Proc. Natl. Acad. Sci. USAVol. 93, pp. 2482-2487, March 1996Cell Biology

Ligand occupancy of the aVf83 integrin is necessary for smoothmuscle cells to migrate in response to insulin-like growth factor I

(growth factors/chemokinesis/atherosclerosis/disintegrin/vitronectin)

JOHN I. JONES, TRACY PREVETrE, AMY GOCKERMAN, AND DAVID R. CLEMMONSDepartment of Medicine, University of North Carolina School of Medicine, Chapel Hill, NC 27599-7170

Communicated by Kenneth M. Brinkhous, University of North Carolina, Chapel Hill, NC, October 26, 1995

ABSTRACT Smooth muscle cells (SMCs) have beenshown to migrate in response to insulin-like growth factor I(IGF-I). However, the mechanism mediating this response hasnot been determined. The migration rates of porcine andhuman vascular SMCs were assessed in a monolayer wound-ing assay. IGF-I and IGF-II induced increases of 141% and97%, respectively, in the number of cells that migrated in 4days. The presence of 0.2% fetal bovine serum in the culturemedium was necessary for the IGFs to stimulate migrationover uncoated plastic surfaces. However, if vitronectin wasused as the substratum, IGF-I stimulated migration by 162%even in the absence of serum. To determine the role ofintegrins in mediating this migration, SMC surface proteinswere labeled with 1251 and immunoprecipitated with specificanti-integrin antibodies. Integrins containing aV (vitronectinreceptor), a5 (fibronectin receptor), and a3 (collagen/laminin receptor) subunits were the most abundant. IGF-Itreatment caused a 73% reduction in a5-integrin subunitprotein and a 25% increase in aV subunit. More importantly,ligand binding of aVf33 was increased by 2.4-fold.We thereforeexamined whether the function of the aVf83 integrin wasimportant for IGF-I-mediated migration. The disintegrinkistrin was shown by affinity crosslinking to specifically bindwith high affinity to aV.83 and not to a5j31 or other abundantintegrins. The related disintegrin echistatin specifically in-hibited 1251-labeled kistrin binding to cVI83, while a struc-turally distinct disintegrin, decorsin, had 1000-fold loweraffinity. The addition of increasing concentrations of eitherkistrin or echistatin inhibited IGF-I-induced migration,whereas decorsin had a minimal effect. The potency of thesedisintegrins in inhibiting IGF-I-induced migration paralleledtheir apparent affinity for the aiV integrin. Furthermore, anaVf83 blocking antibody inhibited SMC migration by 80%o. Insummary, vitronectin receptor activation is a necessary com-ponent of IGF-I-mediated stimulation of smooth muscle mi-gration, and aVj33 integrin antagonists appear to be impor-tant reagents for modulating this process.

Smooth muscle cell (SMC) migration is one of the importantcomponents in the formation of atherosclerotic plaques (1, 2).During restenosis after angioplasty, the accumulation of SMCsin the neointima is believed to be due to the combined processof cell proliferation and directed migration of quiescent cellsfrom arterial media into the intima (1, 3). Both of theseprocesses are potently stimulated by platelet-derived growthfactor (PDGF) and insulin-like growth factor (IGF) 1 (4). SinceIGF-I can be localized in rat (5) or human (6) lesions afterangioplasty and since SMCs possess IGF-I receptors (7), it ispossible that IGF-I acts to stimulate directed SMC migrationafter injury, making an important contribution to lesion de-velopment.

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement" inaccordance with 18 U.S.C. §1734 solely to indicate this fact.

2482

In spite of several important observations regarding IGF-I'srole in mitogenesis (5, 8, 9), its role in stimulating SMCmigration is less well characterized. Recently a study using aBoyden chamber showed that IGF-I stimulates smooth musclemigration, although the stimulation induced by IGF-I was notas great as that induced by PDGF (4). PDGF and IGF-I wereshown to activate different signaling mechanisms, and an intactIGF-I receptor appeared to be necessary for this response tooccur. We have previously shown that IGF-I and IGF-IIstimulate SMC migration in a monolayer wounding assay andthat this response is mediated by the IGF-I receptor andmodulated by IGF binding proteins (10). Growth factors suchas transforming growth factor ( have been shown to modulateintegrin expression on cell surfaces (11), but whether the IGFsmodulate integrin expression has not been analyzed. Likewise,the role of integrin receptor activation in mediating IGF-I-induced migration has not been determined. These studieswere undertaken to determine whether IGF-I and IGF-IIexposure changed integrin receptor abundance and whetherbinding of extracellular matrix (ECM) components to a spe-cific integrin was required for IGF-I to stimulate SMC migra-tion.

METHODSMaterials. Porcine aortic SMCs (pSMCs) were obtained

from thoracic aortas of newborn pigs (age 3 weeks) (2) andwere maintained in culture for 4-6 months as described (10).Human newborn aortic SMCs (hSMCs) were a gift fromSteven Schwartz (University of Washington). They were iso-lated as described (12) and were maintained in Waymouth'smedium supplemented with 10% (vol/vol) fetal bovine serum(FBS). They were shown to contain abundant aVf33 receptorsby flow cytometry (12). Polyclonal antibodies raised againstthe cytoplasmic domains of the human aV, a5, al, a2, a3, (31,(33, and (35 integrin subunits were purchased from Chemicon(Temecula, CA). These antisera are specific for their respec-tive a or , subunit targets but have broad species crossreac-tivity. Highly specific monoclonal antibodies against humanintegrins were obtained from Chemicon (anti-aVf3, cloneLM609), Transduction Laboratories (Lexington, KY) (anti-(3), and Becton Dickinson (anti-f33, clone RUU-PL7F12, andanti-aVf35, clone P1F6). Decorsin and kistrin (13) were giftsfrom Robert Lazarus (Genentech). Echistatin was purchasedfrom Sigma.

