Fibronectin Mediates Attachment of Chicken Myoblasts to a Gelatin-coated Substratum*

Myogenic cells derived from embryonic chicken muscle are conventionally cultured in a horse serum-containing medium. The serum is known to be the source of one or more factors that mediate the attachment of such cells to the gelatin-coated culture dish. In this paper, we demonstrate that fibronectin is the principal horse serum component responsible. In addition, we show that fibronectin synthesized and released into the medium by chicken fibroblasts also promotes the attachment of myogenic cells. Antibodies against highly purified human plasma fibronectin (cold-insoluble globulin) precipitated a single component of horse serum. This horse serum fibronectin was purified by affinity chromatography on gelatin coupled to Sepharose (Engvall, E., and Ruoslahti, E. (1977) Int. J. Cancer 20, l-5). When electrophoresed in sodium dodecyl sulfate-polyacrylamide gels under reducing conditions, human and horse fibronectins both ran as closely spaced doublet bands (M,. = approximately 230,000). Chicken embryo fibroblasts synthesized a protein precipitable by antibodies against human flbronectin. This chicken fibronectin was isolated by affinity chromatography from the medium of fibroblasts cultured in the absence of exogenous fibronectin. When electrophoresed, the newly synthesized chicken fibronectin ran as a single band (not a doublet) of approximately the same mobility as the human and horse fibronectins. Suspended myogenic cells were separated from attached fibroblasts in a serum-free medium and used to assay for attachment-promoting activity. Whether added directly to the medium or used to pretreat the (gelatinized) dishes, all three purified fibronectins promoted the attachment of myogenic cells and allowed their elongation. Horse serum depleted of fibronectin had low attachment-promoting activity. Purified antibodies against human fibronectin inhibited attachment by more than 85% when added to cells suspended in a medium that contained horse serum. Preincubation with these antibodies also removed almost all attachment-promoting activity from horse serum or from gelatinized dishes that had been pretreated with serum.


Myogenic
cells derived from embryonic chicken muscle are conventionally cultured in a horse serum-containing medium.
The serum is known to be the source of one or more factors that mediate the attachment of such cells to the gelatin-coated culture dish. In this paper, we demonstrate that fibronectin is the principal horse serum component responsible.
In addition, we show that fibronectin synthesized and released into the medium by chicken fibroblasts also promotes the attachment of myogenic cells. Antibodies against highly purified human plasma fibronectin (cold-insoluble globulin) precipitated a single component of horse serum. This horse serum fibronectin was purified by affinity chromatography on gelatin coupled to Sepharose (Engvall, E., and Ruoslahti, E. (1977) Int. J. Cancer 20, l-5 Sciences, December 1977 (37). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. $ To whom all correspondence should be addressed.
The effects of antibody could be prevented or reversed by purified fibronectin.
Attachment of cells to noncellular materials (extracellular matrices, basal laminae, collagen fibrils) is fundamental to normal metazoan development (l-4). Studies with cell cultures have shown that many growth-and differentiation-related processes depend on cell-substratum attachment (5-S). Primary cultures of avian myogenic cells in serum-containing media have been widely used in studies of muscle differentiation (reviewed in Refs. 5,9,and 10). Not only cell attachment and spreading, but also proliferation and differentiation of myogenic cells in such cultures, have been thought to require serum (5, 6). The formation of elongated myotubes in clonal cultures of chicken breast muscle cells depends on the presence of substrate-bound collagen (gelatin is as effective as native collagen) and of certain serum components (5,6). In high density cultures, presumably because the cells themselves provide sufficient collagen, serum alone is able to mediate cell attachment to uncoated culture dishes (5); with one exception (ll), substratum attachment and elongation of myogenic cells as well as subsequent myotube formation have not been observed in the absence of serum (5). Hauschka and White (12) showed that the (unidentified) serum component(s) which mediate the attachment of myogenic cells do so by binding to certain segments of the collagen ol (I) chain.
