Proc. Natl. Acad. Sci. USAVol. 75, No. 9, pp. 4408-4412, September 1978Cell Biology

Attachment and spreading of baby hamster kidney cells to collagensubstrata: Effects of cold-insoluble globulin

(fibroblasts/fibronectin/connective tissue)

FREDERICK GRINNELL AND DIANNE MINTERDepartment of Cell Biology, The University of Texas Health Science Center at Dallas, Dallas, Texas 75235

Communicated by Jerome Gross, July 3, 1978

ABSTRACT Studies have been carried out to determine theeffects of cold-insoluble globulin (CIG) on the attachment andspreading of baby hamster kidney cells on various collagensubstrata. Cell attachment to native collagen substrata occurredin the absence of CIG just as fast as attachment to dried collagenor gelatin substrata occurred in the presence of CIG. On theother hand, cell attachment to dried collagen or gelatin wasmarkedly reduced in the absence of CIG. Cell spreading alsooccurred on native collagen in the absence of CIG; however,CIG was absolutely required for cell spreading to occur on driedcollagen or gelatin. Finally, anti-CIG antiserum or lactoperox-idase treatment inhibited cell spreading on CIG-coated sub-strata but not on native collagen substrata. The data are dis-cussed in terms of the interaction of fibroblasts with collagenin situ.

The interaction of fibroblasts with a substratum results in aseries of biochemically discrete steps including contact of thecells to the substratum, formation of initial bonds of attachment,and reorganization of the cell cytoskeleton accompanied byformation of additional bonds of attachment leading to cellspreading (1).With a variety of cell lines including baby hamster kidney

(BHK), HeLa, Chinese hamster ovary (CHO), and L, cellspreading under normal tissue culture conditions requires thatthe substratum surface be coated by a serum factor that hasbeen identified as cold-insoluble globulin (CIG) (2, 3). Theextent of cell spreading is related to the density of the factoradsorbed on the substratum surface (4). On the other hand,several cell strains, including W1-38, MRC-5, and humanconjunctiva cells, have been shown to attach and spread in theabsence of serum or CIG (5-7). Significantly, cell strains gen-erally synthesize and secrete higher levels of the large externaltransformation-sensitive (LETS) protein [cell-surface protein(CSP), fibronectin], which is immunologically related to CIG,than cell lines or transformed cells (8-11). Moreover, inserum-free medium, cells secrete substances beneath themselvesonto the substratum (5, 12-14). These observations have led tothe hypothesis (15) that the ability of fibroblastic cells to spreadin serum-free medium depends upon the extent to which thecells are able to secrete CIG or a CIG-like protein onto thesubstratum. In general, cell strains do not require the additionof exogenous CIG or serum, whereas CIG or serum is requiredfor most cell lines and transformed cells.The adhesion of fibroblasts has also been studied with col-

lagen substrata; however, in all of these studies, the investigatorshave used dried collagen gels prepared at high pH (16-18). Thismethod of preparing collagen gels is quite drastic comparedto usual techniques (19, 20) and might result in the collagenbecoming denatured. Indeed, quantitative binding studies have

shown that CIG has a much higher affinity for denatured col-lagen than for native collagen (21). Therefore, we have inves-tigated the interaction of fibroblasts with native collagen gelsto determine if CIG is required for this interaction to occur.

METHODS AND MATERIALSPreparation of substrata

All substrata were prepared in Falcon 3001 (35-mm) tissueculture dishes. The substrata were prepared by two generalmethods: gelation or adsorption. In the former instance, colla-gen solutions were added to the dishes and then polymerized,thereby forming a gel. In the latter instance, dishes were ex-posed to protein solutions for a short time and then extensivelyrinsed with deionized water. Protein adsorption to the subs-tratum is known to occur almost instantaneously on exposureof the substratum to protein-containing solutions (22-24). Withwhole serum, the adsorbed layer of protein is about 20-50 Athick (5, 25). In general, pure protein solutions result in theformation of a monomolecular layer of adsorbed protein mol-ecules that can be desorbed only under very harsh conditions(26).Dried Collagen Gels. This substratum has been used by

Klebe (16) to demonstrate a serum protein required for cellattachment to collagen. Aliquots (1.0 ml) of a freshly preparedrat tail collagen solution, ca. 2 mg/ml in 0.1% HOAc, weregelled at 220 for 30 min in the presence of an NH3 atmosphere.The final pH was 11. Subsequently, the gels were air dried for48 hrs at 220.