Cell Migration Assay. pSMCs were plated at a density of 2X 104 cells per cm2 in 35-mm diameter six-well plates. The cellswere grown for 4 days, then the medium was changed, and thecultures were maintained for an additional 4 days prior towounding. The confluent monolayers were wounded with asingle edge razor blade as described by Burk (14) and modified

Abbreviations: IGF, insulin-like growth factor; PDGF, platelet-derived growth factor; SMC, smooth muscle cell; ECM, extracellularmatrix; pSMC, porcine aortic SMC; hSMC, human newborn aorticSMC; FBS, fetal bovine serum.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 25

, 202

0

Proc. Natl. Acad. Sci. USA 93 (1996) 2483

by Jones et al. (10, 15). The medium was then replaced withDulbecco's modified Eagle's medium (DMEM) and the treat-ments were added. When experiments were conducted ontissue culture plastic, 0.2% FBS was added to the medium.When vitronectin-coated plates were used, migration wasassessed in serum-free DMEM containing 0.01% bovine se-rum albumin. After 3 days, the number of cells migratingacross the wound line was determined. Each data pointrepresents a mean of at least six 1-mm regions selected priorto observing the effect of a given treatment. Previous exper-iments have shown that the addition of IGF-I to pSMC culturesresults in <10% increase in cell number (16). When woundedmonolayers were analyzed by [3H]thymidine autoradiography,the labeling index of the pSMCs at the wound edge increasedfrom 7 ± 4% to 18 ± 7% after exposure to IGF-I. Therefore,we estimate that <10% of the cells present in the denudedregion at the end of the migration assay resulted from celldivision rather than cell migration. Immunoblot analysis (17)of the residual ECM proteins extracted from the denuded areadetected the presence of fibronectin, plasminogen activatorinhibitor 1, vitronectin, and type IV collagen.When migration was assessed in serum-free medium, the

cells were plated at 6 x 104 cells per cm2 on six-well platescoated with vitronectin. To coat the plates, vitronectin (BectonDickinson) at 5 ,Lg/ml was added in 1.0 ml of phosphate-buffered saline (PBS), and the solution was allowed to standfor 2 hr at 22°C and then aspirated. The cells were removedfrom stock 10-cm plates with trypsin/EDTA, the trypsin wasneutralized with 10% FBS, and the cells were washed twicewith serum-free DMEM. A 10-ml suspension of washed cellsin serum-free medium was applied to the vitronectin-coatedplates and allowed to attach overnight. Wounding assays werethen conducted as described (10). After wounding, approxi-mately 50% of the added vitronectin remained adherent to theplates, and the rest had diffused into the medium.To determine whether vitronectin was one of the active

components of serum that facilitated IGF-I-stimulated migra-tion, FBS was >80% depleted of vitronectin by two serialimmunodepletions using an anti-bovine vitronectin antibody(Telios Pharmaceuticals/BRL) immobilized on protein A-Sepharose. The vitronectin-depleted serum was then used inthe migration assay at a final concentration of 0.2%.Immunoblot Analysis. SMCs were grown to 60-65% con-

fluency, incubated in DMEM with 0.2% FBS with or withoutIGF-I or IGF-II for 48 hr (two 10-cm plates per treatment).The cells were removed from the plates by using a nonenzy-matic cell dispersant (Sigma). The dispersed cells were pel-leted, washed once, and resuspended in 1 ml of PBS containing1 mM benzamidine and 1 mM phenylmethylsulfonyl fluoride.The cells were lysed by freezing at -70°C, thawing, and thensonicating 60 sec. The nuclei were removed from the cell lysateby centrifugation at 1000 x g and 4°C for 5 min. The super-natant was centrifuged at 14,000 x g and 4°C for 20 min, andthe resulting membrane pellet was resuspended in 75 ,ul ofLaemmli SDS/PAGE sample buffer without reducing agentsand then boiled for 10 min. The protein concentration of eachof these membrane samples was determined by the BCAprotein assay (Pierce) with bovine serum albumin dissolved inLaemmli buffer as standard. Six micrograms of protein wasloaded onto 7% polyacrylamide gels for SDS/PAGE. The gelswere transferred onto Immobilon-P filters (Millipore) andprobed with a 1:1000 dilution of anti-a5 or anti-aV antisera.The immunoblots were developed with the ECL chemilumi-nescence detection system according to the manufacturer'srecommendations (Amersham).

Immunoprecipitation of 1251-Labeled Surface Integrins. Todetermine which integrins were present on cultured pSMCsurfaces, confluent pSMC monolayers were surface-labeledwith Na125I and immunoprecipitated with anti-integrin anti-bodies as described (15).

