A Cartilage Explant System for Studies on Aggrecan Structure, Biosynthesis and Catabolism in Discrete Zones of the Mammalian Growth Plate

Matrix Vol. 1311993,pp. 135-147 © 1993 by Gustav Fischer Verlag, Stuttgart· Jena . New York

A Cartilage Explant System for Studies on Aggrecan Structure, Biosynthesis and Catabolism in Discrete Zones of the Mammalian Growth Plate ANNA H. K. PLAAS and JOHN D. SANDY Shriners Hospital for Crippled Children, Tampa Unit, 12 502 North Pine Drive, Tampa, FL, 33 612 and the Department of Biochemistry and Molecular Biology, School of Medicine, University of South Florida, Tampa, FL 33612, USA.

Abstract The structure, biosynthesis and catabolism of aggrecan has been studied in the bovine fetal rib growth plate. Comparative analyses were made on six 1-mm transverse slices which represent the resting zone (slice 6), proliferative zone (slices 5 and 4), upper hypertrophic zone (slice3), middle hypertrophic zone (slice 2) and lower hypertrophic zone (slice 1). Aggrecan was abundant and exhibited very high aggregability in all zones. The aggrecan monomer was similar in structure in the resting and proliferative zones but showed a marked increase in hydrodynamic size in the lower hypertrophic zone; this was apparently due to an increase in the size of substituent glycosaminoglycans and an increase in core protein size as indicated by peptide analysis for G3 domain abundance. Biosynthetic studies with [35 Sl-sulfate showed the rate of synthesis per cell to be highest in the upper hypertrophic zone, and the structure of the newly synthesised molecules to be similar to the resident population in all zones. During explant culture in basal medium both aggregating and non-aggregating forms of aggrecan were released slowly from all zones. Addition of 10 nM retinoic acid to explants stimulated the release of both these forms of aggrecan whereas higher concentrations of retinoic acid (100 nM and 1000 nM) preferentially stimulated the release of the degraded forms. In this regard hypertrophic cells were the most responsive and resting cells were the least responsive. Analysis of the degraded fragments by polyacrylamide gel electrophoresis and by N-terminal sequencing indicated that aggrecan catabolism in all zones of the growth plate is due to the action of aggrecanase, a novel cartilage proteinase which is also active in normal and osteoarthritic articular cartilages (Sandy et al., 1992). These observations are discussed in terms of the role of aggrecan in the extensive matrix remodelling which accompanies chondrocyte hypertrophy in the growth plate. Key words: aggrecan, growth plate, catabolism, retinoic acid.

Introduction Detailed morphological and stereological characterization of growth plate chondrocytes and their surrounding matrix in situ (Hunziker et al., 1987) has shown that the transition from proliferative to hypertrophic state is accompanied by marked changes. During hypertrophy, chondrocytes increase in volume about 10-fold and the average

matrix volume per cell also increases about 3-fold. In the hypertrophic chondrocyte the absolute volumes and surface areas per cell of the organelles involved in matrix synthesis also increase between 2.5- and 5-fold (Hunziker et al., 1987) and this is in keeping with the idea that the hypertrophic cell produces and deposits the extracellular components of this expanding matrix. In addition, isotopic precursor studies with avian growth plate explants (Stocum

136

A. H. K. Plaas and]. D. Sandy

et aI., 1979) and with growing rabbits in vivo (Campo et aI., 1986) have supported this model of hypertrophic cell function. The increase in biosynthetic activity exhibited by hypertrophic cells is accompanied by marked changes in the quality of matrix components surrounding these cells. When compared to the resting zone, the hypertrophic zone matrix is characterized by thin type II collagen fibrils (Poole et aI., 1989) with evidence of pericellular cleavage of type II a-chains by collagenase (Dean et aI., 1985; Dean et aI., 1989), an abundance of the C-propeptide of type II (Alini et aI., 1992) and the appearance of type X collagen (Schmid and Linsenmayer, 1990). Somewhat less is known in terms of the fate of the proteoglycan components of the matrix during this remodelling phase. The large aggregating proteoglycan, aggrecan, is clearly present in all regions of the growth plate (Poole et aI., 1982) and studies on this molecule have primarily focussed on its possible involvement in controlling calcification, which occurs as a late event following hypertrophy (for review see Hunter, 1989). Extraction and characterization of aggrecan from individual zones of the proximal tibia has however shown that its concentration per matrix volume increases in the hypertrophic zone and that its aggregability with hyaluronan is also maintained at a high level (Buckwalter et aI., 1987; Matsui et aI., 1991; Alini et aI., 1992). Indeed it has been suggested (Poole et aI., 1989) that the high aggrecan concentration of the lower hypertrophic zone may be required to initiate focal sites of matrix calcification. There is also an increasing body of evidence for specific changes in aggrecan structure which accompany chondrocyte hypertrophy in the growth plate. Thus in both avian (Kimata et aI., 1974; Shaklee and Conrad, 1985) and human (Byers et aI., 1992 a) growth plates the hypertrophic zone appears to contain monomers bearing unique chondroitin sulfate structures. Further, in the hypertrophic zone of both rat and rabbit (Pita et aI., 1979; Axelsson et aI., 1983) growth plates there is evidence for the presence of unique super-aggregate structures. The mechanism by which disassembly and remodelling of the collagenous network and the accumulation of an aggrecan-rich matrix is co-ordinated by growth plate chondrocytes remains unclear. While altered synthetic activities, such as the production of type X collagen by hypertrophic cells, are clearly involved, it seems likely that distinct temporal patterns for the catabolism of specific matrix components are also necessary. In this regard, a unique pattern of synthesis and distribution of matrix metalloproteinases and TIMPl in the rabbit growth plate has now been described (Brown et aI., 1989). The present study was initiated to extend our understanding of the metabolism of aggrecan during matrix remodelling in the growth plate. We decided to approach this by examining the structure, biosynthesis and catabol-

ism of aggrecan in explants of discrete zones which can be readily prepared from the bovine fetal rib growth plate (Buckwalter et aI., 1987). Explant culture was chosen since this provided an opportunity to examine the metabolic behaviour of the chondrocytes in each growth plate zone with the cells present initially in their existing in vivo matrix.