Hydrated, Native Collagen Gels. The technique for pre-paring native collagen gels composed of cross-striated fibrilsand their microscopic appearance were described by Gross andKirk (19) and by Elsdale and Bard (20). Aliquots (1.0 ml) of afreshly prepared rat tail collagen solution, ca. 2 mg/ml in 0.1%HOAc, were brought to physiological ionic strength and pH at40 by the addition of 10 times concentrated phosphate-bufferedsaline and NaOH and then placed at 370 for 30 min in a hu-midifiedchamber. The gels that formed were about 2mm thick.These substrata were then used immediately.

Gelatin-Coated Substrata. Aliquots (1.0 ml) of a freshlyprepared rat tail collagen solution, ca. 2 mg/ml in 0.1% HOAc,were heated to 50° for 10 min. The solutions were then cooledto 370 and incubated in Falcon dishes for 10 min at 22°, fol-lowing which the dishes were extensively rinsed with deionizedH20.

Native Collagen-Coated Substrata. Aliquots (1.0 ml) of afreshly prepared rat tail collagen solution, ca. 2 mg/ml in 0.1%HOAc, were incubated in Falcon dishes for 10 min at 22°,

Abbreviations: CIG, cold-insoluble globulin; ASF, fetal calf adhesionand spreading factor.

4408

The publication costs of this article were defrayed in part by pagecharge payment. This article must therefore be hereby marked "ad-vertisement" in accordance with 18 U. S. C. § 1734 solely to indicatethis fact.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

6, 2

021

Proc. Natl. Acad. Sci. USA 75 (1978) 4409

following which the dishes were extensively rinsed withdeionized H20.

Preparation of proteins

Rat tail collagen was prepared according to the method of Price(27). CIG was prepared from human plasma by a modificationof the techniques described by Mosher and Blout (28) andMosher (29) that will be reported in detail elsewhere (3). Thespecific spreading activity of the CIG used in these experimentswas 500-700 spreading units per mg of protein (see below).

Assay of cell attachment and spreadingThe methods for measuring cell attachment and spreading ofBHK cells have been described previously (2-4). To quantitatecell attachment, BHK cells (subline BHK-21-13s) were har-vested from suspension cultures by centrifugation and washedand resuspended in adhesion medium (150 mM NaCl/3 mMKCl/1 mM CaCl2/0.5 mM MgCl2/6 mM Na2HPO4; pH 7.3).Incubations of ca. 0.75 X 106 cells in 1.0 ml of adhesion mediumwere placed in various collagen-treated Falcon culture dishes.The adhesion incubation assays were carried out for the timeperiods indicated at 370. CIG was added where designated. Atthe end of the incubations, the flasks were subjected to shakingat 150 rpm on a New Brunswick R-2 reciprocating shaker for10 sec at room temperature, and the cells resuspended by thisprocedure (considered to be nonattached) were removed witha pipette. The turbidities of the starting and final cell suspen-sions were determined at 640 nm with a Bausch and LombSpectronic 70 equipped with digital readout. Cell concentra-tions were calculated from a previously determined linear re-lationship between cell number and turbidity. The percent ofcells attached in an experiment was calculated as the startingnumber of cells in an incubation minus the number of nonat-tached cells, divided by the starting number of cells. The pre-cision and validity of this technique have been established.To determine cell spreading, BHK cells were suspended in