Crosslinking 1251-Labeled Kistrin to SMCs. To analyzedisintegrin binding to integrin receptors, Na'25I was used toradiolabel kistrin (18), and the 125I-labeled kistrin was purifiedby gel-filtration chromatography on Sephadex G-50. SMCswere allowed to grow to 90% confluency in 10% FBS andeither used immediately or, when studying the effects of IGF-I,after exposure to serum-free medium for 24 hr with or withoutthe addition of IGF-I (100 ng/ml). The cell monolayers wererinsed twice with Hepes-buffered saline and then were incu-bated with 125I-labeled kistrin overnight at 4°C in DMEMcontaining 20 mM Hepes (pH 7.3) and 0.1% bovine serumalbumin with increasing (10-10-1O-7M) concentrations ofdecorsin, kistrin, or echistatin. The cells were rinsed twice withminimal essential medium containing no amino acids(GIBCO) plus 20 mM Hepes (pH 7.3), and cell surfaceproteins were crosslinked by adding 0.5 mM disuccinimidylsuberate (Pierce). After 30 min at 4°C, the crosslinking agentwas removed and the reaction was stopped by rinsing and thenincubating the cells with 37.5 mM Tris-HCl, pH 7.4/150 mMNaCl for 10 min. The cells were then lysed with 100mM n-octylglucoside in 37.5 mM Tris-HCl (pH 7.4) containing 150 mMNaCl, 1 mM phenylmethylsulfonyl fluoride, and 1 mM ben-zamidine hydrochloride. Aliquots of the lysates were used tomeasure radioactivity in a y counter (Packard) or immuno-precipitated as described above with anti-integrin antibodies.

RESULTS

IGF-I and IGF-II Stimulate pSMC Migration. Exposure ofthe quiescent SMCs to 0.2% FBS resulted in no stimulation ofmigration (Table 1). However, 0.2% FBS was required for thecells to respond to IGF-I. In four experiments, there was noresponse to IGF-I or IGF-II if serum was omitted from themedium. In contrast when added in the presence of 0.2% FBS,IGF-I induced a 141 + 61% increase in the number of cellsmigrating after 4 days and IGF-II induced a 97 ± 42% increase(Table 1). If migration was quantified after 2 days rather than4 days, the absolute number of cells migrating was reduced byapproximately 20%, but the percent increase induced by IGF-Iwas similar (data not shown). Therefore, we concluded that0.2% FBS was required for the IGFs to have an effect and thatwhen 0.2% FBS was added alone, it represented a reasonablebaseline control since it had no effect compared to serum freemedium.pSMCs Express Multiple Integrins. To define the mecha-

nism by which IGF-I and IGF-II stimulate cellular migration,the abundance of various integrins on pSMC surfaces wasdetermined. Radiolabeling of cell surface proteins followed byimmunoprecipitation was performed (Fig. 1). The most abun-dant integrins were precipitated by antibodies against the aS

and f31 integrin subunits (lanes 4 and 6), but significant signalintensities were detected by the aV, aVf35, a3, and al anti-bodies (lanes 5, 9, 3, and 1, respectively). A faint signal wasdetectible by the f35 antibody. However, none of four com-

Table 1. Cellular migration response to IGF-I and II

No. of % increaseTreatment experiments above control Pvalue

FBS (0.2%) 6 0 5 NS*IGF-I alone (no serum) 3 2 + 7 NS*IGF-I + 0.2% FBS 6 141 + 61 <0.OOitIGF-II alone (no serum) 4 0 + 4 NS*IGF-II + 0.2% FBS 4 97 ± 42 <O.OOit

All results are expressed as the percentage increase in the numberof cells migrating compared to the respective control. The totalnumber of cells migrating per cm of wound area was 11-15 forserum-free medium and 11-18 for 0.2% FBS. The average number ofwound areas counted for each test condition was 14. *, Compared toserum-free control; t, compared to 0.2% serum control.

Cell Biology: Jones et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 25

, 202

0

Proc. Natl. Acad. Sci. USA 93 (1996)

1 2 3 4 5 6 7 8 9 10.0::.

kDa ..

-200 -: ..

, /,13 FI -974 : * 13~/[94

tkl.. ~ - 46 -_t _ .}.

IP: al aX2 a3 a5 ayv Pl P5 (xy P S

(Xy p5

FIG. 1. Immunoprecipitation of 1251-labeled pSMC. Subconfluentcultures of pSMCs were surface-labeled with Na'251, and cell lysateswere immunoprecipitated (IP) with polyclonal antisera raised againstsynthetic peptides containing sequences from the cytoplasmic domainsof ail (lane 1), a2 (lane 2), a3 (lane 3), a5 (lane 4), aV (lane 5), ,B1(lane 6), and j35 (lane 7). In a subsequent experiment, 125I-labeledpSMC proteins were immunoprecipitated with anti-aV (lane 8) andmonoclonal anti-aVj35 (lane 9). The supernatant from the aV,B5immunoprecipitation was then immunoprecipitated again with anti-aV (lane 10). Shown are autoradiographs of dried gels after nonre-ducing SDS/PAGE in 6% gels. The arrows indicate the locations of the1251-labeled integrin subunits in the gels.

