Vol. 8, 1339-1347, December 1997 Cell Growth & Differentiation 1339

3 The abbreviations used are: ECM, extracellular matrix; FN, fibronectmn;RGD, arginine-glycine-aspartic acid; mAb, monoclonal antibody.

Differential Cation Regulation of the a5131 Integrin-mediatedAdhesion of Leukemic Cells to the Central Cell-bindingDomain of Fibronectin1

Zhinan Yin, Elena Gabriele, Alessandra Leprini,Roberto Perils, and Alfonso Colombatti2

DMsione di Oncologia Sperimentale 2, Centro di RiferimentoOncologico, 33081 Aviano, Italy [Z. V., E. G., R. P., A. C.]; Laboratoriodi Biologia Cellulare, Istituto Nazionale per Ia Ricerca sul Cancro, 16132Genova, Italy [A. LI; and Dipartimento di Scienze e TecnologieBiornediche, Universit#{224}di Udine, 35100 Udine, Italy [A. C.]

AbstractNormal and neoplastic leukocytes interact with thecentral cell-binding and carboxyl-terminal regions offibronectin (FN) primarily via the a4fi1 and a5131integrins. By using a unique centrifugation-based cell

adhesion assay and affinity chromatography of the cellsurface-labeled integrins, we show in this study thatthe constitutive a5fi1-dependent attachment of threeleukemic cell lines, BV-173, K562, and Nahm-6, to FN orto a I 1O-kDa central cell-binding fragment of FN wastotally inhibited at 37#{176}Cby preincubation of thesubstrate with either antibodies to the arginine-glycine-aspartic acid-containing region (3Fn-1O module) or tothe synergistic region (3Fn-9 module) of FN. Similarresults were obtained when assays were carried out at4#{176}C,suggesting that energy-dependent events werenot involved. On the other hand, only the antibodyagainst the 3Fn-1O module was able to detach mostfirmly adherent cells. Constitutive cell attachment to

the 11O-kDa fragment was cation dependent, with theorder of efficacy of the cation being Mn2� > Mg2� >

Ca2’�, and Ca2� was only effective for BV-173 cells.Antibodies against the a�3� integrin or the a5 subunit

completely impaired cell attachment of all cell lines,whereas several blocking anti-�31 subunit antibodies,including 4B4, P4CIO, and AIIB2, differentiallyperturbed cell adhesion of BV-173, depending on whichcation was present. These anti-131 blocking antibodies,whose epitopes map to a region distant from the one

Received 7/16/97; revised 9/22/97; accepted 10/21/97.The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely to mdi-cate this fact.1 Supported by funds from Consiglio Nazionale delle Ricerche ProgettoApplicazmoni Cliniche dell Ricerca Oncologica, Associazione Italiana per IaRicerca sul Cancro, Ministero della Sanit#{224}Ricerca Finalizzata Istituto diRicovero e Cura a Carattere Scientifico 1992-i 994, and MinisterodeII’Universit#{224} e Ricerca Scientifica e Tecnologica (grants, 40 and 60%).z. Y. was supported by a fellowship from the Clinical and ResearchTraining Program Center at the Centro di Riferimento Oncologico diAviano, and E. G. was supported by a fellowship from the AssociazioneItaliana per Ia Ricerca sul Cancro.2 To whom requests for reprints should be addressed, at Divisione diOncologia Sperimentale 2, CR0, 33081 Aviano, Italy. Phone: 0039-434-659-301 ; Fax: 0039-434-659-428; E-mail: acolombatti@ets.it.

expected to contain the 131 putative cation bindingsites, seemed to hock the higher activation state of thisintegrin attained in the presence of Mg2� and topreserve it during the subsequent adhesion eventsirrespectively of the presence of the low avidity state-inducing Ca2’ ion. Because the higher binding aviditydisplayed by BV-173 could not be explained by a higher

degree of prechustenng of a5131integnn in this cell linecompared to K562, these findings suggest that cationsmight act as allostenc activators of integrin function.Whether this novel cation-dependent parameter ofa5I�i integrin-FN interaction might contribute to thegrowth control and/or tissue dissemination of leukemiccells remains to be determined.

IntroductionCell adhesion molecules are critical for the physiology ofnormal leukocytes, in which expression of distinct receptorsat different stages of maturation may regulate cell-cell andcell-substrate interactions, and may be important for thetrafficking of leukocytes between tissue compartments (i-4).The same types of receptors may play a similar role in the

process of dissemination of neoplastic cells through various

tissues (5). Furthermore, it is likely that the massive egress

from the bone marrow, which follows malignant transforma-

tion, progression into the leukemic phase, and the transitionto a more aggressive course of the disease, involves multipleinteractions of the malignant cells with cell surface elementsand ECM3 (3) constituents. FN is an ECM constituent that isclassically abundant in the bone marrow and most othertissues and has conventionally served as the prototype mol-ecule for studies of leukocyte-ECM interaction. It is wellknown that cell adhesion with FN is mediated by multiplebinding sites located in the central and carboxyl-terminalregions of the molecule (6-8). The central cell-binding do-main contains the RGD sequence, which may be recognizedby the a5 (9), ca3 (i 0), civ (i i), and a4 (1 2) members of the pi

integrin family and by the integrin aIIbI33 (i 3). One synergisticcell-binding region present within this domain has been pro-posed to cooperate with the RGD sequence for optimal cell

interaction with FN (i4). To ensure optimal cehl-ECM inter-action, heukocytes express several f31 -class integrins whosebinding activity may vary in different cell phenotypes and isnormally susceptible to rapid modulation by intrinsically or

extrinsically induced activation (i 5-i 8). Furthermore, the ac-tivity of these integrin receptors requires divalent cations,consistent with the occurrence of multiple cation-binding

log fluorescence intensity

0z:,0

x z � �CO > u( >0) f� -‘ 0) I�

� ‘-.1 I�3 � ‘.4 I�3 �0) U a� C�I 0)

Fig. 2. Attachment of By-i 73, K562, and NALM-6 cells. Cell adhesion tointact FN and to the 1 05-kDa proteolytic fragment coated at 1 0 pg/mI wascarried out in the presence of either 1 m� Ca2� , 1 m� Mg2� , or 30 �M

Mn2�. In the case of NaIm-6 cells, Mg2� was used at 5 m�i, because at 1mM, the attached cells were below 30%. Attachment to wells coated withdenatured BSA alone is shown for comparison.

1340 a5(3i Integrmn Constitutive Activity and Cation Regulation

a)

E

ci)0

a)>

4�J

a)

131 cc5 cx3 a4 a4137

BV-173

Fig. 1. Cell surface expressionof selected FN integrin receptorsas assessed by flow cytometryusing the following mAbs: 4B4

K 562 (81); P1B5 (a3); HP2.i (a4); JBS5(a5); Act-i (a4(37) and an unre-lated isotype-matched control(open histogram). Histograms in-dicate log fluorescence intensity(X axis) versus relative cell number

(V axis).

