Extracellular, surface, and intracellular proteoglycans produced by human embryo lung fibroblasts in culture (IMR-90).

Medium and cell-associated proteoglycans produced by human embryo lung fibroblasts (IMR-90) in culture were analyzed after 48-h uptake of N ~ z ~ ~ S O ~ . The culture medium contained three proteoglycan fractions which could be separated from one another by density gradient centrifugation and gel chromatography under dissociative conditions. The most  dense  component eluted from Sepharose 2B with peak K,, = 0.40 and contained predominantly heparan sulfate chains with M, = 40,000. A somewhat less dense component had peak K,, = 0.29 and chondroitin sulfate chains with M, zz 40,000. The least dense proteoglycan (p < 1.46) was the predominant proteoglycan component of the medium. It eluted from Sepharose 2B at K,, = 0.68 and was composed primarily of dermatan sulfate-containing chains with M, = 25,000. Cell-associated proteoglycans were solubilized  by treatment with a zwitterionic detergent followed by extraction in 4 M guanidine HCl. By Sepharose 2B chromatography, the cells contained a heparan sulfate proteoglycan fraction similar to the most dense component of the medium. This fraction was removed from the cell surface by trypsin treatment but was not significantly displaced by exogenous heparin. The cells also contained a smaller heparan-sulfate-containing component (Knv = 0.74) which was not sensitive to papain digestion or alkaline treatment and may not be a proteoglycan. This component was diminished by 50% during 24-h chase and probably represents an intracellular storage or degradation pool.

Glycosaminoglycans produced by fibroblasts in culture include hyaluronic acid, chondroitin sulfates, and heparan sulfate. Studies of the production and turnover of these glycosaminoglycans by cultured cells indicate that fibroblasts from specific tissues elaborate a distinct glycosaminoglycan profile (1, 2), and that certain glycosaminoglycans are preferentially located in specific culture compartments. Heparan sulfate, for example, is found at the cell surface (3), and dermatan sulfate is found primarily in the medium (4, 5).
Because the sulfated glycosaminoglycans of cartilage matrix have been demonstrated to be organized into proteoglycan molecules consisting of several polysaccharide chains linked to a core protein (6, 7), it is often assumed that all of the sulfated glycosaminoglycans are in a proteoglycan organization. There are few characterizations of proteoglycans pro-* This work was supported by Grant AG 01214 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
-f Recipient of Research Career Development Award AG 00114 from the National Institutes of Health. duced by noncartilage cells in culture, however, Heparan sulfate from the surface of Chinese hamster cells was fist demonstrated to be proteoglycan by Kraemer and Smith (8). Subsequently, proteoglycan organization for both medium and surface heparan sulfates from human skin fibroblasts (9) and a chondroitin sulfate component from human glial cells (10) was determined. Heparan sulfate proteoglycans from the surface of rat liver cells (11) and a hepatoma (12) have been characterized. Both medium and cellular fractions from cultures of human embryonic skin fibroblasts contained large proteoglycans composed of chondroitin sulfate plus smaller proteoglycan fractions containing heparan sulfate and dermatan sulfate as well (13). Chondroitin sulfate protoglycans released to the medium by adult rat lung cells were of only one size class although cell-associated proteoglycans were of two sizes, the larger containing only chondroitin sulfate and the smaller also containing dermatan sulfate and heparan sulfate (14). Molecular characterizations of proteoglycans produced by cells in culture are valuable in developing ideas about the role of glycosaminoglycans as well as in defining the cellular origins of proteoglycans from specific tissues.
