EXPERIMENTAL
CELL
RESEARCH
167-174 (1991)
193,
Collagen Gel Contraction by Fibroblasts Requires Cellular Fibronectin but Not Plasma Fibronectin HIROAKI
ASAGA, SHIRO KIKUCHI,*
AND KATSUTOSHI
YOSHIZATO?’
Developmental Biology Laboratory, Department of Biology, Faculty of Science, Tokyo Metropolitan University, Setagaya-ku, Tokyo 158, Japan; *Department of Surgery, Kitasato IJniuersity School of Medicine, Sagamihara, Kanagawa 228, Japan; and tMolecular Cell Science Laboratory, Zoological Institute, Faculty of Science, Hiroshima liniuersity, Naka-ku, Hiroshima 730, Japan
this culture contract collagen gels, resulting in hydration and a marked reduction of volume of gels [4]. Fibroblasts remodel and reorganize collagen fibrils into a dermal-like tissue during this process [5-91. Previous reports have shown that contraction of collagen gels is dependent on number of cells [4, 10-121, concentration of serum [4,11,12], and protein synthesis of cells [ll, 121. It is easily considered that binding of fibroblasts to collagen fibrils is a key step in the initial events of the contraction. Actually cell surface receptors for collagen were identified and characterized [ 13-211. However, a mechanism of binding of fibroblasts to collagen has been largely unknown in the process of gel contraction. Recently Gullberg et al. [9] reported a critical role of Pi-integrin collagen receptors in collagen gel contraction. Some investigators including the authors cited above suggested that the generally accepted mode of binding of fibroblasts to collagen fibers through fibronectin (FN) does not work during collagen gel contraction [9, 12, 22, 231. Others presented the evidence that fibronectin-mediated binding of fibroblasts to collagen fibrils is essential to ensuing gel contraction [ll]. We have applied a variety of techniques to elucidate the mechanism of contraction of collagen gels by human fibroblasts. One of the most successful trials was to have obtained monoclonal antibodies (mAbs) which inhibit contraction of collagen gels. Utilizing these mAbs we could demonstrate involvement of cellular FN (cFN) in fibroblast-mediated collagen gel contraction, but not bovine serum FN (sFN) or human plasma FN (pFN) in culture medium. To our knowledge this is the first demonstration of a functional difference between cFN and pFN and of vital roles of cFN in the process of collagen gel contraction.
Fibroblasts embedded in three-dimensional lattices of collagen fibrils have been known to require serum constituents to induce a cell-mediated contraction of collagen gels. The gel contraction was studied with human skin fibroblasts cultured in the presence of fetal bovine serum (FBS). Removal of bovine serum fibronectin (sFN) from FBS did not affect the extent of gel contraction. Gel contraction occurred in serum-free defined media. Therefore, it is concluded that sFN is not required for gel contraction. That cellular FN (cFN) synthesized and secreted by fibroblasts plays a crucial role in gel contraction was suggested by the following experiments: (1) We obtained monoclonal antibodies (mAb A3A5) against fibroblast surface antigens, which suppressed the fibroblast-mediated gel contraction. Immunoblot analyses showed that mAb A3A5 recognizes cFN secreted by human fibroblasts and human plasma FN (pFN), but not bovine sFN in FBS used for culture. (2) Addition of rabbit antisera, which recognize human cFN, to a serum-free gel culture inhibited contraction. Uninvolvement of human pFN in gel contraction was further confirmed by the fact that neither pretreatment of fibroblasts with excess amounts of human pFN nor the presence of excess amounts of human pFN in gels affected the extent of gel contraction. This study seems to be the first demonstration of functional difference between cFN and pFN (or sFN) and proposes a novel mode of binding of fibroblasts with collagen fibrils via cFN during cell-mediated collagen morphogenesis. 8~ 1991
Academic
Press,
Inc.
INTRODUCTION
A culture of fibroblasts in three-dimensional collagen gels (collagen gel culture) originally devised by Elsdale and Bard [l] has been shown to provide a unique model for studying cell to collagen interactions: an in vitro model of collagen morphogenesis [2, 31. Fibroblasts in ’ To whom correspondence dressed.
and
reprint
requests
should
MATERIALS
AND
METHODS
Chemicals, reagents, and culture materials. These were obtained as follows: Dulbecco’s modified Eagle’s medium (DMEM) and RPM1 1640 medium from GIBCO (Grand Island, NY) or Kyokuto Pharmaceutical Industrial Co., Ltd. (Tokyo); fetal bovine serum (FBS) from GIBCO or Japan Biotest Laboratory (Tokyo); ethylenediamine-
be ad-
167 All
Copyright Ccl 1991 rights of reproduction
0014.4827/91 $3.00 by Academic Press, Inc. in any form reserved.
168
ASAGA,
KIKUCHI,
tetraacetic acid (EDTA) and N-(2.hydroxymethyl)piperazine-N-2ethanesulfonic acid (Hepes) from Dojin Chemical Institute (Kumamoto, Japan); streptomycin and penicillin from Meiji Seika Co. (Tokyo); tris(hydroxymethy1) aminomethane (Tris), cycloheximide, trypsin, insulin, glucagon, prolactin, transferrin, hydrocortisone acetate, 5-a-dihydrotestosterone, dexamethasone, p-estradiol, 3,3’,5triiodo-L-thyronine (T,), and prostagrandine F,, from Sigma Chemical Co. (St. Louis, MO); growth hormone from IJCB-Bioproducts (Bruxelles); Sepharose 4B and gelatin-Sepharose 4B from Pharmacia Fine Chemicals (Uppsala); polyethylene glycol 4000 from Merck (Darmstadt); Vectastain ABC kits from Vector Laboratories (Burlingam, CA); Falcon culture dishes and 96.well culture plates from Becton Dickinson Labware (Oxnard, CA); Freund’s complete adjuvant from Difco (Detroit); Millipore filters and nitrocellulose papers from Millipore Japan (Yonezawa, Japan); Centriconfrom W. R. Grace and Co. (Danvers, MA); Athtystar, prepacked affinity columns of protein A, from Kurabo Industries (Osaka); human plasma fibronectin from Hoechst Japan (Tokyo). All other chemicals were of reagent grade and purchased from Wako Pure Chemical Industries (Osaka) or Nacalai Tesque, Inc. (Kyoto). Preparation of FN-free FBS. FN was eliminated from FBS by affinity chromatography on gelatinSepharose 4B according to Ruoslahti et al. [24]. Briefly, Hanks’ balanced salt solution (Hanks’ BSS) was used as a coupling buffer and FBS was passed through the column and collected in glass tubes using a fraction collector. The eluate was concentrated by centrifugation at 3000g for 40 min using Centricon-10 and Hanks’ BSS was added to equalize the final concentration with the original one. Sepharose 4B was substituted for gelatin Sepharose 4B to obtain FBS for a control experiment. Complete elimination of FN was confirmed by polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate (SDSPAGE) [25] and enzyme-linked immunosorbent assay (ELISA) using rabbit antiserum against pFN raised in our laboratory as described below and Vectastain ABC kits. Cell cultures. Human skin fibroblasts were obtained from explants of normal dermis excised as a biopsy at a surgical operation and cultured as described previously [26]. Fibroblasts were grown and maintained in 75.cm* plastic dishes in 15 ml of DMEM containing 10% FBS, 10 mM NaHCO,, 20 mM Hepes, 100 units/ml penicillin, and 100 pg/ml streptomycin in a moist atmosphere of 5% CO,/95% air at 37°C. The population doubling level of cells used in the present study was within 17. Cells were detached from dishes with calciumand magnesium-free Hanks’ BSS (CMF-Hanks’) containing 0.1% trypsin and 1 mM EDTA. Then they were collected by centrifugation at 500g for 5 min and used for experiments after washing with Hanks’ BSS twice. Collagen gel culture.s of fibroblasts. Collagen (Type I) was extracted from calf dermis by digestion with pepsin in a dilute HCI solution, purified, and characterized as described previously [7]. Fibroblasts were grown three-dimensionally in hydrated collagen gels by the method of Elsdale and Bard [l] with slight modifications. The following stock solutions were prepared and kept at 4°C: 0.5% (w/v) collagen solution; 4X concentrated DMEM which contained 80 mM Hepes, 40 mM NaHCO,, 0.4 mg/ml streptomycin, and 400 units/ ml penicillin. Cells were harvested from monolayer cultures by using 0.1% trypsin and 1 mM EDTA in CMF-Hanks’, counted and adjusted to desired cell number, and collected by centrifugation in a plastic tube. The cell pellet was resuspended at 4°C in DMEM containing 0.1 or 0.2% collagen, 20 mM Hepes, 10 mM NaHCO,, 0.1 mg/ml streptomycin, 100 units/ml penicillin, and 10% FBS, which had been prepared by quickly mixing the stock solutions, FBS, and redistilled water. A defined medium specified below was used instead of medium containing 10% FBS in the experiment shown in Fig. 1. The pH was adjusted to 7.4 by adding 1 N NaOH. Media containing cells and collagen were placed in bacteriological dishes with a diameter of 35 or 25 mm: 2 ml of medium containing 5.0 X lo5 fibroblasts and 1 ml of medium containing 2.0 X lo5 cells were placed in 35- and 25.mm
