EXPERIMENTAL

CELL

RESEARCH

199,

134-146 (1992)

Expression of SPARC is Correlated with Altered Morphologies in Transfected F9 Embryonal Carcinoma Cells ELIZABETH Department

of Biological

Structure,

A. EVERITT

AND E. HELENE

School of Medicine,

Inc.

INTRODUCTION

SPARC (secreted protein, acidic and rich in cysteine) is a Ca’+-binding, secreted glycoprotein associated with cells and tissues undergoing remodeling, morphogenesis, migration, or proliferation [l-13]. When added as a

Seattle, Washington

98195

’ Abbreviations used: AFP, a-fetoprotein; AS, antisense; dicAMP, dibutyryl CAMP; DMEM, Dulbecco’s modified Eagle’s medium; ECM, extracellular matrix; ELISA, enzyme-linked immunosorbent assay; FN, fibronectin; hi-FBS, heat-inactivated fetal bovine serum; Met, methionine; MT, metallothionein; NFDM, nonfat dried milk; PE, parietal endoderm; PBS, phosphate-buffered saline; PYS, parieta1 yolk sac endoderm cell line; RA, retinoic acid; RIA, radioimmunoassay; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; S, sense; SSC, salt sodium citrate; VE, visceral endoderm.

I To whom correspondence and reprint requests should be addressed at Department of Biological Structure, SM-20, University of Washington, Seattle, Washington, 98195. Fax: 206-543-1524. 0014-4827192 $3.00 Copyright 0 1992 by Academic Press, Inc. All rights of reproduction in any form reserved.

of Washington,

purified protein to cultured endothelial cells, fibroblasts, and smooth muscle cells, SPARC inhibited the spreading of newly plated cells and induced rounding in monolayers of spread cells [lo]. SPARC belongs to a group of antiadhesive proteins that includes thrombospondin and tenascin; these proteins interfere with the ability of cells to spread in vitro [ 141. In addition, SPARC binds to various extracellular matrix (ECM)2 components, such as type III and V collagen, and thrombospondin [lo]. F9 teratocarcinoma stem cells have been used as a model system for studies on early mammalian development and differentiation [15-171, as they can be stimulated to differentiate into parietal endoderm (PE) in the presence of retinoic acid (RA) and dibutyryl CAMP (diCAMP) [16] or into visceral endoderm (VE) upon aggregation of RA-treated cells plated in petri dishes [18]. One of the changes associated with the differentiation to PE is induced secretion of the ECM proteins, laminin and type IV collagen [16], and the ECM-associated protein, SPARC [I, 10, 191. SPARC mRNA is expressed at nearly undetectable levels in undifferentiated F9 cells, but it increases 20fold after induction of the PE phenotype and represents 0.5% of the poly(A)+ RNA present in PE cells [l, 191. SPARC appears not to be regulated in F9 cells in the same manner as the early RA-responsive genes such as the protooncogene c-myc [20, 211, ERA-l/Hex 1.6 [22, 231, and REX-l [24]. In contrast, it is one of the genes activated relatively late in the differentiation of F9 stem cells to PE [l]. This group includes the ECM proteins laminin and type IV collagen, a-fetoprotein [AFP; 251, and CAD [26].

SPARC (secreted protein, acidic and rich in cysteine) is a Ca2+-binding glycoprotein that has recently been identified as a member of a group of proteins that exert antispreading effects on various cultured cells. In addition, SPARC is induced during the later stages of F9 stem cell differentiation to parietal endoderm (PE). When treated with retinoic acid and dibutyryl CAMP, F9 cells differentiate into PE and SPARC mRNA is increased approximately 20-fold. To determine whether the chronic overexpression or inhibition of expression of SPARC would affect the morphology, attachment, or differentiation of F9 cells, we transfected undifferentiated F9 cells with cDNA encoding SPARC or antisense SPARC and cloned lines that expressed either elevated or reduced levels of SPARC protein. The transfected F9 cells displayed altered morphologies in culture: cells of four overexpressing lines appeared clumped and rounded, whereas those of three underexpressing lines were spread and flat, in comparison to controls. Moreover, the morphological differences persisted during differentiation of the lines to PE. The altered morphology was not due to an increased expression of collagenases and did not affect the ability of the cells to attach and adhere to tissue culture plastic. The altered phenotype of the transfected F9 cells appeared to be directly related to the level of extracellular SPARC. Since overexpression of SPARC induced rounding and aggregation of F9 cells in culture, we propose that SPARC facilitates modulation of cell0 1992 Academic cell or cell-substrate interactions in duo. Press,

University

SAGES

134

DIFFERENTIAL

SPARC

EXPRESSION

In this study we have addressed three questions that we judged relevant to the function of SPARC in a stem cell model system: (1) does the premature production of a secreted protein that associates with ECM components alter the ECM that F9 cells normally encounter and perturb their ability to remain attached or spread, (2) does the secretion of SPARC by F9 cells affect their normal biology; and (3) does the inappropriate and premature expression of SPARC perturb other later ECM proteins in the PE or VE pathway and thereby affect the ability of F9 cells to differentiate? We stably transfected cDNA encoding sense-oriented (S) SPARC and antisense-oriented (AS) SPARC (CRAPS) into undifferentiated F9 cells and report that cells with stably integrated SPARC and CRAPS DNA overexpressed or underexpressed SPARC mRNA and protein, respectively. We show that differences in the expression of SPARC are correlated with altered morphologies of the stable transformants: S lines were clumped and rounded, whereas AS lines were spread and flat, as compared to control lines. Examination of a variety of morphological, biochemical, and molecular features of the transfected lines, including their attachment to culture surfaces, did not reveal changes induced concomitantly by the over- or underexpression of SPARC. We therefore propose that the modulation of cell phenotype was directly related to the altered levels of SPARC and that the secretion of SPARC by F9 cells induced rounded morphologies that did not appear to affect cellular attachment. In vivo this rounding is expected to occur in cells preparing for events that require changes in shape. MATERIALS

AND

METHODS

Cell culture. Murine F9 embryonal carcinoma stem cells were obtained from American Type Culture Collection. F9 stem cells and transfected clones were routinely cultured in DMEM (Sigma) containing high glucose and supplemented with 10% heat-inactivated FBS (hi-FBS) (Hyclone Labs) [27]. F9 cells were grown on loo-mm tissue culture dishes (Corning) coated with 1% gelatin at 37°C in a humidified atmosphere containing 5% CO,; for some experiments, cells were plated directly onto noncoated tissue culture plastic. Cells were passaged every other day with PBS containing 0.01% trypsin and 1 mM EDTA and replated at a dilution of 1:lO. Stocks were maintained in liquid nitrogen, and lines were used between passages i-40. Cells were induced to differentiate to PE with 5 X 10m8M all-trans RA (Sigma) and 1 X 10m3M dicAMP added every second day [28]. Cells that had been allowed to form small floating aggregates in bacteriologic dishes were differentiated to VE in the presence of 5 X lo-* M RA [27]. Mouse PYS-2 cells were provided by Dr. Brigid Hogan (Vanderbilt University, Nashville, TN) and cultured as previously described [lo]. Construction and identification of sense and antisense expression uectors. The plasmid pGEM5.22 [l] (provided by Dr. Brigid Hogan) is a modified pGEM-1 riboprobe vector (Promega) that contains the full-length SPARC cDNA (nt 12-1160). The SPARC insert was excised with BamHI/HindIII and was made blunt by the Klenow fragment of DNA polymerase (Boeringer-Mannheim). One expression vector used in the clonings was pko-neo, a plasmid containing SV40