Glycosaminoglycans at cell surfaces have been implicated in such basic characteristics as cellular adhesion (15) and the related phenomena of aggregation (16) and motility (17). Glycosaminoglycans in extracellular matrix are thought to play an important physiological role in the permeability and hydration of tissues (18). Multiple interactions of fibronectin and collagen with glycosaminoglycans (particularly highly sulfated heparans) could play a major role in the deposition of insoluble extracellular matrix as well as in regulation of interactions between cells and their molecular environment (18)(19)(20). The present report describes the isolation and characterization of three proteoglycan fractions from the medium of lung fibroblasts growing on a polystyrene substratum. In addition, a surface-associated proteoglycan fraction and an intracellular glycosaminoglycan component which may not be proteoglycan are described. These results form a base-line for future investigation of the relationships between specific alterations in cellular behavior and the production and processing of fibroblast proteoglycans. addition of fresh medium containing Na255S04 (5-20 pCi/ml, 0. Zsolation of Proteoglycans-Proteoglycans were obtained from various compartments of two 150-cm2 cultures in the following way. After 48-h uptake of radioisotopes, medium was decanted and dry guanidine hydrochloride added directly to make it 4 M; the solution was then adjusted to 0.1 M NazEDTA, 0.05 M sodium acetate, and 0.005 M benzamidine/HCl and stored at 4 "C. The cells of each flask were rinsed four times with 10 ml of Earle's balanced salt solution. Five ml of 4% Zwittergent (Calbiochem) in 0.05 M sodium acetate with added protease inhibitors was added to one flask and the cells scraped immediately from the growth surface with a rubber hoe. This cell suspension was then placed in the second flask. A 5-ml rinse of the first flask was also added to the second. Cells in 10 ml of 4% Zwittergent were left at room temperature for 20 min. After scraping all cells from the flask, it was rinsed twice with 5 ml of 8 M guanidine hydrochloride bringing the extract conditions to 4 M guanidine hydrochloride and 2% Zwittergent. Extraction was allowed to proceed for 24 h at 4 "C. The extract appeared clear, and centrifugation at 10,000 X g for 30 min did not produce a pellet. Unincorporated labeling isotope was removed by passage of this material through a column (1 X 40 cm) of Sephadex G-50 eluted with 4 M guanidine hydrochloride containing protease inhibitors. Zwittergent was not included in subsequent procedures. In earlier experiments, detergent solubilization was omitted and the cell layer extracted directly with 4 guanidine hydrochloride plus protease inhibitors for 24 h at 4 "C. The viscous particulate material which remained was separated from solubilized components by filtration through paper and centrifugation at 6,000 X g. for 5 min.
The trypsin-removable surface fraction was obtained by treating the cell layer of each flask with BSS containing trypsin (0.1 mg/ml, 15 min, 37 "C, Sigma Chemical Co., two times crystallized). This treatment did not cause release of cells from the culture surface. The trypsin-removable fraction was then decanted, made 4 M with guanidine hydrochloride, and soybean trypsin inhibitor (0.1 mg/ml, Sigma Chemical Co.) was added in addition to the other protease inhibitors. Remaining ceIlular materia1 was then extracted with Zwittergent and 4 M guanidine hydrochloride as described.
The proteoglycan fraction removed by heparin was obtained by incubation of a labeled monolayer for 30 min in BSS with or without added heparin (100 pg/ml, porcine mucosa and sodium salt, grade 11, Sigma Chemical Co). After incubation with heparin, the monolayer was treated with trypsin followed by extraction of cellular material as described above.
Density Gradient Centrifugation-Solid cesium chloride was added to medium and cellular macromolecular fractions in 4 M guanidine hydrochloride, giving an initial density of 1.47. This material was centrifuged at 40,000 rpm for 48 h (10 "C, type 65 rotor, Spinco model L ultracentrifuge). Each tube was divided into equivalent sections and fractions D l (bottom), D2, D3, and D4 (top) were slowly withdrawn from the top. Density was determined using a 100-pl pipette as pyknometer.* Gel Filtration Chromatography-Columns of Sepharose CL-2B and CL-4B (1 X 100 cm) were eluted with 4 M guanidine hydrochloride and 0.1 M sodium acetate, pH 5.8, at a flow rate of 10 ml/h. Analytical Sepharose CL-GB columns (0.7 X 90 cm) were eluted at 5 ml/h. The Vo of 4B and 6B columns was marked by blue dextran; the V, of all columns was marked by vitamin BIZ (Mr = 1350). Aliquots of each fraction (0.1-0.4 ml) were counted in 4 ml of formula 950A scintillation mixture (New England Nuclear). A D l fraction of bovine nasal cartilage was eluted from the Sepharose CL-2B column at a K,, of rose 6B elution position before and after digestion with papain (50 pg/ml, 0.1 M sodium acetate, 10 mM EDTA, and 10 mM 2-mercaptoethanol, pH 5.5,18 h, 60 "C) or treatment with sodium hydroxide (0.2 N NaOH and 1 M NaBH4, 45 "C, 24 h, neutralized with glacial acetic acid) to cleave intact glycosaminoglycans from the core protein.
