Variability in the G3 domain content of bovine aggrecan from cartilage extracts and chondrocyte cultures

ARCHIVES

OF BIOCHEMISTRY

AND BIOPHYSICS

Vol. 297, No. 1, August 15, pp. 52-60, 1992

Variability in the G3 Domain Content of Bovine Aggrecan from Cartilage Extracts and Chondrocyte Cultures Carl Flannery, Victor Stanescu,* Matthias John Gordy, and John Sandy$l

Miirgelin,?

Raymond Boynton,

Tampa Unit, Shriners Hospital for Crippled Children, 12502 North Pine Drive, Tampa, Florida 33612; *URA.584-CNRS, H6pital des Enfants-Mulades, 149 rue de Sevres, 75743 Paris, France; tAbteilung Biophysikalische Chemie, Biozentrum, CH-4056 Basel, Switzerland; and $Department of Biochemistry and Molecular Biology, School of Medicine, University of South Florida, Tampa, Florida 33612

Received February

26, 1992, and in revised form April 13,1992

The content of the globular domains Gl, G2 and 63 on the core protein of high-density (AlDl) aggrecan isolated from newborn and mature bovine cartilage and from cultures of bovine chondrocytes was examined. Quantitation based on the 220 nm absorbance of tryptic marker peptides from each domain isolated by reversed-phase HPLC showed that while the content of Gl and G2 was essentially the same for all samples, the content of 63 varied markedly. The molar yield of G3 and Gl marker peptides indicated that approximately 55% of the Glbearing aggrecan from immature cartilage carried the G3 domain, while for mature cartilage this figure was markedly reduced, at about 35%. Aggrecan prepared from the cell layer matrix of calf chondrocyte cultures had an apparent G3 content similar to newborn cartilage (55%), whereas aggrecan prepared from the medium of these cultures had a markedly higher G3 content, at about 80%. The high content of G3 in cell medium samples compared to cartilage extracts was supported by electron microscopic analysis of AlDl preparations. The G3 content of the two subpopulations of aggrecan present in mature cartilage and separable by flat bed agarose gel electrophoresis was also determined at about 45% (Band I) and 20% (Band II) respectively. These results are discussed in terms of the likely origin of the marked variability in the G3 domain content of aggrecan. o 1992 Academic Press, Inc.

The large aggregating proteoglycan of articular cartilage, aggrecan, is a polydisperse, complex macromolecule 1 To whom correspondence 5.2

should be addressed.

comprised of a protein core to which numerous chondroitin sulfate and keratan sulfate chains and 0- and Nlinked oligosaccharides are covalently attached. The core protein contains three disulfide-bonded globular domains, designated Gl, G2, and G3.2 The Gl domain interacts with hyaluronic acid and link protein to facilitate the formation of large multimolecular aggregates. Specific functions for G2 and G3 are presently unknown, although a region of the G3 domain which has high homology to vertebrate lectins has been expressed and shown to bind specifically to fucose and galactose (1). Aggrecan monomers undergo extensive changes with aging and during the course of joint diseases (for review see Ref. (2)). Some of the age-dependent changes characterized (Refs. (3-6), among others) include an increase in substitution of core protein with keratan sulfate, a decrease in the ratio of chondroitin sulfate (CS)3 to protein, and a decrease in the hydrodynamic size of monomers. While a reduction in CS chain size and/or substitution could clearly explain the latter two changes, proteolytic cleavage of core protein, resulting in CS-substituted species deficient in Gl and/or G3 also contributes to the heterogeneity observed (2, 7-10). Indeed, electron microscopy following glycerol spraying/rotary shadowing of bovine nasal aggrecan has shown that monomers lacking G3 appear to have a significantly shorter protein core than those in which this domain is present (11). Fur’ Aggrecan globular domains are defined in Refs. (15) and (16). 3 Abbreviations used: CS, chondroitin sulfate; IGD, interglobular domain; PSEA, polysulfoethylaspartamide; CsCl, cesium chloride; Hepes, 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid; Mes, 4-morpholineethanesulfonic acid; EDTA, ethylenediaminetetraacetate; PMSF, phenylmethylsulfonyl fluoride; GAG, glycosaminoglycan, FPLC, fast protein liquid chromatography. 0003-9861/92 $5.00 Copyright 0 1992 by Academic Press, Inc. All rights of reproduction in any form reserved.

VARIABILITY

IN THE

G3 DOMAIN

CONTENT

OF BOVINE

AGGRECAN

53

(Gl)

‘y @I

CG31

(G1)

I BGCYGDKDRFF’GVR

TPCVGDKDSSPGVR

Gl)

(Cl) YlWTPR

GNFMYR

i

\I!

