ANALYTICAL
BIOCHEMISTRY
Preparation
94,
308-312 (1979)
of Collagen Substrates for Cell Attachment: Collagen Concentration and Phosphate Buffer
HYNDA
K. KLEINMAN,
Laboratory
of Developmental
National
ERMONA B. MCGOODWIN, AND GEORGE R. MARTIN Biology Institutes
Effect of
STEPHEN I. RENNARD,
and Anomalies, National Institute of Dental Bethesda, Maryland 20014
Research,
of Health,
Received August 2, 1978 We have studied the attachment of cultured Chinese hamster ovary cells to collagen substrates prepared in several ways. The attachment of these cells to collagen required under most conditions either serum or tibronectin purified from serum. Reconstituted collagen substrates required greater amounts of tibronectin than dishes coated by drying a collagen solution, but in each case the amount of fibronectin required was proportional to the amount of collagen on the dish. High levels of phosphate, 0.01 M and above, used as a buffer in heat-reconstituted collagen substrates allowed cell attachment without fibronectin. However, since the cells did not spread under these conditions and were not released from the substrate when incubated with trypsin, binding of cells with such levels of phosphate probably represents nonphysiological adhesion.
Recent studies indicate that the attachment of fibroblasts to collagen substrates is carried out by fibronectin, a large glycoprotein present on the external cell surface (l-3). Cold insoluble globulin, an identical or very similar glycoprotein, is found in large quantities in serum. The fibronectin in serum is required for rapid cell adhesion to collagen in vitro because the trypsin used in passing the cells releases fibronectin from the cell membrane. In vitro fibronectin binds to the collagen substrate and then the cells adhere to the fibronectin-collagen complex and spread. It binds to a specific sequence of amino acids on the collagen chain (4,5) located in the region of the collagenase-sensitive site (6). The physiological role of fibronectin is in question. It binds better to denatured than to native collagen (7-9) possibly because of the inaccessibility of the binding region when present in the helical structure of the native collagen molecule. Moreover, some studies have failed to establish that fibronectin is required for the attachment of all 0003-2697/79/060308-05$02.00/O Copyright All rights
0 1979 by Academic Press, Inc. of reproduction in any form reserved.
cells. Recent studies suggest that hepatocytes (10) and hamster kidney cells (11) do not require fibronectin to bind to collagen substrates reconstituted at neutral pH and 37°C. It was suggested that the fibronectin requirement is limited to the binding of cells to denatured collagen. We report here on the attachment of cells to collagenous substrates prepared in a variety of ways. The data support a role for fibronectin in the attachment of cells to native collagen. High levels of phosphate, which have been used by others to prepare collagen substrates, cause the attachment of cells to collagen, but this is a nonphysiological binding. MATERIALS
AND METHODS
Preparation of collagens. Collagen was prepared by extracting the tendons of rat tails in 0.5 M acetic acid overnight at 4°C. Insoluble material was removed by centrifugation and collagen was precipitated from the supematant fluid with 10% NaCl (w/v). 308
PREPARATION
OF COLLAGEN
COLLAGEN(mgI
FIG. 1. The effect on cell attachment of varying the amount of collagen substrate in the absence and presence of serum. Rat tail tendon collagen was air dried on bacteriological petri dishes before use. The results shown represent the mean of duplicate measurements where duplicates differed by less than 15%.
