the in-tracellular GTPase, Rac, were also studied.

Circular glass coverslips were coated with Fib1 30, Fib1 36–44, or Fib224 in the presence of collagen, or with only collagen following theprocedure that was used for the studies of LF haptotaxis. The LF adheredto the coverslips for 6 hours at 378C in MCDB-201 medium containing5% Serum-Plus, and some coverslips were exposed to 20 ng of PDGF-Aper ml for 4 hours at 378C. The coverslips were then washed with PBS andfixed for 30 minutes in 1% paraformaldehyde at 258C. The LF werepermeabilized for 20 minutes with 0.1% (vol/vol) Triton X-100 in PBSthat also contained 2 mg of bovine serum albumin (BSA) per ml. In someinstances permeabilization was conducted before and in other cases after

McGowan, Holmes, Mecham, et al.: Fibrillins Regulate Lung Fibroblast Migration 437

exposure to the anti-CD51 antibody. After rinsing with PBS containingBSA, the coverslips were incubated with the primary antibody for 2 hoursat room temperature and rinsed with PBS containing BSA. To localize thefibrillins, CD51, and paxillin, the following dilutions of primary antibodieswere used: 1:200 for mouse monoclonal anti-6His-biotin (Qiagen, Valen-cia, CA), 1:200 for goat anti-CD51 (eBioscience, San Diego, CA), and1:1000 for anti-paxillin-TRITC (mouse monoclonal; BD Pharmingen, SanDiego, CA). The LF were exposed to the anti-6His and anti-CD51 primaryantibodies for 1.5 hours, and the coverslips were rinsed and then incubatedfor 1 hour with 1:1,000 dilution of Alexafluor 488 swine anti-goat IgG(InVitrogen, Molecular Probes, Carlsbad, CA). After rinsing, the coverslipswere incubated with a solution containing anti-Paxillin and AlexaFluor633–streptavidin (1:1,000 dilution to visualize the anti-6His; InVitrogen,Molecular Probes). After additional rinses, coverslips were mounted usingmounting medium (0.05 M Tris-HCl pH 8.0, 90% glycerol, and 10 mg/ml1,4,-Diazobicyclo[2,2,2] octane).

To localize collections of Rac, LF were adhered and spread on a separateset of coverlips that were prepared in the same fashion; however, auto-fluorescence was quenched by incubating in 200 mM glycine, 50 mM NaClfor 15 minutes. The fixed coverslips were incubated for 2 hours at 258C witha 1:200 dilution mouse-monoclonal anti-Rac1 (BD Pharmingen) in PBScontaining 0.1% Triton X-100 and 2 mg of BSA per ml. After rinsing withPBS containing Triton X-100 and BSA, the coverslips were incubated witha 1:500 dilution of goat anti-mouse IgG-biotin for 1.5 hours. After rinsingagain, the coverslips were incubated for 30 minutes with a mixture ofphalloidin-FITC (1:800 dilution; Sigma-Aldrich) and AlexaFluor 633–streptavidin (InVitrogen, Molecular Probes), at a 1:1,000 dilution in PBScontaining Triton X-100 and BSA. A final rinse was performed usingPBS and Triton X-100 without BSA and the coverslips were mounted.Three experiments were performed to examine fibrillin and integrin co-localization, and three experiments were performed to examine Rac andphalloidin. In an additional experiment, coverslips were prepared and LFwere adhered in the presence of collagen plus either Fib1 30, Fib1 36–44, orFib2 24. After incubating for 4 hours in the presence of absence of PDGF-A, the cells were fixed and actin was visualized with phalloidin. Usingepifluorescence microscopy, 200 cells were evaluated for each conditionand LF that exhibited either a filamentous or subcortical distribution ofactin were enumerated.

Confocal Microscopy

A Zeiss LSM 510 scanning confocal microscope was used at a 512 3 512pixel resolution, scan speed of 4, and 4 scans were averaged. A 340objective was used, the optical slice was 6.6 mm, and the pinhole was 6.14airy units for the examination of fibrillin and integrin co-localization. A 363objective was used, the optical slice was 3.1 mm, and the pinhole was 2.8 and3.5 airy units, respectively, for the examination of Rac and phalloidin.Representative images for each treatment condition were examined usingImageJ software and saved in the tagged image format (TIF). Compositeswere prepared using either Adobe Photoshop 6 or Adobe Illustrator CS2.

Quantification of Discrete Collections of Rac

Confocal microscopic images showing the fluorescence from anti–Rac-Alexafluor 633 were extracted from the Zeiss MDB files using Image J,converted to an RGB format, and saved as TIF files. These files wereopened in IP Lab (BD-Bioscience, Rockville, MD), and a histogram ofthe number of pixels versus intensity was displayed. Because the intensityvaried among images, a uniform increment, relative to the maximalintensity for each image, was used to ascertain the lower limit of intensityfor segmentation. The pixels within this normalized range of intensitywere identified by segmentation and gave a good representation of theperceived (by eye) boundaries of the discrete collections of Rac. Thesegmented images were uniformly subjected to erosion using a mask,which eliminated small artifactual collections of pixels. The number,centroid, and pixel area of Rac collections was systematically ascertainedfor each LF that was completely positioned within the image boundaries.This procedure provided a uniform set of criteria for excluding the largeperinuclear collections and reduced the subjectivity associated withcounting visually ascertained collections. All of the cells which fellcompletely within the boundaries of the images were analyzed.