AND
YOSHIZATO
dishes, respectively. The dishes were transferred into an incubator at 37°C. Collagen was gelated within 10 min and cells were embedded in the gel. For long-term culture, gels were overlaid with 1.5 ml of fresh medium and media were changed twice a week. When necessary, FNfree FBS, antiserum, IgG, cycloheximide, or human pFN was introduced into the gel culture. The serum-free defined medium contained the following ingredients in DMEM with concentrations indicated: 0.2% collagen, 20 mM Hepes, 10 mMNaHCO,, 100 Kg/ml streptomycin, 100 units/ml penicillin, 10 pg/ml insulin, 10 bg/ml glucagon, 50 rig/ml epidermal growth factor (EGF), 100 rig/ml growth hormone (GH), 0.02 units/ml prolactin (PL), 0.5 Fg/ml transferrin, 100 rig/ml hydrocortisone acetate, 20 rig/ml 5-tu-dihydrotestosterone, 10 pg/ml @-estradiol, 200 rig/ml dexamethasone, 10m9MT,, 710 rig/ml prostagrandine F,,, , 100 mM CuSO,.SH,O, 3 nM H,SeO,, 5 nM ZnSO,. 7H,O. lktermination of fhe c&e& of& contraction. The extent of contraction was quantified during culture by measuring the diameter of gels or by measuring the thickness of gels as described by Guidry and Grinnell [12, 221. In the latter case, thickness of the center of gels was measured by a phase contrast microscope equipped with a dial test indicator. There were not significant differences in rate of gel contraction as determined with the two methods. Production of antibodies. Monoclonal antibodies were produced by the method of Galfre and Milstain [27] with slight modifications. Eight-week-old BALB/c mice were immunized three times with fibroblasts (7 X lo7 cells/mouse) which had been mechanically harvested from culture dishes with rubber policemen and emulsified in Freund’s complete adjuvant or suspended in PBS. Immune spleens were excised, and spleen cells were fused with myeloma cells (P3X63Ag8.653) in the presence of 5O”‘r (w/v) polyethylene glycol. Viable hybridomas were selected in RPM1 1640 supplemented with hypoxantine, aminopterin, thymidine, and 10% FBS. Hybridomas were screened for t,he production of antibodies through the following three assays in this sequence: Assay 1: Hybridomas producing antibodies that react on the cell surface of fibroblasts were identified by the method of ELISA. Fihrohlasts were grown in 96.well culture plates, washed with Hanks’ BSS, blocked with 2% normal horse serum, and then treated with culture supernatant from the hybridomas, washed again, and detected using Vectastain ABC kits. Assay 2: Antibodies that alter spreading of fibroblasts cultured on the collagen fihrils were selected. Trypsinized fihroblasts were washed two times with Hanks’ BSS and suspended in culture supernatant from hyhridomas. Aliquots of the suspension were poured into dishes coated with collagen-fihrils (100 fig collagen/cm’) and cultured. After 6 or 12 h, morphology of fibroblasts was observed by a phase contrast microscopy. Assay 3: Hybridomas producing antihodies which inhibit fibroblast-mediated collagen gel contraction were selected as follows. One-fifth million fihroblasts were preincubated at 10°C for 1 h in 1 ml of culture supernatants of hybridomas and then were cultured in 1 ml of gels containing 0.1% (w/v) collagen, 10% FBS, and 70.3% culture supernatant of hyhridomas. Extent of contraction was measured during culture. Rabbit antisera against human pFN were obtained by an ordinary procedure. Antisera and normal rabbit sera were incubated at 56°C for 30 min in order to inactivate complement systems prior to use in cell culture experiments. Electrophoresis and immunoblotting. SDS-PAGE was performed by the method of Laemmli [25] with modifications. Fibroblasts harvested by rubber policemen were washed with CMF-Hanks’, homogenized with SDS sample buffer (10% SDS, 5% 2-mercaptoethanol, 62.5 mM TrisHCl, pH 6.8), boiled for 5 min, and subjected to SDSPAGE. Liquid samples (FBS or culture supernatants of fibroblasts) were mixed with an equal volume of 2~ concentrated SDS sample buffer, boiled for 5 min, and subjected to SDS-PAGE. SDSPAGE was carried out with a stacking gel of 4% polyacrylamide and a separation gel of 7.5-20% polyacrylamide linear gradient or 7.5% polyacrylamide. For the preparation of stacking gels, lower gel buffer (0.4%
CELL
TO COLLAGEN
SDS and 1.5 M Tris-HCI, pH 8.8) was used instead of upper gel buffer (0.4% SDS and 0.5 M TrissHCl, pH 6.8). After electrophoresis, polypeptides in the get were stained with Coomassie brilliant biue R-250 or electrophoretically transferred to nitracellulose papers by the method of Towbin et al. [28]. Immunochemical detect,ions of polypeptides on papers with mAb or antiserum were carried out by use of Vectastain ABC kits. Immunocytochemical staining of fibroblasts. Fibroblasts were cultured on plain or cobagen-fibril-coated plastic dishes which had been prepared according to Yoshizato et al. 171. Immunocyt,ochemical stainings of fibroblasts were carried out at 3’7°C without fixation using Vectastain ABC kits. Fibroblasts were preincubated with 2% bovine serum albumin (BSA) in Hanks’ BSS for 10 min to block any remaining protein binding sites and incubated for 30 min with culture supernatants of hybridomas. After washing with Hanks’ BSS, they were incubated for 10 min with biotinyIated anti-mouse IgG diluted by Hanks’ BSS containing 1% BSA. After washing with Hanks’ BSS, they were t,reated for 10 min wit,h Vectastain ABC reagents prepared in Hanks’ BSS, washed five times with Hanks’ BSS, and incubated in Hanks’ BSS containing 0.04 ?& H,O, and 0.5 mg/ml 4-chloro-lnaphthol.
L
01 0
RESIJLTS Factors Affecting
Fibroblasts were cultured in collagen gels in a serumfree defined medium containing nutrients, hormones, and growth factors in order to determine which subst,ances are essential for fibroblasts to contract collagen gels. Fibroblasts thus cultured did contract collagen gels although the extent of contraction was less than that in media supplemented with FBS (Fig. 1). Insulin was essential to fibroblast-mediated collagen gel contraction. Fibroblasts do not appear to require the presence of sFN or pFN to interact with collagen fibrils because neither sFN nor pFN was included in the defined medium. This observation should be important because some reported the involvement of sFN or pFN in fibroblast-mediated collagen gel contraction [ll] and others presented the evidence that pFN is not required [E!, 231. Therefore, further experiments were carried out to clarify the role of FN in collagen gel cont,raction. in the Absence of Bovine sFN
Half a million cells were cultured in 2 ml of 0.2% collagen gels containing 10% FBS or sFN-free FBS which had been prepared by passing FBS through a column of gelatin-Sepharose 4B. The absence of bovine sFN did not affect the extent of gel contraction at all (data not shown), supporting the results shown in Fig. 1. Monoe~onal Antibodies
Which Inhibit
2
4 Days
Collagen Gel ~o~tr~ct~on
Collagen Gel Contraction
169
INTERACTIONS
Gel Contraction
In order to analyze the mechanism of fibroblast-mediated collagen gel contraction in more detail, we tried to obtain mAbs which inhibit gel contraction. Ninety-eight hybridoma cell lines which produced the antibodies reacting on fibroblast cell surfaces were obtained. Two clones (A3A5 and BlA4) of them produced
6
8 in
10 Culture
12
14
FIG. 1. Contraction of collagen gels in defined media. Fibroblasts (5 X 10’) were cultured in 2 ml of 0.2% collagen gels containing defined medium. Concentrations of ingredients of the defined medium are described under Materials and Methods. At Day 4, 2 ml of media was overlaid on gels and cultures were maintained up to Day 14 with a medium change every fourth day. (0) DMEM with 10% FBS; (r) defined medium without PI, and GH; (8) defined medium; (A) defined medium without glucagon, EGF, PL, and GH; (0) defined medium without EGF, insulin, PL, and GH; {a) defined medium without glucagon, insulin, PL, and GH; (0) defined medium without glucagon, EGF, insulin, PL, and GH. Each point represents a single determination.