adhesion medium as above, and incubations of ca. 0.5 X 106cells in 1.0 ml of adhesion medium were placed in variouscollagen-treated Falcon culture dishes. The spreading assayswere carried out for the times indicated at 370. CIG was addedwhere designated. At the end of the incubations, the extent ofcell spreading was determined visually with a Zeiss InvertoscopeD inverted microscope equipped with phase contrast objectivesand a Polaroid camera attachment. In instances when wewished to compare cell spreading, activity was determined byvisually observing 100-200 attached cells and estimating thepercentage of spread cells: 5-35% (1+), 40-60% (2+), 65-85%(3+), 90-100% (4+). With experience, it has become possibleto read the qualitative assay with considerable reproducibilityand little variation among three different observers.

It should be pointed out that BHK cells have an absolute re-quirement for CIG in order to spread onto Falcon dishes, al-though they can attach to clean dishes nonphysiologically inmedium with no serum (1). The effect of CIG on cell spreadingon Falcon dishes is the same whether CIG is added to the in-cubation medium or used to pretreat the dishes for 5 min at 220(2-4). One unit of spreading activity has been defined as theamount of serum protein or purified CIG required to promotecomplete spreading of BHK cells on Falcon dishes in a 45-minassay (2, 4).

Spreading experiments were carried out with human fibro-blasts similarly as described above, except that the cells wereharvested from logarithmically growing stationary cultures bytreatment with a trypsin/EDTA solution (Grand Island Bio-logical Company, Grand Island, NY) for 10 min at 37°.

CellsBHK-21-13s cells were the gift of Adrian Chappel, Commun-icable Disease Center, Atlanta, GA. The cells were grown insuspension culture in Eagle's minimal essential medium(spinner modified) with double the concentrations of aminoacids (except for standard glutamine) and vitamins and sup-plemented with N-2-hydroxyethylpiperazine-N'-2-ethane-sulfonic acid (Hepes) buffer (20 mM), 0.1 g of ferric nitrate perliter, 2.0 g of dextrose per liter, 10% tryptose phosphate broth,and 10% fetal calf serum. The final sodium bicarbonate con-centration in the medium was 0.5 g/liter and the pH was ad-justed to 7.2.Human skin fibroblasts were the generous gift of Jim Griffin,

The University of Texas Health Science Center at Dallas. Thecells were grown in stationary culture in McCoy's 5A modifiedmedium (Grand Island Biological Company) supplementedwith Hepes buffer (20 mM, pH 7.2) and 10% fetal calf serum.Human fibroblast cultures were discarded prior to the 20thpassage.

RESULTSTime course of BHK cell attachmentOn dried collagen gels, attachment did not exceed 20% after1 hr of incubation (Fig. 1). Addition of CIG to the mediumresulted in a marked increase in the rate of attachment, andmore than 80% of the cells were attached after 1 hr. This ob-servation is similar to that made previously by others using driedcollagen substrata (16-18). However, the conditions for gelationand subsequent drying of the gels are harsh treatments com-pared to the conditions known to result in formation of nativecollagen gels (20, 21). Therefore, it was possible that the driedcollagen gels were denatured and had partially or completelylost their fibrillar organization. In support of this notion was theobservation that dried collagen gels in contact with filter paperimpregnated with 0.1% trypsin (Sigma type XI) were hydro-lyzed following an overnight incubation at 37°. Hydrated,native collagen gels were unaffected by this treatment (datanot shown). When hydrated, native collagen gels were testedfor cell attachment (Fig. 1), there was no CIG dependence.That is, the rate of cell attachment to hydrated, native collagengels was the same in the presence or absence of CIG in themedium and comparable to the rate of cell attachment to driedcollagen gels in the presence of CIG.

In order to test the possibility that CIG is required for cell

100

80 v

60

40A

.~20

100coC) D80

60

40I20

00 15 30 45 600 15 30 45 '60Minutes

FIG. 1. Time course ofBHK cell attachment on various collagensubstrata. CIG was either omitted from the incubation medium (U)or added at 10 units/ml (@). (A) Dried collagen gel; (B) native collagengel; (C) gelatin coating; (D) collagen coating.

