The intracellular localisation of proteoglycans and their accumulation in chondrocytes treated with monensin

Biochimica et Biophysica Acta

844 (1985) 247-255

247

Elsevier BBA 11408

T h e intracellular localisation of p r o t e o g l y c a n s a n d t h e i r a c c u m u l a t i o n in c h o n d r o c y t e s t r e a t e d with m o n e n s i n L y n d a J. B u r d i t t a, A n t h o n y R a t c l i f f e a, P a t r i c i a R. F r y e r b a n d T i m E. H a r d i n g h a m a,, a Kennedy Institute, Bute Gardens, Hammersmith, London, 14167D W, and b Electron Microscopy, Clinical Research Centre, Harrow, Middlesex, H A l 3UJ (U.K.)

(ReceivedJune 12th, 1984)

Keywords: Proteoglycan;Monensin;(Chondrocyte) Pig laryngeal chondrocytes incubated in the presence of monensin showed inhibition of [3SS]sulphate incorporation and decreased secretion of proteoglycan into the culture medium, but no large decrease in protein synthesis. This lead to the intracellular accumulation of proteoglycan protein core, which was detected in immunoprecipitates of cell extracts. Using the same antiserum protein core was Iocalised by electron microscopy with protein A-coated gold. In control chondrocytes, it was detected only in elements of the Golgi and in secretory vesicles, but following monensin treatment labelling was more intense in the Golgi and extended into the distended cisternae of the rough endoplasmic reticulum. The results suggest that monensin blocks proteoglycan protein core translocation between different elements of the Golgi and that this occurs prior to the major site of chondroitin sulphate synthesis on proteoglycan.

Introduction

Proteoglycan production by chondrocytes requires the coordinated synthesis of protein and carbohydrate components [1,2]. Experiments on the timing of these intracellular processes with rat chondrosarcoma cells showed that there was a considerably delay (approx. 70 min) between protein core synthesis and the addition of chondroitin sulphate chains, but that this was followed by rapid secretion [3,4]. There was thus suggested to be a pool of intracellular protein core that was without chondroitin sulphate chains and this was identified by immunoprecipitation [5]. Intracellular protein core was also detected in chick chondrocytes and it was shown to accumulate in * To whomcorrespondenceshouldbe addressed. Abbreviation: Hepes, 4-(2-hydroxyethyl)-l-piperazineethanesulphonicacid. 0167-4889/85/$03.30 © 1985 ElsevierSciencePublishers B.V.

the presence of monensin [6], which interferes with the intracellular translocation of many secretory proteins [7,8]. Monensin also potently inhibited [3SS]sulphate incorporation into proteoglycans, although it did not greatly inhibit protein synthesis [9-11]. In the present study antibodies to the proteoglycan binding region were used to localise proteoglycans within isolated pig laryngeal chondrocytes and to investigate the change in distribution in cells treated with monensin. Experimental Materials

Materials for cell culture and collagenase (CLII) (EC 3.4.24.3) were obtained from Flow Laboratories (Irvine, U.K.) with the exception of penicillin/streptomycin and glutamine, which were obtained from Gibco Biocult (Paisley, U.K.). Chondroitinase ABC was obtained from Wor-

248 thington (Millipore U.K.), London NW10. Hyaluronidase (bovine testicular) (Type l-S) (EC 3.2.1.35), 6-aminohexanoic acid, phenylmethanesulphonyl fluoride, disodium EDTA, Hepes, protein A-Sepharose CL-4B, sodium deoxycholate, bovine serum albumin (fraction V), insulin (bovine pancreas) and monensin were obtained from Sigma (Poole, Dorset, U.K.). Protein A, Sepharose and Sephadex gels were from Pharmacia (Uppsala, Sweden). T E M E D (N,N,N',N'-tetramethyethylene diamine) was from Kich-Light (Colnbrook, Bucks, U.K.). Acrylamide, N,N'-methylene bisacrylamide (ultrapure) and Nonidet P40 were from Bethesda Research Laboratories (Cambridge, U.K.). Na 1251 (IMS30), [ 35S]sulphate (carrier-free), llaC]methylated protein mixture (M r 14 300-200 000) and L-[3- 3H]serine were obtained from Amersham International (Amersham, U.K.). Lumagel was from Lumac Systems (Basel, Switzerland). Zwittergent 3-14 ( N - t e t r a d e c y l - N , N dimethyl-3-ammonio-l-propanesulphonate) was obtained from Calbiochem-Behring (Bishop's Stortford, Herts., U.K.). Staphylococcus aureus (Cowan 1) was obtained from Miles Scientific (Slough, U.K.) and all reagents for electron microscopy including Lowicryl K4M resin were obtained from Agar Aids, Stansted, Essex, U.K. All other reagents were analytical grade, except guanidine hydrochloride which was purified with activated charcoal and Celite (Hopkin & Williams, Essex, U.K.)