Materials and Methods Ham's F12 medium, gentamycin, penicillin, streptomycin and neomycin were from Gibco. Fetal calf serum was obtained from HyClone. Chondroitinase ABC (protease free from Proteus Vulgaris) and keratanase (from Pseudomonas) were from ICN Immunobiologicals, Lisle, IL. and keratanase II (from Bacillus species KS36) was from Seikagaku America Inc., Rockville MD. eSS]-Na2S04 (carrier free) was from Amersham. DEAE TrisAcryl was from Serva Fine Biochemicals Inc., NY. Superose 6 (HR 10/30), Superose 12 (HR 10/30), Fast Desalting (HRI0110) and prepacked PD 10 columns were from Pharmacia LKB Biotechnology Inc. Retinoic acid (all-trans) (RA) and 4chloro-l-napthol was from Sigma. Precast 4-12% gradient gels (1 mm thick) were obtained from NOVEX, San Diego, CA. Goat anti-Mouse IgM was obtained from Calbiochem, San Diego, CA. Purified bovine decorin and biglycan were a generous gift from Dr. L. C. Rosenberg. All other chemicals used were of the highest purity available. General methods

Proteoglycans were assayed as total GAG as previously described (Farndale et aI., 1982). Collagen was assayed as hydroxyproline (Stegeman and Stalder, 1967) and DNA as described (Kim et aI., 1988). N-terminal sequencing was performed on a 473A sequencer (Applied Biosystems Inc., Foster City, CAl with on-line phenylthiohydantoin analysis. Preparation of aggrecan samples for NHrterminal sequencing and all sequencing procedures were als previously described (Sandy et aI., 1992). Protein was determined using the bicinchoninic acid assay kit from Pierce Chemical Co., Rockford, IL. Isolation and culture of fetal bovine rib growth plates

Ribs were removed from third trimester bovine fetuses (about 10 em tibial length) which had been obtained immediately after slaughter of pregnant cows. Individual growth plates were isolated as previously described (Buckwalter et aI., 1987) and placed in sterile PBS, supplemented with 10units/ml penicillin, 10 !-tg/ml streptomycin and 20 !-tglml neomycin. The cut surface of the epiphyseal end of each growth plate was mounted on a Lancer Series 1000 Vibratome cutting block which was then immersed in PBS.

Extracellular Processing of Aggrecan Six transverse sections of 1 mm were cut from the metaphyseal towards the epiphyseal surface, and these are referred to as slices 1 through 6 in the text. Each slice (80-100 mg wet wt) appeared oval in cross-section with dimensions of about 10 mm (long axis) by 7 mm (short axis).

Histology Histological sections were prepared by a modified version of a previous method (Hunziker et aI., 1987). Briefly, slices were incubated for 3 h at 22°C in 50 mM sodium cacodylate, pH 7.4 containing 2.5% (v/v) paraformaldehyde and 1.6% (w/v) ruthenium hexamine trichloride, washed for 15 min in 50 mM sodium cacodylate, pH 7.4 and embedded in Spurrs resin before sectioning at 2 [lm and staining with 1 % (w/v) toluidine blue.

Explant culture and biosynthetic labelling Growth plate splices were maintained at 37°C in 4 ml of basal medium (Ham's F12 containing 20 mM Hepes, 40 mM NaHC0 3 and 50 [lglml gentamycin) with daily medium changes. To assess rates of proteoglycan synthesis, slices were labelled for 4 h with 3 ml of basal medium (without gentamycin), containing 5 [lCi/ml 35S04. In timecourse studies it was shown that incorporation of radiolabel into proteoglycan was essentially linear over the first 14 h of labelling. Incorporation was terminated by removing the medium and papain digesting the tissue. The digests and medium were fractionated on PD10 columns in 50 mM sodium acetate, pH 6.8. The medium contained less than 0.1 % of the total radiolabelled macromolecules produced over the labelling period and was not further analysed. For determination of release rates, individual slices were transferred to 3 ml of basal medium (without gentamycin) containing 50 [lCi/ml 35S04 and incorporation was allowed to proceed for 4 h at 3 rc. Labelling medium was removed, the slices were washed with basal medium (3 x 10 ml) and maintained in culture for 7 days. In explants supplemented with RA, this was added with daily medium changes as previously described (Campell and Handley, 1987). The media were collected daily and assayed for total GAG (Farndale et aI., 1982) and [35 SJ-macromolecules by PD-10 chromatography. At the end of the chase period, slices were extracted as described below, and the total GAG (resident and radiolabelled) in each culture was calculated. Release rates for both resident and newly synthesised aggrecan were determined graphically by plotting the cumulative percentage (of total) release against time. For most cultures there was an essentially linear relationship between cumulative percentage release and time over the seven days of culture and the rate (% per day) was therefore estimated from the "line of best fit". Due to the high release rates in cultures maintained in 1000 nM RA, linearity was seen for

137

only the first two to four days and the rate was therefore calculated from this period (see Fig. 7).

Extraction and characterization of proteoglycans Explants labelled for 4 h with 50 [lCi/ml of 35S04 were chopped finely on ice and extracted for 24 h at 4°C with 5 ml of 4 M guanidine HCl, 50 mM sodium acetate, pH 7.3 containing 1 mM PMSF, 10 mM N-ethyl maleimide, 5 mM EDT A, 0.3 M aminohexanoic acid, 1 [lg/ml pepstatin and 15 mM benzamidine HC1. Under these conditions about 88% of the total proteoglycan was extracted from zones 6 and 5, and about 95% from the remaining zones. The extracts were adjusted to a density of 1.48 g/ml with cesium chloride and centrifuged for 48 hat 38,000 rpm in a Beckmann 50Ti rotor. Greater than 85% of the aggrecan was recovered in the D 1 fraction (1.55 glml) which was dialysed against water, 1.5 M NaCl and water before analysis. The hydrodynamic size of D1 aggrecan was determined on Sepharose CL2B in 4 M guanidine HCI, 50 mM sodium acetate, pH6.8. To assess the aggregability of aggrecan, portions (200 [lg GAG) were incubated with Healon (5%, w/w) and link protein (5%, w/w), before chromatography on Sepharose CL2B in 0.5 M sodium acetate, pH 6.8. For GAG size determination, portions (500 [lg) of D1 aggrecan were treated with 0.5 ml 50 mM NaOH, 1 M sodium borohydride for 24 h at 45°C, neutralized with acetic acid and evaporated three times in acidified methanol to remove borate salts. The pellet was dissolved in water and run on Superose 6 in 0.5 M pyridinium acetate, pH 5.8.