Statistical Analyses

The data are expressed as mean 6 1 SEM. Only LF that moved more than100 mm during the course of the time-lapse experiment were considered

in the analysis of cellular haptokinesis. At least three separate time-lapseexperiments were performed for each of the coating conditions and thedata for individual cells were pooled across all of the experiments fora particular condition. This enabled us to analyzed at least 56 cells foreach condition. The variance was expressed as one SEM rather than SD,as others have done (40). The data from the analyses of cellular adhesionand migration across the porous filters were normally distributed andwere analyzed using a two-way ANOVA for adhesion and a one-wayANOVA for migration. The haptokinesis was not normally distributedand the nonparametric ANOVA on ranks was used for all of the com-parisons from the time-lapse experiments except for the polarity, whichwas normally distributed. The Student-Newman-Keuls and the Dunn’spost hoc tests were used for the parametric and nonparametric analyses,respectively. The Kruskal-Wallis post hoc test was used for the analysisof polarity. Differences were considered significant when P was lessthan 0.05.

RESULTS

Adhesion of LF to Plastic Coated with Fibrillin Peptides

The wells of microtiter plates were coated with Fib1 30, whichlacks an RGD site, or either Fib1 36–44 or Fib 2 24, which bothcontain an RGD site. Figure 1 shows a representative SDS-PAGE gel after silver staining of Fib 1 30, and demonstrates thatthe isolated peptide migrated as one predominant band. Fib 1 36–44 and Fib2 24 exhibited similar levels of purity (not shown).Figure 2A shows that adhesion in the presence of peptides thatcontained an RGD was significantly higher than in the presenceof Fib1 30, when peptide concentrations were added at concen-trations at or greater than 0.4 mg/ml for Fib1 36–44 and at 4 mg/mlfor Fib2 24. Pre-incubating LF with 20 mg of an anti-aV antibodyper ml significantly reduced adhesion to Fib1 36–44 (Figure 2B).Because the RGD motif promoted adhesion, we also evaluatedthe effect of the RGD motif on the migration of LF.

Fibrillin Peptides Increase LF Migration across

Polycarbonate Filters

Because fibrillin peptides were added to only the lower cham-bers of the chemotaxis apparatus, whereas collagen was addedto the top of the filter, LF adhesion to the top of the filter wasprimarily to collagen. Localization of the fibrillin to the bottomsurface of the filter provided a step-increase in fibrillin concen-tration that could promote directional migration of the LF.

Figure 1. Purity of fibrillin 1

30 peptide. The recombinantbacterial product was sub-

jected to SDS-PAGE and silver

staining. Protein standards areshown on the left (std) and

the lanes labeled as 1 and 5

contain 1 and 5 mg, respec-

tively, of recombinant pep-tide.


Increasing concentrations of Fib1 36–44, Fib1 23–44, or Fib2 24progressively promoted LF migration, whereas Fib1 30, whichlacked the RGD motif, did not increase migration above that

which was observed with only PDGF (Figure 3). Only Fib2 24augmented migration when the fibrillin peptide concentrationwas 0.05 mg/ml. At concentrations equal to or exceeding 0.5 mg/ml,all three RGD-containing fibrillin peptides promoted migrationexcept at 2 mg/ml, at which the effect of Fib1 23–44 was nolonger significant.

Fibrillin Peptides Increase LF Haptotaxis

To more precisely determine why the RGD-containing exons inFib1 and Fib2 increased LF migration, we studied the move-ment of LF on surfaces that were coated with fibrillin peptides.When combined with time-lapse photography, this allows one toassess the movement of individual cells and to evaluate thevelocity and persistence, two parameters that have been linkedto biochemical events within the cell. During the migrationperiod, the LF were exposed to a unidirectional flow of mediumat 0.5 ml/hour. Coating with collagen alone promoted adhesionand allowed the cells to spread. The LF did not migrate in theabsence of PDGF (data not shown), so all of the data that areshown were obtained in the presence of 50 ng of PDGF per ml.Representative cellular paths for LF on only collagen orcollagen with either Fib1 36–44 or Fib2 24 are shown in Figure4. When LF were exposed to only collagen or collagen plus Fib130 (lacks RGD), they migrated with a slower velocity than in

Figure 2. Arginine-glycine-aspartic acid (RGD)-containing fibrillin poly-peptides increase lung fibroblast (LF) adhesion to plastic. (A) Naked

polystyrene wells were coated with increasing concentrations of either

Fib1 36–44, Fib2–24 (contain RGD), or Fib1 30 (lacks RGD). Adherent

LF were enumerated and data were pooled from four separate experi-ments in which the means from quadruplicate wells were determined.

Bars represent mean 6 1 SEM, *P , 0.05 comparing Fib1 36–44 to

Fib1 30, #comparing Fib2 24 to Fib1 30. (B) Inhibition of the adhesionof LF to Fib1 36–44 by anti–aV-integrin was assessed as in A and the

mean 6 1 SEM number of adherent LF are shown (n 5 3, P , 0.05

compared with an 20 mg/ml of nonimmune IgG).