‘JW

‘“1 (4

1

rm I

(‘3 HHAFCFR \ LPGGVVFHYRPGSSR

(G3) KGTVACGIWVVEHAR

R

# R

Time (min)

i

FIG. 1. Reverse-phase HPLC of peptides in tryptic digests of calf aggrecan. Peptides present in PSEA pools I (a), II (b), III (c), and IV (d) were fractionated on C-8 as described (see Experimental Procedures and Ref. (7)) and the eluent was monitored at 220 nm. Typical chromatograms for calf cartilage aggrecan are shown. The peaks indicated corresponded to the peptides shown. Peak areas for both forms of the LPGG peptide (LPGGVVFHYR and LPGGVVFHYRPGSSR) and the (K)GTV peptide (GTVACGEPPVVEHAR and KGTVACGEPPVVRHAR) were combined for quantitation.

54

FLANNERY TABLE Marker

Peptide Content of Aggrecan Calf Articular Cartilage

from

Peptide content

(Azm peak area/ag protein)

Peptide

1

2

3

4

5

6

Gl

GIVF YPIH EGCY ARPN LPGG YSLT TPCV CYAG TVYL HHAF FQGH (K)GTV HVPT ITCT

4250 4733 4447 1538 3633 3978 2141 3799 6830 3551 1172 2579 1382 1127

2533 3565 2524 1586

2563 3225 2016 1431 1648 2856 2009 3686 3579 3062 657 2161 1043 767

2266 3204 2118 1120 1430 4497 1661 3337 3851 2550 585 1864 1059 644

3192 3443 3844 1306 3864 3660 1941 2650 8203 2975 951 ND 1098 973

3063 3130 3790 1247 3506 3530 1715 4443 6954 2520 925 2114 1020 895

G2

G3

It has been determined that both human and rat aggrecan are encoded by a single gene (15, 16) and that discrete structural domains are encoded by separate exons as follows: the Gl domain by exons 3-5, the interglobular domain (IGD) by exon 6, the G2 domain by exons 7 and 8, the keratan sulfate domain by exon 9, the CS domains by exon 10, and the G3 domain by exons 11-16. It has also been shown that the aggrecan gene can undergo alternate splicing of G3-encoding exons (15, 17); thus the extent to which all aggrecan molecules contain regions encoded by all exons has not been directly examined. In order to investigate the globular domain content of aggrecan we have developed a peptide isolation scheme which has identified high yield peptides encoded by exons 4, 5, 6, 7, 8, 12, 14, 15, and 16 (7). This method has been adapted in the present work to examine variations in the globular domain content of high density bovine aggrecan preparations.

I

Domain

IGD

ET AL.

3140 2102 ND 3384 3532 836 2322 1376 781

EXPERIMENTAL Note. Peptides are designated by their four N-terminal residues. Complete sequences are shown in Fig. 1. The data shown are for six separate digests of calf AlDl aggrecan. ND, not determined.

thermore, aggrecan from mature cartilage is composed of a heterogenous population which has been separated into discrete species by both electrophoretic (12,13) and density gradient methods (14). Differences in the core proteins of these two species have been detected in peptide patterns of trypsin digests seIjarated by thin-layer electrophoresis and chromatography (14).

TABLE Marker

Peptide

Content

PROCEDURES

Materials used for peptide isolation were as previously described (7). Sepharose CL-2B and Superose-6 were from Pharmacia/LKB. All other reagents were analytical grade from standard sources. Proteoglycan as total glycosaminoglycan (GAG) was determined by dimethylmethylene blue (18) using shark chondroitin sulfate as standard. Proteoglycan core protein was measured using the micro-bicinchoninic acid kit from Pierce, Inc., with bovine serum albumin as standard. Chromatography on Mono-Q and polysulfoethylaspartamide (PSEA) columns was performed using a Pharmacia/LKB FPLC system. Reversephase HPLC was performed using a Perkin-Elmer 410 pump and LC95 detector with Omega-2 data collection and peak integration. Peptide sequencing was performed on an Applied Biosystems 473A or 477A sequencer with on-line phenylthiohydantoin analysis. Amino acid analysis was done on an Applied Biosystems 420-A instrument with automated

of Aggrecan

II from

Calf Articular

Cartilage

Peptide content (A**,, peak area relative Domain

Peptide

1

2

Gl

GIVF YPIH EGCY ARPN LPGG YSLT TPCV CYAG TVYL HHAF FQGH (K)GTV HVPT ITCT

0.90 1.00 0.94 0.33 0.77 0.84 0.45 0.80 1.44 0.75 0.25 0.55 0.29 0.24

0.71 1.00 0.71 0.45 0.93 0.88 0.59 ND 0.95 0.99 0.24 0.65 0.39 0.22

IGD G2

G3

Note. Peptides are designated by their four N-terminal digests of calf AlDl aggrecan. Data have been normalized of the mean.

to YPIHTPR)