Analysis ofaggrecan fragments from medium of explant cultures by gradient gel electrophoresis D1 aggrecan isolated from explant medium was treated with chondroitinase ABC, keratanase and keratanase II as described (Sandy et aI., 1992). Portions (1-8 [lg of protein) were electrophoresed on a 4-12 % gradient gel and electroblotted onto nitrocellulose. The membranes were blocked overnight in 1 % (w/v) BSA, 0.15 M NaCI, 50 mM Tris HCl, pH 7.5, before incubation with antibody 3B3 (Caterson et aI., 1985) (gift from Dr. B. Caterson) at a dilution of 1:500 in Tris-buffered saline containing 0.05% (w/v) Tween 20, followed by incubation with horseradish peroxidase conjugated goat-anti-mouse IgM (1:5000) and 2.8 mM 4-chloro-1-napthol in 20% (v/v) methanol and 0.018% (v/v) hydrogen peroxide.

Characterization of small proteoglycans Portions of [35 SJ-Iabelled guanidine HCI extracts were applied to PD10 columns in 6 M urea, 50 mM Tris acetate, pH 7.0 (containing protease inhibitors). The macromolecules were applied to DEAE Trisacryl (1 ml bed volume), the bound material was eluted with 2 ml of 4 M guanidine.

138

A. H. K. Plaas and]. D. Sandy

HC1, desalted, separated on 3-15% SDS PAGE gradient gel (Beresford et ai., 1987) and visualized by autoradiography.

Results Compositional analysis Slices from five randomly selected growth plates from a single animal were analyzed for total proteoglycan, total collagen and total DNA. The same trends in composition were seen in all growth plates analyzed, and data from a single growth plate is shown to illustrate these changes (Fig. I). The DNA content per slice decreased markedly from slice 6 to slice 1 consistent with previous stereological estimates indicating a several-fold decrease in cell number per unit volume between proliferating and hypertrophic zones (Hunziker et ai., 1987). The collagen content per slice decreased even more markedly toward the hypertrophic zone. This is probably largely due to a decrease in matrix volume per slice, as described in similar work on fetal tibial growth plate (Alini et ai., 1992), but might also be a result of active degradation and loss of collagen in the hypertrophic matrix. In marked contrast, the proteoglycan content was maintained at a high level throughout the growth plate, supporting the conclusion (Alini et ai., 1992) that the aggrecan content per matrix volume (matrix concentration) actually increases in the hypertrophic zone.

Proteoglycan analysis in growth plate slices The nature of the proteoglycans present in the individual slices was assessed following high yield extraction with guanidine, HCl (see Method for details) and dissociative

•

Ilf{d OH Pro

GAG

cesium chloride density gradient separation of aggrecan from small proteoglycans. In all slices greater than 95% of the GAG was recovered in the D1 fraction, indicating that the majority of GAG in all regions of the growth plate is present on aggrecan. The aggrecan monomer size (Fig. 2, left hand panels) was similar in slices 6, 5 and 4, but showed a distinct increase in slices 3 and 2 with a further increase in slice 1. The peak Kav determined for these samples was consistent with an increase in the mean molecular mass of monomers (Ohno et ai., 1986) from 1.6 X 106 Da (slice 6) to 2.1 X 106 Da (slice3) to 3.0 x 106 Da (slice I). This marked increase in aggrecan size on moving from the resting to the hypertrophic zone was at least partly due to a concomitant increase in the size of the substituent GAGs. Analysis of GAGs (Fig. 2, right panels} also showed a similar size distribution in slices 6, 5 and 4, with an increase in size in slices 3 and 2 and a marked transition to larger chains in slice 1. The same variation in GAG size was observed whether the fractions were assayed for GAG content by the DMMB assay or by hexuronic acid assay (data not shown). The observed increase in GAG size was also consistent with an increase in the GAG: protein ratio {Ilg/Ilg} for aggrecan monomer which increased from about 13 in slice 6 samples to about 24 in slice 1 samples. However, increases in the length of the aggrecan core protein also appeared to contribute to the increase in monomer size observed in the hypertrophic region. Thus, quantitative peptide analysis (Sandy et ai., 1991a) for the G1, G2 and G3 domains on aggrecan from all slices indicated that while the Gland G2 domain content was essentially constant throughout, the G3 domain content increased by a factor of about 1.3 between slice 6 and slice 1 (data not shown). This suggests that the percentage of molecules with the full

~DNA

3500

350

~

.S:!

~

'"Q, ~

~

~ Q

'"

='" ~

0

Q

-< Z Q

0

o 6

5

4

3

Slice Number

2

1

Fig. 1. Biochemical composition of growth plate slices. Six 1 mm slices of a typical growth plate were taken for compositional analysis. Each slice was digested with papain and portions were analysed for GAG, hydroxyproline and DNA as described in the Methods section.

Extracellular Processing of Aggrecan length aggrecan core protein increases significantly on going from the resting to the hypertrophic zone. Aggrecan from all slices showed greater than 85% aggregability (data not shown), consistent with the high and reproducible content of G 1 domain found on peptide mapping of these samples.

Biosynthesis of proteoglycans in explant cultures In order to study proteoglycan synthesis in growth plate zones, individual slices were established in explant culture

139

in serum-free basal medium. To examine the effect of culture period on the biosynthetic activity of cells, the rate of biosynthesis of proteoglycans in each slice was determined by labelling with 35S04 on day 1, day 3 and day 7. The results (Fig. 3) showed that in the resting and proliferative zones (slices 6, 5 and 4) the rate of synthesis (cpm per slice per h) was essentially constant over the seven day culture period. In the hypertrophic zones (slices3, 2 and 1) the rate of synthesis was maintained over the first three days of culture but there was a marked decrease in rate on further maintenance up to day 7. When taken together with the

AGGRECAN MONOMER

GLYCOSAMINOGLYCANS

A 525

A 525 1.0

1.0

SLICE #6

1.0

1.0

SLICE #5

1.0

1.0

SLICE #4

1.0

1.0

SLICE #3

-.-...., 1.0

Fig. 2. Chromatography of aggrecan monomer and substituent GAG chains from individual growth plate slices. Aggrecan and substituent GAGs were prepared from slices as described under Methods. Portions (250 f-lg) of aggrecan were fractionated on Sepharose CL-2B (left hand panels), as described in the Methods. Portions (100 f-lg) of alkaline borohydride released GAGs were fractionated on Superose 6 in 0.5 M pyridinium acetate, pH 6.0 (right hand panels), as described in the Methods.