The precipitate was collected by centrifugation and dialyzed against 1.0 M NaCl, 0.05 Tris-HCl, pH 7.4. Insoluble material was removed, and the soluble material was subsequently dialyzed against 0.1% acetic acid and lyophilized. In addition, collagen from the skin of lathyritic rats was prepared by standard methods (12). Ascaris collagen was a gift from K. Sullivan (NIDR). Preparations of collagen substrates. The dried collagen was dissolved at 4 mg/ml in 0.1% acetic acid and stored frozen in small aliquots until used. Collagen substrates were prepared on tissue culture plates in different ways. In some cases collagen fibers were reconstituted from solution by exposure to NH,OH vapors for a few minutes at room temperature, producing a firm adherent gel (13). In other cases, an aliquot of the stock collagen solution was mixed with l/IO its volume of 2.0 M NaCl in 0.1 M TrisHCl, at 4°C and the pH was adjusted to 7.2 with a dilute NaOH solution. Then 0.5 ml portions were pipetted onto 35mm tissue culture dishes and left at room temperature or at 37°C for 30 min to gel (“Tris” gel). Substrates were also prepared by diluting
SUBSTRATES
309
1 ml of the stock collagen solution with 0.1 ml of 2.0 M NaCl in 0.1 M phosphate, pH 7.4, or with 2.0 M NaCl in 0.1 M HepesHCl,’ pH 7.4. Reconstitution of fibers was achieved by maintaining them at 24 or at 37°C for 15 to 60 minutes. The coated dishes were either used immediately after preparation or, in some cases, were air dried. Dried collagen substrates were also prepared from the rat tail collagen, lathyritic rat skin collagen, or Ascaris collagen. In these studies collagen was dissolved at 1 mg/ ml in 0.5 M acetic acid overnight with stirring at 4°C. Various amounts of collagen were also added to bacteriological petri dishes and air dried. Attachment assay. Dishes coated with the various substrates were incubated for 1 h at 24 or at 37°C in 95% air, 5% CO, with Eagle’s minimal essential medium containing 200 &ml bovine serum albumin (GIBCO) and bovine serum were indicated. Then lo5 Chinese hamster ovary (CHO) cells freshly isolated from culture with trypsin in 0.1 ml of the same buffer were added to each plate and incubated an additional 1.5 h. At the end of the incubation, the unattached cells were removed, combined with cells derived from three saline washes of the plate, and counted in a Coulter electronic cell counter. Where possible, the attached cells were released from the substrates by a 3- to 5-min incubation with 0.1% trypsin, 0.1% ethylenediamine tetraacetate (EDTA), in phosphate-buffered saline, and counted electronically. Fibronectin was purified from bovine serum (Colorado Serum Co.) using a collagen affinity column (7,14). RESULTS
Most cells did not attach to dried collagen substrates in the absence of serum, no matter how much collagen was on the dish (Fig. 1). Low serum (5%) levels stimulated the 1 Abbreviations used: Hepes, 4-(2-Hydroxyethyl)-lpiperazineethanesulfonic acid; CHO, Chinese hamster ovary.
310
KLEINMAN
attachment of cells to plates coated with low levels of collagen, but were not effective with higher levels of collagen (Fig. 1). However, greater amounts of serum (10 and 20%) permitted attachment to collagen over the whole range of collagen concentrations tested (Fig. I). The inhibition of attachment observed at higher levels of collagen could be due to the binding of serum fibronectin by collagen that leaches from the surface of the dish into the medium. Using radioactive collagen we found that up to 40% of the collagen was lost from the surface of a dish within 90 min after adding medium to the dish at 37°C. The attachment of cells to heat-reconstituted collagen substrates was also examined. Since collagen from rat tail tendon formed more stable and adherent gels than collagen from lathyritic rat skin, tendon collagen was used in these experiments. Furthermore, reconstituted collagen gels were found to adhere to tissue culture dishes better than to bacteriological dishes and therefore tissue culture dishes were used with these gels. The attachment of the CHO cells to dried collagen films, to gels formed by exposure of collagen solutions to NHIOH vapors, and to gels reconstituted by raising the temperature using Tris and Hepes buffers was similar (Figs. 2 and 3). Serum and purified fibronectin stimulated the attachment of cells to these substrates at both 24 and 37°C (Fig. 2). Identical binding was observed at 24 and 37°C. Collagen substrates reconstituted at 24 or 37°C showed an equal requirement of serum for cell attachment. The lower temperature should minimize the amount of denaturation occurring in the substrates. The amount of serum required for adhesion and spreading was proportional to the amount of collagen in the gel. Identical results were also obtained when the gels were allowed to air dry and then assayed for cell adhesion (not shown). The attachment and spreading of the cells were dependent upon cellular metabolism as al-
ET AL.
