Endothelial Heparan Sulfate Proteoglycan. I. Inhibitory Effects on Smooth Muscle Cell Proliferation William E. Benitz, Richard T. Kelley, Clay M. Anderson, Diane E. Lorant, and Merton Bernfield Division of Developmental and Neonatal Medicine, Department of Pediatrics, Stanford University School of Medicine, Stanford, California.

Proliferation of smooth muscle cells is an important component of pulmonary arterial morphogenesis, both during normal development and pathologic remodeling. However, little is known of the factors that regulate smooth muscle proliferation in these vessels. To investigate the hypothesis that factors produced by endothelial cells may regulate smooth muscle cell growth, we studied the effects of culture medium conditioned by fetal bovine pulmonary arterial endothelium on proliferation of smooth muscle cells in culture. This conditioned medium contains an inhibitor of smooth muscle proliferation that is degraded by nitrous acid, heparinase, and heparitinase, but resists degradation by protease, boiling, and chondroitin ABC lyase, indicating that the inhibitor is structurally similar to heparin. Inhibitor release occurs in both growing and confluent endothelial cell cultures and in the presence and absence of serum. A growthinhibiting proteoglycan purified to homogeneity from endothelial cell-conditioned medium has physicochemical characteristics similar to those of the prototypic basement membrane heparan sulfate proteoglycan of the Englebreth-Holm-Swarm tumor: an overall size of approximately 106 D, heparan sulfate chains of 60,000 D, and a buoyant density of 1.33 g/m!. Antibody raised against the tumor basement proteoglycan recognizes this endothelial heparan sulfate proteoglycan, and Western blotting after SDS-PAGE demonstrates that the core proteins of both proteoglycans migrate as a doublet at apparent molecular weights of 450,000 and 360,000 D. Heparan sulfate glycosaminoglycan prepared from purified medium proteoglycan is a potent inhibitor of smooth muscle cell growth, exhibiting activity approximately 1,000 times greater than that of heparin. These results indicate that endothelial cells cultured from fetal bovine pulmonary arteries produce a basement membrane heparan sulfate proteoglycan that is a potent inhibitor of smooth muscle proliferation. This proteoglycan may mediate endothelial regulation of smooth muscle growth during development or pathologic pulmonary arterial remodeling.

Acquisition of the muscular investment of the pulmonary arteries during development is characterized by progressive extension of medial smooth muscle into peripheral vessels and increasing thickness of this investment in proximal vessels (1-5). Acquisition of the muscular investment ofthe pulmonary arteries is accelerated during late fetal development in infants who have persistent pulmonary hypertension after birth (6). Pulmonary hypertension due to chronic hypoxia is

Key Words: endothelium, vascular smooth muscle, heparan sulfate, proteoglycan, growth regulation (Received in original form August 28, 1989 and in revised form October 12, 1989) Address correspondence to: William E. Benitz, M.D., Department of Pediatrics-S222, Stanford University School of Medicine, Stanford, CA 94305. Abbreviations: endothelial cells, EC; endothelial cell-conditioned medium, ECCM; endothelium-derived growth inhibitor, EDGI; Englebreth-HolmSwarm, EHS; glycosaminoglycan, GAG; heparan sulfate, HS; heparan sulfate proteoglycan, HSPG; smooth muscle cells, SMC; Tris-urea-Triton buffer, TUT. Am. J. Respir. Cell Mol. BioI. Vol. 2. pp. 13-24, 1990

associated with increased quantities of smooth muscle in the pulmonary arteries, with thickening of the medial muscle layer in the proximal arteries and extension of the smooth muscle investment into peripheral, normally nonmuscular vessels (7-10), both in humans and in laboratory animals. Normal and abnormal development, as well as pathologic remodeling of the pulmonary arteries, requires proliferation of smooth muscle cells or their precursors (11), but the identities of factors that regulate this process remain uncertain. Increased muscularity of these vessels is apparently not a result of direct effects of hypoxia upon the arterial smooth muscle cells themselves, because growth of these cells in vitro is reduced, rather than increased, under hypoxic conditions (12). Heparin inhibits proliferation of aortic (13, 14) and pulmonary arterial (15) smooth muscle cells in vitro as well as pulmonary arterial remodeling in chronically hypoxic animals (8), suggesting that factors that affect smooth muscle cell proliferation may regulate remodeling or morphogenesis of the pulmonary arteries. Because endothelial cells from calf aorta produce a heparin-like inhibitor of smooth muscle cell proliferation (16), we hypothesized that a similar material produced by pulmonary arterial endothelium might

14

AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 21990

regulate smooth muscle cell growth in these vessels. We now show that medium conditioned by endothelial cells cultured from the proximal pulmonary arteries of near-term bovine fetuses contains a heparan sulfate proteoglycan that inhibits smooth muscle cell growth.