1.0

SLICE #2

1.0

1.0

SLICE #1

0

0.5 Kav

1.0

0

0.5 Kav

1.0

140

A. H. K. Plaas and]. D. Sandy

80

r..

D

Oayl

D

_

Oay3

Day7

60

::I 0

..c r..

~

Q.

~

.~

Vi

40

r..

~

Q.

;;:;-

6 ....

~

20

~

Q..

U

0 6

5

4 3 Slice Number

2

DNA data (Fig. 1), it can be calculated that the biosynthetic activity in explanted tissue on day 1 was 0.20, 0.32, 0.55, 0.73, 0.72 and 0.20 cpm X 10- 3 per h per t-tg DNA) for slices6, 5, 4, 3, 2 and 1 respectively. Maximal cellular biosynthetic activity for proteoglycans therefore appeared to be in slices 3 and 2 (upper and middle hypertrophic zones). Structure of newly synthesised proteoglycans in explant cultures

To examine the effect of maintenance in this explant system on the quality of the proteoglycans being synthesised, portions of the guanidine HCI extracts of tissue labelled on days 1 and 7 were combined with hyaluronan and link protein, dialyzed to associative conditions and fractionated on CL2B (Fig. 4). The results showed that the newly synthesised aggrecan in all slices up to 7 days of culture was greater than 85 % aggregatable, which is similar to the resident aggrecan population. In addition, about 10% of the radiolabel in eac!) of these samples (see Fig. 4, inset panels) eluted in the position for small proteoglycans. SDS-PAGE analysis of the small proteoglycans synthesised on day 1 (Fig. 5) showed the presence in all zones of two [35 Sl-labelled species with electrophoretic properties consistent with decorin and biglycan. This identification was based on their electrophoretic mobility relative to standards and on their sensitivity to digestion with chondroitinase ABC:Interestingly, the electrophoretic mobility of both species decreased in the hypertrophic zones, consistent with an increase in size of the newly synthesised molecules. This may be explained by an increase in the size of the substituent GAGs on newly synthesised small proteoglycans in the hypertrophic zone, much as seen for newly synthesised aggrecan (see below).

1

Fig. 3. The effect of explant culture on proteoglycan synthesis in individual growth plate slices. Individual slices (6 through 1) from nine growth plates of a single animal were established in explant culture. The rate of proteoglycan synthesis (35 S0 4 incorporation per slice per hour) was determined (see Method for details) for each slice type on days 1, 3 and 7 of culture. The data shown is the mean ± SO derived from three separate cultures of each slice type.

Portions of the guanidine HCI extracts of tissue labelled on days 1, 3 and 7 were also fractionated on dissociative CL-2B. In all slices, the apparent size of the radiolabelled monomers was greater than the resident population on day 1, but similar to the resident population at later labelling times. To investigate this further, alkaline-borohydride treated samples were fractionated on Superose 6 to assess the size of the substituent GAGs. Consistent with the results obtained for whole aggrecan, the newly synthesised GAGs were larger than the resident on day 1, but similar to the resident at later labelling times (data not shown). It therefore appears that following an early transient period of synthesis of abnormally long GAGs, the aggrecan being synthesised in these explants is similar to the resident population in all zones for up to seven days of culture. Histology ofgrowth plate slices after explant culture

A single growth plate was sectioned as described in the Methods and each slice was cut in half along the short axis. One portion was immediately processed for histology, and the second half was maintained in basal medium for 7 days. The cellular morphology of fresh tissue slices 1 through 6 is shown in Fig. 6. These micrographs clearly show the small rounded cells of the resting zone (slice 6), some cellular enlargement in the proliferative zone (slices 5 and 4) and the progressive columnar arrangement and enlargement of cells through the hypertrophic zone (slices 3, 2 and 1). A comparison of chondrocyte morphology in fresh tissue (upper panel) with that after 7 days in culture (lower panel) showed little change except for an indication of further enlargement of the cells in the hypertrophic zone (slices 2 and 1).

Extracellular Processing of Aggrecan

Day 7

Day! 1.0

1.0

\.

-

11.9% ...........:::::::::.,!!\Iir.

--.---- -.-.""::'"'.-=--.-

"0

141

-~~-

1.0

1.0

.~ -; 8

9.9%

""

Q

Z

.~

0

rI:J

--

II)

rr'l

'-'

;...

:~

( "I

~

-I

--_._----. ........... .

----.,;.:;...;;.;.

1.0

1.0

:0

=

CII:

9.8%

9.3% ·u.... ..... ~:.....

o

0.5

1.

Kav

0.5

Kav

1.0

Fig. 4. The effect of explant culture on the synthesis of aggrecan and small proteoglycans in individual growth plate slices. Individual slices from two growth plates of a single animal were established in explant culture and radiolabelled on day 1 (left hand panels) or day 7 (right hand panels) for 4 h. [35 S]-proteoglycans were extracted, combined with exogenous hyaluronan and link protein (see text for details) and fractionated on Sepharose CL-2B under associative conditions. The profiles shown are from slice 6 (top panels), slice 4 (middle panels) and slice 2 (bottom panels). In each case the inset panels show the profiles plotted on an expanded scale to illustrate the presence of [35 S]-labelled small proteoglycans (shaded area).