Figure 3. Fibrillin polypeptides containing RGD increase LF transmigra-

tion. The regions of the lower surface of a polycarbonate filter that werecontained within the wells were coated with various concentrations of

fibrillin peptides. LF were placed in the upper compartments and PDGF-

AA in the lower compartments of a modified Boyden chamber. After

fixation, the LF that migrated to the bottom of the filter were enumer-ated. The quantities of LF (mean 6 1 SEM, n 5 4 separate analyses, for

each condition) that migrated are shown, *P , 0.05, one-way ANOVA

comparing an RGD-containing peptide with Fib1 30, at each concentra-

tion. PDGF was included in all wells except those shown for mediumonly.In some cases two asterisks are shown because two points overlap.

Figure 4. Haptokinesis of LF on glass coverslips that were coated with

collagen in the absence or presence of fibrillin polypeptides (10 mg per

ml of Fib1 36–44 or 4 mg per ml of Fib2 24). The LF resided at theorigins of the cellular paths shown in the left column at the beginning of

the time-lapse exposure. The x and y coordinates (�) of the cells are

shown at hourly intervals. The right column shows the instantaneous

velocities (the velocity of the cell during the preceding hour), whichvaried during the course of the experiment. Representative results are

shown for one cell for each coating condition.


the presence of Fib1 36–44 (Figure 5A). Addition of Fib1 36–44to collagen significantly increased the velocity of migration byapproximately 3-fold, but there was also a significant (z 2-fold)decrease in persistence. The velocity of migration on coverslips,which had been coated with 4 mg of Fib2 24 per ml, wassignificantly higher than on coverslips coated with Fib1 30 butnot those that were only coated with collagen. The persistencewas significantly higher for LF that were migrating on coverslipscoated with Fib2 24 compared with coverslips coated with eithercollagen alone or with collagen plus Fib1 36–44. The velocity ofmigration on Fib 1 30, which lacks an RGD motif, was similar tothat observed in the presence of only collagen, and significantlyless than in the presence of Fib1 36–44 or Fib2 24. The effects ofdifferent concentrations of Fib2 24 on LF migration are shownin Figure 5B. Coverslips coated with 4 mg of Fib2 24 per mlsupported more rapid migration than did coverslips coated with10 mg per ml.

Fibrillin Peptides Modify How LF Spread during Migration

The shape of cells that had been present on the coverslips for6 hours in the presence of PDGF-AA were compared after

coating with only collagen or collagen and equimolar RGDconcentrations of Fib1 30, Fib1 36–44, or Fib2 24 per ml. Theshapes of representative LF that are spread on collagen, Fib136–44, or Fib2 24 are shown in Figure 6. Lamellipodia thatoccupied the leading edge of the cells were also enumerated.Four parameters were assessed: area, perimeter, the ratio ofperimeter to area (P/A) and number of lamellipodia. Theresults shown in Figure 7 indicate that LF that spread on Fib136–44 had a more convoluted (larger) perimeter, a larger P/A,and more lamellipodia at the leading edge.

The Area of Protrusions in the Leading Edge Is

Increased by Fibrillins

Cellular movement is accompanied by advancement of theleading edge and retraction at the trailing edge. Because thevelocity increased after exposure to Fib1 36–44, we quantifiedadvancement of the leading edge at 1-hour intervals duringa period when a cell was migrating. The area change of theleading edge was expressed relative to the area of the entire cellto normalize for differences in cell size (percent positive flow).Fibrillin 1 36–44 and Fib2 24 each significantly increased the %positive flow relative to that observed when the surface wascoated with only collagen or collagen along with Fib1 30 (Figure8). Lung fibroblasts that migrated on Fib1 36–44 exhibiteda lower polarity than LF that migrated on Fib1 30, Fib2 24, oronly collagen. The percent negative flow (hourly areal change inthe trailing edge relative to total cell area) was not altered byany of the fibrillin peptides (data not shown).

Fibrillin Peptides Co-Localize with aV-Integrins but

Not with Paxillin

Figure 9 (top panel) demonstrates co-localization of 6-His-labeled Fib1 36–44 and anti-CD5–labeled aV-integrin in a peri-nuclear distribution, when LF were permeabilized before addingthe anti-aV antibody. The aV chain has a very similar distribu-tion, indicating that intermolecular interactions could occurbetween the RGD domain of Fib2 24 and aV-integrins. Whenthe anti-CD51 antibody was added before permeabilization(Figure 9, bottom two panels), the aV-integrin was only visualizedat the focal contacts and not in a perinuclear distribution. Paxillin-containing focal contacts were only located at the periphery of thecells, and were not associated with the perinuclear collections ofthe aV-integrin. Paxillin and the aV-chains had similar distribu-tions in the presence of 6-His–labeled Fib1 30.