Materials and Methods Cell Culture Smooth muscle cells (SMC) were cultured from the main pulmonary arteries of near-term bovine fetuses by the explant method (17) and characterized by phase and electron microscopy, as described previously (15). SMC were used for experiments only in the second to fourth passage after harvest from the primary culture. All SMC cultures were carried out in DMEM (GIBCO, Grand Island, NY). Endothelial cells (EC) were cultured from the pulmonary arteries of near-term bovine fetuses and characterized by phase and electron microscopy and by staining with anti-Factor VIII antibody (18) and acetyl-low density lipoprotein (19), as described previously (12). EC were used for experiments only in the second to fifth passage after harvest from primary culture. EC were cultured in medium M199 (GIBCO) supplemented with 10-5 M thymidine (20) or in DMEM. All culture media were supplemented with glutamine (2 mM), penicillin (10 V/ml) , streptomycin (10 p.g/ml), and FCS (Tissue Culture Biologicals, Tulare, CA). Primary cultures were harvested by rinsing the monolayer with Tris (20 mM)-NaCl (140 mM)-EDTA (0.5 mM) buffer (pH 7.4) and releasing the cells by digestion for 3 to 5 min with trypsin (0.05 %)-EDTA (0.5 mM) diluted 1:2 to 1:3 with Tris-NaCI-EDTA. Cells were used immediately for experiments or placed in 10% DMSO in FCS to be frozen in liquid nitrogen. All cultures were incubated in a humidified atmosphere of 95 % air and 5% COz at ~o C. Preparation of Conditioned and Radiolabeled Media Medium supplemented with 10% FCS was conditioned for 48 to 72 h over endothelial cell cultures that had been confluent for at least 2 d. Conditioned medium was collected, centrifuged to remove cells and particulate material, sterilized by filtration through 0.22-p.m Millipore filters, and either used immediately or frozen for use within 6 wk. Control (unconditioned) medium was prepared by incubation of identical medium at ~o C for 48 to 72 h in absence of EC. Metabolic radiolabeling was achieved by addition of carrierfree Hz35S0. (100 p.Ci/rnl; ICN Biochemicals, Irvine, CA) to the medium prior to conditioning. Growth Inhibition Assay Samples of medium digests or proteoglycan fractions were reconstituted for bioassay by exhaustive dialysis against DMEM (5 changes over 4 d). In most cases, fresh FCS (10%) was also added to each sample. The volume of each sample was adjusted to achieve the desired concentration by addition of DMEM and FCS (to a final concentration of 10%). BSA (1 mg/ml; Sigma, St. Louis, MO) was added to protease-treated samples that were not supplemented with serum; this was necessary to maintain attachment of SMC to the culture plates during the growth inhibition assay. Purified glycosaminoglycans (GAG) samples were simply

diluted in at least 10 volumes of complete medium for bioassay, or buffer exchanged on a Sephadex G-50 column to remove digestion reagents (acetic or nitrous acid), if necessary. Growth-inhibiting activity was assessed using the SMC growth assay described previously (15). Briefly, SMC were plated at 25,000 cells/35-mm dish in DMEM with 10% FCS and allowed to attach for 4 to 6 h. The plating medium was then removed, and experimental medium or control (unconditioned) medium was added to replicate dishes. Several dishes were frozen as zero time samples. After 6 d of incub-

as C

--

CD

18

as,c

_.!? i 32 _Cl

The Endothelium-derived Growth Inhibitor Is Heparin-like To determine whether the growth inhibitor produced by fetal pulmonary arterial endothelium is heparin-like, as is that produced by adult bovine and adult rat aortic endothelium (16), the activity of ECCM was compared to that of medium supplemented with heparin, which we have previously shown to inhibit growth of fetal bovine pulmonary arterial SMC in vitro (15). ECCM inhibited SMC growth to a greater extent (62.2 ± 4.2 %; P < 0.01 versus unconditioned medium controls) than did heparin at 100 J.'glrnl (33.6 ± 5.7%; P < 0.05 versus unconditioned medium controls, P < 0.05 versus ECCM). Supplementation of ECCM with heparin (100 J.'g/ml) reduced the growth-inhibiting activity to that ofheparin alone (36.8 ± 5.7 %; Figure 2), suggesting that heparin (acting as a partial agonist) may competitively inhibit the effects of an analogous inhibitor derived from the endothelium. To further explore the nature of this inhibitor, the biologic activity of ECCM was assessed after each of several treatments that degrade heparin-like material (Table 1). As noted above, the inhibitory activity of ECCM was increased after proteolytic digestion, suggesting that the inhibitor is not a polypeptide. Digestion with Flavobacterium heparinase, which selectively cleaves glycosidic linkages between N-sulfated hexosamine and iduronic acid 2-0-sulfate (23), resulted in a substantial reduction in the inhibitory activity of the medium. A similar reduction in inhibitory effect was achieved by digestion with nitrous acid at low pH (26), which cleaves glycosidic linkages adjacent to N-sulfated hexosamine residues (40). Structures susceptible to these treatments are unique to heparin and heparan sulfate (HS) GAG (41). A smaller decrement in the inhibitory activity of ECCM was produced by digestion with Flavobacterium heparitinase, which cleaves glycosidic linkages between sulfated or monosulfated N-acetyl glucosamine and unsulfated iduronic acid or glucuronic acid (24); such linkages are characteristic ofHS GAG, but are rare within heparin-like portions of those

12

C:::l..

0_ (,)

« Z

,c

---I

'Oi

6

C

C

o

100 1 P < 0.025

~ 75

p '" 0.005

.2 CJ A

B

c

o

~G)

E

F

Figure 1. Endothelial cell-conditioned medium (ECCM) contains an inhibitor of smooth muscle cell (SMC) proliferation. Bars indicate the DNA content of sparsely plated SMC cultures after 6 din various media: unconditioned medium (A), medium conditioned for 48 h by confluent endothelial cells (B), unconditioned (C) and conditioned (D) medium after digestion with Streptomyces protease and boiling, and unconditioned (E) and conditioned (F) medium after protease digestion, boiling, and repletion with 10% FCS. Growth in ECCM was significantly reduced, both before and after protease treatment, which increased the inhibitory activity of this medium. The ability of either medium to support SMC growth was obliterated by protease treatment (final DNA content not different from plating density), but was completely restored after serum repletion of unconditioned medium. All differences are significant at P < 0.001.

.D-

EC ~ 50 :::J :;::!E C,c

~ '0 ... 0

25

lfe en

oI

I

A

"

B

c

o

Figure 2. Heparin blocks the effects of ECCM on SMC proliferation. Growth of SMC over 6 d was markedly inhibited by ECCM (B) as compared to unconditioned medium (A). Heparin (100 p.g/ml) inhibited SMC growth to a lesser extent (C) and reduced the inhibitory activity of heparin-supplemented ECCM (D) to that of heparin alone.