Several investigators have isolated serum factors that mediate cell attachment or spreading, or both. These factors, called cell attachment protein (13), cell attachment factor (14), or spreading factor (15), have subsequently been shown to belong to a class of vertebrate serum proteins immunologically related to human cold-insoluble globulin, a dimeric plasma protein with a subunit molecular weight of about 220,000 (16)(17)(18)(19). These plasma and serum proteins are thought to be the circulating forms of a cell surface protein that is the major iodinatable surface component of many normal cell types and that has been called LETS . Fibronectin binds to collagen and even more strongly to gelatin (28) and is believed to be important in cell-substratum attachment since it is present in reduced amounts on the surfaces of less adherent transformed cells (20,21,25,26) and since its addition to transformed cells partially restores their normal morphology and adhesion (22,24). Fibronectin-mediated cell attachment to collagen-coated substrata has been demonstrated for several kinds of cells (13,14). It is thought that cells bind to an insoluble complex of fibronectin and collagen and that 5475 5476 Fibronectin-mediated Myoblast Attachment this makes possible the attachment of cells to the extracellular matrix (13,14). We thought fibronectin the most likely candidate for the attachment-mediating serum component described by Hauschka (5). Accordingly, we purified fibronectin from horse serum and investigated its action on the attachment of chicken myogenic cells that had been cultured in suspension in serumfree medium (29).  (Fig. la). This fibronectin was used to obtain antisera: 0.5.ml samples containing 100 pg of protein and previously dialyzed against BSS were mixed with 0.5 ml of Freund's complete adjuvant (Gibco) and injected intracutaneously into rabbits. Boosts were made after 2 and 4 weeks and antisera were collected after 5 weeks. Anti-human fibronectin IgG was purified from these antisera as follows: 10 mg of human fibronectin purified (cf. Fig. 2) by the method of Chen and Mosesson (17) from fresh human plasma (Swiss Red Cross) were covalently bound to 1 g of CNBr-activated Sepharose 4B (Pharmacia).

Isolation
Twenty milliliters of anti-fibronectin antiserum were passed over a 3-ml column of fibronectin-Sepharose previously equilibrated with NaCl/P, and, after extensive washing with NaCl/P,, the fibronectin-specific IgG was eluted with 4 M MgCb (pH adjusted to 6.3 with NaOH).
Protein-containing fractions were pooled and passed over a Sephadex G-25 (Pharmacia) column equilibrated with NaCl/ P,. This method yielded 12 mg of anti-human fibronectin IgG (4 to 5'% of the total IgG fraction).
Isolation of Horse Serum Fibronectin-Horse serum fibronectin was isolated by an affinity chromatography method slightly modified from that described by Engvall and Ruoslahti (28 The preparation was stable for at least 4 weeks in 4 M urea at 4'C; repeated freezing and thawing led to precipitation. The same method was also used to purify fibronectin from human serum (Swiss Red Cross).
Horse serum free of material precipitable by anti-human tibronectin IgG (Fig. 2)  2. Electrophoretic analysis of horse serum fibronectin precipitated by anti-human fibronectin or isolated by affinity chromatography. The arrows on the left side of the gel indicate the positions of (from top to bottom) fibronectin, serum albumin, IgG heavy chain, and IgG light chains. Immunoprecipitates (with 57 pg of anti-fibronectin IgG) of: 5 (a) and 10 ~1 (b) of horse serum; 4 (d) and 8 pg (e).of horse fibronectin; 10 ~1 of fibronectin-depleted horse serum (g). Controls with 39 pg of preimmune IgG were: 10 ~1 of horse serum (c) and 8 pg of horse fibronectin (f). Lanes h to h show horse fibronectin (1.6,4, and 8 pg, respectively). Lanes 1 and m show human fibronectin isolated by the method of Chen and Mosesson (17) (2.5 and 5 pg). The protein band in the immunoprecipitates of horse serum (a and b) that co-migrates with bovine serum albumin is a contaminant from horse serum; it is seen with control IgG (c). The material in immunoprecipitates of horse serum which did not penetrate the separating gel (a and b) probably consists of aggregated fibronectin or IgG or both; it was not found when immunoprecipitates were electrophoresed in 4 to 12% polyacrylamide gradient gels containing 1% SDS.* ' V. Bieri, personal communication.