3

4

5

6

Mean f SD

0.44 0.51 0.89 0.62 1.14 1.11 0.95 0.20 0.67 0.32 0.24

0.71 1.00 0.66 0.35 0.45 1.40 0.52 1.04 1.20 0.80 0.18 0.58 0.33 0.20

1.00 1.12 0.38 1.12 1.06 0.56 0.77 2.38 0.86 0.28 ND 0.32 0.28

0.98 1.00 1.21 0.40 1.12 1.13 0.55 1.42 2.22 0.81 0.30 0.68 0.33 0.29

0.84 f 0.11 1.00 0.88 f 0.23 0.39 * 0.05 0.82 f 0.27 1.03 f 0.20 0.55 * 0.05 1.04 f 0.24 1.55 k 0.55 0.86 k 0.09 0.24 + 0.04 0.63 f 0.05 0.33 f 0.03 0.24 f 0.03

residues. Complete sequences are shown in Fig. 1. The data shown are for six separate to the content of the Gl peptide YPIHTPR. ND, not determined; SD, standard deviation

VARIABILITY

IN THE

G3 DOMAIN TABLE

CONTENT

OF BOVINE

55

AGGRECAN

III

Marker Peptide Content of Aggrecan from Steer Articular Cartilage Peptide content (Azzo peak area relative

to YPIHTPR)

Total AlDl Domain

Peptide

1

2

3

Gl

GIVF YPIH EGCY ARPN LPGG YSLT TPCV CYAG TVYL HHAF FQGH (K)GTV HVPT ITCT

0.91 1.00 0.95

0.76

0.77 1.00 0.80 0.19 ND

IGD G2

G3

0.34 1.14 0.84 0.38 1.46

1.36 0.87 0.24

0.39 0.18 0.20

1.00 0.62

0.39 0.66 0.75 0.41

ND 0.65 0.52 0.13 0.28 0.23 0.12

1.07 0.21 0.78

1.10 ND 0.11 ND 0.18 0.08

Mean t SD 0.81 xk 0.07* 1.00 0.79 + 0.14* 0.31 k 0.08** 0.90 i 0.24* 0.89 f 0.13* 0.34 + 0.09*** 1.12 k 0.34* 1.04 f 0.30* 0.70 3~0.18* 0.16 t 0.06** 0.33 k 0.067 0.20 f 0.04$ 0.13 + 0.05t

Band I

Band II

0.84

0.80

1.00

1.00

0.73 0.35 1.13 0.87 0.38 1.07 0.42

0.83 0.35 1.15 1.03 0.34

1.11 1.51

0.89

0.91

0.41 0.43 0.28

0.13 0.17

0.11

0.29

0.08

Note. Peptides are designated by their four N-terminal residues. Complete sequences are shown in Fig. 1. The data shown are for three separate digests of steer AlDl aggrecan and for bands I and II. Data have been normalized to the content of the Gl peptide YPIHTPR. ND, not determined; SD, standard deviation of the mean. The difference between the mean normalized value for each peptide for steer total AlDl relative to the equivalent value for calf AlDl aggrecan (bv Student’s t test) was rated not significant (*P > 0.05) or significant (**P < 0.05, ***P < 0.005, tP < 0.001, $P < 0.0005). -I -

gas-phase hydrolysis and microbore separation of derivatized amino acids. Isolation of aggrecanfrom articular cartilage. Aggrecan was prepared from freshly dissected, full depth articular cartilage from the metacarpalphalangeal joints of newborn calves and 6-year-old steers as previously described for cultured cartilage (7). Briefly, portions of cartilage (1 g wet weight) were cryostat sectioned (20 pm) and extracted at 4°C for 50 mM sodium acetate, 48 h with 15 vol (ml/g) of 4 M guanidine-HCl, pH 6.8 (with protease inhibitors). Extracts were finally prepared as isopycnic CsCl gradient fractions AlDl, AlD2, AlD3, and AlD4 (densities of 1.61, 1.50, 1.42, and 1.36 g/ml, respectively). Typically, Al fractions (density 1.61 g/ml) contained 93% of the total GAG recovered from the associative gradient. For calf extracts, AlDl and AlD2 fractions contained 93 and 4%, respectively, of the total GAG recovered from the dissociative gradient. For steer extracts, 82 and 13% of the total GAG recovered from the dissociative gradient was present in fractions AlDl and AlD2, respectively. For analysis, AlDl fractions were sequentially dialyzed against 1 M NaCl and water. Chondrocyte cultures. Chondrocytes were isolated from the metacarpal-phalangeal joints of newborn calves and cultured essentially as described (19). Cells were plated in 60-mm dishes at 7 X lo5 cells/cm’ in 4 ml Ham’s F12 containing 10% (v/v) fetal calf serum, 0.04 M NaHCO,, 0.02 M Hepes, 0.002 M glutamine and maintained for 16 h. Thereafter, medium alone or medium supplemented with 50 pg/ml ascorbic acid was changed daily and collected as three pools, pool A (Day 2-Day 6), pool B (Day ‘i-Day lo), and pool C (Day ll-Day 15). All medium collections were adjusted to pH 6.5 in the presence of 10 mM Mes, 5 mM EDTA, and 0.1 mM PMSF and then stored at -20°C. After 15 days of culture, cell layers were extracted for 24 h in 5 ml of 4 M guanidine, pH 6.5, containing 5 mM EDTA and 0.1 mM PMSF. Extracts were pooled and filtered on glass wool. Aggrecan from the cell layer and medium pools was finally isolated in an AlDl cesium chloride gradient fraction as described above for cartilage extracts. Preparation of aggrecan on cesium sulfate gradients and electrophoresk on horizontal agarose gels. Portions of steer AlDl (5 mg GAG/ml)