ALTERS

MORPHOLOGY

135

promoter and polyadenylation sequences [29]. After removal of the neo insert with HindIII, the SPARC cDNA was ligated to the remaining pko vector with T4 DNA ligase [30]. The second expression vector used for cloning was Zem-3, a plasmid containing the mouse metallothionein (MT) promoter (provided by Dr. Elaine Mulvihill, Zymogenetics Inc., Seattle, WA). The SPARC insert was excised from pGEM5.22 and ligated to linear Zem-3 as described above. Transformation of E. coli strain BSJ 72 was performed as described by Maniatis et al. [30]. Plasmid DNA was purified and digested with restriction enzymes prior to Southern analysis [31]. Blotted membranes were hybridized with a 557-bp BamHIIEcoRI fragment of SPARC cDNA [l] labeled with [32P]dCTP and were washed as previously described [32], prior to exposure to X-Omat AR film (Kodak) with intensifying screens at -70°C. DNA from transformant colonies that hybridized at high stringency to the probe were further characterized by restriction digests to establish the orientation of the SPARC insert. Sense (S) or antisense (AS) orientation of the SPARC insert was determined by restriction mapping with NcoI for the pSV40-containing plasmids, and with KpnI for the pMT-l-containing plasmid. The SV40 plasmid containing the SPARC cDNA in the sense orientation was termed pSV-SPARC, while the antisense plasmid was termed pSV-CRAPS. The MT plasmid was termed pMTSPARC. Cells transfected with pMT-SPARC were cultured with and without 40 PM ZnCl,. Stable transfection of F9 celki with SPARC and CRAPS expression uectors. Undifferentiated F9 cells were plated as described above 1 day prior to transfection. Experimental plasmid DNA was added to cells in a lo-fold molar excess over pko- or pMT-neo. DNA was transfected as calcium phosphate precipitates [33] and was added to the cells in the presence of Lipofectin (BRL). Twenty-four hours after addition of DNA, the cells were rinsed and reincubated in DMEM with 10% hi-FBS. After an additional 24 h, cells were exposed to G418 (Sigma) for 14 days of selection. Stably transfected cell lines were routinely maintained in 100 pg/ml G418; otherwise no antibiotics were added. For determiAnalysis of genomic DNA from stable transformants. nation ofthe presence of exogenous SPARC/CRAPS sequences, genomic DNA was isolated [34] from (i) undifferentiated F9 cells, (ii) clones transfected with pko-neo only (Neo), and (iii) clones transfected with four different pSV-CRAPS, one pSV-SPARC, and one pMT-SPARC expression vector. Southern analysis was performed as described above. X-ray films were exposed for durations less than those necessary to detect hybridization with the endogenous SPARC gene. Isolation and analysis of RNA. Total cytoplasmic RNA was isolated from cells by the method of Chomczynski and Sacchi [35]. Poly(A)+ RNA was prepared by passing total RNA twice over oligo(dT) cellulose columns (Pharmacia-LKB) and eluting with sterile RNase-free water. Northern analysis was performed as previously described [32]. For analysis of poly(A)+ RNA with a single-stranded antisense RNA probe, plasmid pGEM5.22 was transcribed in vitro with SP6 polymerase. Blots were washed at a final stringency of 75°C in 0.1X SSC containing 0.1% SDS. The density of autoradiographic signals on preflashed films was quantitated with a laser scanning densitometer equipped with a peak integrator (Beckman), and values for L32 ribosomal protein RNA (cDNA probe provided by R. Kaspar, University of Washington, Seattle, WA) were used to correct for differences in loading of RNA. Densitometric scans were quantified by performing serial dilutions of RNA from the cells to confirm that the response of the film to the radioactive bands was linear [36]. Immune precipitation of SPARC. Immunoprecipitations from cell culture media were performed as previously described [37], with minor modifications. F9 and transfected cells were released with trypsin/EDTA in PBS, counted, and plated at high density on gelatin-coated tissue culture dishes in DMEM lacking methionine (Met)

136

EVERITT

containing 10% hi-FBS. Cells were incubated with 10 &i/ml [35S]Met (ICN Radiochemicals) for 24-48 h. Media were collected and clarified, and a mixture of protease inhibitors was added. Antibodies were specific for SPARC [9, lo] and have been previously characterized by ELISA, immunoblotting, and radioimmune precipitation [9, lo]. SDS-PAGE was performed for immunoblot analysis and quantitation. RIA of culture medium was Radioimmunoassay (RIA) of SPARC. performed for each of the stable lines according to a protocol modified from Malaval et al. [38]. Briefly, equal cell numbers were plated on gelatin-coated or uncoated cultureware and cells were incubated in DMEM containing 10% hi-FBS for 6 h; medium was collected and clarified. To obtain a standard curve, SPARC protein was dissolved in DMEM containing 10% hi-FBS or in a Tris-saline buffer containing 2% nonfat dried milk (NFDM) at pH 7.9 (Tris buffer), and serial dilutions were made from 250 to 1 rig/ml. ‘251-labeled SPARC in Tris buffer or DMEM containing 10% hi-FBS was added to tubes containing sample or SPARC standard. To this mixture a polyclonal antiSPARC antibody was added for an incubation overnight at 4°C. lmmune complexes were precipitated in a 25% PEG-8000 solution in Tris buffer without NFDM, and radioactivity of the precipitates was counted. Attachment assays. Cell attachment assays were modified from a procedure previously described for F9 cells [39]. Cells were labeled with [3H]thymidine or [3H]leucine (DuPont) in DMEM containing l-10% hi-FBS for 18-20 h prior to the experiment. Cells were then treated with trypsin/EDTA and counted. Equal cell aliquots were added in duplicate to tissue culture cluster wells (Corning) which had been coated with 0.02-0.05 mg/ml fibronectin (FN, Telios Pharmaceuticals). Attachment was allowed to proceed from 30 min to 48 h; aliquots of attached and detached cells were subsequently solubilized, and radioactivity was measured. Detection of proteinuses and activation of metalloproteinase proenzymes. Zymography was used to detect the secretion of proteinases [40, 411 by undifferentiated F9 cells, transfected lines, and their differentiated derivatives. Cells were labeled with [35S]Met as described above, and equal cpm from medium and cell layers were separated on polyacrylamide gels, to which had been added 1 mg/ml type I gelatin (Sigma). Preparation of samples, electrophoresis, and detection of bands were performed as described by Herron et al. [40] and Adler et al. [42]. Proteins from the DMEM and 2% lactalbumin hydrolysateconditioned media of cell lines differentiated with RA, followed by RA/dicAMP, were activated by incubation with plasmin (Sigma), and zymography was performed as described by Adler et al. [42].