fibronectin
IgG in a double immunodiffusion test and which is immunologically related to human fibronectin (Fig. lb). This component was isolated by affinity chromatography (CL Fig. lb). Electrophoresis of the preparation revealed two closely migrating protein bands which co-migrated with human fibronectin (cold-insoluble globulin; M, = -230,000) purified by the method of Chen and Mosesson (17) (Fig. 2). Affinity-purified antibodies against human fibronectin precipitated the same doublet, both from preparations of the affinity-purified material and from whole horse serum (Fig. 2). A doublet is seen after electrophoresis of whole horse serum (Fig. 3), and not only in immunoprecipitates of serum or in purified preparations.
The doublet, therefore, does not arise in the course of the purification or immunoprecipitation procedures. As we do not know whether the two prominent bands in horse serum are also found in fresh plasma or in newly synthesized horse fibronectin, the origin of the doublet is unclear. A similar doublet of even more closely migrating bands can sometimes be detected in preparations of human plasma fibronectin (see Fig. 2 and Ref. 18); in this case, the subunit heterogeneity has been attributed to the action of proteases (18,19).
The yield of horse serum fibronectin was about 0.6 to 0.7 mg of protein/ml of serum (or 0.7 to 0.8% of the total horse serum protein), compared to about 0.2 mg of human fibronectin/ml of serum (or 0.22% of the total serum protein) isolated by the same method. The latter value agrees well with esti-  (17); c, chicken fibronectin (2 pg) isolated from conditioned medium (free of exogenous fibronectin); d, horse serum fibronectin (4 pg); e, medium that had originally been free of fibronectin (LH-medium containing 10% fibronectin-depleted horse serum) after conditioning by fibroblasts (15 ~41, 80 pg of protein); f, fibronectindepleted horse serum (37 pg of protein); g, horse serum (32 pg of protein); h, human serum (36 ,ag of protein). mates of the fibronectin content of human serum based on immunological methods (17). Since fibronectin binding to and recovery from the affinity column is essentially 100% efficient (Ref. 28; cf. also Fig. 2), the fibronectin concentration in horse serum must be at least 3 times that in human serum.
Chicken Fibronectin from Fibroblast Cell Surface and from Conditioned Medium-Cell surface fibronectin (CSP) extracted from chicken fibroblasts that had been cultured in a serum-containing medium contained a major protein band that co-migrated with human plasma fibronectin (Fig. 3). If the cells were first labeled with [""Slmethionine and then extracted, radioactive chicken fibronectin could be specifically precipitated from the extracts with anti-human fibronectin (Fig. 4). This confimed that the extracted fibronectin was (at least partially) of cellular origin. In contrast to the fibronectins isolated from human plasma or horse serum, the immunoprecipitated, newly synthesized, chicken fibronectin ran not as a doublet, but as a single protein band (Fig. 4). Even at shorter exposure times, no evidence of a doublet was detected on the x-ray films.
Fibroblasts proliferated rapidly in LH-medium supplemented with fibronectin-depleted horse serum; after 4 days, the originally fibronectin-free medium contained a component (which ran as a single protein band, not a doublet) with the same electrophoretic mobility as human and horse fibronectin (Fig. 3). This chicken fibronectin is released into the medium by the fibroblasts. It is precipitable with anti-human fibronectin." It was isolated by affinity chromatography (Fig. 3 This suspension containing about 4.5 x 10" cells/ml (estimated by counting in a hemocytometer) was diluted 1:1, 2:3, 1:3, 1:6, and 1:12 with LH-medium.