were incubated with 5% (w/w) hyaluronate for 16 h at 4°C. Aggregated monomers were isolated by rate zonal sedimentation on associative cesium sulfate density gradients essentially as described (20). Briefly, 10 ml linear gradients of 0.15 to 0.5 M CsBSOl were prepared on 1.2-ml cushions of 2 M CszSOI and 1 ml of sample layered on top. Following centrifugation in a Beckman SW 40Ti rotor at 35,000 rpm for 4 h at lO”C, gradients were fractionated and the bottom 3 ml (which typically contained 67% of the total GAG recovered from the gradient) was taken for dissociative isopycnic gradient centrifugation. Gradients were fractionated into four equal portions with 95% of the total recovered GAG present in the Dl fraction (density of 1.61 g/ml). For analysis, Dl fractions were sequentially dialyzed against 1 M NaCl and water. Aggrecan prepared in this way was then electrophoresed on horizontal agarose gels and bands I and II were separated as described (21). Aggrecan from each band was recovered at about 60% yield by a freeze-squeezing procedure. Excised strips were cut into small pieces, homogenized, and then freeze-thawed. Following centrifugation at 17,5OOg,the supernatant was filtered and concentrated using a Millipore Ultrafree MC 30,000 filter unit. Aggrecan was precipitated with ethanol and portions were taken for analytical electrophoresis to assess purity. Characterization of aggrecan samples. Portions of AlDl aggrecan (0.25 mg GAG) from calf and steer cartilage and from the cell layer and medium pools of calf chondrocytes were adjusted to 0.05 M sodium acetate, pH 6.8, and fractionated at 2 ml/h on columns (0.6 X 100 cm) of Sepharose CL-2B eluted with 0.5 M sodium acetate, pH 6.8, containing 0.1% (v/v) Triton X-100. Additionally, CS chains were prepared by alkaline p-elimination from portions (0.1 mg GAG) of the above preparations and steer bands I and II and fractionated at 30 ml/h on a column (1 X 30 cm) of Superose-6 eluted with 0.5 M sodium acetate, pH 6.8. All calf AlDl samples (cartilage extract and cell products) were found to have the same mean monomer size (KAv = 0.31 on CL-PB), mean CS chain size (KAv = 0.59 on Superose-6), and a similar ratio (mg/mg) of GAG/protein (12.5-15.8). Steer AlDl aggrecan had a smaller mean monomer size than calf (KAv = 0.39 on CL-2B) and all steer samples (cartilage extract, band I and band II) had the same CS chain size (KAv

56

FLANNERY

ET AL.

= 0.69 on Superose-6). The ratio (mg/mg) of GAG/protein was 7.7,8.9, and 8.0 for steer AlDl, band I, and band II, respectively. Isolation of aggrecan globular domain marker peptides. Tryptic peptides from the globular domains of aggrecan from calf and steer cartilage, and from calf chondrocyte cell layer and medium pools were isolated as described (7) except that sequencing grade bovine trypsin was used in all digestions. The peptide maps obtained here were different from those in Ref. (7) in that the G2 peptide TYGVR was no longer obtained in good yield and two new markers, TPCVGDKDSSPGVR (GZ) and ITCTDPATYKR (G3), were obtained. Peptide synthesis. Two peptides (YPIHTPR and HVPTIR) were synthesized by solid-phase polyamide methodology using fluorenylmethoxycarbonyl chemistry on an Applied Biosystems 430A peptide synthesizer. Assay of standard solutions was by amino acid analysis. Electron microscopy. Portions of AlDl aggrecan from calf cartilage and chondrocyte medium pool B (from cultures without ascorbate) were analyzed by electron microscopy following glycerol spraying/rotary shadowing. For each sample, the mean length and the ratio of G3 to double-globe (Gl-G2) were determined for 200 monomers. Electron microscopic procedures and methods of data evaluation were as previously described (22).