mercially available anti-(3 integrin antibodies tested was ableto precipitate detectible amounts of (33 integrin from pSMCs(data not shown). In contrast, three of the four antibodiesimmunoprecipitated radiolabeled (33 integrin when humanSMCs were used (see Fig. 8), suggesting that these antibodiesare unable to immunoprecipitate porcine (33. The aV integrinthat was present on pSMC surfaces was only partially precip-itated by the aV(35 antibody, since the supernatant remainingafter aV(35 immunoprecipitation still contained aV integrin,presumably aVP3 (Fig. 1, lanes 8-10). These results suggestthat the a5(31 fibronectin/IGF binding protein 1 receptor (15)was the most abundant integrin and that significant amountsof the aV(33 vitronectin receptor, as well as a3(31 and al(1,were present. We were unable to directly determine therelative amounts of aV(33 vs. aV,(5 in pSMCs.To determine whether IGF-I exposure altered the expres-

sion of two of the most abundant integrins, immunoblotanalysis for the fibronectin and vitronectin receptor a subunitswas performed by using pSMC membrane preparations. Anantibody specific for the a5 subunit showed that there was anapproximately 70% reduction in the fibronectin receptor afterthe cells were exposed for 72 hr to IGF-I and a 45% reductionafter exposure to IGF-II (Fig. 2). In contrast, immunoblotanalysis for the aV subunit of the vitronectin receptor showed

kDa-200

-97

1 2 3 4 5 6

FIG. 2. Alterations in a5 and aV proteins after IGF exposure.Subconfluent SMC monolayers were exposed to IGF-I or IGF-II andthen membrane proteins were extracted. Cell lysates (6 ,tg of totalprotein per sample) were electrophoresed in a SDS/7% gel. Aftertransfer to poly(vinylidene difluoride) filters, the filters were exposedto antibodies against the cytoplasmic domains of aV (lanes 1-3) or a5(lanes 4-6) integrin subunits. Lanes 2 and 6 were obtained fromcultures that were exposed to IGF-I (100 ng/ml) and lanes 3 and 5 arefrom cultures that received IGF-II (100 ng/ml). In response to IGF-I,the abundance of the fibronectin receptor decreased more than 70%as assessed by scanning densitometry, whereas the aV receptor bandintensity increased approximately 25%.

that there were 20% and 25% increases in aV receptors afterexposure to IGF-II and IGF-I, respectively, suggesting that thisreceptor was regulated differently by the IGFs.

IGF-Stimulated Migration over Vitronectin Occurs in theAbsence of Serum and Is Inhibited by Kistrin and Echistatin.Because of the effects of the IGFs on vitronectin receptorexpression, we wondered whether the IGF effect was mediatedthrough a vitronectin receptor. When pSMCs were plated onwells coated with vitronectin in the absence of serum and thenwounded, their migration increased 2.5-fold in response toIGF-I even in the absence of serum (Fig. 3). When thedisintegrins echistatin or kistrin were added, between concen-trations of 10-10 and 10-7 M, the migration response to IGF-Iwas inhibited. However, if these peptides were added alone,they had no effect. Decorsin inhibited the IGF-I effect mini-mally. To determine whether vitronectin was one of the aVf3ligands in serum necessary for IGF-I to stimulate migration,the pSMC migration response to IGF-I was quantified by using0.2% FBS that had been immunodepleted (>80% removed)with anti-vitronectin antibody. The migration response wasreduced by 72 ± 11% (mean of three experiments).

Because of the unavailability of antibodies capable of block-ing porcine integrins, we extended our migration studies tohSMCs. The response of hSMCs to IGF-I and IGF-II wassimilar to that of pSMCs (Fig. 4). Approximately 80% of theincrease in migration of hSMCs induced by IGF-I was alsoinhibited by kistrin and echistatin. Most importantly, a block-ing anti-aV,33 integrin antibody (LM609) caused an equivalentinhibition of IGF-I-stimulated migration.

Kistrin and Echistatin Bind with High Affinity to aV,83Integrin Receptors on SMCs. To determine the specificity ofthese disintegrins for integrin binding and to assess theirrelative affinities for aV integrins, kistrin was radiolabeled,allowed to bind to the pSMC surface, and then crosslinked tocell surface proteins. Anti-integrin antibodies were then usedto immunoprecipitate the crosslinked complexes. As shown inFig. 5, 125I-labeled kistrin could be immunoprecipitated byantibodies directed against the cytoplasmic tail of the aVintegrin. A smaller amount of crosslinked kistrin could beimmunoprecipitated by a 31 antibody, but none was precipi-tated by antibodies against a1, a2, a3, aS, ,B3, ,35, or aV,B5. The

0

4)

0

U)(U0111U-

180

160 -

140 -

120 -

100 -

80

60 -

40

20 -

0-IGFalone

Disintegrin Concentration

1 109M1 1o04M

nKistrin

a10-7 M T

Echistatin Decorsin

180

- 160

- 140

- 120

- 100

- 80

- 60

- 40

- 20

-0

FIG. 3. Effect of the disintegrins on pSMC migration. Increasingconcentrations of kistrin, echistatin, or decorsin were incubated witha constant amount of IGF-I (100 ng/ml), and pSMC migration onvitronectin-coated plates was determined after 4 days. The results areexpressed as percent of increase over control cultures that containedonly serum-free medium. Each bar represents the mean ± SD of sixto nine wounds per experiment from three experiments. IGF-I induceda 149% increase in cell migration compared to control. Both kistrinand echistatin were effective inhibitors of IGF-induced migration at10-8 M, while decorsin was much less potent. The addition of theseconcentrations of kistrin, echistatin, or decorsin had no effect on thebasal rate of cell migration in the absence of added IGF.