Release ofaggrecan from explant cultures In order to examine mechanisms of aggrecan degradation in discrete zones of the growth plate, the medium of explant cultures was monitored for release of newly synthesised and resident aggrecan in a pulse chase protocol (see Method for details). For all slices in basal medium there was an essentially linear release of both GAG and [35 Sl-GAG over the seven days, which amounted to 15 - 20% of the total pres-

ent. Cultures maintained in the presence of increasing concentrations of RA (10 nM, 100 nM and 1000 nM) showed a dose-dependent increase in the rate of release of both GAG and [35 Sl-GAG over this period. To illustrate this, the cumulative release data for slice 4 under basal conditions and RA treatment are given in Fig. 7. When the rate of release (% of total released per day) was calculated (see Methods) for each slice type in basal and

142

A. H. K. Pia as and]. D. Sandy relatively unaffected, even at the highest concentration of RA.

1 2 3 4 5 6 7 8 9 10

Degradation ofaggrecan by growth plate chondrocytes

M

:8

p

:0

Fig. 5. Fluorogram of a 3-15% gradient SOS-PAGE of [35 S]_ labelled small proteoglycans synthesised in growth plate explants. 5 S]-labelled proteoglycans were prepared for electrophoresis as described in the Methods. Samples shown are chondroitinasetreated calf articular 5 S]-small PG (Lane 1), calf articular [35 S]_ small PG (Lane2), growth plate zones6 thru 1 (Lanes3 thru 8 respectively), growth plate zone 1 (Lane 9) and chondroitinasetreated growth plate zone 1 (Lane 10). The migration positions of myosin (M, 200 kOa), phosphorylase B (P, 95.5 kOa), purified bovine biglycan (B) and bovine decorin (0) are indicated.

e

e

RA-stimulated cultures the results shown in Fig. 8 were obtained. The same general effects were seen for both resident GAG and eSS]-GAG. Thus under basal conditions the rate was low and similar in all slices. Responsiveness to RA was zone dependant, with the most pronounced effect on aggrecan release being seen in the hypertrophic zones (slices3, 2 and 1). The proliferative zones (slices 5 and 4) were next most responsive and the resting zone (slice 6) was

To assess the extent to which release of aggrecan in these explants was accompanied by degradation of the core protein with loss of aggregability, radiolabelled medium aggrecan (pool of all medium collections) from slices 2, 4 and 6 was dissociated in 4 M guanidine HCI, reassociated in the presence of carrier calf A1D1, excess hyaluronan and link protein and fractionated on associative Sepharose CL2B (data not shown). The results showed that much of the released aggrecan had retained the capacity to bind hyaluronan. For example for slice 4, the percentage of aggregatable monomers was 72% in basal medium, 68% at 10nM RA, 40% at 100nM RA and 24% at 1000nM RA; indeed chromatography of these samples without addition of exogenous hyaluronan and link protein gave similar data, indicating that intact aggregates were being released under these conditions. The proportion of the medium aggrecan in the aggregating and non-aggregating forms, was then used to calculate the percentage of the total pulse-labelled product which was released in these two forms over the seven days of culture, in slices 2,4 and 6, in the various concentrations of RA (see TableI). Under basal conditions total release was low (17.1 %-24.0%) and the majority of the product from all slices was aggregatable. In the presence of 10 nM RA increased release was seen only in slices 2 and 4 and this was due to enhanced release of both aggregating and nonaggregating forms. At 100 nM RA, increased release of both forms was now seen in all zones. Finally, at 1000nM the majority of the medium product was non-aggregating. Aggregation studies on the molecules retained in the tissue over the culture period in these samples, consistently showed this to be greater than 85% aggregatable. It therefore appears that in contrast to articular cartilage explants, aggrecan release from growth plate explants can result from both the previously described proteolytic removal of the Gl domain, and a second process which does not involve such loss of aggregability. Both processes are clearly promoted by treatment of the tissue with RA.

Table I. Percentage of total [35 S]-labelled-aggrecana • Retinoic Acid concentration (nM)

Agg

Non-Agg

Agg

Non-Agg

Agg

Non-Agg

0 10 100 1000

9.7 10.3 25.4 27.2

7.4 6.4 17.0 47.2

15.4 24.6 34.8 23.4

6.6 11.6 53.3 71.5

17.3 28.4 45.6 26.8

6.7 12.7 36.3 68.3

a

Slice 2

Slice 4

Slice 6

The values shown are the percentage of the [35 S1-labelled aggrecan, formed during the pulse label, which was released into the medium in the aggregating (Agg) and non-aggregating (Non-Agg) forms. For example, to calculate the Slice 4 value for Agg at 100 nM RA, the % release over 7 days (87.4% , taken from Fig. 7, right panel) was multiplied by the % aggregatable (40%, as determined on Sepharose CLlB) to obtain 34.8 % ).

Extracellular Processing of Aggrecan

4

5

2

3

SLICE No

143

1

Fig. 6. Histology of growth plate slices. A longitudinal section (magnification x 100) of the six slices before (upper panel) and after (lower panel) explant culture is shown.

100

100 80

100nM

"C ~

'"~

.. ''<"" .... Qi

"C

80

..

60

~

100nM

~

Qi

60

'r.h<" '"

40

1Il "l

';

....s

l~nM

__--------~~I~nM

20

Basal

0

~ I

1

I

2

I

3

I

4

Days in Culture

I

5

...~....

40

10nM

20

Basal

0

~

r-I

I

6

7

1

2

3

4

5

6

7

Days in Culture

Fig. 7. The effect of RA on the time course of aggrecan release from growth plate slices in explant culture. Growth plate explants (data from slice 4 shown here) were pulse labelled with [35 S]-sulfate and maintained in the presence of various concentrations of retinoic acid with daily medium collections. The percentage distribution of GAG (left panel) and [35 S]-GAG (right panel) between the tissue and the medium was determined as described in Methods.

144

A. H. K. Plaas and J. D. Sandy 50

50

(a) ;...

~

40

r..

lOOOoM

30

.. ~

~

10

I

I

I

I

lOOnM

ol!:::~~~j:!::~:=!:~...J. ......e 100M Basal 6

5

4

3

2

I

40

~

"0

"ii r..

20

(b)

-< ~

rr.. III

lOOOnM

30 20

I

I

--

-

5

4

10

~

0

I

I

1 J. . I l~-- _______

to)

6

Slice Number

~

lOOnM 100M

- - - . Basal

3

I

2

Slice Number

Fig. 8. Sensitivity of different growth plate zones to RA-stimulated release of aggrecan. The rate of release for GAG (left panel) and [35 SJ-GAG (right panel) for each slice type at the indicated concentrations of RA is shown. Details of the calculation of release rates are given in the Methods.