I
,
I
/
2.5
5.0
7.5
SERUM
I 10.0
(%)
FIG. 2. Effect of varying serum concentrations on cell attachment to small and large amounts of airdried collagen substrates and to reconstituted collagen. Rat tail collagen was air dried at low (W) and at high (0) concentrations and the reconstituted collagen (0) was prepared in the presence of Hepes buffer. The results shown represent the means of duplicate measurements where duplicates differed by less than 10%.
ready shown (15) because both were inhibited at 4°C and by N-ethyl maleimide. Different results were obtained on the substrates prepared with varying amounts of phosphate buffer. Cell attachment occurred well without serum (Fig. 3 and 4) and serum produced only minimal stimulation of attachment. Cells attached but did not spread out on the substrates prepared with 0.01 M phosphate (Fig. 3) or on substrates to which phosphate was added after reconstitution. The phosphate effect was not reversed if, after 1 h of incubation, the phosphate-containing medium was removed and replaced with medium lacking phosphate. Few of the cells that attached to the substrate in the presence of this level of phosphate were released after incubation with trypsin for up to 4 h. This is in contrast to cells attached with fibronectin to collagen or to cells cultured in medium. Here the cells became flattened on the substrate and when treated briefly (3 min) with trypsin detached from the substrate. Phosphate was also tested for its ability to promote attachment of cells to a collagen
FIG. 3. Attachment of cells to collagen gelled by four different procedures and to dried lathyritic collagen in the presence and absence of serum. Cells were allowed to adhere 1.5 h to a (a) collagen film without serum, (b) collagen film plus 30% serum, (c) NH,OH vapor-gelled collagen without serum, (d) NH,OH vapor-gelled collagen plus 30% serum, (e) phosphate buffer-gelled collagen without serum, (f) phosphate buffer-gelled collagen plus 30% serum, (g) Hepes buffer-gelled collagen without serum, (h) Hepes buffer-gelled collagen plus 30% serum, (i) Tris buffer-gelled collagen without serum, and (j) Tris buffer-gelled collagen plus 30% serum.
from Ascaris cuticle that does not bind fibronectin (16), to dried collagen substrates, and to bacteriological and tissue culture dishes. Phosphate stimulated the
attachment of cells to Ascaris collagen in the absence of serum but had no effect upon cell attachment in the presence of serum (Fig. 5). It stimulated the attachment of cells to dried collagen substrates in the absence of serum, but not to the same extent as seen with the gelled collagen substrate. Phosphate had no effect upon cell attach-
3 E 6
0
No Added
El
+lO+ M phosphate
Phosphate
Qitl 5
20 -
0
I 10-2
10 3 PHOSPHATE
CONCENTRATION
, 10’ IM)
FIG. 4. Attachment of cells to Hepes gelled collagen prepared in the presence of increasing amounts of phosphate buffer. Serum (30%) was present where indicated. Each bar represents the mean of duplicate measurements where duplicates differed by less than 10%.
,Ascaris
Typel,
NO SERUM
Ascaris +2%
Type
I
SERUM
FIG. 5. Attachment of cells to dried collagen (type I) from rat tail tendon and to Ascan’s collagen substrates in the presence and absence of serum and added phosphate. Each bar represents the mean of duplicate measurements where duplicates differed by less than 20%.