. NaIm6

sites in the extracelluhar portion of their a subunits. Some

general rules for the sensitivity of integrins to divalent cationsand the intimate relationships between higand- and cation-binding have been proposed, but the individual requirementsvary among different integrins and different integrin-ligandinteractions (19-23).

In a previous study, we have described the capability ofseveral human neophastic B cells to differentially recognizevarious ECM molecules including FN (24). Here, we havefurther explored the molecular mechanisms underlying theinteraction of one of these leukemic cells, BV-1 73, with FNwhen compared to other well-studied cell lines such as K562and NaIm-6. The adhesion of these cells to FN was totallydependent on the activity of the a5f31 integrin. In addition, theuse of Ca2�, Mg2�, anti-�1 subunit blocking antibodies andthe behavior of By-i 73 cells under these conditions sug-gested that cations might act as allostenc activators of this

integrin. By disclosing novel traits of the composite molec-

ular mechanisms governing the interaction of leukocyte a5f31

with the central cell-binding domain of FN, the present find-ings imply that tissue dissemination of heukemic cells in vivo

may be affected by a fine-tuned cation regulation of a5131activity.

Results

Constitutive Leukocyte Adhesion to the ROD-containing

and Synergy Regions of the Central Cell-binding Domainof FN. As a first step, we examined the expression of knownEN-binding integrins on the cell surface of leukemic cell lines.

As shown by flow cytometry, BV-1 73, K562, and NaIm-6

expressed high levels of f31 and ca5 chains (Fig. 1) and ex-

pressed intermediate levels of the a4 chain and of the a4f37integrin, whereas the cr3 chain was present almost exclu-sivehy on Nahm-6 cells. Next, the constitutive cell adhesion toEN was investigated; whereas no cell attachment to intact ENwas observed in the absence of cations (data not shown), ahigh degree of BV-1 73 cell adhesion to EN was noted in the

presence of physiological extracellular concentrations ofMn2� and Mg2t A somewhat lower percentage of BV-1 73cells bound in the presence of Ca2�, whereas K562 andNalm-6 cells responded well to both Mg2� and Mn2� butfailed to adhere to intact EN in the presence of up to 5 m�iCa2� (Fig. 2). Cell attachment to the chymotryptic 105-kDaand recombinant 1 10-kDa fragments, comprising the entirecell-binding domain, was equivalent to that seen on intactEN. None of the cell lines bound to a 38-kDa fragment that

included the CS-i cell attachment site, despite the fact that

both BV-1 73 and NaIm-6 cells express a4/31 integrin.Previous reports have identified a region within the central

cell-binding domain of EN, located upstream of the majorRGD recognition sequence and thought to be essential for

optimal attachment, spreading, and migration of fibroblasticcells (14). To determine whether leukemic BV-i73 cells

CEU-BINDING DOMAINL__.______

Cell adhesionsites

Antibodies

333 HFN-7 IST-6 333 HFN-7 1ST-B

Cell Growth & Differentiation 1341

Fig. 3. Perturbation of cell attach-ment by anti-FN mAbs. Top panel,schematic diagram illustrating thecentral cell-binding domain of FNand the approximate epitope local-ization of the antibodies used in thisstudy. The major cell adhesion sitesalso are indicated. Boxed numberscorrespond to Fn3 modules (53).Bottom panel, wells coated with theii0-kDa recombinant fragment (10pg/mI) were preincubated with theindicated antibodies for 1 h at 37#{176}C,and cells were allowed to adhere at37#{176}C(A) or 4#{176}C(C) in the presence of1 mM Ca2� or 1-3 mM Mg2� . Alter-natively, antibodies were added tocells that had been allowed to attachto the substrate for 30 mm at 37#{176}C(B)or at 4#{176}C(D), and then the extent ofresidual cell attachment was as-sessed. In experiments reported in Band D, antibody IST-6 was not ap-plied. Control represents the numberof cells remaining attached in the ab-sence of added antibodies.

8

0z

would similarly use this region in cooperation with the RGDsite, cells were allowed to adhere to the i i 0-kDa fragment

preincubated with antibodies directed against differentepitopes present in the 3En modules contained by this frag-

ment (Fig. 3, top panel). Preincubation of the 1 i 0-kDa frag-ment with antibody 333 (binding near the RGD site; Ref. i 4)caused complete inhibition of cell adhesion (Fig. 3A, bottom

panel). In contrast, two other antibodies, IST-6 (Ref. 25; Fig.3A, bottom panel) and i 937 (data not shown), which bothbind at the boundary between the 3En-7 and 3Fn-8 module,had no effect. Antibody HFN-7, recognizing an epitope onthe 3En-9 modules, greatly reduced attachment of K562,Nahm-6, and BV-i 73 cells to the i i 0-kDa fragment. How-ever, in the case of BV-i 73 cells, this effect was nearlycomplete only when cells were allowed to bind in the pres-ence of Ca2� and was much less pronounced when cellswere allowed to attach in the presence of Mg2�. Next, weexamined the ability of antibodies 333 and HEN-7 to detach

cells firmly bound to the ii0-kDa substrate. Only antibody333 was able to detach most adherent cells in all cell lines,whereas HFN-7 detached about 60% of the K562 cells andonly a minor fraction (20-30%) of Nalm-6 and By-i 73 cells(Eig. 3B, bottom panel).

To distinguish between effects on initial versus firm attach-

ment involving cytoskehetal reorganization and to determine

whether engagement of the a5f31 integrin induced recogni-

tion of the substrate via mechanisms involving lateral cell

membrane mobility, cell adhesion assays were performed at4#{176}C.BV-i 73 cells displayed minimal adhesion to the 1 i 0-

kDa fragment at 4#{176}Cand could not be investigated under

these conditions. Conversely, when K562 and NaIm-6 cellattachment was examined after preincubation of the sub-

strate with anti-EN mAbs, there was a complete inhibition of

cell binding (Fig. 3C, bottom panel). However, whereas both

cell lines were detached by antibody 333, antibody HFN-7was only effective with K562 cells and was only marginally

active with Nalm-6 cells that remained bound to 75% (Fig.3D, bottom panel). Given the fact that increasing the amountof antibodies added did not yield significantly different de-

tachment levels, it seems reasonable to conclude that thecomposite behavior of the various cell lines after the additionof anti-EN mAbs cannot be related to a specific property ofthe antibody-higand interaction, but rather to an intrinsic traitof the different heukemic cell types.