Samples were desalted before treatment by passage through a PD-10 minicolumn of G-25 (Pharmacia). In other cases, samples in 4 M guanidine hydrochloride were directly treated with 0.05 M NaOH for 18 h at 37 "C. The large heparan sulfate proteoglycan was particularly vulnerable to loss on columns if removed from 4 M guanidine hydrochloride. For this reason, all analytical columns have been run in 4 M guanidine hydrochloride; recovery 35S04 was >90%.
Analytical Procedures for Glycosaminoglycans-The quantity of [~S]glycosaminoglycan in each fraction was determined by counting the cetylpyridinium chloride precipitate of protease-digested samples trapped on a Millipore fiiter. The glycosaminoglycan composition of each fraction was determined by digestion with chondroitinase ABC or chondroitinase AC (Miles Laboratories, Inc.), separation of the disaccharides produced by thin layer chromatography, and determination of percentage of 35S-labeled material at each position. The details of these procedures have been previously described (5). Material which does not migrate on the thin layer chromatography plate after digestion is presumed to be heparan sulfate, and the amount of this material corresponds closely to the amount of the original sample which is cleaved to smaller fragments by treatment with nitrous acid. Dermatan sulfate content was calculated by comparison of the amount of 35S04 in the chondroitin 4-sulfate position following digestion with chondroitinase ABC and chondroitinase AC, although it is recognized that this method can somewhat overestimate the iduronic acid residues in a heteropolymer (see Ref. 22).

RESULTS
Extraction of Proteoglycans-The fibroblast cultures were divided into two compartments, the soluble or extracellular material decanted off the monolayer with the culture medium and the material remaining with cells after rinsing of the monolayer. After 48-h uptake of Nai35S04, approximately 90% of the 35S-labeled proteoglycan was in the medium. The cellassociated proteoglycans were solubilized completely by a combination of detergent treatment followed by extraction with 4 M guanidine hydrochloride. When detergent was omitted, only about 25% of the cell-associated proteoglycans were solubilized; the rest remained in a viscous pellet.
Gel Filtration of Fibroblast Proteoglycans: Dissociative Conditions-Proteoglycans of the medium were separated into two classes on the basis of differences in hydrodynamic size by Sepharose 2B fiitration (Fig. la). The larger material ("I) formed a typically broad and indistinct peak, while the smaller material (M-II) formed a sharp peak which included 62% of the radioactivity. This latter peak was characterized by a high percentage of chondroitin 4-sulfate and dermatan sulfate, while the larger material was mostly heparan sulfate and chondroitin 6-sulfate (Table I).
Cell-associated proteoglycans were also separated into two size classes by Sepharose 2B filtration (Fig. lb). The elution positions of these peaks did not correspond to those of the medium, and each peak contained approximately the same amount of radioactivity. The larger of the cell-associated peaks (C-I) was characterized by a high percentage of heparan sulfate. If the cell layer was not treated with detergent prior to extraction with guanidine hydrochloride, the Sepharose 2B profie of that material which was solubilized demonstrated a substantial peak at the V,, (E-I,   density-dependent separation. The glycosaminoglycans of this dense peak were primarily heparan sulfate (Table 11). At a somewhat lower density (02), proteoglycans of two distinct sizes were seen. The larger of these eluted at a K,, of 0.29 and was composed primarily of chondroitin 6-sulfate. The smaller of the D2 proteoglycans contained both heparan sulfate and chondroitin 6-sulfate and also a significant percentage of dermatan sulfate, which was absent in the larger D2 proteoglycan. The D3 fraction demonstrated two proteoglycan peaks similar in elution position to those of D2. At this density, the smaller dermatan sulfate-containing material was the major component. This material (03-14 shows a hydrodynamic size (&" = 0.68) similar to that of the major component of medium prior to density-dependent separation ("11, Fig. la). In other experiments, the densities of D2 and D3 were slightly different and it was noticed that the smaller, dermatan-sulfate-rich component was a major constituent when the D2 density was at 1.46. The lightest fraction, D4, was also enriched for this smaller component, but it was not used for further characterizations.