Methods Pig laryngeal chondrocyte cultures. The cartilage was dissected from fresh pig larynges and adhering tissue and perichondrium were removed. It was washed briefly in absolute ethanol and then in Medium A (Hank's balanced salt solution containing 100 U / m l penicillin, 100 ~ g / m l streptomycin, 200 # g / m l gentamycin and 2 # g / m l fungizone), before slicing into small pieces (1 × 1 ram) and suspending in medium A containing 0.05% hyaluronidase (bovine testicular type 1-S) for incubation in a shaking water bath at 37°C for 30 min. After washing twice with medium A, the slices were resuspended in medium B, (Hams F12 containing 20 mM Hepes, 100 U / m l penicillin, 100 /~g/ml streptomycin, 200 ~ g / m l gentamycin and 2 /~g/ml fungizone) plus 2% foetal calf serum, 0.016

U / m l insulin and 0.25% collagenase and incubated for a further 2 h at 37°C. The digest was then diluted 2.5-times with medium B plus 10% foetal calf serum and placed in 100 mm plastic tissue culture dishes for 16 h, and incubated in 37°C in a humidified atmosphere of 5% CO2/95% air. The released cells were recovered by centrifugation for 10 min at 250 × g and washed twice with medium C (Hams F12 containing 20 mM Hepes, 50 U / m l penicillin, 50 ~ g / m l streptomycin and 100 # g / m l gentamycin) supplemented with 10% foetal calf serum. After resuspending in medium C plus 10% foetal calf serum, cells were filtered through a 20 ktm Nitex screen followed by one further wash in the same medium. The cells were then resuspended as before and plated at a density of 2- 10 6 cells per 35 mm dish in medium C containing 10% foetal calf serum and maintained in a humidified atmosphere of 5 % C O y 95 % air at 37°C. After 24 h, the medium was changed to medium C containing 2% foetal calf serum and 0.016 U / m l insulin, and the cells were maintained for a further 24 h before monensin treatment. Treatment of chondrocytes with monensin. Under the culture conditions used, the cells were only loosely attached to the tissue culture dishes. In order to treat the cells with monensin, they were removed from the dishes by gentle agitation and placed in sterile centrifuge tubes. The cells were recovered by centrifugation and the experimental incubations carried out in 1 ml of Medium C, supplemented with 0.016 U / m l insulin, and containing the appropriate monensin concentration. The effect of monensin on protein synthesis and proteoglycan secretion was determined by incubating the cells with or without monensin for 120 min followed by a further 120 min with the addition of 40/~Ci [3H]serine and 20/~Ci [35S]sulphate per ml of culture medium. At the end of the incubation period, the cells were recovered by centrifugation, the medium was removed, and an equal volume of 8 M guanidine hydrochloride/0.1 M sodium acetate (pH 6.8) was added to it. The cell pellet was extracted by the addition of 1 ml of 4 M guanidine hydrochloride, 0.05 M sodium acetate (pH 6.8) containing 100 mM 6-aminohexanoic acid, 5 mM benzamidine hydrochloride, 1 mM phenylmethanesulphonyl fluoride. 10 mM Na 2EDTA and 1% (w/v) Zwittergent 3-14. The total