Benitz, Kelley, Anderson et al.: Endothelial Heparan Sulfate Inhibits Smooth Muscle Growth

17

TABLE 1

Effects of digestions on the growth-inhibiting activity of endothelial cell-conditioned medium Inhibition of SMC Growth

Significance of Difference from Unconditioned Controls (P Value)

(%)

Test Medium

Endothelial cell-conditioned medium (ECCM) ECCM after protease (ECCM-P) ECCM-P after heparinase ECCM-P after nitrous acid ECCM-P after heparitinase ECCM-P after chondroitinase ABC

54.0 70.4 14.0 16.8 40.0 62.1

± ± ± ± ± ±

< 0.002

2.1 6.3 5.8 9.8 5.6 6.3

molecules (41). In contrast, digestion with Proteus chondroitin ABC lyase, which is specific for chondroitin-4-sulfate, chondroitin-6-sulfate, and dermatan sulfate (25), had no effect on the growth-inhibiting properties of ECCM. These results indicate that the growth inhibition observed in SMC cultures maintained in ECCM is mediated by a heparinlike GAG. Requirements for Endothelium-derived Growth Inhibitor Production by Endothelial Cells To determine whether EDGI is released by EC at low density, as well as by postconfluent cells, medium was harvested from EC cultures and replaced every 24 h after plating the cells at low density (1:8 split), and the capacity of each of these conditioned medium samples to inhibit SMC growth was assessed (Figure 3). All samples inhibited SMC growth, although the inhibitory effects of medium conditioned during the first 2 d after plating was modest. A steady increase in inhibition, roughly parallel to the concomitant increase in cell numbers, was observed until day 6, when the cultures

< 0.001 NS NS < 0.01 < 0.001

~

~

~

'Oi cO~

CQ)_

°0 ==0

O~

~

~~

:SQ)

:c

-

~ 40

10 ~

:et

-Z

Cj

_~

~~20 (,) -

5

~O

Q)O ~ E 0 0 001234567

~ 0 = Q) C) ~ ::i

'0-

"

C

W

Days after Plating of EC Figure 3. Proliferating endothelial cells release growth inhibitor into culture medium. Medium harvested daily from an endothelial cell culture plated at low density (1:8 split) was assayed for growthinhibiting activity (open circles). All conditioned media significantly inhibited SMC growth (P < 0.005 for media harvested at days 1 and 2, P < 0.001 for days 3 through 7). Maximal growth inhibition was achieved in medium collected on day 6, when the culture reached confluence. The density of endothelial cell cultures was determined by measurement of their DNA contents (closed circles).

< 0.05 < 0.002 < 0.005 < 0.01 NS

were fully confluent. Maximal inhibitory activity was observed only in the medium conditioned by fully confluent cultures (collected on days 6 and 7). Nonetheless, because significant inhibitory effects were exerted by medium conditioned on the third and fourth days after plating, before the EC achieved confluence, production of EDGI appears to be unrelated to cell contact. To determine whether platelet-derived enzymes in serum are required for release of EDGI from fetal pulmonary arterial EC, medium supplemented with BSA (0.1 %), with FCS (1, 2.5, 5, or 10%), or with neither material was conditioned over EC monolayers for 48 h. The growth-inhibiting activities ofthese media were assessed after supplementation with FCS to achieve a final concentration of 10%. SMC growth was inhibited by all of these media to a similar extent (60 to 75%), indicating that serum was not required for inhibitor release (Figure 4). There was a modest, but statistically significant, increase in inhibitory activity in medium supplemented with 10% serum, as compared to medium

80

0 0 0 ·

Significance of Difference from Medium Not Subjected to Last Digestion (P Value)

i

i

~

'Oi C 0

.2-

0

.Ci)

x E Q.

c

CD

-

C

U

"If' 0 I

0

I

1

f/)

I

I

I

I

I

6

I

1-1-1-1-1~11.1

11

16

Fraction Number

It)

M

-

-E

~

B

?2

c 0

:;::

--· u

ca ~

M

1

,..

0

x E Q.

U

"If' 0

0

f/)

It)

M

0.0

0.5

Ve/Vt

1.0

C

Figure 8. Purification of HSPG from ECCM. A. Ion exchange chromatograph of ECCM. ECCM was supplemented with medium metabolically labeled with 35S04 , digested with testicular hyaluronidase, and chromatographed on DEAE-Sephacel in 6 M urea at pH 7.4, eluting with a step gradient from 250 to 500 mM NaC!. Fractions containing radiolabeled macromolecules were pooled as indicated by the bar. B. Buoyant density gradient ultracentrifugation of pooled materials from ion exchange chromatography. The pooled fractions eluted from DEAE-Sephacel were buffer exchanged into Tris-acetate, digested with chondroitin ABC lyase, and subjected to buoyant density gradient ultracentrifugation in cesium chloride as described in MATERIALS AND METHODS. Fractions recovered at buoyant densities ~ 1.3 g/rnl were pooled as indicated by the bar.

cluding those with pulmonary hypertension due to chronic hypoxia, chronic pulmonary parenchymal disease, or congenital heart disease (7-10). These morphogenetic processes require proliferation of SMC or their precursors (11), but little is known of the factors that may regulate proliferation of these cells. We have previously demonstrated that heparin inhibits proliferation of SMC from the pulmonary arteries of near-term bovine fetuses (15). Others have reported that EC from systemic arteries of mature animals produce heparinlike inhibitors of smooth muscle proliferation (16, 44). We hypothesized that EC from the fetal pulmonary arteries produce a similar growth inhibitor. We now show that medium conditioned by EC cultured from the proximal pulmonary arteries of near-term bovine fetuses contains an HSPG which inhibits smooth muscle cell growth. Endothelial Basement Membrane Heparan Sulfate Proteoglycan Is an Inhibitor of Smooth Muscle Cell Proliferation Medium conditioned by EC from the pulmonary arteries of near-term bovine fetuses inhibits the proliferation of SMC in vitro. Repletion experiments demonstrate that reduced growth in this conditioned medium does not result from depletion of medium or serum components during conditioning. The inhibitory activity resists destruction by protease digestion and boiling, implying that it is not mediated by a polypeptide. In fact, the inhibitory activity of conditioned medium is consistently increased after proteolytic digestions. It is unlikely that this results from destruction of polypeptide mitogens of endothelial origin (although such mitogens are undoubtedly present), because the FCS used to promote SMC growth in the bioassay system is more than sufficient to maximally stimulate SMC growth throughout the 6-d assay period. This enhancement of growth-inhibiting activity could result from activation of a latent precursor, such as transforming growth factor-J3 (TGF-J3) (45), or from an increase in the effective concentration of inhibitor because of release of multiple active fragments from a single protease-susceptible precursor, as might result from proteolytic liberation of GAG chains from a core protein. Susceptibility of this activity to digestion with heparinase or nitrous acid (Table I) aJld apparent competitive inhibition of its effects by heparin (Figure 2) suggested that the growthinhibiting effects of ~CCM could be mediated by an HSPG. To confirm this tentative identification of the EDGI, HSPG was purified from ECCM. This proteoglycan elutes from DEAE-Sephacel at 350 to 450 mM NaCI (at pH 7.4), is of low buoyant density (modal density, 1.33 g/ml), has a large hydrodynamic size (K. v "J 0.07 on Sepharose CL-4B, corresponding to a molecular weight of circa I~ D), and bears GAG chains of apparent molecular weight of 60,000 D (K" of 0.45 on Sepharose CL-4B). This proteoglycan therefore corresponds to the previously characterized HSPG recovered from medium conditioned by adult bovine aortic en-