Four milliliters of diluted cell suspension were plated/gelatinized 6-cm dish and horse serum was added to of attached cells was roughly proportional to the number of plated cells over a wide range (between lo5 and 1.2 x 10" plated tells/6-cm dish). Attachment of myoblasts after addition of horse serum or horse serum fibronectin occurred rapidly, reaching characteristic values, which depended on the serum or fibronectin concentration, by 2 h (Fig. 6). The attached cells were still rounded at this time, although two short processes were frequently seen, particularly in the presence of purified horse fibronectin (Fig. 7). The elongation of the bipolar myoblasts was complete only after about 24 h (Fig. 7); elongation in cultures supplemented with isolated fibronectin was indistinguishable from that observed with whole horse serum.
Horse serum from which fibronectin was selectively removed by gelatin affinity chromatography lost most of its attachment-promoting activity. In one experiment, 6.1 x lo4 and 1.0 x 10" cells/dish, respectively, were attached at 3 h in 1 and 3% depleted serum, while attached cells numbered 6.3 x lo5 and 7.6 x 10" in 1 and 3% untreated serum. At 20 h, the values for 1 and 3% depleted serum were 7.2 x lo4 and 1.1 x 105, respectively, compared to 7.0 X 10" and 5.4 x 10" in full serum; Table I (A). Cultures were fixed after 3 h (--) and 20 h (---).

experiment.
The activity of fibronectin-depleted serum was fully restored by addition of purified fibronectin.
The number of attached cells continued to increase between 2 and 20 h after addition of horse serum, although at a lower rate than in the first 2 h; this was seen even at low serum concentrations in which attachment during the fist 2 h was well below the maximum (Fig. 6). In contrast, there was a net loss of attached cells during the same period in cultures supplemented with purified fibronectin even at high concentrations (Fig. 6). This difference in behavior between cultures supplemented with whole serum and fibronectin, respectively, could be overcome by adding fibronectin-depleted horse serum together with fibronectin (see below and Table I). Attachment of chicken myoblasts was also mediated by serum or fibronectin previously adsorbed to the gelatinized substratum.
The time course of attachment was similar to that after addition of horse serum or horse fibronectin to the culture medium (not shown); to reach similar attachment levels, however, higher concentrations were necessary. Over the entire range tested (up to 100% horse serum and 400 pg/ ml of fibronectin), attachment increased with increasing concentration of horse serum or fibronectin in the pretreatment solution.

Fibronectin-mediated
Myoblast Attachment  8 shows the dependence of number of attached cells on the concentration of horse serum and horse fibronectin at 3 and 20 h after addition. At 3 h, halfmaximal attachment activity was reached with about 700 pg/ ml of horse serum protein (0.9% horse serum) or 25 pug/ml of purified horse serum fibronectin.
For purified fibronectin, the concentration giving half-maximal activity at 20 h was the same as at 3 h, while with horse serum, attachment had reached a higher absolute level by 20 h and half-maximal activity was seen at about 300 pg/ml of protein (0.4% horse serum). This finding, taken together with the loss of attached cells between 3 and 20 h after addition of fibronectin in the absence of other serum factors (Fig. 6), suggests that horse serum factors other than fibronectin enhance cell attachment, perhaps indirectly by affecting cell metabolism or viability. This synergistic effect is demonstrated in Table I: fibronectindepleted serum enhances, by about lo-fold, the apparent specific activity of purified horse fibronectin measured at 20 h.