Figure 5. Fibrillin 1 36–44 increases the velocity but diminished thepersistence of LF. (A) Comparison of velocity and directional persistence

of LF migrating on a glass surface that was treated with only collagen,

or collagen plus either 10 mg of Fib1 30 per ml, 10 mg of Fib1–36–44,or 4 mg of Fib2 24 per ml. Velocity (open bars) and persistence (hatched

bars) are shown as means and 1 SEM of 56, 91, 62, and 61 cells for only

collagen, or with Fib1 30, Fib1 36–44, or Fib2 24, respectively. **Velocity

for Fib1 36–44 compared with only collagen, Fib2 24, and Fib1 30;#velocity for Fib2 24, compared with Fib1 30: P , 0.05 ANOVA on ranks,

Dunn’s post hoc test; opersistence for Fib1 30 compared with only

collagen; *persistence of Fib1 36–44 compared with only collagen,

Fib1 30, or Fib2 24; tFib2 24 compared with collagen: P , 0.05, ANOVAon ranks, Dunn’s post hoc test. (B) Effects of different concentrations of

Fib2 24 [Fib2 24] along with a constant concentration of collagen on

velocity (open bars) and persistence (hatched bars). Mean 6 1 SEM are

shown using 56, 61, and 77 cells for 10, 4, and 2 mg of Fib2 24 per ml,respectively. *P , 0.05 for 4 compared with 10 mg of Fib2 24 per ml.

ANOVA on ranks, Dunn’s post hoc test.

Figure 6. The shapes of three representative LF that are spread on

collagen only or collagen plus either 10 mg of Fib1 36–44 or 4 mg ofFib2 24 per ml. The leading edges are highlighted in white, and on Fib1

36–44 more than one lamellipodium occupies the leading edge.


Figure 10 shows that Rac is abundant in the region of thenucleus, but is also located at the periphery of the cell, particularlywhere phalloidin-stained subcortical actin is abundant. PDGF-Apromoted the formation of peripheral collections of Rac, whichwere particularly dense in lamellipodia. These dense lamellipo-dial collections of Rac were significantly more abundant in LF onFib1 36–44, than on either Fib2 24, Fib1 30 (Figure 10B). Becausethe antibody that we used recognizes an epitope that is notmodified by interaction with GTP, we were unable to determinewhether Rac was in the active, GTP-bound state. Although itappears from Figure 10A that the distribution of actin varies withthe type of fibrillin peptide that the LF were exposed to, this wasnot evident when the configuration of actin filaments wasanalyzed in a larger number of cells. The large majority offibroblasts contained filamentous rather than subcortical actin.In the absence of PDGF-A, filamentous actin predominated in98.4, 95.1, and 90.4 percent of fibroblasts on Fib1 20, Fib1 36–44,and Fib2 24, respectively. In the presence of PDGF-A, filamen-tous actin predominated in 97.2, 96.6, and 98.6 percent offibroblasts on Fib1 20, Fib1 36–44, and Fib2 24, respectively.

DISCUSSION

Whereas the RGD motifs in both Fib1 36–44 and Fib2 24polypeptides had similar effects on LF adhesion and migrationacross porous filters, distinct differences in haptotaxis wereobserved when LF migrated on a coated glass surface. In thepresence of Fib1 36–44, LF migration was characterized by

greater velocity, more dispersed lamellipodia, and less persis-tence than in the presence of Fib2 24. Because the molarcoating-concentration of RGD was the same for both Fib136–44 or Fib2 24, portions of the peptides outside the immediateRGD motif may influence the cellular response. These influen-ces could involve differences in peptide folding that alter howthe RGD motifs are presented to the LF, or differences incellular association with or response to other portions of thepolypeptides. There may also be differences in the way Fib1 36–44and Fib2 24 promote interactions between cellular integrins andRac GTPase. The precise mechanisms that mediate the variedhaptotactic responses to Fib1 36–44 and Fib2 24 remain un-defined, but our observations indicate that the context of theRGD moiety may be important.

Recent studies by Bax and associates showed that cellularadhesion is influenced by the context in which the RGD motif ispresented (26). Their studies revealed that adhesion is greatlyenhanced when exon 37 of fibrillin 1 is presented in associationwith one or more upstream EGF-like domains (26). Theadhesion of human dermal fibroblasts was primarily mediatedby a1b5–integrin, although aVb3-integrins played a lesser role.They also identified a heparin-binding site in exon 41 that wasnot required for adhesion but promoted the formation of focalcontacts. The EGF-like Exon 36, which Bax and associates(EGF-22 by their terminology) showed enhances adhesion, isalso present on our construct Fib1 36–44. Fib1 36–44 alsocontains the heparin-binding motif in exon 41 that was charac-terized by Bax and coworkers. Although Fib2 24 lacks both theflanking EGF-like domain and the heparin-binding region, weobserved significant adhesion of LF to Fib2 24 at 4 mg per ml,although it was lower than Fib1 36–44 at 0.4 mg per ml.

The adhesion of LF was significantly enhanced by coatingplastic with RGD-containing Fib 1 or Fib 2 peptides, and wasincreased by more than 2-fold, compared with a peptide thatlacked the RGD, when the wells were coated using 4 mg ofpeptide per ml. Others have observed that fibrillins increase the

Figure 8. Fib1 36–44 enhances cellular movement at the leadingedge. The areal changes at the leading and trailing edges of LF were

measured at six 1-hour intervals when the cells were continually

moving. The areal change at the leading edge was divided by the area

of the cell (% positive flow). The distance between the centroids of theprotruding and retracting zones was divided by the length of the cell

along the major axis (polarity). Bars are mean and 1 SEM for 12 cells

randomly selected from four separate experiments for each treatment

group, except Fib1 30, 14 cells. Positive flow: *,#P , 0.05, comparing% positive flow for LF on Fib1 36–44 with LF on Fib 1 30 or only

collagen, respectively; t,1P , 0.05, comparing % positive flow for LF on

Fib2 24 with LF on Fib 1 30 or only collagen, respectively; ANOVA onranks, Kruskal-Wallis post hoc test. Polarity: �P , 0.05, comparing LF

on Fib1 36–44 and LF on only collagen; #P , 0.05, comparing polarity

for LF on Fib1 36–44 and LF on Fib1 30; one-way ANOVA, Student-

Newman-Keuls post hoc test.