2484 Cell Biology: Jones et al.

(x_*

a2M2

Dow

nloa

ded

by g

uest

on

Dec

embe

r 25

, 202

0

Proc. Natl. Acad. Sci. USA 93 (1996) 2485

c0C)0

0.0

00at

0-C)

180

160

140

120

100

80

60

40

20

00.2%Serum

IGF-I IGF-II IGF-I + IGF-I + IGF-I +Kistrin Echistatin LM609

FIG. 4. Effect of disintegrins (10-7 M) and anti-aV133 antibody(LM609, 25 ,tg/ml) on IGF-I (100 ng/ml)-stimulated migration ofhSMCs. Migration assays were performed as described in Fig. 3.Responses to IGF-I and IGF-1I stimulation were indistinguishablefrom pSMCs. The response to IGF-I was >80% inhibited by kistrin,echistatin, and the aV133 antibody.

125I-labeled kistrin appeared to have been crosslinked pre-dominantly to the integrin ,B subunits.

Competition studies showed that the 125I-labeled kistrinbinding and crosslinking were specific. Both unlabeled kistrinand echistatin were potent inhibitors of 125I-labeled kistrinbinding to cell surfaces and could completely inhibit binding athigh concentrations (Fig. 6). The concentrations of kistrin andechistatin necessary to inhibit 125I-labeled kistrin binding tocell surfaces by 50% were both on the order of 10-9 M, whereasdecorsin was approximately 1000-fold less potent. To deter-mine whether unlabeled kistrin and echistatin were equipotentin competing with 125I-labeled kistrin for binding to aV vs. (31integrins, we immunoprecipitated pSMC lysates crosslinked to125I-labeled kistrin in the presence of competing concentra-tions of echistatin and unlabeled kistrin. Both kistrin andechistatin competed for binding to both the aV and the (31integrins with an IC50 of approximately 10-9 M (Fig. 7).While kistrin binds with high affinity to both aV and (31

integrins, on pSMC surfaces most of the binding was to the aVintegrin. Fig. 7, lane 1, shows 125I-labeled kistrin crosslinkingto unselected pSMC surface proteins. Most of the 125I-labeledkistrin that crosslinked to surface proteins (lane 1) could beimmunoprecipitated with anti-aV (lane 2). In contrast, only a

1 2 3 4 5 6 7 8

kDa

200 -

97.4-

69 -

A

V 4-

4*- (33

aP _

lp: al 2 CY3 Ca5 a. p1, f5 a,P5

FIG. 5. Immunoprecipitation of pSMC lysates after crosslinking to1251-labeled kistrin. Subconfluent monolayers of pSMCs were incu-bated overnight with 1251-labeled kistrin and then cell surface proteinswere crosslinked by exposure to disuccinimidyl suberate. Cell lysateswere immunoprecipitated with the same antibodies as in Fig. 1 and theradiolabeled proteins were resolved by SDS/PAGE on 6% gels undernonreducing conditions. An autoradiograph of a dried gel is shown.The specific antibody used to immunoprecipitate (IP) each cell lysateis shown. The positions of the ,31 and 133 integrin subunits crosslinkedto 1251-labeled kistrin are indicated with arrows.

140* Kistrin

120 - O Echistatino Decorsin

0

0

100

80

60

40

20

140

120

100

80

60

40

20

0.1 1.0 10.0 100.0 1000.0Concentration of competing disintegrin, nM

FIG. 6. Competition of disintegrins with 1251-labeled kistrin forbinding to pSMCs. 1251-labeled kistrin (6 x 105 cpm per well) wasincubated with subconfluent monolayers of pSMCs on six-well plateswith the addition of 0 to 10-6mM unlabeled kistrin (0), echistatin (0),or decorsin (c:1). After overnight equilibration at 40C, surface proteinswere crosslinked. After crosslinking, the cells were washed, then lysed,and the amounts of 125I-labeled kistrin in equal aliquots of the lysateswere determined by y scintillation counting. Approximately 4-5% ofthe total 1251-labeled kistrin added in the absence of unlabeled ligandbound to the cells. Total binding (mean ± SD) is shown as a percentof binding in the absence of unlabeled ligand, performed in triplicatein four assays. Nonspecific binding, as demonstrated by binding in thepresence of 1000 nM kistrin, was <1% of total binding.

small portion of the total 125I-labeled kistrin crosslinked topSMCs (i.e., a relatively minor band in lane 1) could beimmunoprecipitated with anti-,Bl (lane 6). By quantitatingband intensities in lane 1 of this gel with a Phosphorlmager, wedetermined that only 11% of the 125I-labeled kistrincrosslinked to pSMC proteins larger than 46 kDa was boundto the (31 integrin, while the remaining 89% was bound to theaV integrin.The identity of the ,B chain associated with the aV kistrin-

binding integrin on pSMCs was apparently not f35, since the 35and aVf35 antibodies precipitated porcine aVf35 surface-

1 2 3 4 5 6 7 8 9 10 11 1213

kDa200-

97.4- _

69-

46 -.

Li;and COnC. 0 0 1O-10 10-8 0 1010 109 10.8 0 10-9 108 7Logand: E EE E EE K K K

IP: 0 (V 131 AV

FIG. 7. Anti-integrin aV and 131 immunoprecipitations of 1251-labeled kistrin crosslinked to pSMCs with kistrin and echistatincompetition. 1251-labeled kistrin was crosslinked to pSMC surfaces, inthe presence of the indicated molar concentrations of echistatin (E) orkistrin (K). The cells were then lysed and the lysates were immuno-precipitated (IP) with antisera directed against aV (lanes 2-5 and10-13) or 131 (lanes 6-9) integrin subunits. Lane 1 contains thecrosslinked total radiolabeled cell lysate, and lanes 2-13 containimmunoprecipitates. SDS/PAGE was performed under nonreducingconditions in 6% polyacrylamide gels, and autoradiographs of thedried gels are shown. Arrows indicate the locations of the 131 and 33integrin subunits, which were the protein bands that were labeled mostintensely after crosslinking with 1251-labeled kistrin.