RESULTS AND DISCUSSION Quantitation of Globular Domains in Aggrecan Preparations To assess the content of Gl, G2, and G3 domains in different samples, marker peptides from each domain were isolated from tryptic digests by a modification (see Experimental Procedures for details) of an established method (7). The chromatograms obtained in the final step of this method and the identification of each peptide are given in Fig. 1. The yield of each peptide (Azzo area units per microgram aggrecan core protein) was then determined from the integrated AzzO peak areas obtained from these reverse-phase chromatograms. To determine the reproducibility of this method, six separate digests of calf AlDl aggrecan were analyzed, and the data are shown in Table I with each peptide indicated by its four N-terminal residues. Variability in these values for a given peptide between digests appeared to be largely a function of variable recovery of peptides from all domains. Thus, high values for Gl peptides (see digest 1) were accompanied by high values for both G2 and G3 peptides in the same digest and low values for Gl peptides (digest 4) were accompanied by low values throughout. This would suggest that relative losses during isolation were similar for peptides from all domains. To account for this variability, and so allow a direct comparison of data obtained in different digests, peak area values were normalized to the value obtained in the identical digest for the Gl domain peptide YPIHTPR. This peptide was chosen as an internal standard since when synthetic YPIHTPR was taken alone through the complete peptide isolation method, recovery was reproducibly high at about 95%. These normalized values for the six calf AlDl samples are shown in Table II. It should be noted that these values do not indicate molar content since the absorption value for each peptide is a function

FIG. 2. Analytical horizontal agarose gel electrophoresis of steer cartilage aggrecan. Portions of steer cartilage aggrecan bands I and II (indicated) which had been isolated from preparative agarose gels (see Experimental Procedures) were electrophoresed with portions of unfractionated steer cartilage aggrecan (T) to assess purity.

both of peptide structure and yield. Derivation of the mean + SD of these normalized values for each peptide (righthand column, Table II) shows the reproducibility of quantitation of each peptide by this method. Domain Content of Steer Cartilage Al Dl Aggrecan and Subpopulations from Preparative Electrophoresis To assess the effect of aging on aggrecan core protein structure, an AlDl preparation from 6-year-old steers was analyzed next. These content data were also normalized to the content of YPIHTPR in the identical digests. A comparison of the mean + SD values for the peptides from calf AlDl (Table II) and steer AlDl (Table III) shows that for seven of the nine peptides isolated from the Gl, IGD, and G2 domains there is apparently no significant change in content with age. This is based on a statistical analysis (Student’s t test) of the significance of the difference between the mean value for each peptide (P values are given in Table III). Since the data are normalized to a Gl peptide (YPIHTPR) this implies that the regions of the GI, IGD, and G2 domains “marked” by these peptides are coordinately expressed and that this applies to aggrecan from both immature and mature cartilage. This result is therefore in agreement with the ac-

VARIABILITY

IN THE

G3 DOMAIN

CONTENT

TABLE

OF BOVINE

57

AGGRECAN

IV

Marker Peptide Content of Aggrecan from Calf Chondrocytes Cultured in the Absence (-) or Presence (+) of Ascorbic Acid Peptide content (AzzOpeak area relative

to YPIHTPR)

- ascorbic acid

+ ascorbic acid

Domain

Peptide

Cell layer

Medium A

Medium B

Medium C

Cell layer

Medium A

Medium B

Gl

GIVF YPIH EGCY ARPN LPGG YSLT TPCV CYAG TVYL HHAF FQGH (K)GTV HVPT ITCT

0.64

0.63

0.80

1.00

1.00

0.87 0.28 0.72 0.73 0.48

0.73 0.25 0.63 0.86 0.43 0.76

0.93 0.32 0.82 0.71 0.75 0.99

0.64 1.00 0.97 0.33 0.82 0.68 0.69

0.64 1.00 0.82 0.26 0.76 0.61 0.52 0.76

0.76 1.00 1.25 0.33 1.05 1.08 0.72 0.56

0.82

1.00

1.15 ND

1.39 0.59

0.58 0.84 0.36 0.60

0.40 0.62 0.31 0.32

IGD G2

G3

0.80 0.93

1.13

1.29

0.51 0.37 0.66 0.33

0.44 0.54 0.72 0.43

0.21

0.41

0.89 0.52 1.57 0.54 0.45

ND

1.33

1.14

0.78 0.54 1.33 0.50 0.69

0.44 1.02 1.40 0.61 0.48

Note. Peptides are designated by their four N-terminal residues. Complete sequences are shown in Fig. 1. Data have been normalized content of the Gl peptide YPIHTPR. ND, not determined.

cepted model of aggrecan supported by glycerol spraying/ rotary shadowing data (11) in which all the Gl-bearing molecules also carry the G2 domain. In contrast, the same comparison of calf and steer showed a statistically significant age-dependent decrease in the content of all four peptides isolated from the G3 domain. This clearly indicates that the percentage of GlG2 bearing aggrecan molecules which carry the G3 domain is reduced with age in bovine cartilage. Further, the values