312
KLEINMAN
ment to or release from bacteriological tissue culture dishes.
or
DISCUSSION
Higher than physiological levels of phosphate are often used in the preparation of reconstituted collagen gels. However, such levels of phosphate permit cells to bind to collagen without fibronectin. Since the cells do not spread out and are not liberated by trypsin, we conclude that this adhesion is not physiological. In the absence of serum, high levels of phosphate also promote adherence to Ascaris collagen, a nonfibronectin binding collagen, but in the presence of serum, phosphate has no effect on cell adherence to this collagen. Phosphate may precipitate as calcium phosphate and the cells then bind to these deposits. Precipitates are seen on the surface of the dish and these are not removed when the medium is changed. Since changing the medium after incubation with phosphate does not remove the adhesion-promoting effect of phosphate, it is likely that these precipitates persist and cause this effect. We have shown that the attachment as well as spreading of cells both to dried collagen films and to reconstituted collagen gels require serum or fibronectin. More serum is required with the gelled collagen because greater amounts of collagen are present, and hence more is solubilized from the substrate and available to bind fibronectin. It is also possible that reconstituted collagen offers fewer binding sites for fibronectin than the collagen films since the majority of the molecules are at internal sites in the fibers. In the absence of serum, slightly more cells attached to collagen substrates reconstituted with NH,OH vapors or with heat using Tris or Hepes buffers than with airdried collagen films. This difference could be due to trapping of cells within the fiber mesh or may represent the adhesion of cells to collagen via a mechanism independent of fibronectin.
ET AL.
In summary, our results indicate that CHO cells require fibronectin to attach to collagen substrates. The amount of fibronectin required is proportional to the amount of collagen used as a substrate. Attachment occurs to a similar level with fibronectin on all substrates except those prepared with 0.01 M or higher levels of phosphate. Cells attach to collagen substrates treated with this level of phosphate in the absence of fibronectin. However, the cells do not flatten and are not liberated by trypsin, both characteristics of physiological cell attachment. Our data are consistent with the requirement of fibronectin for cell attachment to both native and denatured forms of collagen. REFERENCES 1. Klebe, R. J. (1974)Nuture (London) 2.50,248-251. 2. Pearlstein, E. (1976) Nature (London) 262, 497-499. 3. Kleinman, H. K., Murray, J. C., McGoodwin, E. B., and Martin, G. R. (1978)J. Invest. Dermard. 71,9-11. 4. Kleinman, H. K., McGoodwin, E. B., and Klebe, R. J. (1976) Biochem. Biophys. Res. Commun. 72, 426-432. 5. Dessau, W., Adelmann, B. C., Timpl, R., and Martin, G. R. (1978) Biochem. J. 169, 55-59. E. B., Martin, 6. Kleinman, H. K., McGoodwin, G. R., Klebe, R. J., Fietzek, P. P., and Woolley, D. E. (1978)J. Biol. Chem. 253, 5642-5646. 7. Hopper, K. E., Adelmann, B. C. Genter, G., and Gay, S. (1976) Immunology 30, 249-259. 8. Ruoslahti, E., and Engvall, E. (1978) Ann. N. Y. Acad. Sci. 312, 178-191. 9. Jilek, R., and Hbrmann, H. (1978) HoppeSeyler’s
2. Physiol.
Chem.
459, 247-250.
10. Rubin, K., Kjellen, L., and iibrink, B. (1977) Exp. Cell Res. 109, 413-422. 11. Grinnell, F., and Minter, D. (1978) Ann. N. Y. Acad. Sci. 312, 434-435. 12. Bomstein, P., and Piez, K. A. (1%6) Biochemistry
5, 3460-3473.
13. Elsdale, T., and Bard, J. (1972) J. Cell Biol. 54, 626-637. 14. Engvall, E., and Ruoslahti, E. (1977) In?. J. Cancer 20, 1- 17. 15. Klebe, R. J. (1975) J. Cell. Physiol. 86, 231-236. 16. Kleinman, H. K., Murray, J. C., McGoodwin, E. B., Martin, G. R. and Binderman, I. (1978) Calcif. Tissue Abst. Suppi. pp. 61-72.