Cell Biology: Jones et al.

.+s

Dow

nloa

ded

by g

uest

on

Dec

embe

r 25

, 202

0

Proc. Natl. Acad. Sci. USA 93 (1996)

1 2 3 4 5 6 7kDa

200 -

1 2 3 4

kDa200 -

97.4-

69 -

4--

av 131 03 33 4 5

FIG. 8. Immunoprecipitation of hSMC integrins crosslinked to125I-labeled kistrin. Surface proteins of hSMCs were crosslinked to1251-labeled kistrin and cell lysates were immunoprecipitated andresolved by SDS/PAGE on 6% gels. Antibodies used were polyclonalanti-aV, polyclonal anti-131, polyclonal anti-13, monoclonal anti-,33(RUU-PL7F12), monoclonal anti-aV133 (LM609), polyclonal anti-,85,and monoclonal aV135 (P1F6) (lanes 1-7, respectively). The arrow

indicates the position of the crosslinked 133 integrin subunit, which wasprecipitated by the aV and 13 antibodies but not by the 131 or 135antibodies.

labeled with Na1251 (Fig. 1, lanes 7 and 9) but did notprecipitate 125I-labeled kistrin crosslinked to the surfaces ofthese cells (Fig. 5, lanes 7 and 8). Since the anti-133 andanti-aV133 antibodies would not precipitate porcine aV,B3,similar studies were performed with hSMCs that had beenshown to express both aVVl33 and aVP5 (12). All of the125I-labeled kistrin that crosslinked to these cells could beprecipitated with anti-133 and with anti-aV,B3, but not withpolyclonal anti-135 or monoclonal anti-aV135 (Fig. 8). Theseresults strongly suggest that kistrin binds to aV13 but not toaV135 and support the conclusion that kistrin mediates itseffects on pSMCs by blocking aV13.

IGF-I Increases Ligand Binding to Vitronectin Receptors onpSMCs. To determine the effects of IGF-I on vitronectinreceptor ligand binding, we crosslinked 125I-labeled kistrin topSMC monolayers that had been exposed for 24 hr to serum-

free medium with or without added IGF-I (100 ng/ml).Immunoprecipitation of lysates of the crosslinked cell surfaceproteins (Fig. 9) demonstrates a 2.4-fold increase in 1251_labeled kistrin binding to the aV integrin but no change inbinding to the 131 integrin.

DISCUSSIONWe have previously shown that IGF-I and IGF-II are potentstimuli of pSMC migration and that their effects appear to bemediated through the IGF-I receptor (10). In this study we

demonstrate that the SMC migration response to the IGFs isdependent upon the aV,B3 integrin and the presence of ligandsthat can bind to this receptor, such as vitronectin. IGF-I or

IGF-II did not stimulate migration in serum-free medium, butvitronectin coating of plates was sufficient to observe an effectof either IGF-I or IGF-II in the absence of serum. Immu-nodepletion experiments showed a decrease in the cell migra-tion response to IGF-I after the vitronectin concentration inserum was decreased, suggesting that it is one of the activeserum components. However, other aV133 ligands such as

fibrinogen, von Willebrand factor, or osteopontin may also beactive. The importance of the aV,33 vitronectin receptor forIGF-I-stimulated migration is supported by the fact that twodisintegrins that bound to the aV,33 integrin with high affinityand specificity could inhibit the migration response to IGF-Ion vitronectin-coated plates. Both compounds were active atconcentrations that bind to the aV133 receptor (19), and their

W.0,97.4- * 4-3

69-d

46-

IGF: - + - +IP: (LV [I,

FIG. 9. Crosslinking of 125I-labeled kistrin to pSMC cultures thathad been exposed to IGF-1. pSMC lysates were immunoprecipitatedwith anti-aV (lanes 1 and 2) or anti-131 (lanes 3 and 4) after cell surfaceproteins were crosslinked to 1251-labeled kistrin. Prior to crosslinkingthe cultures had been incubated with IGF-I (100 ng/ml) (lanes 2 and4) or serum-free medium alone (lanes 1 and 3) for 24 hr. Theimmunoprecipitates were resolved by nonreducing SDS/PAGE on 6%gels, and crosslinked 1251-labeled kistrin was detected by autoradiog-raphy. Arrows indicate the locations of the 131 and 133 integrin subunits.The intensities of these bands were quantitated by using a Phosphor-Imager (Molecular Dynamics), which determined that IGF-I exposureincreased kistrin binding to aV133 by 2.4-fold (lane 2 vs. lane 1), whilethere was no change in kistrin binding to the 131 integrin (lane 4 vs. lane3).