C. Size exclusion chromatography of HSPG. Pooled high density

fractions from cesium chloride gradients were chromatographed on Sepharose CL-6B as described in MATERIALS AND METHODS. Fractions eluting at the excluded volume of the column were pooled as indicated by the bar.

21

Benitz, Kelley, Anderson et al.: Endothelial Heparan Sulfate Inhibits Smooth Muscle Growth

400 kD300 kD200 kD_ 500 kD -400 kD -300 kD

100 kD-

AS

CD E FG

Figure 9. SDS-PAGE and Western blotting of endothelial HSPG. HSPG prepared by ion exchange chromatography( density gradient centrifugation, and size exclusion chromatography, as described in MATERIALS AND METHODS and Figure 8, were electrophoresed on a polyacrylamide gradient (6 to 15%), both before (lane A) and after digestion with nitrous acid (lane B). Protein and GAG were visualized by silver staining. All staining apparent in lane A is attributable to HSPG (M, > 400 kD) and HS GAG; no protein bands are apparent in either lane. Comparison with staining of BSA standards (not shown) indicated that the purity of the HSPG sample exceeded 99.6 %. Migration of rabbit phosphorylase standards in this gel is indicated at left. Endothelial HSPG (lanes C and D) and low density basement membrane HSPG from Englebreth-Holm-Swarm tumor (provided by Dr. John Hassell; lanes F and G) were electrophoresed on a polyacrylamide gradient (3 to 15 %) both before (lanes C and G) and after (lanes D and F) digestion with heparinase and heparinitase (to remove the GAG chains), and transferred to cationic nylon membranes as described in MATERIALS AND METHODS. Immunolocalization was achieved using antibody raised against the Englebreth-Holm-Swarm tumor HSPG. Migration of rabbit phosphorylase standards in this gel is indicated at right. Bands in lane E (no sample loaded) result from spillover from lane F.

dothelium (43, 46), and also resembles the low density HSPG purified from the basement membranes of the EHS murine tumor (34, 35). Western blotting demonstrated that both the endothelial HSPG and that obtained from tumor basement membranes are recognized by antiserum raised

against the latter material. Both proteoglycans migrated as typical broad smears on SDS-PAGE, and the core protein of each migrated as a predominant band at an apparent M, of 450,000, with a less prominent band at 360,000 D, as previously described for the basement membrane HSPG (47), suggesting that the endothelial HSPG may be homologous to the prototypic basement HSPG of the EHS tumor. Saku and Furthmayr have demonstrated that the activity of antibodies raised against the medium HSPG obtained from aortic endothelial cell cultures is indistinguishable from that of antibodies raised against the tumor HSPG, and that tryptic peptide maps prepared from the core proteins of these proteoglycans are nearly indistinguishable (48), providing further evidence in support of the hypothesis that the endothelial medium HSPG is a member of the family of basement membrane proteoglycans (49). Antigen recognized by antiserum to the EHS tumor proteoglycan has been localized to endothelial basement membrane both in vitro (43) and in vivo (50). We have shown that this endothelial basement membrane HSPG inhibits smooth muscle cell proliferation in vitro, both when tested as an intact proteoglycan (Table 2) and after proteolytic release of its HS GAG chains (Tables 3 and 4). The Growth-inhibiting Activity of Conditioned Medium Resides in Heparan Sulfate Chains This activity is greatly reduced after digestion with heparinase or nitrous acid, which both selectively cleave HS GAG at glycosidic linkages adjacent to sulfaminohexose residues (23,40), which are abundant in heparin-like HS GAG (51). This activity is also reduced, but to a lesser extent, by digestion with heparitinase, which cleaves HS GAG at glycosidic linkages adjacent to N-acetylglucosamine residues, which are characteristic of HS GAG, but unusual in heparin and similar extensively sulfated heparan sulfates (51). These data weigh strongly against TGF-{3 as a possible mediator of these effects, because it should resist degradation by heparinase and should actually be activated by exposure to the acidic conditions of nitrous acid treatment (45). Although reversal of growth inhibition by digestions of conditioned media was

TABLE 3

Effects of purified endothelial heparan sulfate glycosaminoglycan on smooth muscle cell proliferation Inhibition of SMC Growth (%)

Test Medium

Endothelial HS GAG (0.2 JLg/ml) Endothelial HS GAG (0.6 JLg/ml) Endothelial HS GAG (1.7 JLg/ml) Heparin (200 JLg/ml)

24.2 57.3 96.2 26.2

± ± ± ±

3.3 3.4 3.3 10.9

Significance of Difference from Control (P Value)

Significance of Difference from Heparin (P Value)