antibodies which suppressed gel contraction. The clone A3A5 was further characterized. IgGs purified from culture supernatants of the mAb A3A5 showed a dose-dependent inhibition of gel contraction (Fig. 2). When fibroblasts were cultured in collagen gels, cells spread three-dimensionally prior to contraction of gels. However, most fibroblasts in a gel remained round in shape in the presence of supernatants of A3A5 (Fig. 3). It is noteworthy that mAb A3A5 had subtle effects on spreading of fibroblasts cultured on plain or collagen-fibril-coated plast,ic as compared to that of cells cultured within gels (Fig. 4). Fibroblasts were cultured on plain plastic or collagenfibril-coated dishes for 3 days and were immunochemitally stained with mAb A3A5. The surface of fibroblasts and boundaries between fibroblasts on plain plastic were intensely stained with the antibody (Fig. 5A). However, mAb A3A5 could not stain fibroblasts cultured on collagen fibrils (Fig. 50. This unresponsiveness of mAb A3A5 was not due to the absence of antigens recognized by mAb A3A5 on fibroblast surface membrane, because fibroblasts cultured on collagen fibrils contained a protein band which reacts on mAb A3A5 (Fig. 6C). The chemical nature of ant.igens correspon~ng to
170
6
m =
ASAGA,
KIKUCHI,
5
s
t
,
0’ 0
5
10 Hours
I 15 in
20 Culture
I
I
25
30
FIG. 2. Dose-dependent inhibition of collagen gel contractions by mAb A3A5. Fibroblasts (2 X 105) were cultured in 1 ml of 0.1% collagen gels containing DMEM, 10% FBS, and different concentrations of IgG of mAb A3A.5. (0) Without IgG, (0) 20 pg/ml, (0) 50 eglml, (A) 100 fig/ml. Each point represents a single determination.
mAb A3A5 was examined by Western blot analyses. When proteins extracted from fibroblasts were subjected to immunoblotting with mAb A3A5, a single band was detected (Fig. 6A). Molecular weights calculated from electrophoretic mobility of the band were about 230 kDa which is the same as that of FN. Actually, A3A5 reacted with human pFN (lane pFN of A3A5 in Fig. 6A). Moreover, A3A5 reacted on a band corresponding to FN secreted by human fibroblasts but not on any kind of proteins in FBS which was included in culture media (Fig. 6B). If cFN is defined as FN which is synthesized and secreted by fibroblasts, these results indicate that A3A5 recognizes FN of human origin (both cFN and pFN) but not bovine serum FN in culture medium. Results of Fig. 1 through 6 collectively indicate uninvolvement of bovine sFN and involvement of human cFN in collagen gel contraction by human fibroblasts. This statement was further verified in the following experiments. Critical
AND
YOSHIZA’I’O
was dose-dependently inhibited by the antiserum (Fig. 7), clearly showing involvement of human cFN in gel contraction, because the antiserum can recognize human cFN of fibroblasts as shown in Fig. 6A. Figure 7 also suggests uninvolvement of rabbit sFN in the contraction process, because there were no noticeable differences in the extent of gel contraction at different concentrations of rabbit serum from 10 through 50%. When IgGs purified from rabbit anti-human pFN antiserum were introduced into collagen gel culture at concentrations of more than 1 mg/ml, gel contraction was inhibited (data not shown), again confirming involvement of human cFN in gel contraction. Considering the results described above together, we postulate that human cFN is an inevitable protein for human fibroblasts to bind collagen fibrils in collagen gel culture. During culture, fibroblasts are expected to actively synthesize cFN, which can mediate interactions between fibroblasts and collagen fibrils. Actually, we measured that fibroblasts in serum-free DMEM synthesize and release cFN into media at 0.1 rig/cell per day. When synthesis of proteins including cFN was inhibited by cycloheximide, collagen gel contraction was inhibited in a reversible and concentration-dependent manner (data not shown). However, the possibility that human pFN substitutes
Roles of cFN in Gel Contraction
Rabbit antiserum against human pFN was prepared. This preparation reacts on human cFN as well as human pFN (arzti-FN, lanes F and pFN in Fig. 6A). Collagen gel cultures in DMEM without FBS were prepared and were given the rabbit antisera at concentrations of 10, 20, and 50%. The gel culture thus obtained was to contain as FN rabbit sFN from the antisera and human cFN from fibroblasts and was to be supplied with serum factors necessary for the gel contraction such as insulin by the antisera. The antiserum did not show cytotoxicity for fibroblasts at these concentrations. Contraction
FIG. 3. Photomicrographs of fibroblasts cultured within collagen gels in the presence of supernatants of hybridoma cultures. Fibroblasts were suspended in 1 ml of supernatants of hybridomas, incubated at 10°C for 1 h, and then cultured in 1 ml of 0.1% collagen gels containing 10% FBS and 70.3% culture supernatants of hybridomas. Photographs were taken at 3 h in culture. (A) A3A5, (B) myeloma. Magnification: X100.
171
FIG. 4. Photomicrographs of fibroblasts cultured on plain plastic or collagen-fibril-coated plastic in the presence ofmAb. Fibroblasts were treated as described in the legend to Fig. 3 and seeded in tissue culture dishes. (A and 13) On plain dishes, (C and D) on dishes coated with collagen-fibrils at a density of 100 pglcm’. (A and C) With mAb A3A5, (B and D) with supernatant of myeloma cult,ures as controls. Photographs were t.aken at 6 h in culture. Magnificati(~n: ~100.
for human cFN cannot be ruled out, because mAb A3A5 recognizes both human cFN and pFN. Therefore, three further experiments were carried out to test this possibility. Uninvolvement Contraction
of Human
pFN in Collagen Gel
Fibroblasts were preincubated in a medium containing human pFN at the concentrations of 1 and 2 mg/ml
at 4°C for 1 h, rinsed, and cultured on collagen gels containing sFN-free FBS instead of FBS. These concentrations are far beyond the plasma concentration of FN (0.33 mg/ml) [29] and are expected to be enough for competing cFN in binding to hypothetical common receptors for pFN and cFN on the cell surface of fibroblasts. The extent of contraction was the same as those of control experiments without human pFN (data not shown). Moreover, even when fibroblasts preincubated
FIG. 5. Immunocytochemical stains of fibroblasts with mAb. Fibroblasts (2 X 105) were cultured for 3 days in 2 ml of DMEM supplemented with 10% FBS in plain 35-mm cult,ure dishes (A and B) or in dishes coated with collagen fibrils as described in the legend to Fig. 4 (C and D). Cells were then immunostained with mAb A3A5 (A and C) or supernatants of myeloma culture (B and D). Magnification: X100.
172
ASAGA, SDS-PAGE A
F pFN
anti-FN
A3A5 F pFN
F pFN
#.
a.
-.
KIKUCHI,
AND
YOSHIZATO
The addition of human pFN to the collagen gel culture with cycloheximide did not rescue the contractioninducing ability of fibroblasts (data not shown), again supporting the notions that human pFN is not involved in gel contraction and human pFN cannot substitute for human cFN.
-..
94K 68K 43K 30K
DISCUSSION 14.4K
SDS-PAGE c
8QWcY
A3A5 Qf Qv 0” -
me,---.