Figure 7. Fib1 36–44 alters the shape of LF during haptotaxis. LFmigrating on only collagen (open bars), Fib1 30 (vertical striped bars),

Fib1 36–44 (hatched bars), or on Fib2 24 (cross-hatched bars). (A) Area

in mm2 has been divided by 10 to permit use of the same ordinate scaleas perimeter in mm. (B) Perimeter (P) divided by area (A, P/A, mm21)

has been multiplied by 10 to permit use of the same ordinate scale as

the number of lamellipodia. Bars are mean 6 1 SEM for 40 LF migrating

on only collagen (open bars), 52 LF migrating on 10 mg of Fib1 30 perml 1 collagen (vertical striped bars), 33 LF on 10 mg of Fib1 36–44 1

collagen (hatched bars), and 33 LF on 4 mg of Fib2 24 per ml 1

collagen (cross-hatched bars), respectively. In A, P , 0.05 perimeter for

Fib1 36–44 compared with *collagen or with #Fib2 24. In B, *P/A forFib1 36–44 compared with collagen, Fib1 30, or Fib2 24; **number of

lamellipodia for Fib1 36–44 compared with collagen, Fib1 30, or Fib2

24. ANOVA on ranks, Dunn’s post hoc test.


adhesion of bovine chondroblasts and ligamentum nuchaefibroblasts or various transformed epithelial or meshenchymalcell lines (15, 16). As we have observed with rat LF, theadhesion of bovine nuchal ligament fibroblasts is also inhibitedby blocking the function of aVb3-integrins (16).

Fib2 24 supported migration of LF across collagen coatedpolycarbonate filters to a greater extent than Fib1 36–44 ata fibrillin concentration of 0.5 mg per ml. However, migration onFib2 24 and Fib1 36–44 or Fib1 23–44 were similar when thebottom surface of the filter had been coated with 2 mg of eitherpeptide per ml. Others have suggested that the RGD in exon 24of Fib 2 does not participate in cell attachment based onSakamoto and associates finding that a 12–amino acid peptidecontaining the RGD did not disrupt cellular adhesion topurified fibrillin-1 (16). Our findings indicate that the RGD inexon 24 of Fib 2 promotes cell adhesion and migration when it isassociated with the remaining 73 amino acids in this exon. Westudied the direct attachment of LF to peptides containing theRGD motif, whereas Sakamoto and associates used the shortpeptide containing the Fib2 24-RGD as a competitor to disruptattachment (16, 41). We observed co-localization of Fib2 24 andaV-integrin (Figure 9), which is also consistent with (althoughnot proof of) a physical interaction between these two proteins.

Lung fibroblasts clearly migrate more slowly than someother types of cells (neutrophils and some epithelial cells, forexample). Precise quantitation of velocity is more difficult in

slowly moving cells, because cell spreading could be mistakenfor movement, and the velocity is not constant (Figure 4). Othershave observed saltatory movement of more slowly migratingcells (42). We have used the random-walk model, which othershave shown accurately describes the behavior of fibroblastsfrom other sources on surfaces that are coated with extracellularmatrix molecules (43). We used a prominent landmark in thenucleus rather than the cell centroid to plot the location of thecell at different times, because the nuclear land mark is less in-fluenced by alterations in cell shape, in the absence of move-ment. Because cellular movement exhibits a stochastic response,it was necessary to analyze at least 50 cells (44). There was sig-nificant variation in velocity and persistence among the cellsthat migrated on a particular substrate, and this variation wasnot normally distributed. Therefore we used nonparametricstatistical methods, which ranked the cells according to the mag-nitude of either velocity or persistence, and then compared theeffects of the different substrates. This reduced the effects ofcell-to-cell variation, and promoted a meaningful comparison ofthe populations of cells within the various treatment (coatingsubstrate) groups.

Others have shown that integrins influence the velocity andpersistence of migration through several potential mechanisms.Persistence is directly correlated with the coating concentrationof integrins (40). However, this is unlikely to account for thedifferences that we observed, because our coating concentra-

Figure 9. Fibrillins and aV-integrins

co-localize in perinuclear region

and at focal contacts of LF. After

coating coverslips with Fib1 36–44or Fib2 24, LF were allowed to

adhere and spread in the presence

of PDGF-A. After fixation, LF werepermeabilized either before or after

adding anti-CD51. Fib1 36–44 of

Fib2 24 were identified by the 6-His

tag, aV-chains were identified us-ing anti-CD51, and focal contacts

were identified by anti-paxillin.