< 0.02 < 0.001 < 0.001 < 0.001

< 0.001 < 0.001

NS

TABLE 4

Nitrous acid digestion destroys the growth-inhibiting activity of purified endothelial heparin sulfate proteoglycan Inhibition of SMC Growth Test Medium

Heparan sulfate GAG Heparan sulfate GAG after acetic acid digestion Heparan sulfate GAG after nitrous acid digestion

(%)

72.7 ± 5.2 63.6 ± 5.1 -8.1 ± 4.2

Significance of Difference from Control (P Value)

< 0.001 < 0.001 NS

Significance of Difference from HS GAG (P Value)

NS

< 0.001

22

AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 21990

never complete, the residual growth inhibition observed after heparinase or nitrous acid digestion was not statistically significant. However, Castellot and colleagues (16, 42) also observed incomplete reversal of inhibition after heparinase digestion of medium conditioned by aortic EC, suggesting that heparinase digestion may have been incomplete, even though they used heparinase at a concentration approximately 60 times greater than that used in our experiments. Similarly, susceptible substrate may have survived digestion with nitrous acid in complete medium, because of quenching of this reaction by proteins or amino acids in this solution. In view of the observation of Castellot and colleagues (52) that growth-inhibiting activity is largely retained by heparin that has been N-desulfated and then N-reacetylated, some HS fragments that contain no sulfaminoglucose residues (and would therefore resist digestion by these procedures) may be large enough and sufficiently O-sulfated (52, 53) to retain growth-inhibiting activity. This is unlikely, however, because reversal of growth inhibition by purified HS GAG was complete after nitrous acid (Table 4). In contrast, no effect on the inhibitory activity was apparent after treatment with chondroitin ABC lyase, which selectively degrades chondroitin4-sulfate, chondroitin-6-sulfate, and dermatan sulfate (25), demonstrating that reduced growth inhibition after nitrous acid, heparinase, or heparitinase digestion is not simply a nonspecific artifact of glycosaminoglycan degradation. If the residual inhibitory activity of the medium digests (Table 1) is real, it could represent independent effects of another protease-resistant growth inhibitor. Our data do not allow exclusion of the presence of such additional growth-inhibiting materials in ECCM, because many fractions of the ECCM were discarded during the purification and only selected samples (including the purified HSPG) were assayed for growth-inhibiting activity. SMC growth was inhibited by both the partially purified HSPG (Table 2) and by HS GAG prepared from the purified proteoglycan (Table 3), but not by endothelial chondroitin sulfate proteoglycan (Table 2). The inhibitory effects of purified HS GAG, like those of conditioned medium, were destroyed by digestion with nitrous acid (Table 4). These experiments demonstrate that the inhibitor of SMC proliferation produced by EC cultured from the fetal bovine pulmonary arteries, like that produced by EC from aortas of adult rats and cattle (16), contains N-acetylglucosamine, sulfaminohexose, and 2-0-sulfo-iduronic acid, and is therefore a heparin-like HS. While the presence of these structures within the active portions of the molecule is consistent with the hypothesis that the active oligosaccharide sequence in the endothelial HS may be identical to that proposed as the active sequence in heparin (52), these data do not imply that these structures are essential for growth inhibition. The similar activities of the HS GAG and HSPG suggest that the core protein is not essential for growth inhibition. Although the endothelial HS GAG appears to be approximately 1,000 times more potent than beef lung heparin in inhibiting SMC proliferation in vitro (Table 3), it is not clear whether this is a consequence of a 1,000-fold greater frequency of the active oligosaccharide sequence in the EC HS or of the presence of a distinctive oligosaccharide of greater potency. Supplementation of ECCM with heparin reduces its

growth-inhibiting activity to that of the heparin alone, indicating that heparin may act either as an antagonist of EDGI or as a less potent inhibitor of SMC growth, suggesting that the active oligosaccharide sequences in these preparations may differ, or at least that heparin contains some oligosaccharide sequences that bind to the receptors that mediate the growth-inhibiting activity, but do not induce this effect, Determination whether this apparent interaction between heparin and EDGI is a result of competition for the same receptor or binding protein awaits detailed studies of binding of these molecules to their receptor(s), which may be at the cell surface (54) or within the cell (55), requiring accumulation of heparin intracellularly prior to manifestation of this biologic effect (56). The Heparin-like Growth Inhibitor Produced by Endothelium from the Fetal Pulmonary Artery May Differ from that Produced by Adult Aortic Endothelium Modest but statistically significant growth inhibition was apparent in medium conditioned by endothelial cell cultures in the first 2 d after plating at low density, with a steady increase in inhibitory activity as the cultures approached confluence 6 d after plating (Figure 3). Thus, release of heparin-like growth inhibitor by fetal pulmonary arterial endothelium does not require confluence of the monolayer, as is the case for adult aortic endothelium, which produces no detectable growth inhibitor while exponentially proliferating (16). These data do not allow estimation of the quantity of growth inhibitor released into the medium by the fetal EC during any of these conditioning periods, because it is likely that the relationship between percent inhibition of SMC growth and growth inhibitor concentration is nonlinear, as we have previously shown to be the case for heparin in this bioassay (16). Growth inhibitor production may actually be greater (on a per cell basis) in subconfluent cultures, as is suggested by studies of adult bovine EC, which produce more HS per cell after wounding of the monolayer (57). Release of heparin-like growth inhibitor into the medium of EC cultured from calf aorta is dependent upon the action of a platelet-derived endoglycosidase (42), and this activity is recovered with TCA-soluble, ethanol-precipitable isolated material from that medium(16), indicating that this growth inhibitor is a free GAG. In contrast, serum-free or albuminsupplemented mediumconditioned by fetal pulmonary arterial endothelium inhibited growth nearly as well as medium conditioned in the presence of 10% serum (Figure 4), indicating that platelet-derived enzymes are not necessary for release of heparin-like growth inhibitor from these cells. Nonetheless, growth-inhibiting activity was slightly greater in serum-supplemented than in serum-free medium, and was greatest in medium supplemented with 10% serum prior to conditioning, suggesting that some inhibitory material may be released via a serum-dependent mechanism. The small differences between the inhibitory effects of medium conditioned in the presence of albumin or 1 to 5 % serum and medium conditioned in the presence of 10% serum should not be overinterpreted, however, because they may also be related to the fact that each of the former media, but not the latter, were supplemented with fresh serum prior to bioassay. Willems and associates (58) also found that growth inhibitor