249

incorporation of [3H]serine and [35S]sulphate into macromolecules was determined by chromatographing up to 2 ml of the medium or cell extract on columns (0.9 x 25 cm) of Sephadex G-50 in 4 M guanidine hydrochloride, 0.05 M sodium acetate (pH 6.8), to separate unincorporated isotope [4]. Fractions of 1 ml were collected and the distribution of radioactivity was determined by mixing samples with Lumagel and counting in a Searle mark III liquid scintillation spectrophotometer with automatic quench correction. The experimental cultures used for immunoprecipitation were treated with monensin as described and labelled with 40 #Ci [3H]serine per ml of medium. At the end of the incubation period the cells were recovered by centrifugation and the pellet extracted by the addition of 1 ml of extraction buffer (0.15 M NaC1 buffered with 0.01 M sodium phosphate (pH 7.3) containing 0.1% Nonidet P40 and 0.5% sodium deoxycholate). Extracts were stored at -20°C. In experiments where the culture media were analysed for binding region by radioimmunoassay or the cells were used for electron microscopic immunolocalisation, the cultures were incubated for 120 min with or without monensin, the medium was then removed and a second aliquot of the same medium applied for a further 120 min incubation. The effect of monensin on the secretion of binding region was assessed by analysis of the second aliquot of medium. Proteoglycan antiserum. The antiserum used was that described by Ratcliffe and Hardingham [12]. It was raised in rabbits to the hyaluronate binding region of aggregated proteoglycan from pig laryngeal cartilage. The antiserum was shown to be specific for the binding region and did not bind any other cartilage components. Immunoprecipitation of proteoglycan core protein from cell extracts. For immunoprecipitation of proteoglycan core protein, 200 /~1 of the cell extract was incubated with 1 mg bovine serum albumin and 10/~1 of control rabbit serum at 22°C for 90 min. Protein A-Sepharose CL-4B (15 /~1 packed volume) was then added and incubation continued for a further 30 min. The protein ASepharose was removed by centrifugation at 12000 X g for 5 min at 22°C. The supernatant was treated with 10 /tl of anti(binding region) anti-

serum [12] for 90 min at 22°C. A further 15/~1 of protein A-Sepharose CL-4B was added and incubation continued for 30 min at 22°C. Protein A-Sepharose CL-4B was recovered by centrifugation and washed extensively with extraction buffer, extraction buffer containing 0.5 M NaC1 and lastly phosphate-buffered saline alone to remove the detergents from the pellet. Finally, the pellets were heated to 80°C for 5 min in 100/~1 of 50 mM Tris (pH 6.7) containing 20% (v/v) glycerol, 1% (v/v) 2-mercaptoethanol and 1% (w/v) sodium dodecyl sulphate. The protein A-Sepharose was removed by centrifugation as before and the supernatants retained for electrophoretic analysis. Electrophoresis was carried out in vertical slab gels (13.0 cm x 15.0 cm × 10 mm) in 6% (w/v) acrylamide, 2.5% (w/w) bisacrylamide (with respect to the acrylamide concentration) in 0.375 M Tris (pH 8.9), 0.1% sodium dodecyl sulphate with a 1.5 cm 3% (w/v) acrylamide stacking gel in 62.5 mM Tris (pH 6.7), 0.1% sodium dodecyl sulphate. Electrophoresis was at 10 mA for 1 h to load the samples followed by 30 mA for approx. 4 h until the tracking dye (bromophenol blue) was 1 cm from the end of the gel. The gels were fixed in 20% (v/v) acetic acid, 20 % (v/v) isopropanol for 16 h and fluorographed according to the method of Skinner and Griswold [13], using glacial acetic acid as the solvent for 2,5-diphenyloxazole. Gels were dried under vacuum and exposed to preflashed Kodak XR/5 X-ray film at - 7 0 ° C for periods of 2-14 days. Radioimmunoassay procedure. The tissue culture medium was removed from the cells and stored at - 2 0 ° C until analysed. In order to dissociate any proteoglycan aggregates which might be present, the media were diluted in the assay buffer and incubated at 80°C for 15 rain in the presence of 0.025% sodium dodecyl sulphate (SDS), and the radioimmunoassays were carried out as described previously [12]. Electron microscopic immunolocalisation procedure. The electron microscopic studies were performed using a post-embedding labelling technique [14] with protein A-coated gold particles as the immunolabel. The cells were fixed in 1% glutaraldehyde in 0.1 M phosphate buffer (pH 7.2) and embedded in Lowicryl K4M resin by a modification [15] of the method of Roth et al. [16]. The

250 immunolocalisation procedure was carried out on ultrathin sections mounted on parlodion-coated nickel grids, and all reagents were prepared in phosphate-buffered saline containing 1% bovine serum albumin. After digestion with chondroitinase A B C (0.5 U / m l ) for 45 min, at 37°C, the sections were incubated for 2 h with a dilution of the anti-proteoglycan binding region antiserum, washed and incubated with a solution of protein A-gold (12 nm diameter) for 1 h. The sections were washed and lightly stained with uranyl acetate and lead citrate, and viewed in a Philips 300 electron microscope.

50

4

40

~2 O

30

u,J

~

20

~ 0

to

Results The effect of monensin on synthesis and sulphation of macromolecules by chondrocytes The incorporation of [3H]serine and [35S]sulphate into macromolecules was used to determine total protein synthesis and sulphation respectively (Fig. 1). Monensin produced a concentration-dependent inhibition of [35S]sulphate incorporation, but much less effect was observed

!