94K 68K -
FIG. 6. Western blot analyses with mAb A3A5 or rabbit anti-human pFN antiserum. (A) Proteins of fibroblasts. Proteins extracted from fibroblasts (F) and purified human pFN (pFN) were subjected to SDS-PAGE (left) and immunoblotted with mAh A3A.5 (middle) or rabbit anti-human pFN antiserum (right). Eight micrograms of fibroblast proteins and 1 fig of pFN were analyzed. (B) Proteins in FBS and culture medium of fibroblasts. FBS, culture supernatants of fibroblasts (SUP), and human pFN (pFN) were subjected to SDS-PAGE (left) and immunoblotted with mAb A3A5 (right). Ten micrograms of FBS and SUP and 2 pg of pFN were analyzed. (C) Proteins of fibroblasts cultured on collagen fibrils. Fibroblasts (:i X IO5 cells) were cultured for 3 days on 35-mm plastic (PL) or collagen fibrils (CL) coated on 35-mm plastic, washed extensively with PBS, and solublized with SDS sample buffer. Proteins from 3 x 104 cells were subjected to SDS-PAGE and Western blotting with mAb A3A.5. Arrowheads indicate the bands immunochemically detected with mAb A3A5. Arabic numerals with K at the left side indicate molerular weights determined by marker proteins.
as described above were cultured on collagen gels containing excess amounts of human pFN (1 or 2 mg/ml at a final concentration), the extent of gel contraction was identical to those of control experiments without human pFN (data not shown). If human pFN has some role in mediating binding between fibroblasts and collagen fibrils and pFN and cFN have common binding sites on collagen fibrils, the gel contraction was expected to be suppressed in these experiments due to competitive binding of pFN. Or even a stimulatory effect of pFN was expected if human pFN has some role in mediating a binding between fibroblasts and collagen fibrils through binding sites for pFN but not cFN. Therefore, these results clearly indicate that human pFN is not involved in collagen gel contraction by human fibroblasts.
The contraction of collagen gels by fibroblasts was first reported by Bell et al. [4] and is a phenomenon useful in the study of cell to collagen interactions (collagen morphogenesis) [2, 31. The mechanism of fibroblast-mediated collagen gel contraction has been poorly understood and is still controversial. We performed several experiments to study the mechanism of fibroblastmediated collagen gel contraction, focusing on the relationship among the cell surface of fibroblasts, collagen, and FN. On the basis of this study, we present a mechanism of an indirect interact.ion between fibroblasts and collagen in the process of gel contraction via cFN synthesized and secreted by fibroblasts. Neither sFN nor pFN in culture medium plays any role in this mechanism. Fibroblast-mediated collagen gel contraction occurs in serum-free defined medium which does not contain sFN or pFN and removal of bovine sFN from FBS does not affect the extent of gel contraction. These facts indicate that bovine sFN does not play any roles for collagen gel contraction.
P f 0
5
0
I.-.0
5
10 Hours
15 20 in Culture
25
30
FIG. 7. Inhibition of contraction of collagen gels by rabbit antihuman pFN antisera. Fibroblasts (2 X 105) were cultured in 1 ml of 0.1% collagen gels containing DMEM and antisera at concentrations indicated. Normal rabbit sera were used as controls. (0 and 0) lo%, (0 and H) 20%, (A and A) 50%. Open and closed symbols indicate antisera and normal sera, respectively. Each point represents the average of two determinations, whose range was less than 5% of the mean.
CELL
TO COLLAGEN
We obtained mAb A3A5 which inhibits collagen gel contraction. Monoclonal antibody A3A5 recognizes human cFN secreted from flbroblasts and human pFN, but not bovine sFN in FBS. It is important for the discussion of the mechanism of collagen gel contraction that mAb A3A5 reacts on human cFN from fibroblasts but not bovine sFN in culture medium. There exists human cFN secreted by fibroblasts and bovine sFN in culture media of the experiment presented in Fig. 2, but not other kinds of FN. In this gel culture, mAb A3A5 specific to human FN fibroblasts but not bovine sFN has a critical role for collagen gel contraction. Since rabbit anti-human pFN antisera contain rabbit sFN and polyclonal antibodies which react on human cFN but not rabbit sFN, inhibition of gel contraption by the antisera also indicates critical roles of human cFN and that rabbit sFN cannot substitute for human cFN. Pretreatment of fibroblasts with excess amounts of human pFN does not affect the extent of gel contraction. Moreover, the presence of excess amounts of human pFN in a gel has no effect on contraction. These results indicate uninvolvement of human pFN in the process of human fibroblast-mediated collagen gel contraction. In other words, human pFN cannot substitute for human cFN. This notion is also supported from the fact presented in the present paper that human pFN cannot restore the contraction-inducing ability of fibroblasts in the presence of cycloheximide. Other investigators have also reported that bovine sFN and human pFN are not required for collagen gel contraction by human fibroblasts [ 12,231. On the other hand, Gillery et al. [ll] presented evidence suggesting an involvement of bovine sFN. Schafer et al. [23] reported that polyclonal antibodies against human pFN did not inhibit collagen gel contraction by human fibroblasts, which is contrary to the result of Fig. 7 of the present paper. In the experiments of Schafer et al. 1231, they tested the effect of t.he antibodies at a concentration of 200 @g/ml or less. In the present study, the inhibition of collagen gel contraction by anti-human pFN IgGs was observed at concentrations of more than 1 mg/ml. Therefore, the difference in the result between the present study and theirs might be due to the amounts of antibodies used in the experiments. However, the apparent discrepancy between the present study and that of Gillery et al. [ll] cannot, be reasonably explained at present. It is known that fibroblasts of established cell lines or transformed fibroblasts are less able to contract collagen gels than normal fibroblasts obtained in primary culture [lo, 30, 311. This fact can be explained by the present study, because it is generally accepted that established or transformed cells secrete a smaller amount of cFN than the corresponding normal cells [32,33]. The present study emphasizes the difference between cFN and pFN in mediating binding between fibroblasts
INTERACTIONS
173
and collagen fibrils which is the first step for ensuing contraction of collagen gels. Cellular FN and pFN share many common features in structure and function, because both FNs originate from the common gene 1341. However, some differences have been also known. Cellular FN is thought to exist as polymeric forms [35,36]. Molecular weights of A and B chains of cFN are higher in about 10 kDa than those of pFN (37,381. This difference in molecular weights is due to a different splicing of mRNA of FN [34]. Cellular FN contains regions of EDA and ED-B which lack in pFN [39]. At present it is not known which of these differences is responsible for the functional difference between cFN and pFN observed in gel contraction. To obtain an answer for this it is crucial to know what region mAb A3A5 recognizes on molecules of FN. Such investigations are now being undertaken by analyzing affinities of proteolytic fragments of FN for mAb A3A5. Another possibility is that mAb A3A5 binds two epitopes, one nonfunctional present in both cFN and pFN and another functional present in only cFN. The effects of mAb A3A5 on cell morphology are manifested more intensely for cells in collagen gels than those on plain plastic or collagen-fibril-coated plastic. There might be some difference in the mechanism of cell to collagen interaction between three-dimensional culture within collagen gels and conventional two-dimensional culture. Recently, Grinnell et al. 1401 studied effects of peptides with the cell recognition sequence on morphology of cells cultured on collagen-coated surfaces or inside collagen gels and noticed that there are some critical differences between the mechanism of three-dimensional cell to collagen recognition and that of two-dimensional recognition. Monoclonal antibody A3A5 could not stain fibroblasts on collagen-fibril-coated plastic, while the mAb A3A5 could stain those on plain plastic. The immunological unresponsiveness was not due to the absence of antigens on the surface of cells on collagen fibrils. One plausible explanation for this may be that mAb A3A5 recognizes some specific region of cFN which is important for cell to collagen interactions. This region of cFN had been occupied by collagen when mAb A3A5 was added to the assay cultures. Considering these together, we postulate that mAb A3A5 recognizes the collagen binding region, but not the cell binding region, of FN molecules. The present study demonstrates that insulin is important as a factor for the fibroblast-mediated collagen gel contraction. There have been reports indicating that transforming growth factor /3 (TGF~) [S] and plateletderived growth factor (PDGF) [9] stimulate collagen gel contraction by fibroblasts. The function of insulin and these growth factors in the gel contraction remains to be clarified. In relation to the present study, it should be noted that TGFP stimulates fibroblasts to produce cFN [41] and PDGF increases the ability of cells to adhere to
174
ASAGA,
KIKUCHI,
FN [42]. Insulin also may have actions similar to TGFP or PDGF. FBS should contain another unknown factor(s) necessary to collagen gel contraction, because contraction of gels containing FBS proceeds faster and results in a more marked contraction than that in defined medium. We thank Dr. R. Matsuda for his technical cooperation on production of mAbs. We also thank Mr. T. Taira and Mr. K. Tokuda for their cooperation during this study. This work was supported in part by a grant-in-aid for special project research from the Ministry of Education, Science and Culture of Japan (59105003), by a grant-inaid for Developmental Scientific Research (08170034), and by a grant from TERUMO Life Science Foundation (8907).
REFERENCES Elsdale, T., and Bard, J. (1977) J. Cell Biol. 54, 626-637. Harris, A. K., Stopak, D., and Wild, P. (1981) Nature (London) 290,249-251.