Benitz, Kelley, Anderson et al.: Endothelial Heparan Sulfate Inhibits Smooth Muscle Growth

was released by EC from human umbilical veins in serumfree medium supplemented with albumin and transferrin. Release of GAG by endoglycosidase contaminants in the albumin or transferrin used in their cultures is unlikely, because they estimated the molecular weight of the growth inhibitor to exceed 100,000 D, corresponding to the size of the proteoglycan described in this report. Furthermore, we have found that less than 2 % of the GAG present in medium conditioned by fetal bovine pulmonary arterial endothelium is free GAG, as reflected by greater than 98 % precipitation of radiosulfate-Iabeled GAG with TeA (data not shown). These results suggest that fetal human, as well as fetal bovine, EC release the growth inhibitor as a proteoglycan. Potential Function of the Proteoglycan Growth Inhibitor HS GAG prepared from purified endothelial cell basement membrane proteoglycan is a potent inhibitor of smooth muscle cell proliferation in response to serum stimulation. The data presented here do not permit conclusions about the capacity of this material to inhibit growth induced by other stimuli. In addition, serum components may be necessary to mediate the inhibitory effects of this HS on SMC proliferation, by interacting with the HS to activitate either the HS or with a serum constituent, such as the TGF-I3/armacroglobulin complex (59), to produce the proximate inhibitor. If such is the case, activity of this molecule in vivo may be limited to those circumstances in which the subendothelial basement membrane is exposed to plasma or to platelet products, such as the platelet-derived endoglycosidase (42). Until the mechanism by which this molecule exerts its inhibitory activity is identified, extrapolation of the results obtained in our in vitro growth assay system to vascular development and disease must remain tentative. In addition to inhibition of SMC proliferation in vitro, as described here and previously reported by us (15) and others (13, 16,42), heparin-like materials inhibit SMC proliferation in vivo (60), promote modulation into the maintenance of the contractile (nonmitotic) SMC phenotype (61), inhibit migration of SMC (62) in vitro and in vivo (63), and prevent pulmonary arterial remodeling in chronically hypoxic animals (64). Our observation that pulmonary arterial EC produce heparin-like materials capable of inhibiting SMC proliferation indicates that the EC are potentially able to suppress proliferation of SMC (or their precursors), promote phenotypic modulation into the contractile phenotype, and inhibit migration of SMC into nonmuscular vessels. Regulation of these processes by EC in developing arteries could account for the observations that cessation of mitosis (65) and acquisition of the contractile phenotype (66) first occur in the SMC adjacent to the endothelium. Identification of this growth modulator as a basement membrane component places it within the basement membrane that is shared by the endothelium and the underlying medial smooth muscle cells in the small arteries and arterioles, where it is strategically located to mediate these regulatory cell-cell interactions during arterial morphogenesis in development and pathologic remodeling (42). Acknowledgments: This work was supported by Biomedical Research Support Grant RR5353 awarded by the Biomedical Research Support Grant Program, Division of Research Resources, National Institutes of Health; by a Basil

23

O'Connor Starter Scholar Research Award from the March of Dimes Birth Defects Foundation; and by Grant HD-06763 from the National Institutes of Health. Dr. Benitz is the recipient of a Clinician-Scientist Award from the American Heart Association, with funds contributed in part by the California Affiliate. Dr. Kelley was supported by the Stanford Medical Alumni Medical Scholars Program and by a Medical Student Traineeship of the Cystic Fibrosis Foundation. Dr. Anderson was supported by the Stanford Medical Alumni Medical Scholars Program. Dioctadecyl-tetramethyl-indocarbocyanine-Iabeled acetyl-LDL was provided by Dr. Robert Auerbach of the University ofWisconsin. BM-I proteoglycan and antiserum was provided by Dr. John Hassell of the National Institute of Dental Research. We thank Drs. Markku Jalkanen, Alan Rapraeger, and Michael Trautman for helpful discussions, and Susan Cain for technical assistance. References