Cell Biology: Grinnell and Minter

Dow

nloa

ded

by g

uest

on

Mar

ch 1

6, 2

021

4410 Cell Biology: Grinnell and Minter

adhesion to heat-denatured collagen, gelatin samples wereprepared and used to coat the substratum. In control experi-ments, untreated collagen solutions were also used to pretreatthe substratum. The results indicate that CIG is required forcell attachment to heat-denatured collagen-coated substratabut not to native collagen-coated substrata (Fig. 1).BHK cell spreadingExperiments were carried out under conditions similar to thosedescribed in Fig. 1 in order to analyze cell spreading on thevarious collagen substrata. Cells remained rounded and no cellspreading was observed in the absence of CIG on dried collagengels (Fig. 2) or gelatin-coated substrata (Fig. 3). The additionof CIG to the medium resulted in cell spreading (Figs. 2B, 3B,and 3C), and all the cells spread with CIG at concentrations >4units/ml (Figs. 2B and 3C). Pretreatment of the denaturedcollagen substrata for 5 min at 220 with a CIG-containing so-lution also promoted subsequent spreading onto the substrata(Figs. 2C, 2D, and 3D); about 4-fold higher CIG concentrationswere required for complete spreading to be observed (Figs. 2Dand 3D).On native collagen-coated substrata, partial cell spreading

occurred in the absence of CIG in 30 min (Fig. 4A), andspreading was complete by 2 hr (Fig. 4C). However, the ad-dition of CIG to the incubation medium at 10-20 units/mlpromoted the rate of cell spreading, which was best demon-strated after 30 min (compare Fig. 4 B with A). After 2 hr, cellmorphology appeared the same in the presence or absence ofCIG (Fig. 4 D and C). The effect of CIG at early times couldalso be observed after pretreatment of the native collagen-coated substrata with CIG.On hydrated, native collagen gels, cell spreading was dif-

ferent from that observed on any of the other substrata. For one

tt.;..l|iFIG. 3. BHK cell spreading on gelatin-coated substrata. The

incubations were carried out for 45 min. (A) Control; I(B) + CIG in themedium at 2 units/ml; (C) + CIG in the medium at 4 units/ml; (D)substratum pretreated for 5 min at 220 with CIG at 16 units/ml.(X200.)

thing, the shape of spread cells was more bipolar and less tri-angular (compare Fig. 5 with Figs. 2-4; see ref. 19). Moreover,cell spreading never reached 100% even when incubations werecarried out in complete growth medium overnight. Typically,about 25% of the cells were spread after 30 min (Fig. 5 A and

FIG. 2. BHK cell spreading on dried collagen gels. The incuba-tions were carried out for 45 min. (A) Control; (B) + CIG in the in-cubation medium at 5 units/ml; (C) substratum pretreated for 5 minat 220 with CIG at 5 units/ml; (D) substratum pretreated for 5 minat 220 with CIG at 20 units/ml. (X200.)

roG. 4. krHK cell spreading on native collagen-coatea substrata.The incubations were carried out for 30 min (A and B) or 2 hr (C andD). (A and C) Controls; (B and D) + CIG in the medium at 20 units/ml. (X200.)

Proc. Natl. Acad. Sci. USA 75 (1978)

Dow

nloa

ded

by g

uest

on

Mar

ch 1

6, 2

021

Proc. Natl. Acad. Sci. USA 75 (1978) 4411

FIG. 5. BHK cell spreading on hydrated native collagen gels. Theincubations were carried out for 30 min (A and B) or 2 hr (C and D).(A and C) Controls; (B and D) + CIG in the medium at 20 units/ml.(X200.)