BIOCHEMISTRY
Preparation
94,
308-312 (1979)
of Collagen Substrates for Cell Attachment: Collagen Concentration and Phosphate Buffer
HYNDA
K. KLEINMAN,
Laboratory
of Developmental
National
ERMONA B. MCGOODWIN, AND GEORGE R. MARTIN Biology Institutes
Effect of
STEPHEN I. RENNARD,
and Anomalies, National Institute of Dental Bethesda, Maryland 20014
Research,
of Health,
Received August 2, 1978 We have studied the attachment of cultured Chinese hamster ovary cells to collagen substrates prepared in several ways. The attachment of these cells to collagen required under most conditions either serum or tibronectin purified from serum. Reconstituted collagen substrates required greater amounts of tibronectin than dishes coated by drying a collagen solution, but in each case the amount of fibronectin required was proportional to the amount of collagen on the dish. High levels of phosphate, 0.01 M and above, used as a buffer in heat-reconstituted collagen substrates allowed cell attachment without fibronectin. However, since the cells did not spread under these conditions and were not released from the substrate when incubated with trypsin, binding of cells with such levels of phosphate probably represents nonphysiological adhesion.
Recent studies indicate that the attachment of fibroblasts to collagen substrates is carried out by fibronectin, a large glycoprotein present on the external cell surface (l-3). Cold insoluble globulin, an identical or very similar glycoprotein, is found in large quantities in serum. The fibronectin in serum is required for rapid cell adhesion to collagen in vitro because the trypsin used in passing the cells releases fibronectin from the cell membrane. In vitro fibronectin binds to the collagen substrate and then the cells adhere to the fibronectin-collagen complex and spread. It binds to a specific sequence of amino acids on the collagen chain (4,5) located in the region of the collagenase-sensitive site (6). The physiological role of fibronectin is in question. It binds better to denatured than to native collagen (7-9) possibly because of the inaccessibility of the binding region when present in the helical structure of the native collagen molecule. Moreover, some studies have failed to establish that fibronectin is required for the attachment of all 0003-2697/79/060308-05$02.00/O Copyright All rights
0 1979 by Academic Press, Inc. of reproduction in any form reserved.
cells. Recent studies suggest that hepatocytes (10) and hamster kidney cells (11) do not require fibronectin to bind to collagen substrates reconstituted at neutral pH and 37°C. It was suggested that the fibronectin requirement is limited to the binding of cells to denatured collagen. We report here on the attachment of cells to collagenous substrates prepared in a variety of ways. The data support a role for fibronectin in the attachment of cells to native collagen. High levels of phosphate, which have been used by others to prepare collagen substrates, cause the attachment of cells to collagen, but this is a nonphysiological binding. MATERIALS
AND METHODS
Preparation of collagens. Collagen was prepared by extracting the tendons of rat tails in 0.5 M acetic acid overnight at 4°C. Insoluble material was removed by centrifugation and collagen was precipitated from the supematant fluid with 10% NaCl (w/v). 308
PREPARATION
OF COLLAGEN
COLLAGEN(mgI
FIG. 1. The effect on cell attachment of varying the amount of collagen substrate in the absence and presence of serum. Rat tail tendon collagen was air dried on bacteriological petri dishes before use. The results shown represent the mean of duplicate measurements where duplicates differed by less than 15%.