RESULTS

Expression

Vectors: SPARC and CRAPS

Eukaryotic expression vectors capable of generating SPARC and CRAPS mRNA were constructed to overexpress or inhibit the expression of SPARC in undifferentiated F9 cells and their endodermal derivatives (Fig. 1). The choice of the SV40 promoter (Figs. 1A and 1B) was based on previously successful transfection results obtained using the SV40 promoter in F9 cells [43-451 and our own transient transfection assays, in which the relative efficiencies of various promoters in transcribing the chloramphenicol acetyltransferase gene (CAT) was assessed. Prepared as a calcium phosphate precipitate and added to cells in the presence of Lipofectin, pSV,CAT consistently gave a high level of expression.

AND

SAGE A

B

Lacuvivs promter

S”40 early promoter

CRAPS

S”40 ply A addition

1.1 kb

3’

C

hOHply A

Ml-I promoter

SPARC

eg

K

E

aciauon

I.1 kb

K

%

FIG. 1. Description of pSV-SPARC, pSV-CRAPS, and pMTSPARC. (A) pSV-SPARC (5884 bp) was constructed by replacing the neo cDNA of pko-neo (4736 bp) with SPARC cDNA (1148 bp). (B) pSV-CRAPS (5884 bp) was constructed similarly, except that the SPARC cDNA was ligated in the antisense orientation. (C) pMTSPARC (5148 bp) was constructed by ligating a ZEM-3 plasmid (4000 bp) with SPARC cDNA (1148 bp). Thin lines, pBR322 sequences (A) and (B), or pUC 13 sequences (C); thick lines, SV40 DNA; hatched boxes, SPARCICRAPS cDNA, open boxes, lnc UV5 promoter; lined boxes, metallothionein (MT-l) DNA and human growth hormone (hGH) DNA. B, BamHl; Bg, Bglll; E, EcoRl; H, HindIll; K, KpnI; N, Ncol; P, Pstl.

The expression vector containing the MT-l promoter (Fig. 1C) was chosen because it allowed us to induce the transcription of the integrated DNA when ZnCl, was added to the culture medium. Analysis of conditioned medium from F9 cells indicated that the addition of 40 PM ZnCl, to the cultures did not alter the constitutive, low-level production of SPARC. Orientation of inserts in bacterial transformants was determined by restriction digests of plasmid DNA. Restriction maps of both SPARC and pko-neo cDNA in&cated that sense-oriented SV40 plasmids, digested with NcoI, would yield DNA fragments of approximately 0.98 and 4.9 kb (Fig. 1A). One bacterial transformant (No. 10) was identified as containing pSV-SPARC. After transfection into F9 cells, this DNA gave rise to senseoriented clones S-5 and S-8 (Table 1). Antisense-oriented SV4O-plasmid DNA digested with NcoI was predicted to yield two fragments of approximately 0.3 and 5.6 kb (Fig. 1B). Four bacterial transformants, 1, 6, 7, and 11, were identified as having pSV-CRAPS and gave rise to transfected antisense clones AS-l, 2, and 3 (Table 1). The orientation of the third expression vector, pMT-SPARC, was not verified by restriction analysis, as the ligation was designed in the sense orientation. The insert was confirmed in the transfected cells by

137

DIFFERENTIAL SPARC EXPRESSION ALTERS MORPHOLOGY TABLE

1

Transfection, Antibiotic Selection, and Southern Blot Analysis of SPARC and CRAPS Stably Transfected F9 Cell Clones Clones Plasmid 1 6 I 11 10 27

Cell clones derived” AS-3 NA NA AS-l, AS-2 S-5, S-8 s-27

Clones screened* 4 4 3 3 7 1

containing’ cDNA insert 2 1 1 3 7 1

% positived 50 25 33 100 100 100

’ pSV-SPARC plasmid is represented by number 10; this plasmid is carried in F9 cell clones S-5 and S-8. pSV-CRAPS plasmids are represented by numbers 1,6,7, and 11; plasmids 1 and 11 are carried in F9 cell clones AS-3 and AS-l and 2, respectively. Each plasmid was added to one 60-mm dish containing 1.5 X lo3 F9 cells; dishes received a total of 17 pg DNA. pMT-SPARC plasmid is represented by S-27 and S-27+; DNA (5 pg) was transfected onto 5 X lo5 F9 tells/60-mm dish. Cells were cotransfected with pko-neo or pMT-neo for pSVSPARC/CRAPS and pMT-SPARC transfections, respectively. Transfected cells were treated with 400 rg/ml G418 antibiotic for 14 days. AS, antisense orientation; S, sense orientation. * Number of cell clones surviving G418 selection that were chosen for Southern blot analysis of SPARC DNA. ’ Number of clones analyzed by Southern blot that demonstrated the presence of transfected SPARC DNA. d Percentage of total clones screened that were positive for exogenous SPARC cDNA. e NA, cell lines derived from these plasmids were characterized but are not further described in these studies.

df -

FIG. 2. Southern blot analysis of G418-resistant clones. Genomic DNA (2 kg) from cell lines transfected with pko-neo (lane l), pSV-SPARC (lanes 3 and 4), and pSV-CRAPS (lanes 5,6, and 7) was compared with 500-ng pGem5.22 plasmid DNA (lane 2). DNA digested with EcoRI and PstI yielded the two predicted fragments (559 and 518 bp) after the blot was probed with a BamHIIEcoRI fragment of SPARC cDNA. C, controls; S, sense transfectants; AS, antisense transfectants. The two fragments of SPARC/CRAPS are too similar in size to be resolved in this gel and are indicated as one band by the arrow. df, dye front; positions of molecular weight standards are indicated in kb on the left.

entiated lines was prepared and analyzed by Northern blotting, with an antisense SPARC RNA probe. Figure 3A shows the Northern blot of SPARC and L32 ribo-

A

B

118 -

I

2.25

Southern analysis of DNA digested with KpnI (Fig. 1C). pMT-SPARC yielded the transfected sense clone S-27. DNA pko-neo gave rise to the control cell clone, Neo. After transfection and antibiotic selection, cell lines representing six different sense- and antisense-oriented plasmids were selected for Southern analysis. Southern blots of EcoRllPstl(pSV-SPARC/CRAPS; Fig. 2) and KpnI- (pMT-SPARC; data not shown) restricted genomic DNA confirmed that approximately 70% of the neomycin-resistant lines examined contained exogenous SPARC DNA (Table 1). Hybridization and washing of the blots were performed at high stringency, such that the single copy of the endogenous SPARC gene was not detected [19] (Fig. 2, lane 1). We found that variable amounts of exogenous SPARC DNA had been incorporated into cellular DNA (Fig. 2). Quantitation of SPARC mRNA and Protein from Stably Transfected Lines To determine whether SPARC mRNA was being transcribed from exogenous, stably integrated SPARC DNA, poly(A)+ RNA from S, AS, and control undiffer-