B) and 50-75% by 2 hr (Fig. 5 C and D), after which there wasno further change. The addition of CIG to the incubations hadonly a marginal effect on cell spreading even at the highestconcentration tested, 20 units/ml. At most, there was a slightincrease. (Compare Fig. 5 B with A and 5 D with C.) In ex-

periments in which the hydrated native collagen gels were

pretreated with CIG solutions, no effects were observed.Immunological and biochemical analysis of cellspreadingThe freshly prepared collagen utilized in these experiments wasnot purified; therefore, it was possible that CIG might bepresent as a contaminant in the preparations. This could explainthe lack of a CIG requirement for cell attachment andspreading on native collagen substrata if such a CIG contami-nant were lost or inactivated during preparation of the driedcollagen gels or gelatin-coated substrata. However, sodiumdodecyl sulfate gel electrophoretic analysis of the collagenpreparations did not reveal a high molecular weight(210,000-250,000) polypeptide.

Experiments were carried out to compare the immunologicalspecificity of cell spreading on hydrated native collagen gelscompared to CIG-coated dried collagen gels. Antisera preparedin rabbits against the fetal calf adhesion and spreading factor(anti-ASF, 4) and against human CIG (anti-CIG, 3) were foundto crossreact with and inhibit the spreading activity-i.e.,CIG-in rat serum. Clean culture dishes were pretreated withmedium + 10% rat serum for 10 min at 220. BHK cell spreadingon these dishes was found to be inhibited by the addition of theantisera to the spreading assays. The inhibition was 50% withanti-ASF (1:5 dilution) and 30% with anti-CIG (1:5 dilution)(averages of 4 determinations). Preimmune serum was withouteffect. It should be pointed out that the rat tail collagen prep-

arations might contain rat CIG as a contaminant. Neither

anti-ASF; nor anti-CIG was inhibitory toward cell spreading onhydrated native collagen gels, suggesting that rat CIG was notpresent as a contaminant. In marked contrast, anti-CIG (1:5dilution) partially inhibited cell attachment and completelyinhibited cell spreading on denatured collagen gels that hadbeen treated with human CIG at 10 units/ml for 5 min at 220.We have shown elsewhere (3) that, under these conditions, theinhibitory effects of the antisera are directed at adsorbed CIGon the substratum and not on the cells.

Finally, we have found that a brief, 10-min treatment of CIGadsorbed to a substratum with lactoperoxidase (0.01 mg/ml)and NaI (1.0 mM) in phosphate-buffered saline, pH 7.0, fol-lowed by 0.1 mM H202, resulted in an inhibition of thespreading activity of adsorbed CIG. These studies will be re-

ported in detail elsewhere. Similar experiments were carriedout with hydrated native collagen gels, and it was found thatthis treatment did not alter cell attachment or spreading ontothe gels.Physiology of cell adhesionDespite the differences in CIG requirements for cell attachmentand spreading on the various collagen substrata, the physiologyof attachment and spreading was similar in every case. It hasbeen previously shown that CIG-dependent cell attachmentand spreading on tissue culture dishes or dried collagen gels canbe inhibited by sulfhydryl binding reagents, energy metabolisminhibitors, or removal of divalent cations from the incubationmedium (1, 30). Similar experiments were carried out withnative collagen-coated substrata and native collagen gels, anda typical experiment is shown in Table 1. The results indicatethat the addition of N-ethylmaleimide (0.1 mM) or 2,4-dini-trophenol (1 mM), or removal of divalent cations from the in-cubation medium inhibits cell attachment and spreading.Attachment and spreading of human fibroblastsAll of the experiments carried out thus far were with BHK cells,an established cell line. In order to demonstrate CIG depen-dence of cell spreading with tissue culture dishes, it is necessaryto use cells that do not secrete their own CIG-like factors (e.g.,most cell lines). The hypothesis has been presented (15) thatsome cells (e.g., cell strains) may attach and spread on a layerof CIG or a CIG-like substance endogenously produced andelaborated by the cells. If this is correct, diploid human fibro-blasts ought to attach and spread onto dried collagen substrata