Identification of the a5131Integrin as the Receptor-mediating Leukocyte Attachment to FN. Consistent with

the high cell surface expression of a5�1 (Fig. 1) and in spiteof the concurrent expression of a4�1 � the constitutive attach-

ment of By-i 73, K562, and NaIm-6 cells to FN was entirelymediated by the a5�1 integrin, because antibodies against

= �03 � �U5 anti-a�

-J0

I-z0U

0z:D0

(I,-J-JUiU

1342 a5(3i Integrmn Constitutive Activity and Cation Regulation

1 2 34 1 234

116-$dS�$�

I NON-REDUCED I I REDUCED

Fig. 5. Immunoprecipitation and affinity chromatography of “5131 inte-grin. About 2.0 x 1 o� BV-i 73 cells (Lanes 1 and 3) and 1 .0 x 1 0� K562cells (Lanes 2 and 4) were radiolabeled with 1251 extracted, and thenimmunoprecipitated using an antiserum to the cytoplasmic portion of themntegrmn a5 subunit (Lanes 3 and 4). Aliquots of the lysates were incubatedbatchwise with Sepharose-i 1 0-kDa FN recombinant fragment, andbound proteins were eluted with 25 m�i EDTA and analyzed by a 7.5%SOS-PAGE (Lanes 1 and 2). The migration of the (3-galactosidase moles-ular weight marker (1 16) is shown on the left.

Fig. 4. Inhibition of cell attachment to FN by anti-integrin antibodies. Thefollowing antibodies were added at 5 �.tg/ml just before cell plating ontoFN-coated wells (10 �g/rnI): 4B4 (anti-f31), P1 B5 (anti-a3), HP2.1 (anti-a’l),P106 (anti-cr5), JBS5 (anti-cx5f3,), Act-i (anti-a4(37), and B2121 (anti-f33).Control represents cell attachment in the absence of antibodies.

either the cr5 subunit or the a5�3, complex completely

blocked cell attachment (Fig. 4). In contrast, antibodies

against other putative EN-binding integrins, including cr3, a4,

av, and a4137 (Fig. 4 and data not shown), were ineffective.

These data indicated that the other EN-binding integrins of

BV-i 73 and NaIm-6 cells were not expressed in a constitu-tivehy active state such as to significantly contribute to the EN

binding of these cells.

To verify by a different approach that the a5�31 integrin wasthe only receptor complex involved in the interaction of the

cells with EN, detergent extracts of 1251 surface-labeled BV-

i 73 and K562 cells were subjected to affinity chromatogra-

phy on the 1 i 0-kDa recombinant fragment coupled to CNBr-

activated Sepharose. Bound proteins were ehuted with 25 m�EDTA and analyzed by SDS-PAGE under nonreducing andreducing conditions. Two major proteins showing apparentmolecular mass of i20-150 kDa under nonreducing condi-

tions and about 1 30 kDa under reducing conditions bound to

the columns (Eig. 5). In parallel experiments, ahiquots of the

extracts were immunoprecipitated with a pohychonal anti-serum specific for the a5 subunit, and the proteins identified

were found to match the proteins bound to the affinity col-umns in ehectrophoretic mobihities (Fig. 5). Furthermore, no

proteins bound to the affinity columns from anti-a5 antibody-

depleted cell extracts (data not shown).Distinct Effects of Ca2’ and Mg2’ on the Constitutive

Binding Activity of the Leukocyte a5131hntegnn. To further

define the role of cations in the a5j31 mntegrin-mediated at-tachment of the heukemic cells examined here, we initially

compared the dose-dependent ability of the different cationsto support attachment. Because Mn2� has been shown togreatly increase the affinity of several integrins, including

a5�1 (i 9-2i), for their ligands, it was adopted as a reference

cation. As expected, Mn2� supported 50% of the maximal

cell attachment at a significantly lower concentration (6 �M)

than both Mg2� and Ca2�, and this concentration was

equally effective for all three cell lines (Fig. 6). In contrast, the

Mg2� concentrations required to yield 50% of the maximal

adhesion varied widely, ranging from 45 �M for BV-i 73 cells

to 1 .4 mM for NaIm-6. Ca2� only promoted By-i 73 cell

attachment at concentrations similar to those of Mg2� (Fig.

6). However, when the adhesion response was examined in

the presence of optimal ion concentrations but with decreas-

ing amounts of the substrate molecule, Ca2� was found to be

less effective than Mg2t In fact, in the presence of Ca2�,attachment of BV-i 73 cells to the i 05- and i 1 0-kDa EN

fragments required an 8-fold higher higand concentration

than in the presence of Mg2� (data not shown). However, if

the f3i subunit was exposed to the T52/16 activating anti-

body, comparable levels of cell attachment were observedfor all three cell lines, even at cation concentrations well

below the amounts needed to yield detectable cell adhesionand irrespective of the ion present (data not shown). Thisfinding suggests that the differences detected in the consti-

tutive activity of the a5f�1 integrins of the three cell lines could

be overcome by the modulation of the integrin conformationinduced by the artificial activation through the TS2/i 6 anti-

body.The centrifugal cell adhesion assay used in this study is

unique in that it permits the estimation of the relative avidity

to a given substrate by varying the detachment force applied.

Therefore, we next examined the avidity profile of cell attach-ment to the i i 0-kDa EN fragment by varying the centrifugal

forces applied to dislodge the adherent cells (Fig. 7). The

force necessary to detach 50% of the cells was about 50 x

g for BV-i73 cells bound in the presence of 1 mr�i400 x g for Nahm-6, 600 x g for K562, and >700 x g for

BV-i 73 bound in the presence of i m� Mg2� . To note, a

force of 700 x g was the maximal one applicable to retainfully viable cells. Taken together, the ion/higand dose re-

sponses and the differential detachment profiles indicatethat the constitutively active a5j31 integrins of the three Ieu-

kemic cell lines confer to the cells the capability to recognize

#{149}���.BV173/Ca2+ .-. BV173/c02+

0-0 BV173/Mg2+

v�...v K562/M92+

0� . . . � N01m6/Mg2+

80

0z0

-JUi0

Fig. 6. Effect of divalent cations on cell attachment to the central cell-binding domain of FN. Cells were assayed for adhesion to the 1 iO-kDarecombinant fragment coated at 10 �g/ml in the presence of differentcations.

ION CONCENTRA11ON (mM)

100 10005 10

CENTRIFUGAL FORCE (xg)

Fig. 7. Assessment ofthe relative avidity of cell attachment to the centralcell-binding domain of FN by varying the centrifugal force applied todislodge bound cells. Cells were allowed to adhere to the 1 iO-kDa re-combinant fragment coated at 3 pg/mI in the presence of 1 m� Ca2� or1-3 m� Mg2�. Control represents the number of cells remaining attachedto the substrate at the standard centrifugal force of 13 x g.

Cell Growth & Differentiation 1343

EN with a range of relative avidities that decreases in the

order BV-i 73/Mg2’ > K562/Mg2� > NaIm-6/Mg2� > BV-

i 73/Ca2’.