Since affinity-purified fibronectin represents about 0.8% of total horse serum protein, one would expect a 125-fold purification if the activity yield were 100%. The reason for the low recovery (about 25%, based on 30-fold increase in specific activity assayed at 3 h; Fig. 8) in our affinity purification of horse serum fibronectin is not known. Horse serum fibronectin isolated (37) by the method of Klebe (13) had about the same specific activity as the affinity-purified material (not shown). The specific activity of purified human fibronectin was about 2 to 3 times higher than that of horse fibronectin (Table II). Human serum is at least 2 times less active than horse serum (as expected from its approximately 4-fold lower content of fibronectin) and toxic at concentrations over 1%. The chicken fibronectin used in attachment assays was extracted from fibroblast cultures grown in medium containing fibronectin-depleted horse serum. This was necessary to ensure that the extracted fibronectin would be entirely chicken-derived and not contaminated with exogenous serum fibronectin that had adsorbed to the cells. Chicken cell-surface fibronectin isolated in this way promoted attachment of chicken myoblasts; half-maximal activity appeared to be similar to that found for human and horse fibronectin (Table II). After conditioning by fibroblasts, LH-medium supplemented with fibronectin-depleted horse serum became a very potent promotor of myoblast attachment: more myoblasts attached within 20 h with only 3% conditioned medium in the assay medium than with 3% horse serum (Table II). The greatly enhanced activity was at least partially due to release of fibronectin by the fibroblasts: chicken fibronectin isolated from this conditioned medium by affinity chromatography was at least as active in promoting attachment of chicken myoblasts as chicken cell-surface fibronectin (Table II) (Table III). The number of attached cells in cultures treated with anti-fibronectin was only slightly higher than the background level in the absence of horse serum. The specificity of this effect was shown by two control experiments.
1) Preincubation of the antibody with purified horse fibronectin eliminated its action on serummediated attachment (Table III). 2) As part of the experiment shown in Table III, other aliquots of the cell suspension were supplemented both with horse serum (1%) and with either anti-human fibronectin antiserum (1.3%) or antiserum that had first been passed over a fibronectin-Sepharose column (1.3%); attachment after 20 h was, respectively, 2.1 x 10" and 6.8 x lo" cells/dish.
In a separate experiment, anti-human fibronectin likewise inhibited the attachment activity of purified fibronectin.
It is, therefore, not surprising that the inhibition To exclude the possibility that the immunoreaction occurring in the culture medium blocked cell attachment by indirect mechanisms, such as cell agglutination, horse serum was preincubated with anti-human fibronectin IgG and, after removal of the immunoprecipitate, added to a suspension of myoblasts. In this experiment, 35 pg of anti-human fibronectin IgG reduced the attachment activity of 10 ~1 of horse serum by 80% (Table III). This demonstrates that the inhibitory effect of anti-fibronectin is indeed due to specific removal of fibronectin from horse serum.
The myogenic cells we use are homogeneous with respect to developmental stage (virtually all are postmitotic myoblasts), are essentially free of fibroblast contamination, have never been exposed to serum in culture, and do not need to be treated with protease or other dissociating agents in order to obtain a suspension for the attachment test (29). Myoblast attachment depends upon the addition of exogenous fibronectin. This suggests that, if myoblasts themselves produce fibronectin, the amount is insufficient to mediate attachment (cf. Ref. 37).
In a further experiment, gelatinized dishes were pretreated with 30% horse serum; in the absence of a second pretreatment, 1.8 x lo" cells/dish attached by 20 h. A second pretreatment with anti-human fibronectin IgG (0.14 and 0.29 mg/ml) reduced attachment of chicken myoblasts to the serumtreated dishes by up to 90% (2.6 X lo4 and 8 x 19' cells/dish at 20 h, respectively).
Control IgG (0.39 mg/ml) in the second pretreatment was ineffective in blocking adhesion (1.9 x loj cells attached/dish at 20 h). This confirmed that the attachment-promoting serum component which binds to the gelatinized substrate is fibronectin; the antibody evidently inhibits attachment of myoblasts by blocking their substratum binding sites. Taken together with the results on the specific removal of fibronectin from serum by affinity chromatography (see above), the experiments reported in this section establish that fibronectin is the principal serum factor responsible for promoting myoblast attachment.