These results are representative of

three other experiments using Fib136–44 or Fib2 24, and the distri-

bution of antigens was similar for

LF that adhered to either Fib1 30(not shown).


tions were adjusted to provide equal mols of RGD moieties forFib1 36–44 and Fib2 24. Furthermore, the persistence was notsignificantly altered by varying the concentration of Fib2 24.However, differences in concentration may contribute to thehigher persistence that was observed on Fib1 30 plus collagen,because both coating proteins were present at 10 mg per ml,compared with 10 mg per ml for the collagen-only condition. Wedid observe a concentration-related inverse correlation withvelocity of migration on Fib2 24, because the velocity wasgreater on 4 mg than on 10 mg of Fib2 24 per ml (Figure 5B).Therefore, it is more likely that fibrillin-peptide structuraldifferences account for the greater persistence on Fib2 24 thanon Fib1 36–44 (Figure 5A). There are two structural differencesbetween Fib1 36–44 and Fib2 24 that may contribute to theobserved differences in LF migration. Fib1 36–44 has an EGF-like domain just upstream to exon 37, which produces a hairpinloop in the protein and promotes exposure of the RGD tointegrins (45). In addition, exon 41 of Fib1 36–44 encodes for anarginine-arginine dipeptide heparin-binding motif that pro-

motes the stabilization of focal plaques (26). Pankov andassociates showed that persistence is inversely correlated withthe intracellular levels of activated Rac (9). If the regionsencoded by exon 36 and/or exon 41 in Fib1 36–44 promoteRac-activity, then their absence in Fib2 24 may foster greaterpersistence. We observed that LF migrating on Fib2 24 exhibitfewer lamellipodia at the leading edge (Figure 7) and greaterpolarity (Figure 8) than LF migrating on Fib1 36–44, which isconsistent with the findings of Pankov and coworkers (9).Furthermore, Figure 10 demonstrates that in the presence ofPDGF-A, LF have significantly more collections of Rac, as wellas more lamellopodia when migrating on Fib1 36–44 comparedwith Fib 2 24. This suggests that Fib1 36–44 and PDGF-A mayinfluence Rac to promote less persistent migration than on Fib224.

Guo and associates have recently shown that PDGF pro-moted the p21-activated kinase activation of Rac, resulting ingreater guanalyl cyclase activity (46). This process fostered theformation of lamellipodia and a greater velocity of migration.

Figure 10. Rac is differentially distributed in the presence of Fib 1–36–44. Coverslips

were coated with either Fib1 30, Fib1 36–44, or Fib2 24 and LF were allowed to adhere

and spread in either the presence of the absence of PDGF-A. After fixation, Rac1 was

immunostained and actin filaments were identified using phalloidin. (A) In the absence ofPDGF-A the distribution of Rac was primarily perinuclear but became more diffuse in the

presence of PDGF-A. Discrete collections of Rac (arrows) were observed in the protrusions

of LF that had spread on Fib1 36–44; collections of Rac were less frequently observed in LFthat had spread on Fib2 24. (B) Discrete collections of Rac were enumerated (see

MATERIALS AND METHODS). The mean 6 SEM number of Rac collections per cell from

analyzing 20, 26, and 24 for LF on Fib1 30, Fib1 36–44, and Fib2 24, respectively, are

shown. *P , 0.01 Fib1 36–44 compared with Fib1 30 and Fib2 24, one-way ANOVA,Student-Newman-Keuls post hoc test.


Therefore Fib1 36–44 may enhance the effect of PDGF,resulting in greater velocity and lower persistence than for LFmigrating on Fib 2 24. Our observation that Fib1 36–44 supportsgreater areal change (% positive flow) at the leading edgesuggests that there is more turnover of integrin–fibrillin peptidecontacts and actin reorganization, a known effect of PDGF infibroblasts (7). The precise features of Fib1 36–44 that couldfoster this process require further elucidation.

Disruption of the elastic fiber network is one of the hall-marks of pulmonary emphysema (47, 48). Whereas the levels ofelastin may be restored in emphysematous lungs, the elasticfiber network remains discontinuous and dysfunctional. Rat LFdeposit tropoelastin as they migrate in vitro (49). Structural andfunctional restoration of the network likely involves directionalmigration of interstitial LF along contiguous and continuouspathways. Our studies provide new insight into how fibrillininteracts with LF and could assist in the development of newstrategies for restoring a functional elastic fiber network inpulmonary emphysema.

Conflict of Interest Statement: None of the authors has a financial relationshipwith a commercial entity that has an interest in the subject of this manuscript.

References

1. Burri P, Shittny JC. Morphogenesis of the mammalian lung: aspects of

structure and extracellular matrix. In: Massaro DJ, Massaro GD,Chambon P., editors. Lung development and regeneration. New York:Marcel Dekker; 2004. pp. 275–310.

2. Burri PH. Structural aspects of prenatal and postnatal development and

growth of the lung. In: McDonald JA, editor. Lung growth and devel-opment. New York: Marcel Decker; 1997. pp. 1–35.

3. Mariani TJ, Pierce RA. Development of lung elastic matrix. In: Gaultier

C, Bourbon JR, Post M, editors. Lung development. New York:Oxford University Press; 2002. pp. 28–45.

4. Bostrom H, Willetts K, Pekny M, Leveen P, Lindahl P, Hedstrand H,

Pekna M, Hellstrom M, Gebre-Medin S, Schalling M, et al. PGDF-Asignaling is a critical event is lung alveolar myofibroblast developmentand alveogenesis. Cell 1996;85:863–873.