The mechanism ofaggrecan catabolism in growth plate explants

Aggrecan in seven day medium collections from slices 2, 4 and 6 (basal and 1000nM RA-treated) was purified by CsCI gradient centrifugation and processed for N-terminal analysis by deglycosylation and Superose 12 fractionation as previously described for synovial fluid samples (Sandy et aI., 1992). For basal samples, the major products were recovered from the Vo of the Superose 12. The predominant sequence obtained with this material was VEVxxP which is the expected N-terminal of the bovine G1 domain; this result is consistent with the high degree of aggregability observed for these samples. A second, quantitatively minor sequence of ARGxVIL was also found and this is generated by cleavage of a Glu-Ala bond in the interglobular domain corresponding to the Glu 373-Ala 374 bond of the human protein (Doege et aI., 1991). The same analysis of aggrecan fragments in the medium of 1000 nM RA-treated slices gave evidence of a wider range of sizes of the core fragments on Superose 12 (data not shown). Analysis of the fragments in the Vo gave the same two N-terminal sequences described above (VEVxxP and ARGxVIL) in approximately equal yield. Analysis of smaller core fragments, which were recovered from the included volume of the column, gave two other N-terminal sequences. The predominant sequence obtained was AGEGPXGILE and a quantitatively minor sequence found was LGQxxxVxY. These N-terminals are generated by cleavage of bovine core protein at positions which correspond to the Glu 1819-Ala 1820 bond and the Glu 1919Leu 1920 bond in the human protein (Doege et aI., 1991). The electrophoretic separation of the core protein fragments prepared from RA-treated growth plate explants is shown in Fig.9 (lane 1). Four major immunoreactive species were found with apparent molecular masses at about 200, 170, 135 and 100 kDa. Further, an essentially identical fragment pattern was obtained with the catabolic

products released from all growth plate zones (data not shown). These core fragments have been shown by others with bovine articular explants (Hic et aI., 1992; Loulakis et aI., 1992) to indeed represent the species defined by the N-

1

2

3

4

200---

97----

69---_

46----

j

30---22---Fig. 9. Analysis of aggrecan fragments released from growth plate and articular cartilage explant cultures. Portions of deglycosylated core protein preparations from culture media (see Methods for detail) were electrophoresed on 4-12% gradient gels, electro blotted and detected with monoclonal antibody 3B3 (which recognizes the core protein-bound terminal 6sulfated disaccharide generated by chondroitinase ABC). Samples shown are: lane 1, bovine rib growth plate (slice 2) maintained in 100 nM retinoic acid; lane 2, bovine articular cartilage maintained in 20 Vlml IL-1u; lane 3, bovine articular cartilage maintained in 1000 nM retinoic acid; lane 4, bovine articular cartilage maintained in basal medium. Samples shown in lanes 2-4 were prepared from articular explants as described (Sandy et aI., 1991 a).

Extracellular Processing of Aggrecan terminal sequences described above. This similarity in the pattern of aggrecan core protein cleavage seen in growth plate and articular cartilage explants is further illustrated in Fig. 9. Thus, the same four major immunoreactive core protein products seen in growth plate (lane 1) also appear to be generated in bovine articular cartilage explants under basal conditions (lane 4), IL-1-stimulated catabolism (lane 2) or RA-stimulated catabolism (lane 3).

Discussion The present study was undertaken to examine the structure and metabolism of aggrecan in discrete zones of the mammalian growth plate. Our compositional data (Fig. 1) generally confirm the results of a similar study undertaken in the bovine proximal tibial physis (Matsui et aI., 1991), and therefore support the view that the matrix aggrecan concentration increases markedly in the hypertrophic zone, where the collagen content is at a minimum. Our determination of aggrecan monomer size (Fig. 2) showed that molecules in the resting and proliferative zones were similar in hydrodynamic size, whereas those in the upper and particularly in the lower hypertrophic zone were significantly larger. Evidence in support of this was obtained by two independent methods. Firstly, the GAG chains on aggrecan from the hypertrophic zones were significantly larger than in other zones (Fig. 2) and secondly, quantitative peptide mapping suggested an increase in the mean aggrecan core size in the hypertrophic zone. A similar increase in the size of aggrecan in the hypertrophic zone of foetal sheep growth plate has also recently been described (Byers et aI., 1992b). The increase in the hydrodynamic size of matrix aggrecan along with its increased concentration in the hypertrophic zones would be most readily explained by the addition of a new population of "hypertrophic" molecules to the matrix during the transition from resting cells to hypertrophic cells. Indeed, evidence for this was obtained from the biosynthetic studies with explants. Thus, maximal proteoglycan synthesis per cell was observed in the upper hypertrophic zone (Fig. 3), where the unusually large monomers are first detected. Moreover, the newly synthesised molecules in the hypertrophic zone show the same increase in GAG size as is seen with the resident population. Further, this variation in the size of the newly synthesised GAGs is also seen after 7 days of explant culture, indicating that the unusual aggrecan structure produced by hypertrophic cells is a stable biosynthetic marker for this cell type. In this regard it is interesting that the apparent size of decorin and biglycan synthesised in explants (Fig. 5) also increased markedly in the hypertrophic zone, presumably due to the synthesis of abnormally long GAGs. Our results also showed that the aggrecan extracted from all zones exhibited a high degree of aggregability, and there

145

is no evidence from the present study for the accumulation of degraded or smaller forms of aggrecan in the hypertrophic zones of the growth plate. The absence of degraded forms of aggrecan in these extracts does not of course suggest that aggrecan is not being actively degraded in growth plate cartilage in vivo. Thus from studies with articular cartilage explants (Sandy and Plaas, 1986; Sandy et aI., 1991 a; Sandy et aI., 1991 b) and synovial fluid analysis (Sandy et aI., 1992), it is clear that degraded forms of aggrecan are lost rapidly from the tissue by diffusion. Therefore, in order to study the capacity of different growth plate zones to degrade aggrecan, and to define the nature of the protease involved, we have used growth plate cartilage explant methods. The advantage of explants over isolated cells for this purpose is that the cells are present in their in vivo matrix and their capacity to degrade that matrix can be directly evaluated by collection and structural analysis of the catabolic products in the culture medium. In basal medium we found that the rate of release of aggrecan from growth plate explants was very slow (about 1-2 % per day) and the majority of the product had retained the capacity to bind hyaluronan. However, following addition of RA to the culture medium, a high yield of degraded aggrecan was obtained from all growth plate zones. Since RA appears to accelerate the normal pathway of aggrecan catabolism (Ilic et aI., 1992; Morales and Roberts, 1992) we have analysed in detail the products of growth plate explants treated with this agent. All zones of the growth plate exhibited some sensitivity to RA at 100-1000nM (Fig. 8), although the hypertrophic zone slices were particularly responsive at the higher concentration. Most significant was the finding, based on N-terminal sequencing and electrophoretic analysis (Fig. 9), that the proteinase which is active in the growth plate, both under basal conditions and in the presence of RA, is the same enzyme that promotes aggrecan catabolism in articular cartilage explants (Sandy et aI., 1991 b; Hic et aI., 1992; Loulakis et aI., 1992). However, in contrast to articular explants, growth plate explants release a high proportion of aggrecan which retains aggregability, and much of which appears to be released as intact aggregates (Table I). Release in this form may be the result of the high porosity of the collagen network in the extended interterritorial matrix of growth plate relative to the densely packed fibrils of articular cartilage (Eggli et aI., 1985). The present findings therefore indicate that, unlike articular cartilage, the matrix of growth plate has a high content of proteoglycan aggregates which are relatively free to diffuse. This would appear to support the idea (Alini et aI., 1992) that aggregates which are free of interactions with type II collagen may promote the formation of focal concentrations of aggrecan in calcifying sites in the lower hypertrophic zone of the growth plate. In summary, the present study has provided the first