CELL
RESEARCH
167-174 (1991)
193,
Collagen Gel Contraction by Fibroblasts Requires Cellular Fibronectin but Not Plasma Fibronectin HIROAKI
ASAGA, SHIRO KIKUCHI,*
AND KATSUTOSHI
YOSHIZATO?’
Developmental Biology Laboratory, Department of Biology, Faculty of Science, Tokyo Metropolitan University, Setagaya-ku, Tokyo 158, Japan; *Department of Surgery, Kitasato IJniuersity School of Medicine, Sagamihara, Kanagawa 228, Japan; and tMolecular Cell Science Laboratory, Zoological Institute, Faculty of Science, Hiroshima liniuersity, Naka-ku, Hiroshima 730, Japan
this culture contract collagen gels, resulting in hydration and a marked reduction of volume of gels [4]. Fibroblasts remodel and reorganize collagen fibrils into a dermal-like tissue during this process [5-91. Previous reports have shown that contraction of collagen gels is dependent on number of cells [4, 10-121, concentration of serum [4,11,12], and protein synthesis of cells [ll, 121. It is easily considered that binding of fibroblasts to collagen fibrils is a key step in the initial events of the contraction. Actually cell surface receptors for collagen were identified and characterized [ 13-211. However, a mechanism of binding of fibroblasts to collagen has been largely unknown in the process of gel contraction. Recently Gullberg et al. [9] reported a critical role of Pi-integrin collagen receptors in collagen gel contraction. Some investigators including the authors cited above suggested that the generally accepted mode of binding of fibroblasts to collagen fibers through fibronectin (FN) does not work during collagen gel contraction [9, 12, 22, 231. Others presented the evidence that fibronectin-mediated binding of fibroblasts to collagen fibrils is essential to ensuing gel contraction [ll]. We have applied a variety of techniques to elucidate the mechanism of contraction of collagen gels by human fibroblasts. One of the most successful trials was to have obtained monoclonal antibodies (mAbs) which inhibit contraction of collagen gels. Utilizing these mAbs we could demonstrate involvement of cellular FN (cFN) in fibroblast-mediated collagen gel contraction, but not bovine serum FN (sFN) or human plasma FN (pFN) in culture medium. To our knowledge this is the first demonstration of a functional difference between cFN and pFN and of vital roles of cFN in the process of collagen gel contraction.
Fibroblasts embedded in three-dimensional lattices of collagen fibrils have been known to require serum constituents to induce a cell-mediated contraction of collagen gels. The gel contraction was studied with human skin fibroblasts cultured in the presence of fetal bovine serum (FBS). Removal of bovine serum fibronectin (sFN) from FBS did not affect the extent of gel contraction. Gel contraction occurred in serum-free defined media. Therefore, it is concluded that sFN is not required for gel contraction. That cellular FN (cFN) synthesized and secreted by fibroblasts plays a crucial role in gel contraction was suggested by the following experiments: (1) We obtained monoclonal antibodies (mAb A3A5) against fibroblast surface antigens, which suppressed the fibroblast-mediated gel contraction. Immunoblot analyses showed that mAb A3A5 recognizes cFN secreted by human fibroblasts and human plasma FN (pFN), but not bovine sFN in FBS used for culture. (2) Addition of rabbit antisera, which recognize human cFN, to a serum-free gel culture inhibited contraction. Uninvolvement of human pFN in gel contraction was further confirmed by the fact that neither pretreatment of fibroblasts with excess amounts of human pFN nor the presence of excess amounts of human pFN in gels affected the extent of gel contraction. This study seems to be the first demonstration of functional difference between cFN and pFN (or sFN) and proposes a novel mode of binding of fibroblasts with collagen fibrils via cFN during cell-mediated collagen morphogenesis. 8~ 1991
Academic
Press,
Inc.
INTRODUCTION
A culture of fibroblasts in three-dimensional collagen gels (collagen gel culture) originally devised by Elsdale and Bard [l] has been shown to provide a unique model for studying cell to collagen interactions: an in vitro model of collagen morphogenesis [2, 31. Fibroblasts in ’ To whom correspondence dressed.
and
reprint
requests
should
MATERIALS
AND
METHODS
Chemicals, reagents, and culture materials. These were obtained as follows: Dulbecco’s modified Eagle’s medium (DMEM) and RPM1 1640 medium from GIBCO (Grand Island, NY) or Kyokuto Pharmaceutical Industrial Co., Ltd. (Tokyo); fetal bovine serum (FBS) from GIBCO or Japan Biotest Laboratory (Tokyo); ethylenediamine-
be ad-
167 All
Copyright Ccl 1991 rights of reproduction
0014.4827/91 $3.00 by Academic Press, Inc. in any form reserved.
168
ASAGA,
KIKUCHI,
tetraacetic acid (EDTA) and N-(2.hydroxymethyl)piperazine-N-2ethanesulfonic acid (Hepes) from Dojin Chemical Institute (Kumamoto, Japan); streptomycin and penicillin from Meiji Seika Co. (Tokyo); tris(hydroxymethy1) aminomethane (Tris), cycloheximide, trypsin, insulin, glucagon, prolactin, transferrin, hydrocortisone acetate, 5-a-dihydrotestosterone, dexamethasone, p-estradiol, 3,3’,5triiodo-L-thyronine (T,), and prostagrandine F,, from Sigma Chemical Co. (St. Louis, MO); growth hormone from IJCB-Bioproducts (Bruxelles); Sepharose 4B and gelatin-Sepharose 4B from Pharmacia Fine Chemicals (Uppsala); polyethylene glycol 4000 from Merck (Darmstadt); Vectastain ABC kits from Vector Laboratories (Burlingam, CA); Falcon culture dishes and 96.well culture plates from Becton Dickinson Labware (Oxnard, CA); Freund’s complete adjuvant from Difco (Detroit); Millipore filters and nitrocellulose papers from Millipore Japan (Yonezawa, Japan); Centriconfrom W. R. Grace and Co. (Danvers, MA); Athtystar, prepacked affinity columns of protein A, from Kurabo Industries (Osaka); human plasma fibronectin from Hoechst Japan (Tokyo). All other chemicals were of reagent grade and purchased from Wako Pure Chemical Industries (Osaka) or Nacalai Tesque, Inc. (Kyoto). Preparation of FN-free FBS. FN was eliminated from FBS by affinity chromatography on gelatinSepharose 4B according to Ruoslahti et al. [24]. Briefly, Hanks’ balanced salt solution (Hanks’ BSS) was used as a coupling buffer and FBS was passed through the column and collected in glass tubes using a fraction collector. The eluate was concentrated by centrifugation at 3000g for 40 min using Centricon-10 and Hanks’ BSS was added to equalize the final concentration with the original one. Sepharose 4B was substituted for gelatin Sepharose 4B to obtain FBS for a control experiment. Complete elimination of FN was confirmed by polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate (SDSPAGE) [25] and enzyme-linked immunosorbent assay (ELISA) using rabbit antiserum against pFN raised in our laboratory as described below and Vectastain ABC kits. Cell cultures. Human skin fibroblasts were obtained from explants of normal dermis excised as a biopsy at a surgical operation and cultured as described previously [26]. Fibroblasts were grown and maintained in 75.cm* plastic dishes in 15 ml of DMEM containing 10% FBS, 10 mM NaHCO,, 20 mM Hepes, 100 units/ml penicillin, and 100 pg/ml streptomycin in a moist atmosphere of 5% CO,/95% air at 37°C. The population doubling level of cells used in the present study was within 17. Cells were detached from dishes with calciumand magnesium-free Hanks’ BSS (CMF-Hanks’) containing 0.1% trypsin and 1 mM EDTA. Then they were collected by centrifugation at 500g for 5 min and used for experiments after washing with Hanks’ BSS twice. Collagen gel culture.s of fibroblasts. Collagen (Type I) was extracted from calf dermis by digestion with pepsin in a dilute HCI solution, purified, and characterized as described previously [7]. Fibroblasts were grown three-dimensionally in hydrated collagen gels by the method of Elsdale and Bard [l] with slight modifications. The following stock solutions were prepared and kept at 4°C: 0.5% (w/v) collagen solution; 4X concentrated DMEM which contained 80 mM Hepes, 40 mM NaHCO,, 0.4 mg/ml streptomycin, and 400 units/ ml penicillin. Cells were harvested from monolayer cultures by using 0.1% trypsin and 1 mM EDTA in CMF-Hanks’, counted and adjusted to desired cell number, and collected by centrifugation in a plastic tube. The cell pellet was resuspended at 4°C in DMEM containing 0.1 or 0.2% collagen, 20 mM Hepes, 10 mM NaHCO,, 0.1 mg/ml streptomycin, 100 units/ml penicillin, and 10% FBS, which had been prepared by quickly mixing the stock solutions, FBS, and redistilled water. A defined medium specified below was used instead of medium containing 10% FBS in the experiment shown in Fig. 1. The pH was adjusted to 7.4 by adding 1 N NaOH. Media containing cells and collagen were placed in bacteriological dishes with a diameter of 35 or 25 mm: 2 ml of medium containing 5.0 X lo5 fibroblasts and 1 ml of medium containing 2.0 X lo5 cells were placed in 35- and 25.mm