The precipitate was collected by centrifugation and dialyzed against 1.0 M NaCl, 0.05 Tris-HCl, pH 7.4. Insoluble material was removed, and the soluble material was subsequently dialyzed against 0.1% acetic acid and lyophilized. In addition, collagen from the skin of lathyritic rats was prepared by standard methods (12). Ascaris collagen was a gift from K. Sullivan (NIDR). Preparations of collagen substrates. The dried collagen was dissolved at 4 mg/ml in 0.1% acetic acid and stored frozen in small aliquots until used. Collagen substrates were prepared on tissue culture plates in different ways. In some cases collagen fibers were reconstituted from solution by exposure to NH,OH vapors for a few minutes at room temperature, producing a firm adherent gel (13). In other cases, an aliquot of the stock collagen solution was mixed with l/IO its volume of 2.0 M NaCl in 0.1 M TrisHCl, at 4°C and the pH was adjusted to 7.2 with a dilute NaOH solution. Then 0.5 ml portions were pipetted onto 35mm tissue culture dishes and left at room temperature or at 37°C for 30 min to gel (“Tris” gel). Substrates were also prepared by diluting
SUBSTRATES
309
1 ml of the stock collagen solution with 0.1 ml of 2.0 M NaCl in 0.1 M phosphate, pH 7.4, or with 2.0 M NaCl in 0.1 M HepesHCl,’ pH 7.4. Reconstitution of fibers was achieved by maintaining them at 24 or at 37°C for 15 to 60 minutes. The coated dishes were either used immediately after preparation or, in some cases, were air dried. Dried collagen substrates were also prepared from the rat tail collagen, lathyritic rat skin collagen, or Ascaris collagen. In these studies collagen was dissolved at 1 mg/ ml in 0.5 M acetic acid overnight with stirring at 4°C. Various amounts of collagen were also added to bacteriological petri dishes and air dried. Attachment assay. Dishes coated with the various substrates were incubated for 1 h at 24 or at 37°C in 95% air, 5% CO, with Eagle’s minimal essential medium containing 200 &ml bovine serum albumin (GIBCO) and bovine serum were indicated. Then lo5 Chinese hamster ovary (CHO) cells freshly isolated from culture with trypsin in 0.1 ml of the same buffer were added to each plate and incubated an additional 1.5 h. At the end of the incubation, the unattached cells were removed, combined with cells derived from three saline washes of the plate, and counted in a Coulter electronic cell counter. Where possible, the attached cells were released from the substrates by a 3- to 5-min incubation with 0.1% trypsin, 0.1% ethylenediamine tetraacetate (EDTA), in phosphate-buffered saline, and counted electronically. Fibronectin was purified from bovine serum (Colorado Serum Co.) using a collagen affinity column (7,14). RESULTS
Most cells did not attach to dried collagen substrates in the absence of serum, no matter how much collagen was on the dish (Fig. 1). Low serum (5%) levels stimulated the 1 Abbreviations used: Hepes, 4-(2-Hydroxyethyl)-lpiperazineethanesulfonic acid; CHO, Chinese hamster ovary.
310
KLEINMAN
attachment of cells to plates coated with low levels of collagen, but were not effective with higher levels of collagen (Fig. 1). However, greater amounts of serum (10 and 20%) permitted attachment to collagen over the whole range of collagen concentrations tested (Fig. I). The inhibition of attachment observed at higher levels of collagen could be due to the binding of serum fibronectin by collagen that leaches from the surface of the dish into the medium. Using radioactive collagen we found that up to 40% of the collagen was lost from the surface of a dish within 90 min after adding medium to the dish at 37°C. The attachment of cells to heat-reconstituted collagen substrates was also examined. Since collagen from rat tail tendon formed more stable and adherent gels than collagen from lathyritic rat skin, tendon collagen was used in these experiments. Furthermore, reconstituted collagen gels were found to adhere to tissue culture dishes better than to bacteriological dishes and therefore tissue culture dishes were used with these gels. The attachment of the CHO cells to dried collagen films, to gels formed by exposure of collagen solutions to NHIOH vapors, and to gels reconstituted by raising the temperature using Tris and Hepes buffers was similar (Figs. 2 and 3). Serum and purified fibronectin stimulated the attachment of cells to these substrates at both 24 and 37°C (Fig. 2). Identical binding was observed at 24 and 37°C. Collagen substrates reconstituted at 24 or 37°C showed an equal requirement of serum for cell attachment. The lower temperature should minimize the amount of denaturation occurring in the substrates. The amount of serum required for adhesion and spreading was proportional to the amount of collagen in the gel. Identical results were also obtained when the gels were allowed to air dry and then assayed for cell adhesion (not shown). The attachment and spreading of the cells were dependent upon cellular metabolism as al-
ET AL.