relative potencies in inhibiting cellular migration were similarto their potencies in inhibiting binding to pSMC surfaces. Incontrast, decorsin, which has high affinity for the Ilb/IIIaplatelet integrin receptor but low affinity for aVf33, was arelatively poor inhibitor of migration. The majority of the1251-labeled kistrin that bound was crosslinked to aV133 ratherthan to the other abundant integrins on the pSMC surface,suggesting that its effects were specific for this receptor (Fig.5). A specific antibody to human aV,B3 blocked hSMC migra-tion as much as did kistrin. Finally, vitronectin receptor ligandbinding was increased by IGF-I exposure. Thus, these resultssuggest that aV133 occupancy is required for IGF-I to stimulatecell migration.Our results suggest that aV13, rather than aVf5, is the

integrin that mediates the IGF effect on SMC migration. Bothintegrins bind vitronectin, and both integrins have been shownto mediate cell migration (20, 21). The 135 integrin wasdetected in immunoprecipitates of surface-labeled pSMC ly-sates but was not detected in lysates crosslinked to 125I-labeledkistrin, suggesting that kistrin did not bind to aV135. Sincekistrin blocks IGF-stimulated pSMC migration, aVf35 is notlikely to mediate this effect in pSMC. Our data using humanSMCs more strongly support involvement of aVP3 rather thanaV/35, since crosslinked 1251-labeled kistrin could specificallybe immunoprecipitated with aV133 antibodies but not thoseagainst aVP35. The importance of aVP33 in the migration ofhSMCs was confirmed by the ability of anti-aV133 to blockIGF-stimulated migration in these cells. It is unlikely thataV136 mediates the IGF stimulation or kistrin inhibition ofpSMC migration, since this integrin has been reported tospecifically bind to fibronectin and not to vitronectin (22). Wehave not ruled out the presence of the aV138 vitronectinreceptor in our cells; however, this integrin has not beendetected in vascular tissue (23). The identity of the ,13 integrinthat contributed a minor proportion of kistrin binding is notknown. None of the antibodies against a-integrin subunitsprecipitated 125I-labeled kistrin with a band corresponding tothe 131 subunit, even though all of these antibodies precipitate125I-labeled 131 integrins from the same cells (Fig. 5).

2486 Cell Biology: Jones et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 25

, 202

0

Proc. Natl. Acad. Sci. USA 93 (1996) 2487

The conclusion that a vitronectin receptor mediates IGFstimulation of pSMC migration is further strengthened by theobservation that expression and, in particular, ligand bindingof this receptor was increased after exposure to IGF-I. Incontrast, IGF-I induced a 73% reduction in the fibronectinreceptor. Since fibronectin receptors were down-regulated byIGF-I, the findings suggest either that a531 is less importantthan aV,B3 for IGF-I-stimulated migration or that a5f31 bind-ing to fibronectin inhibits IGF-I-stimulated SMC migration.This induction of the aV subunit with concomitant down-regulation of the a5 subunit is similar to that seen in fibroblasts3-4 days after wounding during the time of maximal cellularmigration (24). It has been proposed that induction of aVf33may be more important for migration whereas a5031 may bemore important for retraction (11). Similarly f3 integrinsappear to be more important for migration of SMCs comparedto (31 integrins (25). In addition to IGF-I, transforming growthfactor ,B and PDGF have also been shown to induce the aV,33receptor in cultured SMCs (11, 26). Therefore, more than onegrowth factor may use this receptor to facilitate migration.The relative potencies of kistrin, echistatin, and decorsin in

blocking IGF-I-stimulated migration on vitronectin-coatedplates were comparable to their relative affinities for thevitronectin receptor. The IC50 value for echistatin inhibition ofvitronectin binding to pure aVf33 in vitro is 10.1 ± 0.4 nM (27).Our studies show that kistrin has a similar affinity for aVP3integrin and inhibits migration with a similar potency. Incontrast, decorsin is a poor inhibitor of 125I-labeled kistrinbinding and pSMC migration. These data verify that kistrinand echistatin are useful reagents for blocking aVI33 integrinbinding to ECM components.Our studies raise important questions regarding the mech-

anism by which binding and activation of the IGF receptor islinked to induction of aVf33 and to down-regulation of a531.Recently, Vuori and Ruoslahti (28) reported that 100 nMinsulin, a concentration that binds to the IGF-I receptor,stimulates the association of phosphorylated IRS-1 with theaVP3 receptor. IRS-1 is a major phosphotyrosine-containingdocking protein that is phosphorylated by the IGF-I receptorand has a role in IGF-I receptor signal transduction. If thisinteraction occurred in SMCs, it would support our conclusionthat cooperative interactions between these two receptorsmediate IGF-I/insulin-induced SMC migration. PDGF hasbeen shown to induce the phosphorylation of a 190-kDaprotein that binds to the aVI33 integrin, suggesting that theremay be an interaction between other growth-factor-inducibleproteins and integrin function in other systems (29). BothIGF-I and IGF-II have been shown to stimulate migration ofother cell types. Melanoma cells (30) and normal cell typessuch as bronchial epithelial (31) and endothelial cells (9, 32)show increased migration in response to IGF-I. However, theactual ECM components and integrins that mediate thesemigration responses have not been defined.

In summary, IGF-I induction, an important component ofthe SMC response to injury, stimulates pSMC migration bymechanisms that involve the aV(33 vitronectin receptor andinduces an increase in vitronectin receptor ligand binding.Vitronectin receptor activation pathways and IGF-I receptorsignal transduction pathways may interact to facilitate migra-tion. Further studies to determine the molecular mechanismsby which vitronectin, IGF-I, and their receptors cooperate ininfluencing SMC motility may be important in understandingthe role of the IGFs in atherogenesis.