The distinct buoyant densities, glycosaminoglycan compositions, and elution positions suggest the presence of three different proteoglycan fractions in the medium of lung fibro-

Glycosaminoglycan composition ofproteoglycans from medium after CsCl density gradient centrifugation
Values are expressed as percentage of total radioactivity in each fraction following digestion with chondroitinase ABC/chondroitinase AC and separation of disaccharide products by thin layer chromatography as described under "Experimental Procedures." Fractions refer to Deaks from SeDharose 2B as shown in Fig. 2  can fractions before and after papain digestion or NaOH treatment. Peaks were collected after 2B chromatography and run on Sepharose 6B in 4 M guanidine HC1 before (-) and after 1---) papain digestion or alkaline treatment. Samples c, b, and c are fractions collected as in Fig. 1; d is from the large peak of Fig. 46 The smallest proteoglycan of medium ("11) was the major component of D3 and D4 and eluted from Sepharose 6B a t K,, = 0.30 (Fig. 3a). Papain digestion or alkaline treatment shifted the elution position to a peak at K,, = 0.45, corresponding to a chain length of approximately 25,000 daltons.
Further Characterization of Proteoglycans of the Cell Layer-The Sepharose 2B profiie of proteoglycans extracted from the cell layer by detergent/guanidine hydrochloride showed two peaks of %-labeled material having K,, = 0.40 and 0.74 (Fig. lb). Density gradient centrifugation of this material under conditions similar to that performed on the medium (starting density = 1.47) did not reveal additional components (data not shown). Both components were primarily in the most dense fraction ( p = 1.54, 75% of "5S-labeled material) with the remaining material of both peaks found in the D2 ( p = 1.45). Very little material of either size was recovered from the D3 or D4.
The larger cellular component ((2-1) was composed primarily of heparan sulfate and eluted in the Vo from Sepharose 6B.
The smaller cellular component (C-11) eluted from Sepharose 6B at a €Caw of 0.42 initially, and this elution position was not shifted by papain digestion or alkaline treatment (Fig. 3c). By these criteria, therefore, the cellular ["S]glycosaminoglycans of peak I1 may not be proteoglycans, but are polysaccharide chains with M , E 28,000.
Variations in the relative heights of the two cell-associated peaks have been occasionally noticed. Although they usually contain approximately equal amounts of "SOd, there were several experiments in which the first peak was substantially larger than the second (compare Fig. Ib and Fig. 4a). This difference could not be related to brand of culture flask, passage number of the cultures, or washing of the cell layer prior to extraction. Although of great interest, factors which control the relative amounts of these two components have not been determined.
Location of Cell-associated Proteoglycans-The Sepharose 4B profiie of cell-associated components shows two peaks with K,, = 0.20 and 0.60. (Fig. 4a). Approximately 50% of the cell-associated proteoglycans were released by gentle trypsin treatment (0.1 mg/ml, 15 min) which did not detach the cells or increase trypan blue permeability. Most of this trypsinremovable material eluted as a single peak on Sepharose 4B, with ICav = 0.20 (Fig. 4b). A small percentage was retarded even more. Following trypsin treatment, the proteoglycans of the remaining cell layer were extracted by detergent/guanidine hydrochloride. They eluted from Sepharose 4B in the usual two positions (Fig.  4c); however, the first peak was greatly diminished. The trypsin-removable material eluted from Sepharose 6B in the V, and at K,, = 0.35 following alkaline/borohydride treatment (Fig. 3 4 . Glycosaminoglycan analysis of the trypsin-removable material indicated that it was -85% heparan sulfate. Following a 24-h chase period, the elution characteristics of proteoglycans released to the medium, removable from the surface with trypsin and remaining with the cells, were determined. The apparent size of each of the components was not altered during the chase interval. Although both size classes were lost from the cell, material of the larger size class appeared in the medium (Fig. 5a) and net loss from the culture was less than 10%. This material of the chase medium eluted from Sepharose 6B in the VO and at K,, = 0.33 following alkaline/borohydride treatment. Proteoglycan of similar size remained with the cell layer during chase and could be released by trypsin treatment (Fig. 5b). The smaller intracellular component (Fig. 5c) was diminished by 50% during the chase interval (data not shown); however, nothing of this size appeared in the medium. This material (Fig. 5c) eluted from Sepharose 6B at K,, = 0.45, and its position was not shifted by alkaline treatment.