Cation Modulation of the �J1 Subunit Activity. BecauseBy-i 73 cells were the only cells attaching to EN substrates

in the presence of either Mg2� or Ca2�, we examined

whether their cation-regulated attachment was differentially

affected by antibodies against the �3i and a5 subunits. Asdescribed above, antibodies against the a5�1 complex oragainst the a5 subunit completely abolished cell attachmentin the presence of either cation (Fig. 4). On the other hand,whereas the anti-f3i antibody 4B4 was able to completelyprevent the attachment of K562 and Nahm-6 cells in thepresence of Mg2� and the attachment of BV-i 73 cells in thepresence of Ca2� , it reduced the attachment of BV-173 cellsin the presence of Mg2� by <45% (Fig. 4). This finding raised

the possibility that the differential inhibition of BV-i 73 celladhesion by 4B4 in the presence of Ca2� and Mg2� could bea unique property of this antibody. Therefore, we used twoother function-blocking anti-f3i chain mAbs, P4Ci 0 andAIIB2, whose epitopes are distinct but largely overlappingwith that of the 4B4 (26). Both mAbs P4C1 0 and AIIB2inhibited BV-i 73 cell attachment to the 1 i 0-kDa fragment inthe presence of Mg2� in a fashion that was comparable tothat of mAb 4B4 (45 and 52%, respectively). Thus, we nextevaluated whether the insufficient blocking activity of anti-body 4B4 in the presence of Mg2� was due to a loweramount of antibody bound to the cells. This was examined byincubating BV-i73 cells with 1251-labeled 4B4 in the pres-ence of either Mg2� or Ca2�. Both the number of lgG mol-ecules bound to the BV-i73 cell surface and the dissociationkinetics of the bound lgGs were identical, irrespective of theion present. The ability of mAb 4B4 to inhibit attachment inthe presence of Mg2� was not a mere quantitative effect,because even a 5-fold increase of the amount of IgG added(when compared to that yielding optimal inhibition in the

-J

0 100

I

50

40

20

presence of Ca2�), still reduced BV-i 73 cell adhesion to only43% in the presence of Mg2�.

If the efficacy of mAb 4B4 would only depend on whetherthe a5�1 integrin is more active, such as in BV-i73 cells in thepresence of Mg2�, or less active, such as in BV-i 73 cells inthe presence of Ca2�, the extent of inhibition of cell attach-ment would be determined by the cation present during thecell-substratum interaction. Alternatively, if the presence of

Mg2� or Ca2� during the preincubation step qualitativelymodulated the intrinsic activation state of a5f31 , and this

activation state was locked by the binding of mAb 4B4 to thef3i integrin subunit, inhibition of cell adhesion would be dif-

ferently affected, depending on the cation present when theantibody is added. To distinguish between these possibili-ties, BV-173 cells were preincubated with antibody 4B4 inthe presence of either Ca2� or Mg2�, followed by removal ofthe cation by washing with EDTA, and then allowed to attachto the substrate in the presence of the same cation or a

heterologous cation. When cells were allowed to interact

with the higand in the presence of Mg2�, the 4B4 antibodyinhibited cell attachment to about 50%, irrespective ofwhether the cells were preincubated with the antibody in thepresence of Mg2� or Ca2� (Fig. 8). On the other hand, whenthe cells were allowed to attach in the presence of Ca2�, the4B4 antibody yielded complete blockage of cell adhesiononly if the preincubation step was carried out in the presenceof Ca2�. These results suggest that the presence of Mg2� or

Ca2� during the preincubation step qualitatively modulated

the integrin conformation such that the effects of mAb 4B4binding would be to hock the intrinsic activation state of a5f31.

To further analyze whether reorganization of the cytoskel-

eton conditioned the differential cation response, cells were

treated with cytochalasin D and then allowed to attach to theii0-kDa fragment. The finding that both treated and controlcells had similar levels of adhesion with homologous cations

suggested that the distinct cation sensitivity/response is un-

EJ Pre-PBS EXJ Pre-Ca2� Pre_Mg2+

-J

Ca2+ Mg2+

Fig. 8. Effect of preincubation with Ca2� and Mg2� on the inhibition ofBV-173 cell attachment to FN by the anti-(3i antibody 4B4. Cells wereincubated for 45 mm with antibody 4B4 (5 �g/m� at 37#{176}Cin either PBS(Pro-PBS), 1 mM Ca2� (Pre-Ca�), or 1 m� Mg2� (Pre-Mg��); washed with1 mM EDTA; and finally plated onto FN-coated wells (10 �g/m� in thepresence of the homologous or heterologous cation. No cell attachmentwas detected if cations were omitted after the EDTA treatment. Controlrepresents the number of cells remaining attached after a preincubation inPBS with either cation and assayed in the absence of antibody 4B4.

1344 a5f3i Integrin Constitutive Activity and Cation Regulation

120

100

80

::200

BV173

� -I-

- �L I

likely to be a result of postreceptor events involving remod-eling of cytoskehetal components.

Cell Surface Distribution of a5131. The extent of receptorclustering is currently considered to represent a pivotal fac-tor in the regulation of the functional activity of integrins. Toinvestigate whether the distinct substrate avidities intrinsi-cally displayed by the leukemic cell lines might be explainedby a different cell surface organization of the a5�31 integrin,BV-i 73 and K562 cells were stained at 0#{176}Cwith an antibodyagainst the a5 subunit and examined by confocal fluores-cence microscopy. These experiments revealed that bothcell lines presented equivalent levels of reactivity and a sim-ihar spotty membrane immunofluorescence (Fig. 9). Thus,

within the resolution limits of this approach, the extent of

integrin prechustering did not seem to be a determining pa-rameter to take into account to explain the differences ob-served in the adhesion behavior of the cells.

DiscussionIn this investigation, we have analyzed the mechanisms in-volved in the constitutive a5�3,-mediated attachment of Ieu-kemic cells to EN using a unique centrifugation-based celladhesion assay that permits the precise assessment of thenumber of cells bound and allows for the quantitative deter-mination of the relative substrate binding avidity of the at-tached cells (28, 29). Leukemic cells used the major RGD-containing cell-binding region in the 3Fn-iO repeat and acell-binding site(s) located within the 3Fn-9 repeat for opti-mal Ca2�- and Mg2�-mediated attachment to EN throughthe a5f31 integrin. There is now compelling evidence also inthe case of hemopoietic cells indicating that the a5f31-medi-ated cell adhesion to EN may be mediated by a concurringparticipation of the RGD site and additional synergistic site(s)within the central cell-binding domain (30). Nuclear magnetic

resonance studies using a recombinant EN polypeptide in-

dicate that the two regions are exposed on the same face

and are easily accessible for a single a5f31 mntegrin complex

(3i), which might thus bind to the synergy region through the

a5 subunit and to the RGD site through the pi subunit (32).

In fact, as also indicated from the present study, antibodies

against epitopes mapping near the RGD site within the cell-

binding region and antibodies against epitopes within the

synergistic region fully prevented cell adhesion, suggesting

that both regions are necessary for the initial cell attachment

to EN via a5�1.