I
,
I
/
2.5
5.0
7.5
SERUM
I 10.0
(%)
FIG. 2. Effect of varying serum concentrations on cell attachment to small and large amounts of airdried collagen substrates and to reconstituted collagen. Rat tail collagen was air dried at low (W) and at high (0) concentrations and the reconstituted collagen (0) was prepared in the presence of Hepes buffer. The results shown represent the means of duplicate measurements where duplicates differed by less than 10%.
ready shown (15) because both were inhibited at 4°C and by N-ethyl maleimide. Different results were obtained on the substrates prepared with varying amounts of phosphate buffer. Cell attachment occurred well without serum (Fig. 3 and 4) and serum produced only minimal stimulation of attachment. Cells attached but did not spread out on the substrates prepared with 0.01 M phosphate (Fig. 3) or on substrates to which phosphate was added after reconstitution. The phosphate effect was not reversed if, after 1 h of incubation, the phosphate-containing medium was removed and replaced with medium lacking phosphate. Few of the cells that attached to the substrate in the presence of this level of phosphate were released after incubation with trypsin for up to 4 h. This is in contrast to cells attached with fibronectin to collagen or to cells cultured in medium. Here the cells became flattened on the substrate and when treated briefly (3 min) with trypsin detached from the substrate. Phosphate was also tested for its ability to promote attachment of cells to a collagen
FIG. 3. Attachment of cells to collagen gelled by four different procedures and to dried lathyritic collagen in the presence and absence of serum. Cells were allowed to adhere 1.5 h to a (a) collagen film without serum, (b) collagen film plus 30% serum, (c) NH,OH vapor-gelled collagen without serum, (d) NH,OH vapor-gelled collagen plus 30% serum, (e) phosphate buffer-gelled collagen without serum, (f) phosphate buffer-gelled collagen plus 30% serum, (g) Hepes buffer-gelled collagen without serum, (h) Hepes buffer-gelled collagen plus 30% serum, (i) Tris buffer-gelled collagen without serum, and (j) Tris buffer-gelled collagen plus 30% serum.
from Ascaris cuticle that does not bind fibronectin (16), to dried collagen substrates, and to bacteriological and tissue culture dishes. Phosphate stimulated the
attachment of cells to Ascaris collagen in the absence of serum but had no effect upon cell attachment in the presence of serum (Fig. 5). It stimulated the attachment of cells to dried collagen substrates in the absence of serum, but not to the same extent as seen with the gelled collagen substrate. Phosphate had no effect upon cell attach-
3 E 6
0
No Added
El
+lO+ M phosphate
Phosphate
Qitl 5
20 -
0
I 10-2
10 3 PHOSPHATE
CONCENTRATION
, 10’ IM)
FIG. 4. Attachment of cells to Hepes gelled collagen prepared in the presence of increasing amounts of phosphate buffer. Serum (30%) was present where indicated. Each bar represents the mean of duplicate measurements where duplicates differed by less than 10%.
,Ascaris
Typel,
NO SERUM
Ascaris +2%
Type
I
SERUM
FIG. 5. Attachment of cells to dried collagen (type I) from rat tail tendon and to Ascan’s collagen substrates in the presence and absence of serum and added phosphate. Each bar represents the mean of duplicate measurements where duplicates differed by less than 20%.