f

11 SPARC w2-

c

23

As

s

451

2.00

f 1.75 z -m 1.50 ‘= g 1.25 2

1. 00

9 1::: co

0.25 WS

NE0

AS-3

S-8

S-27+

Cell Line

FIG. 3. Stably transfected undifferentiated F9 cell lines exhibit enhanced or diminished SPARC mRNA. (A) Autoradiogram of Northern riboprobe blot performed with poly(A)+ RNA from cells transfected with pSV-SPARC (lane 4, S-8), pSV-CRAPS (lane 3, AS-3), pMT-SPARC under induction by &Cl, (lane 5, S-27+) or pko-neo (lane 2, Neo) expression vectors. Poly(A)+ RNA from PYS cells is shown in lane 1. RNA (5 pg) was resolved on agarose gels, blotted onto nitrocellulose, and hybridized with an antisense RNA probe for SPARC (2.2 kb) RNA, and a cDNA probe for L32 (0.5 kb) RNA. C, controls; AS, antisense lines; S, sense lines. (B) Densitometric plot of cell lines depicted in A. Values are normalized to the arbitrary value of Neo = 1 and are plotted as the mean of duplicate scans + SEM.

138

EVERITT A

C

As

s

SPARC: 132.

Neo

AS-3

AS-2

S-27+

Cell Line

FIG. 4. F9 lines differentiated to PE phenotype exhibit altered levels of SPARC mRNA. (A) Autoradiogram of Northern blot performed with poly(A)+ RNA from lines AS-3 and AS-2 (lanes 2 and 3, respectively), S-27+ (lane 4), or Neo (lane l), differentiated for 12 days with RA/dicAMP. RNA (5 rg) was resolved on agarose gels, blotted onto nitrocellulose, and hybridized with a cDNA probe for SPARC (2.2 kb) and L32 (0.5 kb) RNA. C, control; AS, antisense lines; S, sense lines. (B) Densitometric plots of cell lines depicted in A. Values are normalized to the arbitrary value of Neo = 1 and are plotted as the mean of duplicate scans f SEM.

somal protein mRNA from two control cell lines, PYS and Neo, and from representative AS and S lines. In using an antisense-oriented riboprobe, we anticipated detecting extremely low levels of SPARC mRNA in the AS lines. As can be seen in AS-3, the levels of endogenous SPARC mRNA were clearly reduced (Fig. 3A, lane 3). Although the amount of L32 mRNA present in AS-3 was low, the amount of L32 mRNA present in the positive control (PYS cell line; Fig. 3A, lane 1) was much lower, but the hybridization signal for SPARC mRNA in the PYS line was extremely high. We therefore concluded that there was enough undegraded mRNA present in these samples for an accurate quantitative evaluation of the various transcripts. The two S lines depicted in Fig. 3 are also good examples of the range of overexpression of mRNA induced in the various lines by the use of the two different promoters or as a consequence of different copy numbers or integration sites of the foreign DNA. In this case, the overexpressed SPARC mRNA levels varied from a slight increase (line S-8) to greater than twofold (line S-27+ = S-27 line + ZnCl,), relative to the control levels expressed by Neo and evaluated by duplicate densitometric scans (Figs. 3A and 3B). When S and AS lines were differentiated to PE with RA/dicAMP and poly(A)+ RNA was prepared and analyzed for SPARC and L32 transcripts (Fig. 4A), the AS lines continued to display diminished expression of SPARC mRNA during differentiation, relative to control levels, as determined by normalization of duplicate

AND

SAGE

densitometric scans. Since these two lines carry pSVCRAPS DNA, the SV promoter might be activated in differentiated F9 cells, as has been reported by Gorman et al. [44]. However, in the mRNA evaluated from S-27+ (mRNA transcribed from pMT-SPARC), expression did not rise above that displayed by the control line Neo (Figs. 4A and 4B). To determine whether SPARC transcripts originating from transfected DNA were being translated and processed correctly as SPARC protein, we measured the production of SPARC from the transfected ceil lines by five different methods. Two methods, indirect immunofluorescence and immunoblot analysis with antiSPARC antibodies, showed enhanced or diminished expression of SPARC within S and AS cells, respectively, and in conditioned culture medium from the lines. In another assay, SPARC was shown to be present on the cell surfaces of control and S lines, as the addition of antibodies against SPARC peptide 1.1 [37] to the culture medium of transfected lines, followed by replacement of medium with medium containing non-hi-FBS, resulted in O-5% lysis of control lines F9 and Neo, O-2% lysis in AS lines, and 40-100% lysis in S lines. Quantitative measurements of SPARC protein were performed by immune precipitation and RIA. Protein recovered from immune precipitates derived from equal numbers of cells demonstrated that S and AS lines secreted SPARC into the culture medium in quantities above or below control levels, respectively (Fig. 5A). The precipitated protein migrated at the correct molecular weight for SPARC, kf, = 40,000-43,000 [ 11. An additional band of apparent M, 46,000 is believed to represent an alternate glycosylation product. RIA was performed to determine more precisely the amounts of SPARC secreted by the various cell lines. Equal numbers of cells were plated for 6 h on gelatincoated or uncoated cultureware, and conditioned medium was assayed for SPARC with polyclonal antiSPARC antibodies. The amount of SPARC produced by the F9 control line was given a value of 1, and that produced by the transfected lines was adjusted to a percentage of control for each experiment. The Neo line secreted levels of SPARC equal to those measured for F9, and the various S and AS lines secreted SPARC above or below control levels, as shown in Fig. 5B. Among the four S and three AS lines maintained throughout these studies, there was variability in the secreted levels of SPARC, fluctuations that did not necessarily correlate with amounts of integrated SPARC DNA or with the abundance of SPARC mRNA (compare Figs. 2 and 3 with Figs. 5A and 5B). For example, SPARC mRNA in the S-8 line is present at levels approximately 11% above those of Neo-transfected cells (Fig. 3B). However, the S-8 values for SPARC in the RIA were higher than control levels by 47 and 30%, depending on whether the

DIFFERENTIAL

NE0 S-5 dye--

SPARC

EXPRESSION

S-S AS-1 AS-2 AS-S CELL LINE

FIG. 5. SPARC protein assayed from conditioned medium of stably transfected FS cell lines. (A) Immunoprecipitation. Equal numbers of stably transfected cells were incubated with [?S]Met, and equal volumes of culture medium were subjected to immunoprecipitation with polyclonal anti-SPARC IgG. Protein was resolved on a 5% stacking/lo% separating gel in the presence of 50 mM dithiothreitol. C, control; AS, antisense transfectants; S, sense transfectants. Medium from line Neo, lane 1; AS-l, lane 2; AS-2, lane 3; S-8, lane 4; S-5, lane 5. Positions of the molecular weight standards are indicated (1M, X 10m3). Dye, dye front. The bands at 43,000-45,000 that appear in each lane correspond to control, reduced, or enhanced levels of SPARC secreted from the transfected cell lines. (B) Radioimmunoassay. Medium from equal numbers of cells was assayed in triplicate for SPARC, with ‘251-labeled SPARC and a polyclonal antibody specific for a synthetic peptide fragment of SPARC. Dark bars, cultureware left uncoated (n = 5); hatched bars, cultureware coated with gelatin (n = 4). Values are normalized to values of SPARC production by F9 cells = 1 and represent the mean + SEM. Inset, values for pMTSPARC line S-27+ are derived from duplicate scans of autoradiograms representing an immune precipitation and an immunoblot and are plotted relative to the levels of SPARC produced by untransfected F9 cells in those assays. AS, antisense; S, sense; Neo, control; S-27 and S-27+, sense line pMT-SPARC i ZnCI,.