The a5f31 integrin may be expressed in different activation

states, resulting in increased higand-binding affinity and/or

avidity, and these different states may be under the influence

of the physiological status of the cells (33) or of a cell type-

specific regulation (34). For instance, it has been found thatK562 cells probed with soluble EN bound the higand with low

affinity, which, however, could be increased 20-fold by pre-

treatment with the activating anti-ni 8A2 mAb (35). Con-versehy, nonactivated anchorage-dependent Chinese ham-

ster ovary cells growing on a EN-containing ECM showed aa5�1-dependent higand affinity intermediate between the two

values observed with K562 cells. The centrifugation assayused herein also provides the capability of discriminatingslight differences in the strength/avidity of cell-substrate ad-hesion. Therefore, although in a distinct experimental setting,the wide range of constitutive avidity states of the a5f31

integrin in different hemopoietic cells detected here with

immobilized EN concurs with the previous findings obtainedwhen using soluble EN. In addition, the fact that maximalBV-i73 cell attachment was largely independent of the cat-ion present during the cell-substrate interaction suggests

that recognition of the two distinct EN cell-binding regions

within the central cell-binding domain occurs irrespectively

of the intrinsic cell binding avidity. This observation is onlyapparently at variance with the conclusion reached by Danenet al. (30), who demonstrated that the dependence on the

synergy site for binding to the RGD site was affected by the

activation state of a5�1of K562 cells. In fact, in this hatterstudy and with the standard cell adhesion assay used, un-activated K562 cells barely attached to EN, and a very strongactivation of a5�1 by a combined treatment with Mn2�, TS2/16, and phorbol i 2-myristate i 3-acetate was necessary toovercome the need of the synergy site, whereas synergysite-dependent adhesion was promoted by a treatment witheach individual stimuhatory agent.

A number of physiologically relevant regulatory mecha-nisms affect integrin function, and the status of their consti-tutive activity may be critically determined by divahent cations(i 9-23). We find that K562 and Nahm-6 cells fail to attach to

EN in the presence of Ca2t whereas BV-i 73 cells display asignificant constitutive adhesion in the presence of physio-

logical concentrations of Ca2t Results obtained elsewherewith another hematopoietic cell line, HL-60, have shown that

the a5j31-mediated adhesion to EN is inhibited by Ca2� (36).This different behavior might be correlated with the kinase

activity of the Bcr/AbI oncogene, which has recently been

reported to stimulate integrin function (37); although bothBV-i73 and K562 have the hybrid gene, the tyrosine kinaseactivity might differ between the two cell lines. Therefore, at

(;��- � �

-4�� �. .#{176}� �4�’I’�� #{182}. �

4�_,. ,�

)

Cell Growth & Differentiation 1345

Fig. 9. Cell surface distributionof a5(31 . Confocal fluorescencemicroscopy images of By-i 73(left) and K562 (right) cells incu-bated with the P1 D6 anti-aS sub-unit antibody.

least three functional patterns, exemplified by BV-1 73, K562/

Nalm-6, and HL-60 cells, are defined when a5131 integrins of

leukemic cells interact with EN in the presence of Ca2�.

Several explanations may be put forward to account for why,

at relatively high functional levels (in the presence of 1 mr�i

Mg2 � ), only about 50% inhibition, irrespective of the amount of

4B4 blocking mAb added, could be reached dunng the subse-

quent BV-i 73 cell-substrate interaction when the lower func-

tion-inducing Ca2� cation was present: (a) the a5f31 integrin of

BV-i 73 cells, but not that of K562 and Nalm-6 cells, is ex-

pressed on the cell surface in a preclustered form (conferring

higher constitutive ligand avidity); (b) the a5f31 integrin of BV-i 73

is posttranslationally modified in a different way compared to

that of K562 and Nalm-6; (c) different cytoskeletal-related pos-

treceptor occupancy events are involved; (a) more integrin mol-

ecules are also activated in BV-1 73 cells at suboptimal condi-

tions (Ca2 � ); and (e) the BV-i 73 membrane environment

renders the a5�1 integrin of BV-i73 cells more susceptible to

the presence of cations compared to the other cell lines. The

first possibility seems unlikely, because the apparent cell sur-

face distribution of a5�1 in BV-i 73 and K562 was similar when

examined by indirect immunofluorescence in conjunction with

confocal laser microscopy. Although it is not known whether

the addition ofGAG chains (38) may affect the functional activity

of a5131 , the gel migration pattern of the BV-i 73 a5�1 integrin

immunoprecipitated and affinity-purified onto EN columns did

not differ in terms of glycosylation/glycanation from that of

K562. The third possibility is unlikely in light of the finding that

cytochalasin D did not affect BV-i 73 cell attachment in the

presence of either Ca2� or Mg2� cations. Instead, the possi-

bility that BV-i 73 cells possess a greater pool of activable a5131

integrin receptors, which might represent a stable phenotypic

trait (34), or that this integrin has a higher cation sensitivity are

plausible hypotheses and deserve further investigation.

This study highlights an important role for the pi integrin

subunit in regulating FN binding in BV-i 73 cells. It is known that

some activating antibodies increase receptor affinity, probably

by changing the conformation of this subunit (39), whereas

other antibodies function by affecting postreceptor events (40-

41). Separate classes of cation binding sites have been re-

ported for the f33 (42) and the �5J�1 integnn (4i), and cation

binding is currently thought to be associated with the induction

and/or maintenance of a favorable integnn conformation. At

physiological concentrations, Ca2 � is known to exert opposite

effects on the binding activity of two related integrins, a�f31 and

avP3(20), suggesting that a substantial contribution in the cat-

ion regulation of the function and the way integrins are orga-

nized at the cell surface (43) is provided by the �3 subunits.

Recently, a structural explanation for the involvement of 13sub-

units was provided for the �33 chain for which the formation of

a ternary complex between ligand, cation, and a highly con-

served region of the �33 subunit has been postulated (45). In

addition, some inhibitory anti-f3i subunit antibodies were do-

scnbed that might inhibit higand binding by stabilizing the higand

unoccupied state of the integrin, with the higand acting as an

allosteric inhibitor of mAb binding (46). By analogy, the partial

inhibition of attachment of BV-173 cells preincubated with 4B4

in the presence of Mg2� and assayed for cell adhesion in the

presence of Ca2� suggests that the binding of 4B4 to the f31

subunit could lock the higher activation state reached under

optimal cation conditions (i.e., in the presence of Mg2�j. Thus,

the cation bound to the �3i subunit could act as an allostenc

activator ofthe �3i subunit and/or ofthe whole integnn function.

The implication of the distinct divalent cation requirements

for adhesion of leukemic cell lines to EN is that physiological

stimuli (cytokines, growth factors, and so forth) activate cells

to different levels; therefore, changes in divalent cation con-

centrations could modulate the strength of higand interaction

so as to switch the cells from a sessile to a migratory con-

dition. This mechanism is also relevant in light of the recent

evidence suggesting a specific involvement of the c�5�1 in-

tegrin in the growth control of leukocytes through interaction

with EN (47, 48).