These experiments have described proteoglycan components of similar elution position which can be removed from the cell layer by trypsin and also are released from the cell layer and remain at the cell surface during incubation in unlabeled medium. Each of these components demonstrates an elution in the Vo from Sepharose 6B prior to alkaline/ borohydride treatment and elution at K,, P 0.35 following this treatment. Similar elution characteristics were also shown by the D l component separated from medium by density centrifugation. All of these components have a high heparan sulfate content. These results suggest that human lung fibroblasts in culture are producing a large heparan-sulfate proteoglycan which accumulates at the cell surface (from which it can be released by gentle trypsin treatment) and is also released continuously by the cells. As such, it is found as one of several proteoglycans in the medium during labeling and as the major proteoglycan of the medium during chase. The elution characteristics of this proteoglycan indicate that it could have a molecular weight as large as lo6 daltons and is certainly much larger than the 75,000-dalton membrane-associated heparan-sulfate proteoglycan of rat liver cells (11) or an ascites hepatoma (12). Heparan sulfate chains of the latter proteoglycans had molecular weights of -14,000 and 21,000, respectively (three to four chains per proteoglycan), whereas the fibroblast species has polysaccharide chains of -40,000 daltons (up to 25 chains/proteoglycan). A more accurate determination of molecular weight cannot be made until the component is available in greater quantity. We estimate that one 150-cm2 flask of IMR-90 fibroblasts has no more than 1 pg of this component associated with the cells.
The chase experiment (Fig. 5) also suggests that the smaller component of the cell layer (C-11, Fig. l b )   containing 100 pg/ml of heparin. Under these conditions, 10-15% of the total cetylpyridinium chloride-precipitable "S-labeled material of the cell layer was released in the presence of heparin, whereas only 5% was released in BSS alone (not shown). The amount of trypsin-removable material subsequently removed from BSS and heparin-treated cells was similar. The heparin-released 35S-macromolecules eluted from Sepharose 4B in the region of the large trypsin-removable proteoglycan (Fig. 6a). These results parallel those of Kjellen et al. (24), who concluded that two distinct pools of extracellular proteoglycan are present, one releasable by heparin and another releasable by trypsin. Attachment via a specific receptor was suggested for the heparin-releasable proteoglycan. The ['Hlserine-labeling profile in Fig, 6a was similar to that from BSS alone (not shown), indicating that exogenous heparin is not releasing a major surface protein fraction. Incidentally, the serine data demonstrate that trypsin treatment releases only -4%'of the labeled protein associated with the cells.
Although we have generally c o n f i i e d t h e heparin-release results seen with rat liver cells, there is a significant difference. The amount of additional 35S-labeled cetylpyridinium chlo-ride-precipitable material released in the presence of exogenous heparin was only 5-10% of the total in the cell layer (also Ref. 25) instead of 30-50%, as in rat liver cells (24). If exogenous heparin is releasing surface molecules from a receptor, then it must be concluded either that fibroblasts have fewer such receptors, that the larger proteoglycan of fibroblasts does not interact as well with the receptor, or possibly that the additional heparan sulfate chains per fibroblast proteoglycan provide multiple binding sites which have greatly enhanced binding avidity.