1.00 0.79 0.30 0.78 0.84 0.50 0.50

to the

obtained (Table III relative to Table II) suggest that this reduction in G3 content may be of the order of 35-45s. In addition, the two separable subpopulations of mature bovine aggrecan were analyzed following isolation from flat-bed agarose gels. Separation of these subpopulations (Band I and Band II) is shown in Fig. 2. The normalized data for Bands I and II (Table III, right-hand columns) indicated that the two subpopulations were very similar to each other and to the starting material in their contents

iyj wltbout

ascorbk

acid

I

I

with aseorbk

add

I

VJ WI

FIG. 3. Histogram showing the molar yield of the Gl peptide YPIHTPR and the G3 peptide HVPTIR. The molar yield of HVPTIR and YPIHTPR per mole of aggrecan core protein was determined as described under Experimental Procedures. The histograms for calf and steer cartilage AlDl aggrecan show the mean values for six and three separate analyses, respectively, and the error bars show the standard deviation of the mean.

58

FLANNERY

FIG. 4. Histogram showing the molar ratio of the G3 peptide HVPTIR and the Gl peptide YPIHTPR. The molar ratio data shown are derived directly from the molar yield data for the two peptides as shown in Fig. 3.

of Gl/IGD/GB domain protein. Thus for eight of the nine peptides isolated from the N-terminal domains (with TVYL being the exception) there was good agreement in the content values for the two populations. In contrast, all four of the G3 peptides were clearly depleted in Band II relative to Band I aggrecan, suggesting that the faster migrating component is reduced in G3 domain. Indeed, since the G3 content data for the starting material were, for all peptides, intermediate between the Band I and Band II data, it appears that the electrophoresis achieves a separation of G3-rich molecules (Band I) and G3-poor molecules (Band II). In this regard, electrophoretic separation of human aggrecan into Band I and Band II species and immunodetection of the EGF-like sequence of the G3 domain (17) have also demonstrated the presence of G3-bearing molecules in both bands (23). The presence of G3 in both populations therefore indicates that the basis for their separation by electrophoresis includes variation in both core protein length and glycosylation patterns. Domain Content of Aggrecan Secreted from Chondrocyte Cultures Previous rotary shadowing (11) and peptide quantitation (7) of extracted bovine aggrecan has indicated that only 30-50% of tissue molecules carry the G3 domain, and chemical analysis of molecules of different size and buoyant density (24) has suggested that this low content is a result of extracellular proteolysis at the C-terminal of aggrecan, resulting in the generation of G3-deficient species. The extent to which G3 deficiency also results from the usage of alternate mRNA transcripts for aggrecan core or the “fast” removal of G3 domain during biosynthesis or secretion, however, remains unknown. In order to investigate this we have determined the globular domain content of aggrecan secreted by calf

ET AL.

chondrocytes in high-density cultures (see Experimental Procedures for details). The rationale for this approach was that cell culture, unlike explant culture, provides a means of isolating a population of newly synthesized molecules which have not been exposed to extended periods of processing in the cartilage matrix. Thus, in explant cultures proteolytic removal of the Gl domain (7,8) results in the release of aggrecan into the medium, whereas aggrecan from the medium of chondrocyte cultures retains the ability to bind to hyaluronate (25). Newborn calf chondrocytes were maintained with or without ascorbic acid, and the aggrecan was purified from medium and cell layer compartments separately. Ascorbate was varied in order to determine whether the domain content was influenced by the extent of aggrecan deposition in the cell layer matrix, which is markedly promoted by the inclusion of ascorbate. Thus, chondrocytes cultured in the absence of ascorbic acid synthesized about 60 mg of aggrecan in 15 days, 9% of which was present in the cell layer. In contrast, chondrocytes cultured in the presence of ascorbic acid synthesized a similar amount of aggrecan, 46% of which was present in the cell layer. A comparison of the normalized data for these samples (Table IV) to the mean + SD for calf AlDl (Table II) suggested that the Gl/IGD/GS content of chondrocyte products was not markedly different to the calf AlDl preparation and that there was no clear effect of ascorbate inclusion nor of culture distribution (cell layer or medium). In addition, the apparent G3 content of cell layer aggrecan, deposited in the presence or absence of ascorbate (Table IV), was also similar to the mean + SD data for calf AlDl (Table II). On the other hand, the apparent G3 content of medium aggrecan in both cultures was consistently, and in some cases markedly, increased over the calf AlDl preparation. For example the G3 data for Medium B (plus ascorbate) showed an apparent increase of 4.25-fold for FQGH, 2.22fold for KGTV, 1.85fold for HVPT, and 2.O-fold for ITCT. These results therefore show that calf chondrocytes, under these conditions of culture, secrete aggrecan with a much higher G3 content than the aggrecan present in the tissue from which the cells were isolated. In addition, the markedly higher G3 content of aggrecan in the medium relative to that in the cell layer suggests that retention in the cell layer may be accompanied by rapid proteolytic removal of G3 domain and that molecules which diffuse into the medium “escape” from this proteolytic processing. Results supporting the observation that newly synthesized aggrecan molecules may undergo such rapid extracellular processing have also been described by Thonar et al. (26), who detected “CS-poor” monomers extracellularly in bovine chondrocyte cultures within 15 min after synthesis.