AND
YOSHIZATO
dishes, respectively. The dishes were transferred into an incubator at 37°C. Collagen was gelated within 10 min and cells were embedded in the gel. For long-term culture, gels were overlaid with 1.5 ml of fresh medium and media were changed twice a week. When necessary, FNfree FBS, antiserum, IgG, cycloheximide, or human pFN was introduced into the gel culture. The serum-free defined medium contained the following ingredients in DMEM with concentrations indicated: 0.2% collagen, 20 mM Hepes, 10 mMNaHCO,, 100 Kg/ml streptomycin, 100 units/ml penicillin, 10 pg/ml insulin, 10 bg/ml glucagon, 50 rig/ml epidermal growth factor (EGF), 100 rig/ml growth hormone (GH), 0.02 units/ml prolactin (PL), 0.5 Fg/ml transferrin, 100 rig/ml hydrocortisone acetate, 20 rig/ml 5-tu-dihydrotestosterone, 10 pg/ml @-estradiol, 200 rig/ml dexamethasone, 10m9MT,, 710 rig/ml prostagrandine F,,, , 100 mM CuSO,.SH,O, 3 nM H,SeO,, 5 nM ZnSO,. 7H,O. lktermination of fhe c&e& of& contraction. The extent of contraction was quantified during culture by measuring the diameter of gels or by measuring the thickness of gels as described by Guidry and Grinnell [12, 221. In the latter case, thickness of the center of gels was measured by a phase contrast microscope equipped with a dial test indicator. There were not significant differences in rate of gel contraction as determined with the two methods. Production of antibodies. Monoclonal antibodies were produced by the method of Galfre and Milstain [27] with slight modifications. Eight-week-old BALB/c mice were immunized three times with fibroblasts (7 X lo7 cells/mouse) which had been mechanically harvested from culture dishes with rubber policemen and emulsified in Freund’s complete adjuvant or suspended in PBS. Immune spleens were excised, and spleen cells were fused with myeloma cells (P3X63Ag8.653) in the presence of 5O”‘r (w/v) polyethylene glycol. Viable hybridomas were selected in RPM1 1640 supplemented with hypoxantine, aminopterin, thymidine, and 10% FBS. Hybridomas were screened for t,he production of antibodies through the following three assays in this sequence: Assay 1: Hybridomas producing antibodies that react on the cell surface of fibroblasts were identified by the method of ELISA. Fihrohlasts were grown in 96.well culture plates, washed with Hanks’ BSS, blocked with 2% normal horse serum, and then treated with culture supernatant from the hybridomas, washed again, and detected using Vectastain ABC kits. Assay 2: Antibodies that alter spreading of fibroblasts cultured on the collagen fihrils were selected. Trypsinized fihroblasts were washed two times with Hanks’ BSS and suspended in culture supernatant from hyhridomas. Aliquots of the suspension were poured into dishes coated with collagen-fihrils (100 fig collagen/cm’) and cultured. After 6 or 12 h, morphology of fibroblasts was observed by a phase contrast microscopy. Assay 3: Hybridomas producing antihodies which inhibit fibroblast-mediated collagen gel contraction were selected as follows. One-fifth million fihroblasts were preincubated at 10°C for 1 h in 1 ml of culture supernatants of hybridomas and then were cultured in 1 ml of gels containing 0.1% (w/v) collagen, 10% FBS, and 70.3% culture supernatant of hyhridomas. Extent of contraction was measured during culture. Rabbit antisera against human pFN were obtained by an ordinary procedure. Antisera and normal rabbit sera were incubated at 56°C for 30 min in order to inactivate complement systems prior to use in cell culture experiments. Electrophoresis and immunoblotting. SDS-PAGE was performed by the method of Laemmli [25] with modifications. Fibroblasts harvested by rubber policemen were washed with CMF-Hanks’, homogenized with SDS sample buffer (10% SDS, 5% 2-mercaptoethanol, 62.5 mM TrisHCl, pH 6.8), boiled for 5 min, and subjected to SDSPAGE. Liquid samples (FBS or culture supernatants of fibroblasts) were mixed with an equal volume of 2~ concentrated SDS sample buffer, boiled for 5 min, and subjected to SDS-PAGE. SDSPAGE was carried out with a stacking gel of 4% polyacrylamide and a separation gel of 7.5-20% polyacrylamide linear gradient or 7.5% polyacrylamide. For the preparation of stacking gels, lower gel buffer (0.4%
CELL
TO COLLAGEN
SDS and 1.5 M Tris-HCI, pH 8.8) was used instead of upper gel buffer (0.4% SDS and 0.5 M TrissHCl, pH 6.8). After electrophoresis, polypeptides in the get were stained with Coomassie brilliant biue R-250 or electrophoretically transferred to nitracellulose papers by the method of Towbin et al. [28]. Immunochemical detect,ions of polypeptides on papers with mAb or antiserum were carried out by use of Vectastain ABC kits. Immunocytochemical staining of fibroblasts. Fibroblasts were cultured on plain or cobagen-fibril-coated plastic dishes which had been prepared according to Yoshizato et al. 171. Immunocyt,ochemical stainings of fibroblasts were carried out at 3’7°C without fixation using Vectastain ABC kits. Fibroblasts were preincubated with 2% bovine serum albumin (BSA) in Hanks’ BSS for 10 min to block any remaining protein binding sites and incubated for 30 min with culture supernatants of hybridomas. After washing with Hanks’ BSS, they were incubated for 10 min with biotinyIated anti-mouse IgG diluted by Hanks’ BSS containing 1% BSA. After washing with Hanks’ BSS, they were t,reated for 10 min wit,h Vectastain ABC reagents prepared in Hanks’ BSS, washed five times with Hanks’ BSS, and incubated in Hanks’ BSS containing 0.04 ?& H,O, and 0.5 mg/ml 4-chloro-lnaphthol.
L
01 0
RESIJLTS Factors Affecting
Fibroblasts were cultured in collagen gels in a serumfree defined medium containing nutrients, hormones, and growth factors in order to determine which subst,ances are essential for fibroblasts to contract collagen gels. Fibroblasts thus cultured did contract collagen gels although the extent of contraction was less than that in media supplemented with FBS (Fig. 1). Insulin was essential to fibroblast-mediated collagen gel contraction. Fibroblasts do not appear to require the presence of sFN or pFN to interact with collagen fibrils because neither sFN nor pFN was included in the defined medium. This observation should be important because some reported the involvement of sFN or pFN in fibroblast-mediated collagen gel contraction [ll] and others presented the evidence that pFN is not required [E!, 231. Therefore, further experiments were carried out to clarify the role of FN in collagen gel cont,raction. in the Absence of Bovine sFN
Half a million cells were cultured in 2 ml of 0.2% collagen gels containing 10% FBS or sFN-free FBS which had been prepared by passing FBS through a column of gelatin-Sepharose 4B. The absence of bovine sFN did not affect the extent of gel contraction at all (data not shown), supporting the results shown in Fig. 1. Monoe~onal Antibodies
Which Inhibit
2
4 Days
Collagen Gel ~o~tr~ct~on
Collagen Gel Contraction
169
INTERACTIONS
Gel Contraction
In order to analyze the mechanism of fibroblast-mediated collagen gel contraction in more detail, we tried to obtain mAbs which inhibit gel contraction. Ninety-eight hybridoma cell lines which produced the antibodies reacting on fibroblast cell surfaces were obtained. Two clones (A3A5 and BlA4) of them produced
6
8 in
10 Culture
12
14
FIG. 1. Contraction of collagen gels in defined media. Fibroblasts (5 X 10’) were cultured in 2 ml of 0.2% collagen gels containing defined medium. Concentrations of ingredients of the defined medium are described under Materials and Methods. At Day 4, 2 ml of media was overlaid on gels and cultures were maintained up to Day 14 with a medium change every fourth day. (0) DMEM with 10% FBS; (r) defined medium without PI, and GH; (8) defined medium; (A) defined medium without glucagon, EGF, PL, and GH; (0) defined medium without EGF, insulin, PL, and GH; {a) defined medium without glucagon, insulin, PL, and GH; (0) defined medium without glucagon, EGF, insulin, PL, and GH. Each point represents a single determination.