DISCUSSION
Previous studies of sulfated glycosaminoglycans (S-GAGS) produced by cultured human embryo lung fibroblasts ( 5 ) , skin fibroblasts (2), and glial cells (4,27) have demonstrated certain definite characteristics for their composition, distribution, and turnover. In all of these studies, the glycosaminoglycans produced included heparan sulfate, chondroitin 4-and 6-sulfate, and dermatan sulfate. Hyaluronic acid was also produced by each of these cell types. The sulfated glycosaminoglycans were distributed as soluble molecules in the culture medium, as cell-associated molecules which were released by gentle proteolysis or by treatment with EDTA, and as intracellular molecules. In each of these studies, there have been certain S-GAGS associated with particular culture compartments. The medium contained primarily chondroitin sulfates of which a great deal was dermatan sulfate; the cell-associated surface compartment was greatly enriched for heparan sulfate; and the intracellular pool contained all types of glycosaminoglycans. Under chase conditions, most of the cellular [35S]GAGs moved out of the cell rapidly although a residual stable pool enriched for heparan sulfate was present. The trypsin-removable [35S]GAGs left the surface with an initial half-life varying from 7-22 h, followed by much slower turnover rates. Sulfated glycosaminoglycans appeared in the medium, and heparan sulfate was an increasing percentage of the ["SIGAGs of chase medium. The present report describes three distinct proteoglycan fractions produced by human lung fibroblasts in culture and suggests that their location and composition now make it possible to describe ['%]GAG turnover in terms of proteoglycans.
The major proteoglycan fraction secreted by lung fibroblasts is a rather small monomer with high dermatan sulfate content (M-ZZ, Fig. l a ) . Its low buoyant density suggests that it is made up of few carbohydrate chains attached to a core protein. Although this is the major proteoglycan fraction secreted during "%04 labeling, it is not found in the medium after chase. Lung fibroblasts do not demonstrate a high amount of intracellular dermatan sulfate, and the smaller intracellular pool (C-ZZ, Fig. lb) does not act like a precursor to this medium component. Thus, for lung fibroblasts, it seems most reasonable to suggest that the dermatan sulfate-containing proteoglycan is synthesized and rapidly secreted without intervening intracellular or surface accumulation. A large chondroitin sulfate proteoglycan is also found in the medium. Report of this proteoglycan from lung fibroblasts added to previous reports of such a proteoglycan from glid cells (10) and skin fibroblasts (13) reinforces the conclusion that large chondroitin sulfate proteoglycans are a general product of noncartilage cells. Aggregation behavior influenced by hyaluronic acid was reported for the glial and skin fibroblast molecules (10, 13). The chondroitin sulfate proteoglycan Seems to be a greater percentage of the secreted S-GAGS of skin fibroblasts (13) than of lung fibroblasts, as reported in this study.
The third proteoglycan fraction identified in the medium of lung fibroblast cultures contains a large, high density heparan sulfate monomer. A similar molecule was also found in the medium after chase incubation and was removable from the cell surface by gentle proteolysis even after a 24-h chase incubation. This was the only heparan sulfate-containing proteoglycan demonstrated, although it is recognized that distinct heparan sulfate species of similar size having variable sulfation characteristics could be combined in the one fraction. Free heparan sulfate polysaccharides were not identified at the cell surface or in the extracellular medium. This suggests that the surface and extracellular heparan sulfate movements previously defined by following only ["SIGAG (5) resulted from movement of proteoglycan.
The trypsin-removable fraction was never 100% heparan sulfate, and the other ["'SIGAG generally present is chondroitin 6-sulfate (5). The presence of a minor amount of the large chondroitin sulfate proteoglycan in surface material would not have been detected by filtration chromatography alone because the proteoglycan sizes and polysaccharide chain lengths are quite similar. The simplest suggestion, therefore, is that proteoglycans at the surface of fibroblasts are of both the chondroitin and heparan sulfate types. The existence of both chondroitin and heparan sulfate polysaccharide chains on the same core protein has not been rigorously excluded, however. It is not clear whether the small amount of proteoglycan which remains attached to cells after trypsin treatment ( Fig. 4c, peak I) represents an intracellular precursor pool or a trypsin-inaccessible surface fraction.