I. Hislop, A., and L. Reid. 1972. Intra-pulmonary arterial development during fetal life-branching pattern and structure. J. Anal. 113:35-48. 2. Hislop, A., and L. Reid. 1973. Pulmonary arterial development during childhood: branching pattern and structure. Thorax 28: 129-135. 3. Kaufman, O. Y. 1984. Intrauterine development of the pulmonary vessels in man. Arkh. Anat. Gistol. Embriol. 46:T584-T587. 4. Levin, D. L., A. M. Rudolph, M. A. Heymann, and R. H. Phibbs. 1976. Morphological development of the pulmonary vascular bed in fetal Iambs. Circulation 53: 144-151. 5. O'Neal, R. M., R. C. Ahlvin, W. C. Bauer, and W. A. Thomas. 1957. Development of fetal pulmonary arterioles. Arch. Pathol. 63:309-315. 6. Murphy, J. D., M. Rabinovitch, J. D. Goldstein, and L. M. Reid. 1981. The structural basis of persistent pulmonary hypertension of the newborn infant. J. Pediatr. 98:962-967. 7. Geggel, R. L., and L. M. Reid. 1985. The structural basis ofPPHN. Clin. Perinatol. II :525-549. 8. Hales, C. A., R. L. Kradin, R. D. Branstetter, and Y. Zhu. 1983. Impairment of hypoxic pulmonary artery remodeling by heparin in mice. Am. Rev. Respir. Dis. 128:747-751. 9. Hislop, A. A., and L. M. Reid. 1976. New findings in pulmonary arteries of rats with hypoxia-induced pulmonary hypertension. Br. J. Exp. Pathol. 57:542-554. 10. Meyrick, B., and L. Reid. 1981. The effect of chronic hypoxia on pulmonary arteries in young rats. Exp. Lung Res. 2:257-271. II. Meyrick, B., and L. Reid. 1978. The effect of continued hypoxia on rat pulmonary arterial circulation: an ultrastructural study. Lab. Invest. 38: 188-200. 12. Benitz, W. E., J. D. Coulson, D. S. Lessler, and M. Bernfield. 1986. Hypoxia inhibits proliferation of fetal pulmonary arterial smooth muscle cells in vitro. Pediatr. Res. 20:966-972. 13. Hoover, R. L., R. Rosenberg, W. Haering, and M. J. Karnovsky. 1980. Inhibition of rat arterial smooth muscle cell proliferation by ~arin. II. In vitro studies. Circ. Res. 47:578-588. 14. Reilly, C. F., L. S. Fritze, and R. D. Rosenberg. 1986. Heparin inhibition of smooth muscle cell proliferation: a cellular site of action. J. Cell. Physiol. 129: 11-19. 15. Benitz, W. E., D. S. Lessler, J. D. Coulson, and M. Bernfield. 1986. Heparin inhibits proliferation of fetal vascular smooth muscle cells in the absence of platelet-derived growth factor. J. Cell. Physiol. 127:1-7. 16. Castellot, J. J., M. L. Addonizio, R. Rosenberg, and M. J. Karnovsky. 1981. Cultured endothelial cells produce a heparinlike inhibitor of smooth muscle cell growth. J. Cell BioI. 90:372-379. 17. Ross, R. 1971. The smooth muscle cell. II. Growth of smooth muscle in culture and formation of elastic fibers. J. Cell Bioi. 50: 172-186. 18. Striker, G. E., J. M. Harlan, and S. M. Schwartz. 1980. Human endothelial cells in vitro. Methods Cell BioI. 21A: 135-151. 19. Voyta, J. C., D. P. Via, C. E. Butterfield, and B. R. Zetter. 1984. Identification and isolation of endothelial cells based on their increased uptake of acetylated-Iow density lipoprotein. J. Cell Bioi. 99:2034-2040. 20. Ryan, U. S., and G. Maxwell. 1986. Isolation, culture, and subculture of bovine pulmonary artery endothelial cells: mechanical methods. J. Tissue Culture Methods 10:3-5. 21. Labarca, C., and K. Paigen. 1980. A simple, rapid, and sensitive DNA assay procedure. Anal. Biochem. 102:344-352. 22. Carlson, D. M. 1968. Structures and immunochemical properties of oligosaccharides isolated from pig submaxillary mucins. J. Bioi. Chem. 243:616-626. 23. Linker, A., and P. Hovingh. 1984. Structural studies on heparin. Tetrasaccharides obtained by heparinase digestion. Carbohydr. Res. 127:75-94. 24. Hovingh, P., and A. Linker. 1974. The disaccharide repeating-units of heparan sulfate. Carbohydr. Res. 37:181-192. 25. Suzuki, S. 1972. Chondroitinases from Proteus vulgaris and Flavobacterium heparinum. Methods Enzymol. 28:911-917. 26. Lindahl, U., G. Backstrom, L. Jansson, and A. Hallen. 1973. Biosynthesis of heparin. II. Formation of sulfamino groups. J. BioI. Chem. 248: 7234-7241.

24

AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 21990

27. Rapraeger, A., and C. Yeaman. 1989. A quantitative solid-phase assay for identifying radiolabelled glycosaminoglycans in crude cell extracts. Anal. Biochern. 169:361-365. 28. Rapraeger, A., M. Jalkanen, E. Endo, J. Koda, and M. Bernfield. 1985. The cell surface proteoglycan from mouse mammary epithelial cells bears chondroitin sulfate and heparan sulfate glycosaminoglycans. J. Bioi. Chern. 260: II 046- I 1052. 29. Wasteson, A. 1971. Properties of fractionated chondroitin sulfate from ox nasal septa. Biochern. J. 122:477-485. 30. Borders, C. L., and M. A. Raferty. 1968. Purification and partial characterization of testicular hyaluronidase. J. BioI. Chern. 243:3756-3762. 31. Koda, J. E., A. Rapraeger, and M. Bernfield. 1985. Heparan sulfate proteoglycans from mouse mammary epithelial cells. J. Bioi. Chern. 260: 8157-8162. 32. Merril, C. R., M. L. Dunau, and D. Goldman. 1981. A rapid sensitive silver stain for polypeptides in polyacrylamide gels. Anal. Biochern. 110:201-207. 33. Towbin, H., T. Staehelin, and J. Gordon. 1979. Electrophoretic transfer of proteins from polyacrylamide gels to nitrocellulose sheets: procedure and some applications. Proc. Natl. Acad. Sci. USA 76:4350-4354. 34. Hassell, J. R., W. C. Leyshon, S. R. Ledbetter et al. 1985. Isolation of two forms of basement membrane proteoglycans. J. Bioi. Chern. 260: 8098-8105. 35. Hassell, J. R., P. G. Robey, H.-J. Barrach, J. Wilczek, S. I. Rennard, and G. R. Martin. 1980. Isolation of a heparan sulfate-containing proteoglycan from basement membrane. Proc. Natl. Acad. Sci. USA 77: 4494-4498. 36. Sorensen, K., and U. Brodbeck. 1986. A sensitive protein assay method using microtiter plates. Experientia 42: 161-162. 37. Chandrasekhar, S., M. A. Esterman, and H. A. Hoffman. 1987. Microdetermination of proteoglycans and glycosaminoglycans in the presence of guanidine hydrochloride. Anal. Biochern. 161:103-108. 38. Smith, R. L., and E. Gilkerson. 1979. Quantitation of glycosaminoglycan hexosamine using 3-methyl-2-benzothiazolone hydrazone hydrochloride. Anal. Biochem. 98:478-480. 39. Sokal, R. R., and F. J. Rohlf. 1981. Biometry. 2nd ed. W. H. Freeman and Co., New York. 40. Shively, J. E., and H. E. Conrad. 1976. Formation of anhydrosugars in the chemical depolymerization of heparin. Biochemistry 18:3932-3942. 4 I. Poole, A. R. 1986. Proteoglycans in health and disease: structures and functions. Biochem. J. 236: 1-14. 42. Castellot, J. J., L. V. Favreau, M. J. Karnovsky, and R. D. Rosenberg. 1982. Inhibition of vascular smooth muscle cell growth by endothelial cell-derived heparin: possible role of a platelet endoglycosidase. J. Bioi. Chern. 257:11256-11260. 43. Oohira, A., T. N. Wight, and P. Bornstein. 1983. Sulfated proteoglycans synthesized by vascular endothelial cells in culture. J. BioI. Chem. 258:2014-2021. 44. Charnley-Campbell, J. H., and G. R. Campbell. 1981. What controls smooth muscle phenotype? Atherosclerosis 40:347-357. 45. Roberts, A. B., K. C. Flanders, P. Kondaiah et al. 1988. Transforming growth factor beta: biochemistry and roles in embryogenesis, tissue repair and remodeling, and carcinogenesis. Recent Prog. Horm. Res. 44: 157-197. 46. Kinsella, M. G., and T. N. Wight. 1988. Structural characterization of heparan sulfate proteoglycan subclasses isolated from bovine aortic endothelial cell cultures. Biochemistry 27:2136-2144. 47. Kato, M., Y. Koike, S. Suzuki, and K. Kimata. 1988. Basement membrane proteoglycan in various tissues: characterization using monoclonal antibodies to the Englebreth-Holm-Swarm mouse tumor low density heparan