5. Bostrom H, Gritli-Linde A, Betsholtz C. PDGF-A/PDGF alpha-receptor

signaling is required for lung growth and the formation of alveoli butnot for early lung branching morphogenesis. Dev Dyn 2002;223:155–162.

6. Hauck CR, Hsia DA, Schlaepfer DD. Focal adhesion kinase facilitates

platelet-derived growth factor-BB-stimulated ERK2 activation re-quired for chemotaxis migration of vascular smooth muscle cells.J Biol Chem 2000;275:41092–41099.

7. Boyle SN, Michaud GA, Schweitzer B, Predki PF, Koleske AJ. A critical

role for cortactin phosphorylation of Abl-family kinases in PDGF-induced dorsal wave formation. Curr Biol 2007;17:445–451.

8. Schneller M, Vuori K, Ruoslahti E. AlphaVbeta3 integrin associates

with activated insulin and PDGFbeta receptors and potentiates thebiological activity of PDGF. EMBO J 1997;16:5600–5607.

9. Pankov R, Endo Y, Even-Ram S, Araki M, Clark K, Cukierman E,

Matsumoto K, Yamada KM. A Rac switch regulates random versusdirectionally persistent cell migration. J Cell Biol 2005;170:793–802.

10. Vukmirica J, Monzo P, Le Marchand-Brustel Y, Cormont M. The

Rab4A effector protein Rabip4 is involved in migration of NIH 3T3fibroblasts. J Biol Chem 2006;281:36360–36368.

11. Popova SN, Barczyk M, Tiger D-F, Beersten W, Zigrino P, Aszodi A,

Miosge N, Forsberg E, Gullberg D. a11b1 integrin-dependentregulation of periondontal ligament function in the erupting mouseincisor. Mol Cell Biol 2007;27:4306–4316.

12. Kielty CM, Sherratt MJ, Shuttleworth CA. Elastic fibres. J Cell Sci

2002;115:2617–2626.13. Trask TM, Trask BC, Ritty TM, Abrams WR, Rosenbloom J, Mecham

RP. Interaction of tropoelastin with the amino-terminal domains offibrillin-1 and fibrillin-2 suggests a role for the fibrillins in elastic fiberassembly. J Biol Chem 2000;275:24400–24406.

14. Hubmacker D, Tiedemann K, Reinhardt DP. Fibrillins: from biogenesis

of microfibrils to signaling functions. In: Schatten GP, editor. Current

topics in developmental biology. Amsterdam: Elsevier; 2006. pp. 75,93–123.

15. Pfaff M, Reinhardt DP, Sakai LY, Timpl R. Cell adhesion and integrin

binding to recombinant human fibrillin-1. FEBS Lett 1996;384:247–250.16. Sakamoto H, Broekelmann T, Cheresh DA, Ramirez F, Rosenbloom J,

Mecham RP. Cell-type specific recognition of RGD- and non-RGD-containing cell binding domains in fibrillin-1. J Biol Chem 1996;271:4916–4922.

17. Porst M, Plank C, Bieritz B, Konik E, Fees H, Dotsch J, Hilgers KF,

Reinhardt DP, Hartner A. Fibrillin-1 regulates mesangial cell attach-ment, spreading, migration and proliferation. Kidney Int 2006;69:450–456.

18. Nakamura T, Lozano PR, Ikeda Y, Iwanaga Y, Hinek A, Minamisawa S,

Cheng CF, Kobuke K, Dalton N, Takada Y, et al. Fibulin-5/DANCEis essential for elastogenesis in vivo. Nature 2002;415:171–175.

19. Lamaire R, Bayle J, Mecham RP, Lafaytis R. Microfibril-associated

MAGP-2 stimulates elastic fiber assembly. J Biol Chem 2007;282:800–808.

20. Thomassin L, Werneck CC, Broekelmann TJ, Gleyzal C, Hornstra IK,

Mecham RP, Sommer P. The Pro-regions of lysyl oxidase and lysyloxidase-like 1 are required for deposition onto elastic fibers. J BiolChem 2005;280:42848–42855.

21. Ramirez F, Sakai LY, Dietz HC, Rifkin DB. Fibrillin microfibrils:

multipurpose extracellular networks in organismal physiology. Phys-iol Genomics 2004;19:151–154.

22. Corson GM, Charbonneau NL, Keene DL, Sakai LY. Differential ex-

pression of fibrillin-3 adds to microfibril variety in human and avian,but not rodent connective tissues. Genomics 2004;83:461–472.

23. Kielty CM, Baldock C, Lee D, Rock MJ, Ashworth JL, Shuttleworth

CA. Fibrillin: from microfibril assembly to biochemical function.Philos Trans R Soc Lond B Biol Sci 2002;357:207–217.

24. Handford PA, Downing AK, Reinhardt DP, Sakai LY. Fibrillin: from

domain structure to supramolecular assembly. Matrix Biol 2000;19:457–470.

25. Bax DV, Bernard SE, Lomas A, Morgan A, Humphries J, Shuttleworth

CA, Humphries MJ, Kielty CM. Cell adhesion to fibrillin-1 moleculesand microfibrils is mediated by alpha 5 beta 1 and alpha v beta 3integrins. J Biol Chem 2003;278:34605–34616.