We thank Dr. Walker H. Busby, Jr., who prepared the iodinated

kistrin that was used in these studies. We thank Dr. Robert Lazarusof Genentech, Inc. for his generous gift of decorsin and kistrin and Dr.Steven Schwartz (University of Washington) for his gift of hSMCs. Wethank Ms. Leigh Elliott for her help in preparing the manuscript. Thisstudy was supported by grants from the National Institutes of Health(HL-26309 and DK02024).

1. Jackson, C. L. & Reidy, M. A. (1992) Ann. N.Y. Acad. Sci. 667,141-150.

2. Ross, R. (1971) J. Cell Biol. 50, 172-186.3. Epstein, S. E., Siegall, C. B., Biro, S., Fu, Y. M., Fitzgerald, D. &

Pastan, I. (1991) Circulation 84, 778-787.4. Bornfeldt, K. E., Raines, E. W., Nakano, T., Graves, L. M.,

Krebs, E. G. & Ross, R. (1993) J. Clin. Invest. 93, 1266-1274.5. Cercek, B., Fishbein, M. C., Forrester, J. S., Helfant, R. H. &

Fagin, J. A. (1990) Circ. Res. 66, 1755-1760.6. Grant, M. B., Wargovich, T. J., Ellis, E. A., Caballero, S., Man-

sour, M. & Pepine, C. J. (1994) Circulation 89, 1511-1517.7. Pfeifle, B., Boeder, H. & Ditschuneit, H. (1987) Endocrinology

120, 2251-2258.8. Clemmons, D. R. (1985) Endocrinology 117, 77-83.9. Grant, M., Jerden, J. & Merimee, T. J. (1987) J. Clin. Endocrinol.

Metab. 65, 370-371.10. Gockerman, A., Jones, J. I., Prevette, T. & Clemmons, D. R.

(1995) Endocrinology 136, 4168-4173.11. Basson, C. T., Kocher, O., Basson, M. D., Asis, A. & Madri, J. A.

(1992) J. Cell. Physiol. 153, 118-128.12. Liaw, L., Skinner, M. D., Raines, E. W., Ross, R., Cheresh, D. A.,

Schwartz, S. M. & Giachelli, C. M. (1995) J. Clin. Invest. 95,713-724.

13. Krezel, A. M., Wagner, G., Seymour-Ulmer, J. & Lazarus, R. A.(1994) Science 264, 1944-1946.

14. Burk, R. R. (1973) Proc. Natl. Acad. Sci. USA 70, 369-372.15. Jones, J. I., Gockerman, A., Busby, W. H., Jr., Wright, G. &

Clemmons, D. R. (1993) Proc. Natl. Acad. Sci. USA 90, 10553-10557.

16. Elgin, R. G., Busby, W. H. & Clemmons, D. R. (1987) Proc. Natl.Acad. Sci. USA 84, 3254-3258.

17. Jones, J. I., Gockerman, A., Busby, W. H., Camacho-Hubner, C.& Clemmons, D. R. (1993) J. Cell Biol. 121, 679-687.

18. Dennis, M. S., Henzel, W. J., Pitti, R. M., Lipari, M. T., Napier,M. A., Deisher, T. A., Bunting, S. & Lazarus, R. A. (1990) Proc.Natl. Acad. Sci. USA 87, 2471-2475.

19. Gould, R. J., Polokoff, M. A., Friedman, P. A., Huang, T. F.,Holt, J. C., Cook, J. J. & Niewainoski, S. (1990) Proc. Soc. Exp.Biol. Med. 195, 168-172.

20. Kim, J. P., Zhang, K., Chen, J. D., Kramer, R. H. & Woodley,D. T. (1994) J. Biol. Chem. 269, 26926-26932.

21. Leavesley, D. I., Schwartz, M. A., Rosenfeld, M. & Cheresh,D. A. (1993) J. Cell Biol. 121, 163-170.

22. Busk, M., Pytela, R. & Sheppard, D. (1992) J. Biol. Chem. 267,5790-5796.

23. Moyle, M., Napier, M. A. & McLean, J. W. (1991) J. Biol. Chem.266, 19650-19658.

24. Clark, R. A. (1993) Am. J. Med. Sci. 306, 42-48.25. Clyman, R. I., Mauray, F. & Kramer, R. H. (1992) Exp. Cell Res.

200, 272-284.26. Janat, M. F., Argraves, W. S. & Liau, G. (1992) J. Cell. Physiol.

151, 588-595.27. Scarborough, R. M., Rose, J. W., Naughton, M. A., Phillips,

D. R., Nannizzi, L., Arfsten, A., Campbell, A. M. & Charo, I. F.(1993) J. Biol. Chem. 268, 1058-1065.

28. Vuori, K. & Ruoslahti, E. (1994) Science 266, 1576-1578.29. Bartfeld, N. S., Pasquale, E. G., Geltosky, J. E. & Languino, L. R.

(1993) J. Biol. Chem. 268, 17270-17276.30. Stracke, M. L., Engel, J. D., Wilson, L. W., Rechler, M. M.,

Liotta, L. A. & Shiffmann, E. (1989) J. Biol. Chem. 264, 21544-21549.

31. Shoji, S., Ertl, R. F., Linder, J., Koizumi, S., Duckworth, W. C. &Rennard, S. I. (1990) Am. J. Respir. Cell Mol. Biol. 2, 553-557.

32. Nakao-Hayashi, J., Ito, H., Kanayasu, T., Morita, I. & Murota, S.(1992) Arteriosclerosis 92, 141-149.

Cell Biology: Jones et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 25

, 202

0