Materials and MethodsCell Lines and Reagents. K562 and BV-173 are Ph1-positive cell linesderived from patients during a myeloid (K562) or lymphoid (By-i 73) blast

crisis phase of chronic myelogenous leukemia. Nalm-6 is an immatureleukemic B-cell line. FN was purified from human plasma, and the 105-kDa chymotryptic fragment comprising the entire cell-binding domain of

FN was kindly provided by Dr. Staffan Johansson (Biomedical Center,

University of Uppsala, Uppsala, Sweden). A bacterial recombinant proteinof 1 10 kDa corresponding to the entire cell-binding domain of FN was

1346 a5(3i Integrin Constitutive Activity and Cation Regulation

purified as described previously (49). mAbs against various mntegrin sub-units were obtained as follows: 484 (Ortho Diagnostic Systems, Raritan,NJ; anti-(3i); Act-i (Dr. S. Shaw, National Cancerlnstitute, NIH, Bethesda,MO; anti-a4(3�); P4C1O (anti-(31), P1 B5 (anti-a3), HP2/1 (anti-a4), P106

(anti-aS), JBS5 (anti-a5(31), and B2i2i (anti-(33; all from Chemicon Inter-

national, Inc., Temecula, CA); L230 (anti-av; Dr. Robert Pytela, The LungBiology Center, Department of Medicine, University of Califomia, San

Francisco, CA); AIIB2 (anti-(3i ; Dr. Caroline Damsky, Department of Sto-matology and Anatomy, University of Califomia, San Francisco, CA); and

TS2/16(anti-(3i ; Dr. Francisco Sanchez-Madrid, Servicio deimmunologia,

Hospital de Ia Princess, Madrid, Spain). mAbs against selected epitopeswithin the cell-binding domain of FN were: 1937 (Chemicon); 333 (Dr. K.Vamada, National Institute of Dental Research, NIH, Bethesda, MD); HFN-

7.1 (American Tissue Type Culture Collection); and IST-6 (Dr. LucianoZardi, Laboratory of Cell Biology, 1ST Genova, Italy). A polyclonal anti-

serum to a synthetic peptide from the integrin subunit a5 cytoplasmicportion was kindly obtained from Dr. Guido Tarone (Department of Ge-

netics, Biology and Chemistry, University of Torino, Tonno, Italy).IndIrect Immunofluorescence. Isotype-matched nonimmune murine

IgGs were used to determine unspecific binding. Antibody-labeled cells

were analyzed with laser confocal microscopy (Diaphot 200 Nikon) usinga x60 phase/fluorescence objective or with a FACSCan flow cytometer,and the data were processed with the Lasersharp MRC-1024 (Bio-Rad

Laboratories, Hercules, CA) or with the Lysis II program (Becton Dickin-

son, Mountain View, CA), respectively.Cell Adhesion Assay. The detailed procedure for the centrifugal ad-

hesion assay used in this study has been described previously(24, 28, 29),and only minor modifications were introduced here. In the present series

of cell adhesion experiments, cation concentrations used were roughlyequivalent to the range of physiological extracellular concentrations for

the different cations (50, 51): Mn2�, 10-30 �i,i; Mg2�, 0.3-1.5 mM; and

Ca2�, 0.5-i .5 mM. r5S]Methionine-labeled cells were washed in Ca2�-

and Mg2�-free PBS followed by 50 m�i Tris-HCI (pH 7.2)and 150 m� NaCI(Tris-buffered saline) containing 1 m� EDTA to remove all extracellularcations and then aliquoted into microplate wells (Falcon Flexiplate III;Falcon, Becton Dickinson; at about i0� cells/well) coated with the sub-

strate to be tested. Assays were repeated at least three times, and theaverage values or values from a representative case that did not deviate>1 0% from the mean of the three are reported. Removal of unbound cellsby centrlfugation at a force of 1 3 x g was determined to be adequate to

yield reproducible binding and negligible (2-8%) nonspecific attachmentto denatured BSA used as a saturating agent. A series of time course testsindicated that up to a 20-30-mm incubation of the cells in contact with the

substrate after centrifugation was sufficient to yield maximal levels of

stable cell adhesion. In experiments aimed at examining the effects of

anti-integnn antibodies, various antibodies were added directly to thewells just before plating the cells. In contrast, to analyze the effects ofdivalent cations on the inhibitory activity of anti-f3i subunit antibodies,

cells were incubated for 45-60 mm at 37#{176}Cin PBS in the presence of the

optimal concentration of antibody and 1 m� Ca2� or Mg2� and thenwashed once with PBS and 1 mp�i EDTA and resuspended in PBS con-taming the heterologous cation, i.e., cells preincubated with 4B4/Ca2�were resuspended in Mg2� and vice versa. In other instances, antibodiesagainst different regions of the cell-binding domain of FN were used toprevent the interaction of the cells with the substrate or to detach already

adherent cells. In experiments aimed at determining whether postreceptorevents affected the differential cation response of the leukemic cell lines,

cells were pretreated with the actmn-disrupting drug cytochalasmn 0 (20�g/ml for 30 mm) and then used for cell adhesion assays.

Radlolabellng of Cell Surface Proteins and Immunoprecipitatlon.Cells were surface-labeled with 1251 according to the following protocol.

Lactoperoxidase (100 units/mI; Sigma), 1 mCVmI 1251 (Amersham), and 10�Iof a 0.06% solution of hydrogen peroxide were added to about iO� cellsevery 10 mm three times. Unbound 1251 was removed by sequentialwashing with PBS, and the cells were lysed in extraction buffer composedof 50 m� Tris-HCI (pH 7.6), 0.15 M NaCI, 0.1 M octyl-(3-o-glucoside (Cal-

biochem), 1 mp,i CaCI2, 1 mp�i MgCI2 and protease inhibitors, 2 mp,i phen-ylmethylsulfonyl fluoride, 5 m�,i N-ethylmaleimide, and 1 m� p-aminoben-

zamldine (Sigma). Lysates were cleared at 15,000 rpm for 20 mm andprecleared with normal rabbit serum and Sepharose-protein A resin (Phar-

macia). The antiserum to the aS integrin chain (52) was then added alongwith Sepharose-proteln A, and the mixture was incubated ovemight at 4#{176}C

under gentle shaking. The immunocomplexes were centrifuged, and thebeads were washed extensively before being removed from the Sepha-rose-protein A resin by boiling for 5 mm in 2x sample buffer [0.125 M

Tns-HCI (pH 6.8), 1% glycerol, and 2% SOS]. Samples were analyzed bySOS-PAGE on 7.5% polyacrylamide gels, and the dried gels were auto-

radiographed on MP films (Amersham).Affinity Chromatography. The recombinant 1 10-kDa fragment was cou-

pled to CNBr-activated Sepharose (Pharmacia) according to the manu-facturer’s instructions. Cleared detergent extracts of 1251 surface-labeledcells were incubated with the 1 1 0-kDa Sepharose slurry under constantmixing overnight at 4#{176}C.The beads were then packed into columns thatwere subsequently washed with extraction buffer with 25 m� octyl-(3-D-glucoside and were finally eluted with 25 m� EDTA. Eluted peak fractions

were concentrated and analyzed by SOS-PAGE as described above.

Cell Surface Binding of Antibodies. Binding of 1251-Iabeled anti-(31mAb 4B4 to cells was measured using trace amounts (10 ng) of 4B4 labeled

according to the chlorarnine-T procedure. A typical binding assay was in a

so-it1 volume composed of 25 �sI of cells (2 x i0� cells/assay) and radiola-beled mAb with or without an excess (1000-fold) of unlabeled antibody inPBS. Aftera3o-min incubation at 37#{176}C,thecells werewashed twice, and thecell pellet with bound antibody was counted in a gamma counter.