146

A. H. K. Plaas and]. D. Sandy

detailed structural evidence for the presence in the hypertrophic zone matrix of growth plate of a "hypertrophic" form of aggrecan. This species aggregates with hyaluronan, has a high content of the G3 domain and is substituted with unusually long chondroitin sulfate chains. Such chains may carry the non-reducing terminal 6-sulfated disaccharide structure (3B3 epitope) described as being adjacent to hypertrophic cells in the human growth plate (Byers et ai., 1992a). In addition this work has clearly indicated that aggrecan catabolism in the growth plate is catalysed by a proteinase, now referred to as aggrecanase, which is also responsible for aggrecan release from articular cartilage into the synovial fluid in both normal and osteoarthritic joints (Sandy et ai., 1992). While the nature of this enzyme remains to be established, it does not appear to be either stromelysin-1(MMP-3) or 72kDa gelatinase (MMP-2), based on the cleavage specificity for aggrecan of these enzymes in solution studies (Flannery et ai., 1992). Finally, this paper describes a mammalian growth plate culture system, which could be used to further examine the temporal sequence of matrix changes which accompany chondrocyte hypertrophy. With this in mind, work is now in progress to establish explant culture conditions which are permissive to hypertrophy of proliferative zone cells, much as has been shown with explants of whole chicken tibiae in serum-free conditions (Cole et ai., 1992). Footnotes 1 Abbreviations used are: TIMP, Tissue Inhibitor of Metalloproteinases, RA, retinoic acid; GAG, glycosaminoglycan; PBS, phosphate-buffered saline; EDTA, ethylenediaminetetra acetic acid; BSA, bovine serum albumin; DMMB, dimethylmethylene blue; SDS-PAGE, sodium dodecyl sulfate polyacrylamide gel electrophoresis; IL-la, Interleukin-1 a.

Acknowledgements We gratefully acknowledge the excellent technical assistance of Raymond Boynton. Additionally, we would like to thank YoungJo Kim for DNA analysis, Tim Ganey for histology, Carl Flannery for core protein isolation and Dr. Peter Neame for N-terminal sequence analysis. This work was supported by grants 15959 and 15960 from the Shriners of North America and AR 38580 from NIH.

References Alini, M., Matsui, Y., Dodge, G. R. and Poole, R. A.: The extracellular matrix of cartilage in the growth plate before and during calcification. Changes in composition and degradation of Type II Collagen. Calcified Tiss. Int. 50: 327 -335, 1992. Axelsson, 1., Berman, 1. and Pita, J. c.: Proteoglycans from Rabbit Articular and Growth Plate Cartilage. Ultracentrifugation, Gel Chromatography and Electron Microscopy. J. BioI. Chem. 258: 8915-8921,1983. Beresford,]. N., Fedako, N. S., Fisher, L. W., Midura, R.J., Yanagishita,M., Termine,J.D. and Gehron-Robey,P.: Analy-

ses of the proteoglycans synthesised by human bone cells in vitro.]. Bioi. Chern. 262: 17164-17172,1987. Brown, C. C., Hembry, R.M. and Reynolds,].J.: Immunolocalization of metalloproteinases and their inhibitors in the rabbit growth plate.]. Bone ]t. Surgery 71-A(4): 580-593, 1989. Buckwalter,J.A., Rosenberg,L.C., and Ungar,R.: Changes in proteoglycan aggregates during cartilage mineralization. Calcif. Tiss. International. 41: 228 - 236, 1987. Byers,S., Caterson,B., Hopwood,].J. and Foster,B.K.: Immunolocation analysis of glycosaminoglycans in the human growth plate.]. Histochem. Cytochern. 40: 275-282, 1992a. Byers, S., van Rooden,J. C., Hopwood,].]. and Foster, B. K.: Changes observed in proteoglycan structure as growth plate chondrocytes mature and hypertrophy. Trans. Orth. Res. Soc. 17: 31, 1992b. Campbell,M.A. and Handley, c.J.: The Effect of Retinoic Acid on Proteoglycan Turnover in Bovine Articular Cartilage Cultures. Arch. Biochem. Biophys. 258: 143-155,1987. Campo,R.D. and Romano,J.E.: Changes in cartilage proteoglycans associated with calcification. Calif. Tiss. International. 39: 175 -184, 1986. Caterson,B., Christner,J.E. and Couchman,].R.: Production and characterization of monoclonal antibodies directed against connective tissue proteoglycans. Fed. Proc. 44: 386-393, 1985. Cole,A.A., Luchene,L.J., Linsenmayer, T.F. and Schmid, T.M.: The influence of bone and marrow on cartilage hypertrophy and degradation during 30-day serum-free culture of the embryonic chick tibia. Devel. Dynamics 193: 277-285, 1992. Dean,D.D., Ofelia,M.E., Berman,I., Pita,J.C., Carreno,M.R., Woessner,].F. and Howell,D.S.: Localization of Collagenase in the Growth Plate of Rachitic Rats. ]. Clin. Invest. 76: 716-722,1985. Dean,D.D., Muniz, O.E. and Howell,D.S.: Association of collagenase and tissue inhibitor of metalloproteinase (TIMP) with hypertrophic cell enlargement in the growth plate. Matrix 9: 366-375,1989. Doege, K., Sasaki, M., Kimura, T. and Yamada, Y.: Complete coding sequence and deduced primary structure of the human cartilage large aggregating proteoglycan, aggrecan. ]. BioI. Chern. 266: 894-902, 1991. Eggli,P.S., Hermann, W., Hunziker,E.B. and Schenk,R.K.: Matrix compartments in the growth plate of the proximal tibia of rats. Anat. Rec. 211: 246-257, 1985. Farndale,R.W., Sayers,C.A. and Barrett, A.].: A direct Spectrophotometric Microassay for Sulfated Glycosaminoglycap.s in Cartilage Cultures. Connect. Tiss. Res. 9: 247-248, 1982. Flannery,C.R., Lark,M.W. and Sandy,].D.: Identification of a Stromelysin Cleavage Site Within the Interglobular Domain of Human Aggrecan. Evidence for Proteolysis at this Site In Vivo in Human Articular Cartilage.]. BioI. Chem. 267: 1008-1014, 1991. Hunter, G. K.: Role of proteoglycans in the provisional calcification of cartilage. Clin. Ortho. Related Res. 262: 256-280, 1989. Hunziker,E.B., Schenk,R.K. and Cruz-Orive,J.-M.: Quantitation of chondrocyte performance in growth-plate cartilage during longitudinal bone growth. ]. Bone ]t. Surgery 69-A(2): 162-173,1987. Ilic, M. Z., Handley, C.J., Robinson, H. C. and Mok, M. T.: Mechanism of catabolism of aggrecan by articular cartilage. Arch. Biochem. Biophys. 294: 115 -122, 1992 Kim,Y.-J., Sah,R.L.Y., Doong,].Y.H. and Grodzinsky,A.J.: Fluorometric Assay of DNA in Cartilage Explants Using Hoechst 33258. Anal. Biochem.174: 168-176,1988.