Table 1. Physiology of cell attachment and spreading*

Cell spreadingNative Nativecollagen collagen-coated

Conditions gels substrata

Control 2+ 2++ MalNEt, 0.1 mM Ot Ot+ MalNEt, 0.01 mM 1+ 1++ MalNEt, 0.001 mM 2+ 2++ DNP, 1 mM Ot Ot+ DNP, 0.1 mM 2+ 1++ DNP, 0.01 mM 2+ 2+0.16 M NaClt Ot Ot

* N-Ethylmaleimide (MalNEt) or 2,4-dinitrophenol (DNP) wasadded to the incubations, as indicated. The extent of cell spreadingwas determined after 60 min (native collagen gels) or 45 min (nativecollagen-coated substrata).

t The extent of attachment was not quantitated in these experiments;however, it was obvious that no attachment at all occurred.Cells rinsed and resuspended in 0.16 M NaCl.

Cell Biology: Grinnell and Minter

Dow

nloa

ded

by g

uest

on

Mar

ch 1

6, 2

021

4412 Cell Biology: Grinnell and Minter

FIG. 6. Human fibroblast cell attachment and spreading. Theincubations were carried out for 60 min on tissue culture plastic (A)or dried collagen gels (B and C). (C) + CIG in the incubation mediumat 10 units/ml. (X200.)

in the absence of added CIG. Indeed, this is exactly what occurs,

as shown by the experiment shown in Fig. 6. Attachment andspreading occurred in the absence of serum or CIG on tissueculture substrata (Fig. 6A) or dried collagen gels (Fig. 6B), andthere was little, if any, effect of adding CIG (10 units/ml) tothe incubation medium (Fig. 6C).

DISCUSSIONThis paper presents quantitative studies of cell attachment andspreading on hydrated native collagen gels. The findingsdemonstrate that BHK cells attach and spread on these sub-strata, which are composed of cross-striated fibrils, without theaddition of CIG to the incubation medium. On the other hand,very little attachment and no spreading of cells occurs on driedcollagen gels or gelatin-coated substrata unless CIG is addedto the medium or used to pretreat the substrata. It is unlikelythat the difference between adhesion to native collagen sub-strata and dried collagen or gelatin substrata can be accountedfor by the presence of CIG as a contaminant in the collagenpreparations. Antibodies directed against CIG do not inhibitcell spreading on native collagen gels but do inhibit cell at-tachment and spreading on CIG-coated denatured collagensubstrata. Moreover, iodination of tyrosine residues inhibits theactivity associated with adsorbed CIG gels but not native col-lagen gels. Nevertheless, the underlying physiology of adhesionwith native collagen substrata is the same as that previouslyobserved with CIG-coated tissue culture dishes (1) or driedcollagen substrata (30)-i.e., a requirement for free SH groups,active energy metabolism, and divalent cations.The simplest interpretation of the data is that native collagen

in the fibrillar form (hydrated, native gels) and microfibrillarform (native collagen-coated substrata) contains sites withwhich BHK cells can directly interact via appropriate cellsurface receptors. These sites are apparently lost on dried col-lagen gels (partially or completely denatured fibrils) or substratacoated by gelatin (dispersed, randomized molecules), becauseBHK cells do not appear to interact directly with these substrata.Therefore, BHK cells may recognize collagen in some physicalorganizations but not others. Nevertheless, BHK cells are ableto attach to CIG-coated dried collagen or gelatin substrata,suggesting that CIG acts as an adhesive bridge in these cases.These findings are consistent with the observation that CIGbinds to gelatin much better than to native collagen (21). This

interpretation suggests that in situ fibroblasts may interact withnative collagen fibrils directly and not via a CIG bridge.However, if there are regions in situ where collagen has becomelocally altered so as to permit CIG binding, cells probably attachto these regions via a CIG bridge.