cells were on gelatin-coated or noncoated cultureware (Fig. 5B). These values represent a significant increase in a protein that is normally expressed at nearly undetectable levels. Figure 5B also depicts the inductive effect that the addition of ZnCl, had on the MT promoter of the pMTSPARC plasmid. Shown in the inset are the SPARC protein levels for lines S-27 and S-27+; line S-27 exhibited a level of SPARC slightly above that of control, but in the presence of ZnCl, SPARC rose to levels greater than twofold those of control lines F9 and Neo. Finally, RIA, immune precipitation, and immunoblot experiments confirmed that the SPARC protein originating from exogenous SPARC DNA was recognized by antibodies specific to SPARC. S and AS Cells Exhibit Altered Cell Morphologies The most striking effect of transfection of SPARC and CRAPS cDNA was the altered morphology of the

ALTERS

MORPHOLOGY

139

lines. Relative to control lines F9 and Neo, S lines uniformly displayed a more aggregated and clumped morphology, whereas AS lines demonstrated a flatter and more spread phenotype. Deviations from the closely packed, undifferentiated F9 and Neo morphology in the S and AS lines are shown in Fig. 6A. The altered S and AS morphologies persisted when the lines were plated on cultureware coated with gelatin (Fig. 6B); similar results were found when cells were plated on FN or type I collagen (Vitrogen-100, Collagen Corp.). The S lines remained clumped and aggregated on these substrates, and the AS lines remained more spread than the control lines F9 and Neo. There appeared to be an association between the level of over- or underexpression of SPARC and the degree of morphological change. For example, the rounded morphology of the S-27 line was enhanced in the presence of ZnCl,, i.e., the addition of ZnCl, to the culture medium rendered the morphology typical of this S line more readily noticeable. The ZnCl,-induced S-27+ line is one of the lines exhibiting both the highest levels of SPARC and the most prominent rounded phenotype. Addition of ZnCl, to the culture medium of other cell lines did not contribute to the appearance of the altered morphologies. Moreover, addition of ZnCl, did not result in any obvious changes in the levels of expression of SPARC by the cells, except in the pMT-SPARC line, in which expression was predictably enhanced. The morphological changes persisted during differentiation of the S and AS lines to the PE phenotype (Fig. 7). Differentiated control F9 cells assumed characteristic PE phenotypes when treated with RA/dicAMP for 8-12 days: cells appeared singular and highly migratory, grew out and away from the stem cell clusters, and developed elongate cell processes. Additionally, the cells formed the adherent embryoid bodies typical of endodermal aggregates (Fig. 7, F9). In contrast, differentiated AS lines gave rise to extremely flattened, separated cells, with few or no embryoid body formations or cell-cell contacts. Cell processes were both more numerous and more elongated than those observed in the control lines F9 and Neo (Fig. 7, AS-2). It appeared that the AS construct inhibited the cells from forming embryoid bodies and promoted the flat and migratory phenotype which appears prior to embryoid body formation during the differentiation of F9 to PE. S lines differentiating to PE appeared strikingly different. Cells initially became closely apposed and subsequently aggregated into tightly compacted embryoid bodies; few if any individual cells were present (Fig. 7, S-27). The high level of SPARC expressed by the undifferentiated S lines prior to differentiation thus appears to have predisposed the differentiating cells toward the formation of the closely packed embryoid bodies.

140

EVERITT

AND

SAGE

FIG. 6. Undifferentiated S and AS lines display altered morphologies. Cells were treated with trypsin and EDTA, dispersed as single cells, counted, and replated. AS, antisense lines; S, sense lines. (A) Uncoated cultureware. a, F9, untransfected; b, Neo, control line; c, AS-I; d, AS-2; e, S-8; f, S-27+. (B) Gelatin-coatedcultureware. a, F9, untransfected; b, Neo, control line; c, AS-2; d, AS-3; e, S-27+; f, S-8. The background of e and f appears darker due to the altered focal plane necessary to see the cells in clumps. Magnification is 100X.

Characterization Morphology

of Cell Lines Exhibiting

Altered

Experiments were performed to determine whether the changes observed in morphology were a direct consequence of altered expression of SPARC in the S and AS lines or the result of an indirect effect on cell-cell or cell-substrate interactions that were sensitive to SPARC. Various culture conditions were changed to assess whether the morphologies would revert to the parental F9 phenotype or would cause the parental F9 cells to assume the S or AS phenotype. As summarized in Table 2, none of the factors assessed in culture caused a reversion of S or AS lines to an F9 phenotype or caused F9 cells to appear more as S or AS lines; however, a few culture modifications either diminished or enhanced the S and AS morphologies. Coating the cultureware with gelatin (Fig. 6B), high density platings,

incubation times of ~24 h, changes in FBS concentration, and use of high trypsin/low EDTA solutions to passage the cells were factors that minimized the differences between the morphologies of S and AS lines, compared to control. Other changes in culture had no apparent effect (Table 2, I). We also asked whether biochemical or morphological characteristics typical of F9 cells had been perturbed by altered expression of SPARC (Table 2, II). Surprisingly, the ability of the S and AS lines to express normal mRNA levels of specific endodermal markers of differentiation (AFP, type IV collagen, and laminin) was unaffected. Thus, cells were able to express appropriate gene products associated with the PE and VE phenotypes, despite their morphological alteration during differentiation. Cortical cellular actin and actin mRNA levels were similarly unaffected. The increased proteolytic activity of PE cells [ 421, and the possibility that levels of

DIFFERENTIAL

SPARC

EXPRESSION

ALTERS

MORPHOLOGY

141

FIG. 6-Continued

reted, matrix-degrading enzymes might have been ?red relative to SPARC, prompted us to perform zy‘graphy on S and AS lines. Labeled conditioned mem and cell extracts from undifferentiated lines, and

plasmin-activated conditioned medium from both undifferentiated and differentiated lines, showed no vi 3riation from control levels with respect to secreted co11agenases or procollagenases (data not shown). Immune Iblot

analysis also showed no differences among the lines in the expression of the secreted proteinase transin (murine stromelysin). Hence, the changes in morphology among the cell lines were not due to perturbations in the levels of matrix-degrading enzymes such as collagenases or transin. The ability of the S and AS lines to attach in culture was also not affected (Fig. 8). Between 30 and 60 min, cells attached to FN-coated substrates at rates equal to or exceeding control levels. These values were equal to the values for control lines at later time points (6, 24, and 48 h). Although transfection of SPARCXRAPS DNA had resulted in striking changes in cell shape, the relative levels of SPARC did not alter the strength of S or AS cell-substrate contacts that ostensibly functioned normally to promote cell attachment. DISCUSSION