Extracellular Processing of Aggrecan Kimata, K., Okayama, M., Oohira, A. and Suzuki, S.: Heterogeneity of proteochondroitin sulfates produced by chondrocytes at different stages of cytodifferentiation. J. Bio!. Chern. 249: 1646-1653,1974. Loulakis, P., Shrikhande, A., Davis, G. and Maniglia, C.A.: N-terminal sequence of proteoglycan fragments isolated from medium of Interleukin-1 treated articular cartilage cultures. Biochem.J. 284: 589-593, 1992. Matsui, Y., Alini, M., Webber, C. and Poole, R. c.: Characterization of aggregating proteoglycans from the proliferative, maturing hypertrophic and calcifying zones of the cartilaginous physis. J. Bone Jt. Surgery 73-A(7}: 1064-1074,1991. Morales, T.I. and Roberts, A.B.: The Interaction Between Retinoic Acid and Transforming Growth Factors-~ in Calf Articular Cartilage Organ Cultures. Arch. Biochem. Biophys. 293: 79-84, 1992. Ohno, H., Blackwell, J., Jamieson, A. M., Carrino, D. A. and Caplan, A. I.: Calibration of the relative molecular mass of proteoglycan subunit by column chromatography on Sepharose CL-2B. Biochem.J. 235: 553-557, 1986. Pita,]. c., Muller, F.J., Morales, S. M. and Alarcon, E.J.: Ultracentrifugal characterization of proteoglycans from rat growth cartilage. J. Bioi. Chern. 254: 10313-10320,1979. Pacifici,M., Golden,E.B., Adams,S.L. and Iwamoto,M.: Retinoic Acid and Basic FGF differentially regulate chondrocyte maturation. Trans. Ortho. Res. Soc. 17: 155, 1992. Poole, A. R., Pidoux, I. and Rosenberg, L.: Role of proteoglycans in endochondral ossification: Immunofluorescent localization of link protein and proteoglycan monomer in bovine fetal epiphyseal growth plate. J. Cell Bioi. 92: 249- 260,1982. Poole, A. R., Matsui, Y., Hinek, A. and Lee, E. R.: Cartilage macromolecules and the calcification of cartilage matrix. Anatomical Rec. 224: 167-179,1989. Poole,A.R.: The Growth Plate: Cellular Physiology, Cartilage

147

Assembly and Mineralization. In: Cartilage: Molecular Aspects, ed. by. Hall, B. and Newman, S., CRC Press Inc., London, 1991, pp.179-211. Sandy,J. D. and Plaas, A. H. K.: Age-related Changes in the Kinetics of Release of Proteoglycans from Normal Rabbit Cartilage Explants.J. Orthop. Res. 4: 263-272, 1986. Sandy,].D., Boynton,R.E. and Flannery,C.R.: Analysis of the Catabolism of Aggrecan in Cartilage Explants by Quantitation of Peptides from the Three Globular Domains. J. Bioi. Chern. 266: 8198-8205, 1991a. Sandy,]. D., Neame, P.J., Boynton, R. E. and Flannery, C. R.: Catabolism of Aggrecan in Cartilage Explants. Identification of a Major Cleavage Site Within the Interglobular Domain. J. Bioi. Chern. 266: 8683-8685,1991 b. Sandy,J. D., Flannery, C.R., Neame, P.J. and Lohmander, L. S.: The Structure of Aggrecan Fragments in Human Synovial Fluid. J. Clin. Invest. 89: 1512-1516,1992. Schmid, T.M. and Linsenmayer, T.F.: Immunoelectron microscopy of Type X collagen. Supramolecular forms within embryonic chick cartilage. Dev. Bioi. 138: 53-62, 1990. Shaklee,P.N. and Conrad,H.E.: Structural changes in the large proteoglycan in differentiating chondrocytes from the chick embryo tibiotarsus. J. Bioi. Chern. 260: 16064-16070, 1985. Stegemann, H. and Stalder, K.: Determination of Hydroxyproline. Clin. Chim. Acta 18: 267-273, 1967. Stocum,D.L., Davis,R.M., Leger,M. and Conrad,H.E.: Development of the tibiotarsus in the chick embryo: Biosynthetic activities of histologically distinct regions. J. Embryol. Exp. Morph. 54: 155 -170,1979. Anna H. K. Plaas, Ph. D., Orthopaedic Research Laboratory, Shriners Hospital for Crippled Children, 12502 North Pine Drive, Tampa, FL 33 612, USA.