AcknowledgmentsWe thank Ors. Caroline Oamsky, Staffan Johansson, Robert Pytela, Fran-

cisco Sanchez-Madrid, Stephen Shaw, Guido Tarone, Kenneth Vamada,and Luciano Zardi for providing us with antibodies and FN fragments. Thetechnical assistance of Maria Teresa Mucignat and Paola Spessotto isgratefully acknowledged.

References1 . Springer, T. A. Adhesion receptors of the immune system. Nature(Lond.), 346: 425-434, 1990.

2. Hemler, M. E. VLA proteins in the integrin family: structures, functions,and their role on leukocytes. Annu. Rev. Immunol., 8: 365-400, 1990.

3. Hynes, R. 0. Integrins: versatility, modulation, and signaling in cell

adhesion. Cell, 69: 11-25, 1992.

4. Shlmizu, V., and Shaw, S. Lymphocyte adhesion mediated by VIA ((31)mntegnns. Functional roles of multiple extracellular matrix and cell-surfaceligands. Chem. Immunol., 50: 34-54, 1991.

5. Fawcett, J., and Harris, A. L Cell adhesion molecules and cancer. Curr.Opin. Oncol., 4: 142-148, 1992.

6. Pierschbacher, M. 0., and Ruoslahti, E. The cell attachment activity offibronectin can be duplicated by small synthetic fragments of the mole-cules. Nature (Lond.), 309: 30-33, 1984.

7. Bemardi, P., Patel, V. P., and Lodish, H. Lymphoid precursor cells adhere

to two different sites on fibronectin. J. CeO Biol., 105: 489-498, 1987.

8. Humphries, M. J., Komoriya, A., Akiyama, S. K., Olden, K., andYamada, K. M. Identification of two distinct regions of the type Ill con-nesting segment of human plasma fibronectin that promote cell type

specific adhesion. J. Biol. Chem., 262: 6886-6892, 1987.

9. Pytela, R., Pierschbacher, M. 0., and Ruoslahti, E. Identification andisolation of a 1 40 kO cell surface glycoprotemn with properties expected of

a fibronectin receptor. Cell, 40: 191-198, 1985.

10. Elices, M. J., Urry, L A., and Hemler, M. E. Receptor functions for theintegrin VLA-3: fibronectin, collagen, and laminin binding are differentiallyinfluenced by Arg-GIy-Asp peptide and by divalent cations. J. Cell Biol.,112: 169-180, 1991.

11 . Vogel, B. E., Tarone, G., Giancotti, F. G., Gailit, J., and Ruoslahti, E.A novel fibronectin receptor with an unexpected subunit composition

(a,,(3i). J. Biol. Chem., 265: 5934-5937, 1990.

12. Sanchez-Aparicio, P., Dominguez-Jimenez, C., and Garcia-Pardo, A.Activation of the a4(31 integrin through the (31 subunit induces recognitionof the RGDS sequence in fibronectin. J. Cell Biol., 126: 271-279, 1994.

13. Pierschbacher, M. 0., and Ruoslahti, E. Influence of stereochemistry

of the sequence Arg-GIy-Asp-Xaa on binding specificity in cell adhesion.J. Biol. Chem., 262: i7294-i7298, 1987.

Cell Growth & Differentiation 1347

14. Aota, S., Nagai, T., and Vamada, K. M. Characterization of regions of

fibronectin besides the arginine-glycine-aspartic acid sequence required

for adhesive function of the cell-binding domain using site-directed mu-tagenesis. J. Biol. Chem., 266: 15938-15943, 1991.

15. Wilkins, J. A., Stupack, 0., Stewart, S., and Caixia, S. (3i Integrin-

mediated lymphocyte adherence to extracellular matrix is enhanced byphorbol ester treatment. Eur. J. Immunol., 21: 517-522, 1991.

16. Arroyo, A. G., Sanchez Mateos, P., Campanero, M. R., Padura, I.

Dejana, E., and Sanchez-Madrid, F. Regulation of the VIA integrin-ligandinteractions through the (31 subunit. J. Cell Biol., 117: 659-670, 1992.

17. Kovach, N. L., Carlos, T. M., Vee, E., and Harlan, J. M. A monoclonalantibody to (31 mntegrin (C029) stimulates VLA-dependent adherence ofleukocytes to human umbilical vein endothelial cells and matrix compo-

nents. J. Cell Biol., 1 16: 499-509, 1992.

18. van do Wiel van Kemenade, E., van Kooyk, V., do Boor, A. J.,

Huijbens, R. J., Weder, P., van do Kasteele, W., Melief, C. J., and Figdor,

C. G. Adhesion of T and B lymphocytes to extracellular matrix and endo-thelial cells can be regulated through the (3 subunit of VLA. J. Cell Biol.,117: 461-470, 1992.

19. Gailit, J., and Ruoslahti, E. Regulation of the fibronectin receptor

affinity by divalent cations. J. Biol. Chem., 263: 12927-12932, 1988.

20. Kirchhofer, 0., Grzesiak, J., and Pierschbacher, M. 0. Calcium as apotential physiological regulator of mntegrin-mediated cell adhesion.J. Biol. Chem., 266: 4471-4477, i99i.

21 . Masumoto, A., and HomIer, M. E. Multiple activation states of VLA-4.

Mechanistic differences between adhesion to CS1/fibronectin and to vas-cular cell adhesion molecule-i . J. Biol. Chem., 268: 228-234, 1993.

22. Shimizu, V., and Mobley, J. L Distinct divalent cation requirements formntegrin-mediated CD4� T lymphocyte adhesion to ICAM-i , fibronectin,

VCAM-i, and invasin. J. Immunol., 151: 4106-4115, 1993.

23. Smith, J. W., Piotrowicz, R. S., and Mathis, 0. A mechanism for dNalent

cation regulation of (33-intogrins. J. Biol. Chom., 269: 960-967, 1994.

24. Segat, 0., Pucillo, C., Marotta, G., Penis, R., and Colombatti, A.

Differential attachment of human neoplastic B cells to purified extracel-lular matrix molecules. Blood, 83: 1586-1594, 1994.

25. Camemolla, B., Leprini, A., Allemanni, G., Saginati, M., and Zardi, L

The inclusion of the type Ill repeat ED-B in the fibronectin molecule

generates conformational modifications that unmask a cryptic sequence.J. Biol. Chem., 267: 24689-24692, 1992.

26. Takada, V., and Puzon, W. Identification of a regulatory region of

intogrin (31 subunit using activating and inhibiting antibodies. J. Biol.

Chem., 268: 17597-17601, 1993.

27. Miyamoto, S., Teramoto, H., Coso, 0. A., Gutkind, J. S., Burbolo,P. 0., Akiyama, S. K., and Vamada, K. M. Intogrin function: molecularhierarchies of cytoskoletal and signaling molecules. J. Cell Biol., 131:

791-805, 1995.