FIG. 7. S and AS lines treated with RA/dicAMP for 12 days reCells were treated ta in altered morphologies during differentiation. and replated on noncoated tissue culture dishes wi Ith trypsin/EDTA were as described in the legend to Fig. 6; medium and RA/dicAMP ch ranged every 2 days. S, sense; AS, antisense lines, a, F9, control line; is 100X. h AS-2; c, S-27. Magnification

We have isolated several lines of stably transfected F9 cells that constitutively overexpressed or underexpressed SPARC. The cell lines were strikingly altered in their morphology but otherwise appeared to be similar to parental and control Neo cells by all of the biochemical and molecular criteria that we evaluated. Following transfection, the lines did not spontaneously differentiate into PE or VE but maintained characteristics typical of F9 cells. They divided every 11-12 h (a cell cycle time that corroborates that previously reported for F9 cells) [46], they gave rise to undifferentiated tumors when injected into compatible host mice, they expressed high levels of E-cadherin (uvomorulin) [47], and they did not synthesize tissue or urokinase plasminogen activator (data not shown). Furthermore, upon addition of RA or RA/dicAMP, the cells increased expression of VE-associated AFP or PE-associated type IV collagen and laminin (data not shown). Hence, transfection of the late differentiation gene SPARC did not prevent or abort the differentiation of F9 cells into their endoderma1 derivatives. Several of the late gene products of differentiation (e.g., AFP, laminin, type IV collagen, and SPARC) clearly appear not to function as the early RAinducible genes (see Introduction) in the initiation of the differentiation process that indicates commitment to the endodermal lineage (431. The later proteins seem more associated with the phenotype of fully differentiated PE or VE. Although we have not strictly shown that the changes in SPARC mRNA resulted from the transcription of exogenously transfected SPARCXRAPS cDNA, it is highly unlikely that an increased or decreased transcription of the endogenous SPARC gene would have accounted for our results. We characterized and maintained four SPARC overexpressing lines, derived from transfections of two different expression vectors, and

DIFFERENTIAL

TABLE Conditions

Altered

EXPRESSION

ALTERS

143

MORPHOLOGY

2

Tested for Their Ability to Modify the Morphologies of S and AS Cell Lines

I. Condition A. B. C. D. E. F. G. H. I. J. K. L.

SPARC

tested

Tissue culture plastic Coating on cultureware + G418 antibiotic Cell plating density* Time of incubation’ + exogenous SPARCd Soluble factor?’ Concentration of FBS’ f anti-peptide antibodiesB f low trypsinjhigh EDTAh Differentiation (RA/dicAMP) Attachment/adhesion II. Proteins

assayed

M. Actin” N. Metalloproteinases/transinj 0. Endoderm-specific markersA

Effect 0” 1 0 2 2 0 0 2 0 2 0 0 Effect 0 0 0

a Values: 0 indicates no change was observed between S and AS cell lines when the parameter was changed or assayed. A value of 1 inclcates that a minimal change was observed in S and AS morphologies, and 2 indicates that S and AS cell morphologies were modified by the factor indicated. * Morphologies became most apparent when cells were plated at sparse densities and were allowed to proliferate in culture for 24-48 h. ‘Altered S and AS morphologies developed during incubation; within O-18 h after plating there were no observable differences in the appearances of single cells of different lines. d Purified SPARC, synthetic peptides of SPARC, and recombinant SPARC were added to the cell lines in culture in quantities previously described as being sufficient to elicit a rounded phenotype in other cell types [lo, 371. ‘Parental F9 cells were incubated in the same culture dish and medium as S-27+ cells for various time intervals and photographed. ‘Raising and lowering the percentage of FBS added to DMEM altered the morphologies of all the lines, including the parental F9 and Neo. FBS (2.5-5%) induced a more flattened and spread cell phenotype, whereas 15-20% FBS induced more clumped, aggregated morphologies in all lines. ‘Addition of anti-peptide SPARC antibodies [37] to the culture medium of the lines did not cause a reversal of S or AS phenotypes to the control F9 phenotype. h Cells needed to be dispersed and replated as a single cell suspension for the characteristic morphologies to appear. ’ Distribution of actin and abundance of mRNA in the cell lines were tested by direct immunofluorescence on cells and by Northern analysis of poly(A)+ RNA, respectively. j Zymography was performed on cells and culture medium, unactivated and activated with plasmin, from undifferentiated lines and on activated medium from differentiated lines. Immunoblot analysis of transin was performed on culture medium from undifferentiated cell lines. k PE markers SPARC, B2-laminin, and (~1 (IV) collagen were examined by Northern blotting with poly(A)+ RNA from differentiated lines, followed by scanning densitometry. Differences in abundance of mRNA for these gene products among control, S, or AS lines did not correlate with the production of SPARC protein.

lhr

‘1

24hr

48lw

Time of incubation FIG. 8. Attach men1t and adhesion of S and AS lines. Cells were labeled overnight with [3H]thymidine or leucine, treated with trypsin/EDTA, counted, and plated in duplicate on FN-coated cultureware. After the designated incubation interval, medium was aspirated and wells were washed thrice with PBS. Cells from medium and washes, as well as adherent cells, were solubilized separately in 1 N NaOH, and incorporated radioactivity was measured. Values shown are normalized to control (Neo, white bars) as 100%; black bars, sense S-5; hatched bars, sense S-27+; striped bars, antisense AS-l; dotted bars, antisense AS-3 (mean -t SEM of duplicate experiments).

three CRAPS underexpressing lines, from two different bacterial transformants. The number of lines isolated provides evidence against the possibility that one common integration site in the F9 genome could have been responsible for expression of the same protein in all the S or AS cell lines. The various lines were selected according to their levels of SPARC expression. Increases in SPARC in the S lines ranged from 12 to 340% above control, whereas SPARC in the AS lines was inhibited from 28 to 32% of the normally low F9 levels; these values fall within a biological range in which one might expect to see alterations in cell behavior or morphology. It has recently been shown that modest decreases in protein from transfected genes can exert extreme morphological effects on cells. When the antisense-oriented cDNA for the cell surface proteoglycan syndecan was transfected into mouse mammary epithelial cells, expression of syndecan in cloned cell lines ranged from ~10 to >40% of control levels. At the lower level of expression, the cells exhibited major morphological changes, but lines that expressed syndecan at higher levels appeared unaltered from their normal epithelial cell morphology [48]. Since undifferentiated F9 cells secrete extremely low levels of SPARC, the two- to threefold increases (or decreases) we saw with our S and AS lines represent significant changes in the transfected cells. There appears to be no single morphology associated with cell types that normally secrete high levels of SPARC. Subconfluent bovine aortic endothelial (BAE)