312
KLEINMAN
ment to or release from bacteriological tissue culture dishes.
or
DISCUSSION
Higher than physiological levels of phosphate are often used in the preparation of reconstituted collagen gels. However, such levels of phosphate permit cells to bind to collagen without fibronectin. Since the cells do not spread out and are not liberated by trypsin, we conclude that this adhesion is not physiological. In the absence of serum, high levels of phosphate also promote adherence to Ascaris collagen, a nonfibronectin binding collagen, but in the presence of serum, phosphate has no effect on cell adherence to this collagen. Phosphate may precipitate as calcium phosphate and the cells then bind to these deposits. Precipitates are seen on the surface of the dish and these are not removed when the medium is changed. Since changing the medium after incubation with phosphate does not remove the adhesion-promoting effect of phosphate, it is likely that these precipitates persist and cause this effect. We have shown that the attachment as well as spreading of cells both to dried collagen films and to reconstituted collagen gels require serum or fibronectin. More serum is required with the gelled collagen because greater amounts of collagen are present, and hence more is solubilized from the substrate and available to bind fibronectin. It is also possible that reconstituted collagen offers fewer binding sites for fibronectin than the collagen films since the majority of the molecules are at internal sites in the fibers. In the absence of serum, slightly more cells attached to collagen substrates reconstituted with NH,OH vapors or with heat using Tris or Hepes buffers than with airdried collagen films. This difference could be due to trapping of cells within the fiber mesh or may represent the adhesion of cells to collagen via a mechanism independent of fibronectin.
ET AL.
In summary, our results indicate that CHO cells require fibronectin to attach to collagen substrates. The amount of fibronectin required is proportional to the amount of collagen used as a substrate. Attachment occurs to a similar level with fibronectin on all substrates except those prepared with 0.01 M or higher levels of phosphate. Cells attach to collagen substrates treated with this level of phosphate in the absence of fibronectin. However, the cells do not flatten and are not liberated by trypsin, both characteristics of physiological cell attachment. Our data are consistent with the requirement of fibronectin for cell attachment to both native and denatured forms of collagen. REFERENCES 1. Klebe, R. J. (1974)Nuture (London) 2.50,248-251. 2. Pearlstein, E. (1976) Nature (London) 262, 497-499. 3. Kleinman, H. K., Murray, J. C., McGoodwin, E. B., and Martin, G. R. (1978)J. Invest. Dermard. 71,9-11. 4. Kleinman, H. K., McGoodwin, E. B., and Klebe, R. J. (1976) Biochem. Biophys. Res. Commun. 72, 426-432. 5. Dessau, W., Adelmann, B. C., Timpl, R., and Martin, G. R. (1978) Biochem. J. 169, 55-59. E. B., Martin, 6. Kleinman, H. K., McGoodwin, G. R., Klebe, R. J., Fietzek, P. P., and Woolley, D. E. (1978)J. Biol. Chem. 253, 5642-5646. 7. Hopper, K. E., Adelmann, B. C. Genter, G., and Gay, S. (1976) Immunology 30, 249-259. 8. Ruoslahti, E., and Engvall, E. (1978) Ann. N. Y. Acad. Sci. 312, 178-191. 9. Jilek, R., and Hbrmann, H. (1978) HoppeSeyler’s
2. Physiol.
Chem.
459, 247-250.
10. Rubin, K., Kjellen, L., and iibrink, B. (1977) Exp. Cell Res. 109, 413-422. 11. Grinnell, F., and Minter, D. (1978) Ann. N. Y. Acad. Sci. 312, 434-435. 12. Bomstein, P., and Piez, K. A. (1%6) Biochemistry
5, 3460-3473.
13. Elsdale, T., and Bard, J. (1972) J. Cell Biol. 54, 626-637. 14. Engvall, E., and Ruoslahti, E. (1977) In?. J. Cancer 20, 1- 17. 15. Klebe, R. J. (1975) J. Cell. Physiol. 86, 231-236. 16. Kleinman, H. K., Murray, J. C., McGoodwin, E. B., Martin, G. R. and Binderman, I. (1978) Calcif. Tissue Abst. Suppi. pp. 61-72.