VARIABILITY

IN THE

G3 DOMAIN

Molar Content of G3 and Gl Domains in Different Samples The content data presented above are comparative only and provide no information on the molar abundance of these domains in different samples. To assess molar content, the Gl peptide YPIHTPR and the G3 peptide HVPTIR were isolated from five separate trypsin digests of calf cartilage aggrecan by collection of 0.2-min fractions during the final reverse-phase run. Peptides were dried for amino acid compositional analysis and an extinction coefficient (Azzo peak area/pm01 peptide) was determined for each peptide based on arginine content. Values of 720 + 43 (mean ? SD) for HVPTIR and 1180 f. 136 (mean + SD) for YPIHTPR were obtained. The validity of these extinction coefficients was supported by C-8 chromatography of standard solutions of the synthetic peptides and these coefficients were used to calculate picomole yields of these two peptides directly from reverse-phase chromatogram peaks. The yield of YPIHTPR and HVPTIR (moles peptide per mole core protein digested) for each of the aggrecan preparations described is shown in Fig. 3. It should be

z

CONTENT

OF BOVINE

AGGRECAN

59

noted that the values shown here were based on a core protein molar estimate calculated from a protein assay of whole proteoglycan (assayed relative to serum albumin as standard) and an assumption that the mean molecular mass of the aggrecan core protein was 220 kDa. Despite these approximations the values obtained, ranging from 0.55 to 1.05 for YPIHTPR (Gl) and 0.15 to 0.92 for HVPTIR (G3), were in reasonable agreement with other available data in this area (7, 11). To simplify interpretation of this data, these molar values were next used to calculate the relative molar content of G3 domain and Gl domain in different samples (Fig. 4). This then represents the best estimate from peptide quantitation of the percentage of Gl-bearing aggrecan monomers in these samples which also carry the G3 domain. It is clear that the chondrocyte medium aggrecan has the highest apparent G3 content which is in the range of 70-100% for the first 10 days of culture (medium A and B). Calf AlDl and cell layer aggrecan appeared to be of similar structure with about 55% G3 content. The lowest content was clearly in the steer aggrecan which was as low as 18% in the Band II sample.

0.2(b)

Lwth (nm) FIG. 5. Histogram of monomer sizes determined by glycerol spraying/rotary shadowing of calf cartilage (a) and chondrocyte medium (b) aggrecan. Measurements were made on approximately 200 monomers from whole aggrecan populations from calf cartilage and from medium pool B (from cultures without ascorbate). N/N,, denotes the fraction of molecules with a monomer length of +lO nm. Average values f standard deviation of the mean are shown.

60

FLANNERY

Electron Microscopic Analysis Glycerol spraying/rotary shadowing of aggrecan samples has previously been used to provide estimates of domain abundance (11). In order to assess the validity of the present peptide quantitation method for determining G3 domain content (see Fig. 4), the identical samples of AlDl aggrecan from calf cartilage and chondrocyte medium pool B (from cultures without ascorbate) were visualized by glycerol spraying/rotary shadowing. For each sample, the ratio of G3 to double-globe (Gl-G2) and the mean length were determined for 200 monomers. For chondrocyte medium aggrecan the G3:double-globe ratio was found to be 0.59:1, which was significantly higher than the equivalent value for calf cartilage aggrecan (0.40:1). Consistent with this, the mean particle length of medium aggrecan (300 + 75 nm; Fig. 5b) was markedly higher than the equivalent value for calf cartilage aggrecan (192 + 105 nm; Fig. 5a). The reduction in G3 content and the marked size heterogeneity of the cartilage aggrecan (Fig. 5a) are consistent with extracellular proteolysis in the tissue, leading to variation in the length of the CS domain (24). Domain analysis of aggrecan by microscopy therefore supported the data obtained from the peptide isolation methods (see Fig. 4). In conclusion, the results provided suggest that G3 removal may occur in both “fast” and “slow” phases during the extracellular processing of aggrecan. The initial or fast phase may occur in the pericellular space and result in an approximate 50% loss of G3, much as is seen for aggrecan in immature cartilages. This may then be followed by a slow phase which occurs during cartilage aging, and which results in a mature cartilage matrix in which only 20-30% of the aggrecan molecules carry the G3 domain. The results do not, however, exclude the possibility that G3 removal also occurs intracellularly or that the deficiency in G3 domain is a result of altered mRNA transcripts. In the cell culture system described here the G3 content of secreted molecules appears to be determined by partitioning of the product between the cell layer and the medium; thus matrix-associated molecules appear to “lose” G3, whereas medium molecules appear to retain the G3 domain. This would seem to support the idea obtained from explant culture experiments (7) that the G3 domain does not provide a high affinity of binding for aggrecan to cartilage matrix. On the other hand the cell culture system may provide a means of examining fast G3 turnover in the cell layer, which may in turn indicate functions for the G3 domain. In addition, isolation and characterization of the putative C-terminal fragments and determination of the precise points of core protein cleavage using previously described methods (8, 9, 27) will be required to provide definitive information on the mechanism(s) of G3 removal.