144

EVERITT

cells secrete an M, 43,000 protein shown to be identical to SPARC [l]; these cells appear polygonal and spread in culture. Murine Leydig cells maintain a round morphology and secrete SPARC in uiuo and when plated in vitro on Matrigel, an extract of basement membrane matrix [49,50]. However, they assume a flat and spread morphology on tissue culture plastic that is associated with a 3.5fold increase in SPARC mRNA over the levels expressed on Matrigel [50]. PYS and fetal calf ligament (FCL) fibroblast cells, which secrete significant amounts of SPARC prior to confluence, appear spread and flat in uitro. In contrast, PE cells in uiuo express extremely high levels of SPARC and appear distinctly round on Reichert’s membrane (see Fig. 3B in [9]), the basement membrane which they secrete. PYS cells, which are similar to PE cells in uitro, secrete large quantities of laminin and type IV collagen, basement membrane components to which they also adhere. These proteins collectively present a substrate for the attachment and spreading of PE cells that F9 cells transfected with SPARC would not encounter, as they do not synthesize these components. Since transfection of the single gene SPARC clearly does not convert an F9 cell to a PE cell, there is no reason to expect that the morphology of the S-transfected lines would be that of PE. Furthermore, as shown in Fig. 7, AS cells differentiate into spread cells of the PE phenotype, and S cells differentiate into rounded cells that form embryoid bodies, but both transfected cell types initiate transcription of the PE-associated products laminin and type IV collagen. Thus, cell shape in this instance cannot be equated with the acquisition of a specific differentiated phenotype. There presently appears to be no obvious correlation between a cell type and its ability to change shape in response to exogenous SPARC [lo]. For example, BAE and FCL cells round in response to SPARC in vitro, but PYS cells do not; however, all of these cells express high levels of SPARC prior to confluence. Normal mouse skin fibroblasts are sensitive to SPARC, whereas the mouse fibroblast cell line 3T3 is not. Since we initially observed that F9 cells do not respond to purified SPARC added in culture, we performed the stable transfections as described in this report. The differences between our previously reported results with SPARC and F9 cells and those presented in this study can be addressed as follows. First, we have found no established cell line that responds to exogenous, purified SPARC, whether the cell line expresses SPARC (PYS) or does not (F9). Second, it is formally possible that the effect of SPARC on cell morphology is mediated intracellularly, prior to the export of SPARC to the cell surface. We and others have not strictly proven that SPARC exerts its biological function as a secreted protein. If F9 cells have no receptor for SPARC, addition of exogenous SPARC might not result in a percepti-

AND

SAGE

ble change in their morphology. Third, if SPARC exerts an extracellular effect through a receptor on F9 cells, the AS lines might have diminished their expression of this receptor and would therefore be refractile to the addition of exogenous SPARC. Finally, S cell lines that secrete high levels of SPARC protein will accumulate SPARC in their extracellular environment at concentrations above those that could be achieved by the single addition of exogenous SPARC. The ability of SPARC to bind to ECM components and evoke shape change through cell-matrix interactions [lo] might in fact may be the critical difference between the addition of exogenous SPARC to the lines and their secretion of SPARC de nouo. The addition of exogenous SPARC generally occurs at a single time point during cell incubation, during which SPARC could become inactive, degraded, metabolized, or processed such that it became inactive. Continuous secretion of SPARC by the S lines would tend to minimize these losses. In addition, as shown in Table 2, the cell lines exhibit their changes in morphology over 24-48 h in uitro, presumably the result of the chronic secretion or lack of secretion of SPARC. Therefore, the concentration of SPARC, the secretion of SPARC into matrix, the processing of SPARC intracellularly, the possibility for the formation of complexes between SPARC and ECM molecules, and the interaction and presentation of SPARC through serum-binding proteins are reasonable explanations for the response of cells to endogenous SPARC (or to the lack of SPARC). They also speak to the failure of added SPARC to elicit a change in cell morphology. The strongest evidence that the altered morphology of the S and AS lines is due to SPARC, as either a primary or a secondary effect, is the number of S and AS lines that exhibited the same effects. In addition, the opposing S and AS constructs and the subsequent SPARCKRAPS effect on the cells reinforce our claim that SPARC is inducing the morphological effect in the lines. Whether this effect is due to the concentration, localization, processing, complex formation, or presence intracellularly or extracellularly of SPARC has not been addressed in these experiments. Undifferentiated AS lines presented morphologies that were markedly different from those of the S lines, and the morphologies of control lines F9 and Neo were intermediate between S and AS cells. Thus, the AS and S lines become appropriate controls for each other, and they corroborate the results in vitro that SPARC added to certain cell types in culture causes them to round up and remain unspread [lo]. Conversely, inhibition of expression by the AS lines of the small amount of SPARC that F9 cells typically produce causes them to become more spread and flat on the culture substrate in comparison to their parental F9 cells.

DIFFERENTIAL

SPARC

EXPRESSION

An interesting finding was that the attachment of the S and AS lines to various substrata was unaltered. The S and AS cells, round or spread, were able to maintain normal mechanisms for adhesion. Although the changes in shape are presumably correlated with alterations in cell-substrate contacts, adherence properties of transfected F9 cells were quantitatively not diminished in strength when compared to those of control F9 and Neo cells. The apparent relationship between the amount of SPARC and the altered morphology could be either a primary effect of SPARC itself or one of the influences of SPARC on other aspects of cell spreading. If a cell is unable to spread effectively on a given substrate, it might exploit other mechanisms; e.g., perturbation of normal cell-substrate interactions by the overexpression of SPARC might predispose cells to adhere to other cells. However, examination of cell surface uvomorulin showed that this intercellular adhesion molecule was expressed at levels that were inconsistent with those of SPARC in S and AS cells (E. A. Everitt and E. H. Sage, manuscript submitted). We have shown that stably transfected F9 cell lines that overexpress SPARC behave as other cells cultured in the presence of exogenous SPARC. This protein clearly affects the ability of cells to spread on various substrates without altering either specific proteinases or the strength of adherence of the cells to their substrata. Perturbation of spreading appears to be attributable to SPARC itself, possibly as a consequence of changes in the synthesis of ECM proteins such as plasminogen activator inhibitor [51]. Future experiments will focus on defining the pathway and signaling mechanisms by which these alterations in cell shape occur. We thank P. Bornstein, W. Couser, B. Hogan, R. Kaspar, T. Lane, L. Matrisian, R. Moon, and J. Yost for generously providing their reagents. Also, we thank J. Yost for assistance with the SPARC RIA, M. Nameroff for assistance with cell cycle analyses and reading of the manuscript, members of the Sage lab for suggestions, and P. Hasselaar and M. Nameroff for stimulating discussions. A portion of this work was presented in poster form at the 1989 and 1990 American Society for Cell Biology meetings, Houston, TX, and San Diego, CA, respectively, and appears as abstracts (E. Everitt and H. Sage, 1989, J. Cell Biol. 109, 64a; 1990, J. Cell Biol. 111, 17a). This work was supported by a Public Health Service National Research Service Award 5T32-GM07270 grant from National Institutes of Health to E.A.E. and National Institutes of Health Grant GM-40711 to E.H.S.