antibodies which suppressed gel contraction. The clone A3A5 was further characterized. IgGs purified from culture supernatants of the mAb A3A5 showed a dose-dependent inhibition of gel contraction (Fig. 2). When fibroblasts were cultured in collagen gels, cells spread three-dimensionally prior to contraction of gels. However, most fibroblasts in a gel remained round in shape in the presence of supernatants of A3A5 (Fig. 3). It is noteworthy that mAb A3A5 had subtle effects on spreading of fibroblasts cultured on plain or collagen-fibril-coated plast,ic as compared to that of cells cultured within gels (Fig. 4). Fibroblasts were cultured on plain plastic or collagenfibril-coated dishes for 3 days and were immunochemitally stained with mAb A3A5. The surface of fibroblasts and boundaries between fibroblasts on plain plastic were intensely stained with the antibody (Fig. 5A). However, mAb A3A5 could not stain fibroblasts cultured on collagen fibrils (Fig. 50. This unresponsiveness of mAb A3A5 was not due to the absence of antigens recognized by mAb A3A5 on fibroblast surface membrane, because fibroblasts cultured on collagen fibrils contained a protein band which reacts on mAb A3A5 (Fig. 6C). The chemical nature of ant.igens correspon~ng to
170
6
m =
ASAGA,
KIKUCHI,
5
s
t
,
0’ 0
5
10 Hours
I 15 in
20 Culture
I
I
25
30
FIG. 2. Dose-dependent inhibition of collagen gel contractions by mAb A3A5. Fibroblasts (2 X 105) were cultured in 1 ml of 0.1% collagen gels containing DMEM, 10% FBS, and different concentrations of IgG of mAb A3A.5. (0) Without IgG, (0) 20 pg/ml, (0) 50 eglml, (A) 100 fig/ml. Each point represents a single determination.
mAb A3A5 was examined by Western blot analyses. When proteins extracted from fibroblasts were subjected to immunoblotting with mAb A3A5, a single band was detected (Fig. 6A). Molecular weights calculated from electrophoretic mobility of the band were about 230 kDa which is the same as that of FN. Actually, A3A5 reacted with human pFN (lane pFN of A3A5 in Fig. 6A). Moreover, A3A5 reacted on a band corresponding to FN secreted by human fibroblasts but not on any kind of proteins in FBS which was included in culture media (Fig. 6B). If cFN is defined as FN which is synthesized and secreted by fibroblasts, these results indicate that A3A5 recognizes FN of human origin (both cFN and pFN) but not bovine serum FN in culture medium. Results of Fig. 1 through 6 collectively indicate uninvolvement of bovine sFN and involvement of human cFN in collagen gel contraction by human fibroblasts. This statement was further verified in the following experiments. Critical
AND
YOSHIZA’I’O
was dose-dependently inhibited by the antiserum (Fig. 7), clearly showing involvement of human cFN in gel contraction, because the antiserum can recognize human cFN of fibroblasts as shown in Fig. 6A. Figure 7 also suggests uninvolvement of rabbit sFN in the contraction process, because there were no noticeable differences in the extent of gel contraction at different concentrations of rabbit serum from 10 through 50%. When IgGs purified from rabbit anti-human pFN antiserum were introduced into collagen gel culture at concentrations of more than 1 mg/ml, gel contraction was inhibited (data not shown), again confirming involvement of human cFN in gel contraction. Considering the results described above together, we postulate that human cFN is an inevitable protein for human fibroblasts to bind collagen fibrils in collagen gel culture. During culture, fibroblasts are expected to actively synthesize cFN, which can mediate interactions between fibroblasts and collagen fibrils. Actually, we measured that fibroblasts in serum-free DMEM synthesize and release cFN into media at 0.1 rig/cell per day. When synthesis of proteins including cFN was inhibited by cycloheximide, collagen gel contraction was inhibited in a reversible and concentration-dependent manner (data not shown). However, the possibility that human pFN substitutes
Roles of cFN in Gel Contraction
Rabbit antiserum against human pFN was prepared. This preparation reacts on human cFN as well as human pFN (arzti-FN, lanes F and pFN in Fig. 6A). Collagen gel cultures in DMEM without FBS were prepared and were given the rabbit antisera at concentrations of 10, 20, and 50%. The gel culture thus obtained was to contain as FN rabbit sFN from the antisera and human cFN from fibroblasts and was to be supplied with serum factors necessary for the gel contraction such as insulin by the antisera. The antiserum did not show cytotoxicity for fibroblasts at these concentrations. Contraction
FIG. 3. Photomicrographs of fibroblasts cultured within collagen gels in the presence of supernatants of hybridoma cultures. Fibroblasts were suspended in 1 ml of supernatants of hybridomas, incubated at 10°C for 1 h, and then cultured in 1 ml of 0.1% collagen gels containing 10% FBS and 70.3% culture supernatants of hybridomas. Photographs were taken at 3 h in culture. (A) A3A5, (B) myeloma. Magnification: X100.
171
FIG. 4. Photomicrographs of fibroblasts cultured on plain plastic or collagen-fibril-coated plastic in the presence ofmAb. Fibroblasts were treated as described in the legend to Fig. 3 and seeded in tissue culture dishes. (A and 13) On plain dishes, (C and D) on dishes coated with collagen-fibrils at a density of 100 pglcm’. (A and C) With mAb A3A5, (B and D) with supernatant of myeloma cult,ures as controls. Photographs were t.aken at 6 h in culture. Magnificati(~n: ~100.
for human cFN cannot be ruled out, because mAb A3A5 recognizes both human cFN and pFN. Therefore, three further experiments were carried out to test this possibility. Uninvolvement Contraction
of Human
pFN in Collagen Gel
Fibroblasts were preincubated in a medium containing human pFN at the concentrations of 1 and 2 mg/ml
at 4°C for 1 h, rinsed, and cultured on collagen gels containing sFN-free FBS instead of FBS. These concentrations are far beyond the plasma concentration of FN (0.33 mg/ml) [29] and are expected to be enough for competing cFN in binding to hypothetical common receptors for pFN and cFN on the cell surface of fibroblasts. The extent of contraction was the same as those of control experiments without human pFN (data not shown). Moreover, even when fibroblasts preincubated
FIG. 5. Immunocytochemical stains of fibroblasts with mAb. Fibroblasts (2 X 105) were cultured for 3 days in 2 ml of DMEM supplemented with 10% FBS in plain 35-mm cult,ure dishes (A and B) or in dishes coated with collagen fibrils as described in the legend to Fig. 4 (C and D). Cells were then immunostained with mAb A3A5 (A and C) or supernatants of myeloma culture (B and D). Magnification: X100.
172
ASAGA, SDS-PAGE A
F pFN
anti-FN
A3A5 F pFN
F pFN
#.
a.
-.
KIKUCHI,
AND
YOSHIZATO
The addition of human pFN to the collagen gel culture with cycloheximide did not rescue the contractioninducing ability of fibroblasts (data not shown), again supporting the notions that human pFN is not involved in gel contraction and human pFN cannot substitute for human cFN.
-..
94K 68K 43K 30K
DISCUSSION 14.4K
SDS-PAGE c
8QWcY
A3A5 Qf Qv 0” -
me,---.