ET AL. and Dr. Joachim Sasse (peptide syntheses). This work was supported by the Shriners Hospitals of North America and by Grant AR 38580 from the National Institutes of Health.

REFERENCES 1. Halberg, D. F., Proulx, G., Doege, K., Yamada, Y., and Drickamer, K. (1988) J. Biol. Chem. 263,9486-9490. 2. Hardingham, T., and Bayliss, M. (1990) Semin. Arthritis Rheum. 20, 12-33. 3. Thonar, E. J.-M. A., and Sweet, M. B. E. (1981) Arch. Biochem. Biophys. 208, 535-547. 4. Inerot, S., and Heineganl, D. (1983) Collagen Relat. Res. 3, 245-262. Rheum. 5. Bayliss, M. T., and Ali, S. Y. (1981) Semin. AFthFitiS ll(Supp1. l), 20-21. 6. Roughley, P., Santer, V., and White, R. (1981) Semin. ArthFitiS Rheum. ll(Supp1. l), 16-17. 7. Sandy, J. D., Boynton, R. E., and Flannery, C. R. (1991) J. Biol. Chem. 266,8198-8205. 8. Sandy, J. D., Neame, P. J., Boynton, R. E., and Flannery, C. R. (1991) J. Biol. Chem. 266,8683-8685. 9. Sandy, J. D., Flannery, C. R., Neame, P. J., and Lohmander, L. S. (1992) J. Clin. Inuest. 89, 1512-1516. Cartilage and Osteoarthritis 10 Sandy, J. D. (1992) in Articular (Kuettner, K., Schleyerbach, R., and Hascall, V. C., Eds.), pp. 2133, Raven Press, New York. M., 11. Paulsson, M., Morgelin, M., Wiedemann, H., Beardmore-Gray, Dunham, D., Hardingham, T., Heinegird, D., Timpl, R., and Engel, J. (1987) Biochem. J. 245, 763-772. 12. McDevitt, C., and Muir, H. (1971) Anal. Biochem. 44, 612-622. 13. Stanescu, V., Maroteaux, P., and Sobczak, E. (1977) Biochem. J. 163,103-109. 14. Heinegbrd, D., Wieslander, J., Sheehan, J., Paulsson, M., and Sommarin, Y. (1985) Biochem. J. 225, 95-106. 15. Doege, K., Sasaki, M., Kimura, T., and Yamada, Y. (1991) J. Biol. Chem. 266,894-902. 16. Doege, K., Sasaki, M., Horigan, E., Hassell, J. R., and Yamada, Y. (1987) J. Biol. Chem. 262, 17,757-17,767. 17. Baldwin, C. T., Reginato, A. M., and Prockop, D. J. (1989) J. Biol. Chem. 264, 15,747-15,750. 18. Farndale, R. W., Sayers, C. A., and Barret, A. J. (1982) Connect. Tissue. Res. 9, 247-248. 19. Plaas, A. H. K., Neame, P. J., Nivens, C. ,M., Reiss, L. (1990) J. Biol. Chem. 265, 20,634-20,640. 20. Kimata, K., Kimura, J., Thonar, E. J.-M. A., Barrach, H.-J., Rennard, S. I., and Hascall, V. (1982) J. Biol. Chem. 257, 3819-3826. 21. Stanescu, V. (1990) in Methods in Cartilage Research (Maroudas, A., and Kuettner, K., Eds.), pp. 44-46, Academic Press, New York. H. (1987) in Methods in Enzymology 22. Engel, J., and Furthmayr, (Cunningham, L. W., Ed.), Vol. 145, pp. l-47, Academic Press, San Diego. 23. Stanescu, V., Chaminade, sue Res. 26,283-293.

F., and Do Pham, T. (1991) Connect.

7%

C., Margolis, R., Pal, S., and 24. Rosenberg, L., Wolfenstein-Todel, Strider, W. (1987) J. Biol. Chem. 251, 6439-6444. 25. Sandy, J. D., and Plaas, A. H. K. (1989) Arch. Biochem. Biophys. 271,300-314.

ACKNOWLEDGMENTS

B., Wang, C., and 26. Thonar, E. J.-M. A., Bjornsson, S., Matijevitch, Kuettner, K. E. (1988) Trans. Orthop. Res. Sot. 13, 47.

We acknowledge the expert contributions of Dr. Peter Neame and Ms. Carmen Young (peptide sequencing and compositional analyses)

27. Flannery, C. R., Lark, M. W., and Sandy, J. D. (1992) J. Bid. Chem. 267, 1008-1014.