Holland, P. W. H., Harper, S. J., McVey, J. H., and Hogan, B. L. M. (1987) J. Cell Biol. 105, 473-482.

6.

Mann, K., Deutzmann, R., Paulsson, FEBSLett. 218,167-172.

7.

Findlay, D. M., Fisher, L. W., McQuillan, C. I., Termine, and Young, M. F. (1988) Biochemistry 27,1483-1489.

2. 3.

127,373-387. 4.

Zung, P., Domenicucci, C., Wasi, S., Kuwata, (1986) Biochem. Cell Biol. 64, 356-362.

F., and Sodek, J.

M., and Timpl,

R. (1987) J. D.,

8. Wewer, U. M., Albrechtsen, Termine, 9. 10. 11. 12. 13.

14. 15. 16. 17. 18.

R., Fisher, L. W., Young, M. F., and J. D. (1988) Am. J. Pathol. 132,345-355.

Sage, H., Vernon, R. B., Decker, J., Funk, S., and Iruela-Arispe, M. L. (1989) J. Histochem. Cytochem. 37,819-829. Sage, H., Vernon, R. B., Funk, S. E., Everitt, E. A., and Angello, J. (1989) J. Cell Biol. 109, 341-356. Sage, H., Decker, J., Funk, S., and Chow, M. (1989) J. Mol. Cell Cardiol. 21,13-22. Howe, C. C., Kath, R., Mancianti, M. L., Herlyn, M., Mueller, S., and Cristofalo, V. (1990) Exp. Cell Res. 188, 185-191. Young, M. F., Day, A. A., Dominquez, P., McQuillan, C. I., Fisher, L. W., andTermine, J. D. (1990) Connect. Tissue Res. 24, 17-28. Sage, E. H., and Bornstein, P. (1991) J. Biol. Chem. 266,1483114834. Strickland, S., and Mahdavi, V. (1978) Cell 15, 393-403. Strickland, S., Smith, K. K., and Marotti, K. R. (1980) Cell 21, 347-355. Martin, G. R. (1980) Science 209, 768-776. Hogan, B. L. M., Taylor, A., and Adamson, E. (1981) Nature

291,235-237. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28.

Mason, I. J., Murphy, D., Miinke, M., Francke, U., Elliott, R. W., and Hogan, B. L. M. (1986) EMBO J. 5, 1831-1837. Dean, M., Levine, R. A., and Campisi, J. (1986) Mol. Cell Biol. 6, 518-524. Griep, A. E., and DeLuca, H. F. (1986) Pr~c. Natl. Acad. Sci. USA 83, 5539-5543. LaRosa, G. J., and Gudas, L. J. (1988) Proc. Natl. Acad. Sci. USA 85,329-333. LaRosa, G. J., and Gudas, L. J. (1988) Mol. Cell Biol. 8, 39063917. Hosler, B. A., LaRosa, G. J., Grippo, J. F., and Gudas, L. J. (1989) Mol. Cell Biol. 9, 5623-5629. Grover, A., and Adamson, E. D. (1986) Dev. Biol. 114,492-503. Rao, G. N., Church, R. L., and Davidson, J. N. (1988) FEBS Lett. 232, 238-242. Grover, A., Oshima, R. G., and Adamson, E. D. (1983) J. Cell Biol. 96, 1690-1696. Morita,

A., Kitagawa,

Y., and Sugimoto, E. (1985) J. Cell Physiol.

125,263-272. 29.

Van Doren, K., Hanahan, 50,606-614.

30.

Maniatis, T., Fritsch, E. F., and Sambrook, J. (Eds.) (1982) Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Sambrook, J., Fritsch, E. F., and Maniatis, T. (Eds.) (1989) Molecular Cloning: A Laboratory Manual, second ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Iruela-Arispe, M. L., Hasselaar, P., and Sage, H. (1991) Lab. Invest. 64, 174-186. Graham, F. L., and van der Eb, A. (1973) Virology 52,456-467. Ausubel, F. M., Brent, R., Kingston, R. E., Moore, D. D., Smith, J. A., and Struhl, K. (Eds.) (1987) Current Protocols in Molecular Biology, Vols. 1 & 2. Wiley, New York.

31. Mason, I. J., Taylor, A., Williams, J. G., Sage, H., and Hogan, B. L. M. (1986) EMBO J. 5, 1465-1472. Termine, J. D., Kleinman, H. K., Whitson, S. W., Conn, K. M., McGarvey, M. L., and Martin, G. R. (1981) Cell 26, 99-105. Sage, H., Tupper, J., and Bramson, R. (1986) J. Cell Physiol.

145

MORPHOLOGY

5.

REFERENCES 1.

ALTERS

32. 33. 34.

D., and Gluzman,

Y. (1984) J. Virol.

146 35.

EVERITT Chomczynski,

162,

P., and Sacchi, N. (1987) Anal. Biochem.

AND 44.

156-159. 36. 37.

Laskey, R. A., and Mills, A. D. (1975) Eur. J. Biochem. 56,335341. Lane, T. F., and Sage, E. H. (1990) J. Cell Biol. 111,3065-3076.

38.

Malaval, L., Fournier, Res. 2,457-465.

B., and Delmas, P. D. (1987) J. Bone Min.

39.

Tienari, J., Lehtonen, E., Vartio, Cell Res. 182, 26-32.

40.

Herron, G. S., Banda, M. J., Clark, E. J., Gavrilovic, Werb, Z. (1986) J. Biol. Chem. 261, 2814-2818.

41.

Brenner, C. A., Adler, R. R., Rappolee, D. A., Pedersen, and Werb, Z. (1989) Genes Deu. 3,848-859.

42.

Adler, R. R., Brenner,

T., and Virtanen,

I. (1989) Exp.

Mtiller,

46. 47.

J., and R. A.,

48. 49. 50.

C. A., and Werb, Z. (1990) Development

110,211-220. 43.

45.

51.

R., and Wagner, E. F. (1984) Nature

Received August 26, 1991 Revised version received November

13, 1991

311,438-442.

SAGE Gorman, C. M., Rigby, P. W. J., and Lane, D. P. (1985) Cell 42, 519-526. Sassone-Corsi, P., Duboule, D., and Chambon, P. (1985) In Molecular Biology of Development: Cold Spring Harbor Symposia on Quantitative Biology, Volume L, pp. 747-752. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Kurki, P., Laasonen, A., Tan, E. M., and Lehtonen, E. (1989) Development 106,635-640. Schuh, R., Vestweber, D., Riede, I., Ringwald, M., Rosenberg, U. B., Jackie, H., and Kemler, R. (1986) Proc. Natl. Acad. Sci. USA 83, 1364-1368. Kato, M., and Bernfield, M. (1990) J. Cell Biol. 111,263a. Vernon, R. B., and Sage, H. (1989) Biol. Reprod. 40,1329-1340. Vernon, R. B., Lane, T. F., Angello, J. C., and Sage, H. (1991) Biol. Reprod. 44, 157-170. Hasselaar, P., Loskutoff, D. J., Sawdey, M., and Sage, E. H. (1991) J. Biol. Chem. 266, 13,178-13,184.