sulfate proteoglycan. J. Cell Bioi. 106:2203-2210. 48. Saku, T., and H. Furthmayr. 1989. Characterization of the major heparan sulfate proteoglycan secreted by bovine aortic endothelial cells in culture. Homology to the large molecular weight molecule of basement membranes. J. Bioi. Chern. 264:3514-3523. 49. Klein, D. J., D. M. Brown, T. R. Oegema et al. 1988. Glomerular basement membrane proteoglycans are derived from a large precursor. J. Cell BioI. 106:963-970. 50. Clowes, A. W., M. M. Clowes, A. M. Gown, and T. N. Wight. 1984. Localization of proteoheparan sulfate in rat aorta. Histochemistry 80: 379-384. 51. Gallagher, J. T., M. Lyon, and W. P. Steward. 1986. Structure and function of heparan sulphate proteoglycans. Biochem. J. 236:313-325. 52. Castellot, J. J., D. L. Beeler, R. D. Rosenberg, and M. J. Karnovsky. 1984. Structural determinants of the capacity of heparin to inhibit the proliferation of vascular smooth muscle cells. J. Cell. Physiol. 120: 315-320. 53. Wright, T. C., J. J. Castellot, M. Petitou, J.-C. Lormeau, J. Choay, and M. J. Karnovsky. 1989. Structural determinants of heparins growth inhibitory capacity. Interdependence of oligosaccharide size and charge. J. BioI. Chem. 264:1534-1542. 54. Lankes, W., A. Griesmacher, J. Griinwald, R. Schwartz-Albiez, and R. Keller. 1988. A heparin-binding protein involved in inhibition of smooth muscle cell proliferation. Biochem. J. 251 :831-842. 55. Castellot, J. J., K. Wong, B. Herman et al. 1985. Binding and internalization of heparin by vascular smooth muscle cells. J. Cell. Physiol. 124: 13-20. 56. Reilly, C. F., L. M. Fritze, and R. D. Rosenberg. 1986. Heparin inhibition of smooth muscle cell proliferation: a cellular site of action. J. Cell. Physiol. 129:11-19. 57. Kinsella, M. G., and T. N. Wight. 1986. Modulation of sulfated proteoglycan synthesis by bovine aortic endothelial cells during migration. J. Cell Bioi. 102:679-687. 58. Willems, C., G. C. B. Astaldi, P. G. DeGroot et al. 1982. Media conditioned by cultured human vascular endothelial cells inhibits the growth of vascular smooth muscle cells. Exp. Cell Res. 139: 191-197. 59. McCaffrey, T. A., D. J. Falcone, C. F. Brayton, and L. A. Agarwal. 1989. Transforming growth factor-13 activity is potentiated by heparin via dissociation of the transforming growth factor-l3/a,-macroglobulin inactive complex. J. Cell Bioi. 109:441-448. 60. Majesky, M. W., S. M. Schwartz, M. M. Clowes, and A. W. Clowes. 1987. Heparin regulates smooth muscle S phase entry in the injured rat carotid artery. Circ. Res. 61 :296-300. 61. Campbell, J. H., and G. R. Campbell. 1986. Endothelial cell influences on vascular smooth muscle phenotype. Annu. Rev. Physiol. 48:295-306. 62. Majack, R. A., and A. W. Clowes. 1984. Inhibition of vascular smooth muscle cell migration by heparin-like glycosaminoglycans. J. Cell. Physiol. 118:253-256. 63. Clowes, A. W., and M. M. Clowes. 1986. Kinetics of cellular proliferation after arterial injury. IV. Heparin inhibits rat smooth muscle mitogenesis and migration. Circ. Res. 58:839-845. 64. Hales, C. A., R. L. Kradin, R. D. Branstetter, and Y. Zhu. 1983. Impairment of hypoxic pulmonary artery remodeling by heparin in mice. Am. Rev. Respir. Dis. 128:747-751. 65. Yurokova, Z., P. Hadjiisky, J. Renais, and L. Scebat. 1976. On the proliferative activity of arterial smooth muscle cells in rat postnatal ontogenesis-auto-historadiograph~c~tudies. Atherosclerosis 23: 297-304. 66. Campbell, G. R., andJ. H. Charnley. 1976. Ultrastructure of differentiating vascular smooth muscle. In Vascular Neuroeffector Mechanisms, Second International Symposium. J. A. Bevan, editor. Karger, Basel. 1-9.