Elucidation of the Functions of Fibronectin is Facilitated by the Use of Culture Conditions
Free of Exogenous Fibronectin-Fibronectin has been reported to promote cell-substratum attachment (13,14), morphological changes associated with spreading (22, 24, 38), cell aggregation (23), and cell motility (39). One difficulty that has become apparent in studies of fibronectin is that this glycoprotein does not have a unique location on the cell surface or in the immediate surroundings of the cell. Aside from that (presumably largely newly synthesized) fibronectin found intracellularly, fibronectin may occur either intimately associated with the cell surface (20,23) or as a protein released from the cell (25,40). Released fibronectin, in turn, is found both as soluble protein (40) and as part of an insoluble matrix (26). Even if one knows that fibronectin is involved in a given activity, it is difficult to ascertain which class or classes of the variously located fibronectin molecules is responsible. Compounding this problem is the possibility that fibronectin in different locations may be structurally distinct. The readily soluble fibronectin released by cells into the medium and the fibronectin extractable from cell surfaces appear to have identical primary structures (40). It is known, however, that some fibronectin molecules occur as polymers linked by disulfide bridges (41) and that fibronectin can be cross-linked by transglutaminase (42,43). The degree of cross-linking might influence the biological activity.
The Attachment Assay-To investigate attachment and spreading activity of serum or serum factors, several different attachment assays with cells from established lines (13-1538) and with primary cells (44) have been developed. The cells are obtained either from monolayer cultures or tissues by dissociation with proteases (13, 14) or from suspension cultures grown in medium free of divalent cations (15,38). The cell suspensions are plated on untreated (15,38,44) or collagen-coated (13, 14) dishes, and serum or serum factors are either added to the culture medium (15,38) or used to pretreat the culture dishes (13-X,38,44,45,. "lppreciable attachment in the absence of added serum or fit ronectin complicates the analysis of cell substratum attachment; in two instances with trypsin-treated cells, this was overcome by washing collagenized dishes with urea before use (13,14). Moreover, certain cells attach to untreated dishes even faster in the absence than in the presence of serum (38) or "spreading factor" (15), which in this case promotes cell spreading only (15). It is an advantage of our culture system that both substratum attachment and elongation of the suspended primary chicken myoblasts strongly depend on the presence of serum or isolated fibronectin.
Because of differences among the various assay systems, comparison of our data with those of others is difficult. Nevertheless, the active concentrations of serum and fibronectin as well as the time course of myoblast attachment appear to be similar to those observed for other cell types and assay systems (13-15, 38, 44).
Although fibronectin alone is able to mediate both attachment and elongation of chicken myoblasts, there is some cell loss over longer times. Our studies reveal that other serum factors synergistically enhance fibronectin-mediated myoblast attachment and elongation between 3 and 20 h; with the aid of our assay system, we hope to elucidate their nature.
An additional complication in studies with cell cultures, which we believe has received too little attention heretofore, is the problem of distinguishing between effects of exogenously supplied fibronectin and that produced by the cells being studied. Serum, generally added at some stage as a supplement to the culture medium, is one obvious source of exogenous fibronectin.
A second potential source are cells others than Fibronectin and Collagen in Muscle Development-Hauschka (5) demonstrated that at least one factor present in horse serum was required for the attachment of myogenic cells to collagen-coated culture plates. From our work, it appears that Hauschka's active horse serum component is the horse serum form of fibronectin.
Strongly supporting this notion is the finding (45) that a "cell attachment protein" from fetal calf serum (13), (which, like the horse serum fibronectin described here, is immunologically closely related to human fibronectin (19)) binds to the very same fragments of the a,(I) chain of collagen that had earlier been shown to promote myogenesis in clonal cultures of chicken myogenic cells (12). Homologous (chicken) fibronectin is also active in promoting myoblast attachment to gelatinized plates. This is an important piece of information if it is to be proposed (37) that fibronectin-mediated anchorage of cells to an extracellular matrix or basal lamina may be part of normal myogenesis in the chicken. Achnowledgments-We are grateful to Drs. J.-C. Perriard and M.