28. Lotz, M. M., Burdsal, C. A., Erickson, H. P., and McCIay, 0. R. Cell

adhesion to fibronectin and tenascin: quantitative measurements of initialbinding and subsequent strengthening response. J. Cell Biol., 109: 1795-1805, 1989.

29. Pems, R., Kuo, H. J., Glanville, R. W., Leibold, S., and Bronner Fraser,M. Neural crest cell interaction with type VI collagen is mediated by

multiple ccoperative binding sites within triple-helix and globular domains.Exp. Cell Res., 209: 1 03-i 1 7, 1993.

30. Danen, E. H. J., Aota, S., van Kraats, A. A., Vamada, K. M., Ruiter,0. J., and van Muijen, G. N. P. Requirement for the synergy site for cell

adhesion to fibronectin depends on the activation state of integrin a5(31.J. Biol. Chem., 270: 21612-21618, 1995.

31. Leahy, 0. J., Aukhil, I., and Erickson, H. P. 2.0 A crystal structure of

a four-domain segment of human fibronectin encompassing the AGO loopand synergy region. Cell, 84: 155-164, 1996.

32. Obara, M., and Voshizato, K. Possible involvement of the interactionof the aS subunit of a5(31 integrin with the synergistic region of the central

cell-binding domain of fibronectin in cells to fibronectin binding. Exp. CellRes. 216: 273-276, 1995.

33. Phillips, 0. R., Charo, I. F., and Scarborough, R. M. GPIlb-IIla: theresponsive integrin. Cell, 65: 359-362, 1991.

34. O’Tcole, T. E., Katagiri, V., Faull, R. J., Peter, K., Tamura, R.,Quaranta, V., Loftus, J. C., Shattil, S. J., and Ginsberg, M. H. Integrin

cytoplasmic domains mediate inside-out signal transduction. J. Cell Biol.,124: 1047-1059, 1994.

35. Faull, A. J., Kovach, N. L, Harlan, J. M., and Ginsberg, M. H. Affinity

modulation of integrin ��(3i : regulation of the functional response by

soluble fibronoctin. J. Cell Biol., 121: 155-162, 1993.

36� Davis, G. E, and Camatlllo, C. W. Regulation of lntegrin-med�ted my-

eloid cell adhesiOn to fibronectin: influence of disulfide reducing agents,divalont cations and phorbol ester. J. Immunol, 151: 7138-7150, 1993.

37. Bazzoni, G., Carlesso, N., Griffin, J. 0., and HomIer, M. E. Bcr/AbI

expression stimulates integrin function in hematopoietic cell lines. J. CIin.Invest., 98: 521-528, 1996.

38. Veiga, S. S., Elias, M. C. 0. B., Gremski, W., Porcionatto, M. A., daSilva, R., Nader, H. B., and Brentani, R. R. Post-translational modifications

of #{176}5(31integrin by glycosaminoglycan chains. The ��(3i integrin is afacultative proteoglycan. J. Biol. Chem., 272: 12529-12535, 1997.

39. Arroyo, A. G., Garcia-Pardo, A., and Sanchez-Madrid, F. A highaffinity conformational state on VLA intogrin heterodimers induced by an

anti-(31 chain monoclonalantibody. J. Biol. Chem., 268: 9863-9868, i993.

40. Luque, A., Gomez, M., Puzon, W., Takada, V., Sanchoz-Madns, F.,

and Cabanas, C. Activated conformations of very late activation integrinsdetected by a group of antibodies (HUTS) specific for a novel regulatoryregion (355-425) of the common (31 chain. J. Biol. Chem., 271: 1 i067-11075, 1996.

41 . Gomez, M., Luque, A., del Pozo, M. A., Hogg, N., Sanchez-Madrid, F.,and Cabanas, C. Functional relevance during lymphocyte migration and

cellular localization of activated (31 integrins. Eur. J. Immunol., 27: 8-16,

1997.

42. Hu, 0. 0., Barbas, C. F., Ill, and Smith, J. W. An allosteric Ca2�

binding site on the (33 integrins that regulates the dissociation rate for the

RGO ligands. J. Biol. Chem., 271: 21745-21751, 1996.

43. Mould, A. P., Akiyama, S. K., and Humphries, M. J. Regulation ofintegrin a5(31-flbronectin interactions by divalent cations. J. Biol. Chem.,

270: 26270-26277, 1995.

44. Stulver, I., Ruggeri, Z., and Smith, J. W. Oivalent cations regulate the

organization of intogrins a�(3� and a�(3s on the cell surface. J. Cell.Physiol., 168: 521-531, 1996.

45. O’Souza, S. E, Haas, T. A., Piotrowicz, R. S., Byers-Ward, V.,

McGrath, 0. E., Soule, H. A., Ciernewski, C., Plow, E. F., and Smith, J. W.Ligand and cation binding are dual functions of a discrete segment of theintogrin (33 subunit: cation displacement is involved in ligand binding. Cell,

79: 659-667, 1994.

46. Mould, A. P., Akiyama, S. K., and Humphries, M. J. The inhibitoryanti-(3i mntogrin monoclonal antibody 1 3 recognizes an opitope that isattenuated by ligand occupancy. J. Biol. Chem., 271: 20365-20374, 1996.

47. Sugahara, H., Kanakura, V., Furitsu, T., Ishihara, K., Oritani, K., Ikeda,

H., Kitayama, H., Ishikawa, J., Hashimoto, K., Kanayama, V., and Matsu-zawa, V. Induction of programmed cell death in human hematopoietic celllines by fibronectin via its interaction with very late antigen 5. J. Exp. Med.,

179: 1757-i766, 1994.

48. Lundell, B. I., McCarthy, J. B., Kovach, N. L, and Verfaille, C. M.Activation-dependent a5(31 intogrin-mediated adhesion to fibronectin do-

creases proliferation of chronic myologenous leukemia progenitors andK562 cells. Blood, 87: 2450-2458, 1996.

49. Camemolla, B., Balza, E., Siri, A., Zardi, L, Nicotra, M. R., Bigotti, A.,

and Natali, P. G. A tumor-associated fibronectin isoform generated byaltemative splicing of messenger RNA precursors. J. Cell Biol., 108:1139-1148, 1989.

50. Olinger, M. L Disorders of calcium and magnesium metabolism.Emerg. Med. Clin. North Am., 7: 795-822, 1989.

51 . Smith, J. W., P,otrowicz� R. S., and Mathis, 0. A mechanism for dhialent

cation regulation of (33-integrins. J. Biol. Chem., 269: 960-967, 1994.

52. Defihippi, P., Silengo, L, and Tarone, G. 0�(3i intogrin (laminin recep-

tor� is down-regulated by tumor necrosis factor a and interleukin-i (3 inhuman endothelial cells. J. Biol. Chem., 267: 18303-18307, 1992.

53. Bork, P., and Koonin, E V. Protein sequence motifs. Curr. Opin.Struct. Biol., 6: 366-375, 1996.