26. Bax DV, Mahalingam Y, Cain S, Mellody K, Freeman L, Younger K,

Shuttleworth CA, Humphries MJ, Couchman JR, Kielty CM. Celladhesion to fibrillin-1: identification of an Arg-Gly-Asp-dependentsynergy region and heparin-binding site that regulaltes focal adhesionformation. J Cell Sci 2007;120:1383–1392.

27. Ritty TM, Broekelmann TJ, Werneck C, Mecham RP. Fibrillin-1 and -2

contain heparin-binding sites important for matrix deposition and thatsupport cell attachment. Biochem J 2003;375:425–432.

28. Skoglund P, Keller R. Xenopus fibrillin regulates directed convergence

and extension. Dev Biol 2007;301:404–416.29. Brettell LM, McGowan SE. Basic fibroblast growth factor decreases

elastin production by neonatal rat lung fibroblasts. Am J Respir CellMol Biol 1994;10:306–315.

30. Chen H, Jackson S, Doro M, McGowan S. Perinatal expression of genes

that may participate in lipid matabolism by lipid-laden lung fibro-blasts. J Lipid Res 1998;39:2483–2492.

31. Ritty TM, Broekelmann T, Tisdale C, Milewicz DM, Mecham RP. Pro-

cessing of the fibrillin-1 carboxyl-terminal domain. J Biol Chem 1999;274:8933–8940.

32. Rye HS, Dabora JM, Quesada MA, Mathies RA, Glazer AN. Fluoro-

metric assay using dimeric dyes for double- and single-stranded DNAand RNA with picogram sensitivity. Anal Biochem 1993;208:144–150.

33. Postlethwaite AE, Seyer JM. Fibroblast chemotaxis induction by human

recombinant interleukin-4. Identification by synthetic peptide analysisof two chemotactic domains residing in amino acid sequences 70–88and 89–122. J Clin Invest 1991;87:2147–2152.

34. Pompe T, Kobe F, Salchert K, Jorgensen B, Oswarld J, Werner C.

Fibronectin anchorage to polymer substrates controls the initial phaseof endothelial cell adhesion. J Biomed Mater Res 2003;67A:647–657.

35. Paborsky LR, Dunn KE, Gibbs CS, Dougherty JP. A nickel chelate mi-

crotiter plate assay for six-histadine-containing proteins. Anal Bio-chem 1996;234:60–65.

36. Stokes DL, Lauffenburger DA, Williams SK. Migration of individual

microvessel endothelial cells: stochastic model and parameter mea-surement. J Cell Sci 1991;99:419–430.

37. Othmer HG, Dunbar SR, Alt W. Models of dispersal in biological

systems. J Math Biol 1988;26:263–298.38. Li Y, Lin JL, Reiter RS, Daniels K, Soll DR, Lin JJ. Caldesmon mutant

defective in Ca21-calmodulin binding interferes with assembly of


stress fibers and affects cell morphology, growth and motility. J CellSci 2004;117:3593–3604.

39. Zicha D, Genot E, Dunn GA, Kramer IM. TGFb1 induces a cell-cycle-dependent increase in motility of epithelial cells. J Cell Sci 1999;112:447–454.

40. Ware MF, Wells A, Lauffenberger DA. Epidermal growth factor altersfibroblast migration speed and directional persistence reciprocallyand in a matrix-dependent manner. J Cell Sci 1996;111:2423–2432.

41. D’Arrigo C, Burl S, Withers AP, Dobson H, Black C, Boxer M. TGF-beta1 binding protein-like modules of fibrillin-1 and -2 mediateintegrin-dependent cell adhesion. Connect Tissue Res 1998;37:29–51.

42. Soll D, Voss E. Two- and three-dimensional computer systems for ana-lyzing how animal cells crawl. In: Soll DR, Wessels D, editors. Motionanalysis of living cells. New York: Wiley-Liss; 1998. pp. 22–52.

43. Dickinson RB, Tranquillo RT. Optimal estimation of cell movementindices from the statistical analysis of cell tracking data. AIChE J1993;39:1995–2010.

44. Skinner MP, Raines EW, Ross R. Dynamic expression of alpha 1 beta 1and alpha 2 beta 1 integrin receptors by human vascular smoothmuscle cells: alpha 2 beta 1 integrin is required for chemotaxis acrosstype I collagen-coated membranes. Am J Pathol 1994;145:1070–1081.

45. Yuan X, Downing AK, Knott V, Handford PA. Solution structure of thetransforming growth factor b-binding protein-like module, a domainassociated with matrix fibrils. EMBO J 1997;16:6659–6666.

46. Guo D, Tan Y-C, Wang D, Madhusoodanan KS, Zheng Y, Maack T, ZhangJJ, Huang X-YA. Rac-cGMP signaling pathway. Cell 2007;128:341–355.

47. Mahadeva R, Shapiro SD. Chronic obstructive pulmonary disease 3:experimental animal models of pulmonary emphysema. Thorax 2002;57:908–914.

48. Niewoehner DE, Kleinerman J. Morphometric study of elastic fibers innormal emphysematous human lungs. Am Rev Respir Dis 1977;115:15–21.

49. Czirok A, Zach J, Kozel B, Mecham RP, Davis EC, Rongish BJ. Elasticfiber macro-assembly is a heirarchical cell motion-mediated process.J Cell Physiol 2006;207:97–106.


Documents

PDF (1424 KB)