It is important to understand how the trypsin-removable heparan sulfate proteoglycan is attached to the cell surface. The fact that it is not removed by washing, high salt, pH changes, or EDTA suggests that it is f i i y attached (25). Furthermore, the need for detergent treatment in order to solubilize the component from whole cells suggests tight association with membrane. The boundary between cell membrane and extracellular matrix cannot be clearly defined, however, and certain components, including glycosaminoglycans and the glycoprotein fibronectin, exist within this boundary. Heparan sulfate has been shown to bind preferentially to fibronectin (19,20), but the organization of these interactions is not yet defined. Fibronectin may be linked to a cell membrane component with secreted heparan sulfate bound to the fibronectin. On the other hand, heparan sulfate attached to a core protein which is an integral membrane constituent may be the cell surface component to which fibronectin binds.
Close association between glycosaminoglycans and fibronectin at the surface of hamster fibroblasts (NIL-8) was demonstrated by their ability to be cross-linked by a photoactivatable reagent bound to added fibronectin (28). Large, nonproliferating late passage IMR-90 lung fibroblasts do not organize a fibrillar array of fibronectin, whereas proliferating, early passage cultures do have such a fibronectin network when visualized by indirect immunofluorescence (29). However, the late passage cells accumulate trypsin-removable heparan sulfate in quantities even greater than the early passage cells (30), indicating that fibronectin fibrils are not necessary for accumulation of the surface-associated proteoglycan. Similar elution characteristics for extracellular and trypsin-released proteoglycan do not necessarily indicate that trypsin treatment is digesting some attachment protein other than the core protein. If trypsin is clipping only a limited piece of the core protein from the proteoglycan, e.g. a cleavage just outside the membrane lipids into which the core protein inserts, the hydrodynamic size of the proteoglycan would not be noticeably different although the core protein would be smaller. It has been recently reported that the core protein of a heparan sulfate proteoglycan solubilized by detergent from Swiss mouse 3T3 cells has M , = 20,000-30,000, whereas the core protein of the same proteoglycan removed from the cell by proteolytic methods contains fewer hydrophobic amino acid residues and has M, = 10,000-15,000 (31). All of the results are consistent with cell surface proteoglycan being synthesized as an integral membrane component which is subsequently either shed to the medium or endocytosed and degraded.
The intracellular pool of nonproteoglycan [35S]GAGs (C-11, Fig. 1b) is interesting. Because proteoglycans are synthesized by addition of monosaccharide moieties to the core protein (32), free glycosaminoglycans of this size would not be expected to add directly to a proteoglycan. Furthermore, this fraction diminished by only 50% over the 24-h chase interval and did not appear at the surface or in the medium, indicating that it is not primarily a precursor of secreted material. The presence of an intracellular pool of large glycosaminoglycans (Mr = 28,000) is somewhat surprising because this chain length does not represent breakdown to single sugars or oligosaccharides produced by endoglucuronidase activity (33). The absence of smaller intracellular fragments of [35S]GAG may indicate desulfation prior to digestion of glycosidic linkages. An intracellular pool of heparan sulfate has been described in 3T3 mouse fibroblasts (34), but these glycosaminoglycans were smaller (Mr = 6,000). Although it is assumed that the C-I1 glycosaminoglycans are from an intracellular pool, only their solubility in 4 M guanidine hydrochloride (Fig.  IC) and insensitivity to trypsin (Fig. 4c) have actually been demonstrated.
Proteoglycans from various tissues of adult bovine lung were recently isolated and characterized (35). Gas exchange tissue contained both a chondroitin and a heparan sulfate proteoglycan fraction, while pleura contained only a heparan sulfate proteoglycan; bronchiolar proteoglycan was composed primarily of chondroitin sulfate and was similar in size to the proteoglycan of nasal cartilage. The major proteoglycan isolated from porcine lung was also similar in size and composition to the cartilage proteoglycan (36). None of these tissues demonstrated a chondroitin/dermatan fraction similar to the smaller secreted proteoglycan ("11) of human embryo lung fibroblasts in culture. The major secreted proteoglycan of cultured lung fibroblasts thus could be unique either to embryonic cells or to human cells. Alternatively, it could represent a cleaved or nonaggregated form of the molecular structure found in intact tissue.