94K 68K -
FIG. 6. Western blot analyses with mAb A3A5 or rabbit anti-human pFN antiserum. (A) Proteins of fibroblasts. Proteins extracted from fibroblasts (F) and purified human pFN (pFN) were subjected to SDS-PAGE (left) and immunoblotted with mAh A3A.5 (middle) or rabbit anti-human pFN antiserum (right). Eight micrograms of fibroblast proteins and 1 fig of pFN were analyzed. (B) Proteins in FBS and culture medium of fibroblasts. FBS, culture supernatants of fibroblasts (SUP), and human pFN (pFN) were subjected to SDS-PAGE (left) and immunoblotted with mAb A3A5 (right). Ten micrograms of FBS and SUP and 2 pg of pFN were analyzed. (C) Proteins of fibroblasts cultured on collagen fibrils. Fibroblasts (:i X IO5 cells) were cultured for 3 days on 35-mm plastic (PL) or collagen fibrils (CL) coated on 35-mm plastic, washed extensively with PBS, and solublized with SDS sample buffer. Proteins from 3 x 104 cells were subjected to SDS-PAGE and Western blotting with mAb A3A.5. Arrowheads indicate the bands immunochemically detected with mAb A3A5. Arabic numerals with K at the left side indicate molerular weights determined by marker proteins.
as described above were cultured on collagen gels containing excess amounts of human pFN (1 or 2 mg/ml at a final concentration), the extent of gel contraction was identical to those of control experiments without human pFN (data not shown). If human pFN has some role in mediating binding between fibroblasts and collagen fibrils and pFN and cFN have common binding sites on collagen fibrils, the gel contraction was expected to be suppressed in these experiments due to competitive binding of pFN. Or even a stimulatory effect of pFN was expected if human pFN has some role in mediating a binding between fibroblasts and collagen fibrils through binding sites for pFN but not cFN. Therefore, these results clearly indicate that human pFN is not involved in collagen gel contraction by human fibroblasts.
The contraction of collagen gels by fibroblasts was first reported by Bell et al. [4] and is a phenomenon useful in the study of cell to collagen interactions (collagen morphogenesis) [2, 31. The mechanism of fibroblast-mediated collagen gel contraction has been poorly understood and is still controversial. We performed several experiments to study the mechanism of fibroblastmediated collagen gel contraction, focusing on the relationship among the cell surface of fibroblasts, collagen, and FN. On the basis of this study, we present a mechanism of an indirect interact.ion between fibroblasts and collagen in the process of gel contraction via cFN synthesized and secreted by fibroblasts. Neither sFN nor pFN in culture medium plays any role in this mechanism. Fibroblast-mediated collagen gel contraction occurs in serum-free defined medium which does not contain sFN or pFN and removal of bovine sFN from FBS does not affect the extent of gel contraction. These facts indicate that bovine sFN does not play any roles for collagen gel contraction.
P f 0
5
0
I.-.0
5
10 Hours
15 20 in Culture
25
30
FIG. 7. Inhibition of contraction of collagen gels by rabbit antihuman pFN antisera. Fibroblasts (2 X 105) were cultured in 1 ml of 0.1% collagen gels containing DMEM and antisera at concentrations indicated. Normal rabbit sera were used as controls. (0 and 0) lo%, (0 and H) 20%, (A and A) 50%. Open and closed symbols indicate antisera and normal sera, respectively. Each point represents the average of two determinations, whose range was less than 5% of the mean.
CELL
TO COLLAGEN
We obtained mAb A3A5 which inhibits collagen gel contraction. Monoclonal antibody A3A5 recognizes human cFN secreted from flbroblasts and human pFN, but not bovine sFN in FBS. It is important for the discussion of the mechanism of collagen gel contraction that mAb A3A5 reacts on human cFN from fibroblasts but not bovine sFN in culture medium. There exists human cFN secreted by fibroblasts and bovine sFN in culture media of the experiment presented in Fig. 2, but not other kinds of FN. In this gel culture, mAb A3A5 specific to human FN fibroblasts but not bovine sFN has a critical role for collagen gel contraction. Since rabbit anti-human pFN antisera contain rabbit sFN and polyclonal antibodies which react on human cFN but not rabbit sFN, inhibition of gel contraption by the antisera also indicates critical roles of human cFN and that rabbit sFN cannot substitute for human cFN. Pretreatment of fibroblasts with excess amounts of human pFN does not affect the extent of gel contraction. Moreover, the presence of excess amounts of human pFN in a gel has no effect on contraction. These results indicate uninvolvement of human pFN in the process of human fibroblast-mediated collagen gel contraction. In other words, human pFN cannot substitute for human cFN. This notion is also supported from the fact presented in the present paper that human pFN cannot restore the contraction-inducing ability of fibroblasts in the presence of cycloheximide. Other investigators have also reported that bovine sFN and human pFN are not required for collagen gel contraction by human fibroblasts [ 12,231. On the other hand, Gillery et al. [ll] presented evidence suggesting an involvement of bovine sFN. Schafer et al. [23] reported that polyclonal antibodies against human pFN did not inhibit collagen gel contraction by human fibroblasts, which is contrary to the result of Fig. 7 of the present paper. In the experiments of Schafer et al. 1231, they tested the effect of t.he antibodies at a concentration of 200 @g/ml or less. In the present study, the inhibition of collagen gel contraction by anti-human pFN IgGs was observed at concentrations of more than 1 mg/ml. Therefore, the difference in the result between the present study and theirs might be due to the amounts of antibodies used in the experiments. However, the apparent discrepancy between the present study and that of Gillery et al. [ll] cannot, be reasonably explained at present. It is known that fibroblasts of established cell lines or transformed fibroblasts are less able to contract collagen gels than normal fibroblasts obtained in primary culture [lo, 30, 311. This fact can be explained by the present study, because it is generally accepted that established or transformed cells secrete a smaller amount of cFN than the corresponding normal cells [32,33]. The present study emphasizes the difference between cFN and pFN in mediating binding between fibroblasts
INTERACTIONS
173
and collagen fibrils which is the first step for ensuing contraction of collagen gels. Cellular FN and pFN share many common features in structure and function, because both FNs originate from the common gene 1341. However, some differences have been also known. Cellular FN is thought to exist as polymeric forms [35,36]. Molecular weights of A and B chains of cFN are higher in about 10 kDa than those of pFN (37,381. This difference in molecular weights is due to a different splicing of mRNA of FN [34]. Cellular FN contains regions of EDA and ED-B which lack in pFN [39]. At present it is not known which of these differences is responsible for the functional difference between cFN and pFN observed in gel contraction. To obtain an answer for this it is crucial to know what region mAb A3A5 recognizes on molecules of FN. Such investigations are now being undertaken by analyzing affinities of proteolytic fragments of FN for mAb A3A5. Another possibility is that mAb A3A5 binds two epitopes, one nonfunctional present in both cFN and pFN and another functional present in only cFN. The effects of mAb A3A5 on cell morphology are manifested more intensely for cells in collagen gels than those on plain plastic or collagen-fibril-coated plastic. There might be some difference in the mechanism of cell to collagen interaction between three-dimensional culture within collagen gels and conventional two-dimensional culture. Recently, Grinnell et al. 1401 studied effects of peptides with the cell recognition sequence on morphology of cells cultured on collagen-coated surfaces or inside collagen gels and noticed that there are some critical differences between the mechanism of three-dimensional cell to collagen recognition and that of two-dimensional recognition. Monoclonal antibody A3A5 could not stain fibroblasts on collagen-fibril-coated plastic, while the mAb A3A5 could stain those on plain plastic. The immunological unresponsiveness was not due to the absence of antigens on the surface of cells on collagen fibrils. One plausible explanation for this may be that mAb A3A5 recognizes some specific region of cFN which is important for cell to collagen interactions. This region of cFN had been occupied by collagen when mAb A3A5 was added to the assay cultures. Considering these together, we postulate that mAb A3A5 recognizes the collagen binding region, but not the cell binding region, of FN molecules. The present study demonstrates that insulin is important as a factor for the fibroblast-mediated collagen gel contraction. There have been reports indicating that transforming growth factor /3 (TGF~) [S] and plateletderived growth factor (PDGF) [9] stimulate collagen gel contraction by fibroblasts. The function of insulin and these growth factors in the gel contraction remains to be clarified. In relation to the present study, it should be noted that TGFP stimulates fibroblasts to produce cFN [41] and PDGF increases the ability of cells to adhere to
174
ASAGA,
KIKUCHI,
FN [42]. Insulin also may have actions similar to TGFP or PDGF. FBS should contain another unknown factor(s) necessary to collagen gel contraction, because contraction of gels containing FBS proceeds faster and results in a more marked contraction than that in defined medium. We thank Dr. R. Matsuda for his technical cooperation on production of mAbs. We also thank Mr. T. Taira and Mr. K. Tokuda for their cooperation during this study. This work was supported in part by a grant-in-aid for special project research from the Ministry of Education, Science and Culture of Japan (59105003), by a grant-inaid for Developmental Scientific Research (08170034), and by a grant from TERUMO Life Science Foundation (8907).
REFERENCES Elsdale, T., and Bard, J. (1977) J. Cell Biol. 54, 626-637. Harris, A. K., Stopak, D., and Wild, P. (1981) Nature (London) 290,249-251.