Finally, the experiments carried out with the human fibro-blast cell strain are consistent with previous studies that havedemonstrated cell strains to be able to attach and spread in theabsence of serum or CIG (5-7), presumably by the elaborationof CIG or a CIG-like molecule (15).Note Added in Proof. Unlike BHK cells, CHO cells have been foundto require CIG for attachment to native collagen gels. However, if therewas phosphate present during collagen polymerization or in the in-cubation medium, CHO cells attached to but did not spread on nativecollagen in the absence of CIG (H. Kleinman, personal communica-tion). On the other hand, we have found that BHK cells attach andspread on native collagen gels even if Hepes buffer is used in place ofphosphate during collagen polymerization and subsequent incuba-tions.

We are indebted to Dr. Andrew Kang and Dr. Jerry'Gross for theirhelpful discussions during the course of this research and in preparationof the manuscript. This research was supported by a grant from theNational Institutes of Health, CA14609.

1. Grinnell, F. (1976) in Membranes and Neoplasla: New Ap-proaches and Strategies, ed. Marchesi V. T. (Alan R. Liss, NewYork), pp. 227-236.

2. Grinnell, F. (1976) Exp. Cell Res. 102, 51-62.3. Grinnell, F. & Hays, D. G. (1978) Exp. Cell Res., in press.4. Grinnell, F., Hays, D. & Minter, D. (1977) Exp. Cell Res. 110,

175-190.5. Taylor, A. C. (1961) Exp. Cell Res. Suppl. 8, 154-173.6. Witkowski, J. A. & Brighton, W. D. (1971) Exp. Cell Res. 68,

372-380.7. Rajaraman, R., Rounds, D. E., Yen, S. P. S. & Rembaum, A. (1974)

Exp. Cell Res. 88, 327-339.8. Vaheri, A. & Ruoslahti, E. (1975) J. Exp. Med. 142,530-535.9. Mosher, D. F. (1977) J. Supramol. Struct. 6, 551-557.

10. Yamada, K. M., Yamada, S. S. & Pastan, I. (1977) J. Cell Blol. 74,649-654.

11. Olden, K. & Yamada, K. M. (1977) Cell 11, 957-969.12. Yaoi, Y. & Kanaseki, T. (1972) Nature (London) 237, 283-

285.13. Pegrum, S. M. & Maroudas, N. G. (1975) Exp. Cell Res. 96,

416-422.14. Stamatoglou, S. C. (1977) J. Ultrastruct. Res. 60,203-211.15. Grinnell, F. (1978) Int. Rev. Cytol. 53,65-144.16. Klebe, R. J. (1974) Nature (London) 250,248-251.17. Kleinman, H. K., McGoodwin, E. B. & Klebe, R. J. (1976) Bio-

chem. Biophys. Res. Commun. 72, 426-432.18. Pearlstein, E. (1976) Nature (London) 262,497-499.19. Gross, J. & Kirk, D. (1958) J. Biol. Chem. 233, 355-360.20. Elsdale, T. & Bard, J. (1972) J. Cell Biol. 54,626-637.21. Engvall, E. & Ruoslahti, E. (1977) Int. J. Cancer 20, 1-5.22. Vroman, L. (1972) Bull. NY Acad. Med. 48,302-310.23. Baier, R. E. (1972) Bull. NY Acad. Med. 48, 257-272.24. Olsen, D. A. & Kletschka, H. D. (1973) Prog. Surf. Membr. Sci.

6,331-364.25. Rosenberg, M. D. (1960) Biophys. J. 1, 137-159.26. Dillman, W. J., Jr. & Miller, I. F. (1973) J. Coll. Int. Sci. 44,

221-241.27. Price, P. J. (1975) Tissue Cult. Assoc. Man. 1, 43-44.28. Mosher, D. F. & Blout, E. R. (1973) J. Biol. Chem. 248,6896-

6903.29. Mosher, D. F. (1975) J. Biol. Chem. 250,6614-6621.30. Klebe, R. J. (1975) J. Cell Physiol. 86,231-236.

Proc. Natl. Acad. Sci. USA 75 (1978)

i

Dow

nloa

ded

by g

uest

on

Mar

ch 1

6, 2

021