0

0.01

0.1

MONENSIN

10

( , M

1

)

Fig. 2. Effect of monensin on the secretion of [35S]sulphateand [3H]serine-labelled macromolecules by chondrocytes. Duplicate cultures were incubated as in Fig. 1 and the proportion of [35S]sulphate- (e) and [3H]serine- (i) labelled macromolecules secreted into the culture medium was determined as described in Methods.

on [ 3 H]serine incorporation into protein. Thus with 1 . 1 0 -6 M monensin, protein synthesis by the chondrocytes was inhibited by only 20%, whereas

100

Z 01 <

I

O u

80

1 O0

-

~

u.J

Z ~

N

U

Z u --

~,~

4o

5o

0 U <

10 ~

~

~

2O

zg

I 0

II

I

I

I

0 O!

0 .I

1.0

11 0

0

(V M )

Fig. 1. E f f e c t o f monensin on the total incorporation o f [3SS]sulphate and [3H]serine into macromolecules by chondrocytes. Duplicate cultures were incubated with different concentrations of monensin for 4 h and [35S]sulphate (0) (20 / x C i / m l ) and [3H]serine (1) (40 /~Ci/ml) were added for the last 2 h. T h e radioactive incorporation into macromolecules was determined as described in Methods.

zo

u uJ m

MONENSIN

4o

0

I

0

11

I

0.01

I

0.1

MONEI"ISIN

I

|

1.0

10

(p M )

Fig. 3. E f f e c t o f monensin on the secretion of binding region by chondrocytes. Duphcate cultures were incubated with different concentrations of monensin. The culture medium was changed after 2 h, and the content of proteoglycan binding region was determined in the medium by radioimmunoassay after a further 2 h incubation.

251 sulphation was decreased by 75%. Only at 1 • 10 -5 M monensin (the highest concentration used) was protein synthesis inhibited to a large extent (40%), and at this concentration sulphate incorporation was 90% inhibited. The effect of monensin on the secretion of newly synthesised macromolecules by chondrocytes was determined by measuring the distribution of [35S]sulphate- and [3H]serine-labelled macromolecules in the medium and the cell extracts (Fig. 2). Although at 1 . 10 7 M monensin [35S]sulphate incorporation was decreased by 66% compared with control values (Fig. 1), the proportion secreted was only 20% less than in controls (Fig. 2). This confirmed previous reports that chondroitin sulphate synthesis and sulphation were more sensitive to inhibition by monensin than the subsequent steps leading to secretion [9,10]. An inhibition radioimmunoassay using the antiserum to the hyaluronate binding region was used to determine the total amount of aggregating proteoglycan secreted into the medium over the last 2 h of a 4 h incubation with monensin (Fig. 3). This analysis was independent of radioactive-labelling techniques and also showed the secretion of aggregating proteoglycan to be inhibited by monensin in a dose-dependent manner. Inhibition was almost complete (93%) in the presence of 1 • 10 -5 monensin.

(2) lntracellular localisation of accumulated proteoglycan core protein Analysis of [3H]serine-labelled cell extracts and their immunoprecipitates on SDS-polyacrylamide gel electrophoresis (Fig. 4a) provided direct evidence for the intracellular accumulation of proteoglycan core protein, with a molecular weight greater than 200 000, in cells treated with monensin. Maximum accumulation was found at 1 " 1 0 - 6 M monensin, with slightly less at 1 . 1 0 -s M. This may have resulted from significant inhibition of total protein synthesis reducing the rate of accumulation at the higher monensin concentration (Fig. 1). Analysis of dual labelled ([3H]serine and [35S]sulphate) cell extracts by SDS-polyacrylamide gel electrophoresis (results not shown) showed that the protein core band did not label with [35S]sulphate in control cells, which was in agreement with the results of Kimura et al. [5]. The

immunoprecipitated proteoglycan protein core band was thus devoid of chondroitin sulphate chains and in monensin-treated cells it remained unlabelled by [35S]sulphate. The protein core accumulated within these cells was thus free of chondroitin sulphate chains. Using the anti-binding region serum and the immunoelectron microscopic localisation technique described by Ratcliffe et al. [14]. The intracellular distribution of proteoglycan core protein was investigated in normal and monensintreated cells. In control cells, the labelling of binding region was not abundant, but was localised over the Golgi region and in membrane-bound secretory-type vesicles (Fig. 4b). Other compartments of the cell were not labelled and this included the cisternae of the rough endoplasmic reticulum. Chondrocytes which had been treated for 4 h with 1 . 1 0 -8 M monensin showed no significant change in this distribution, but with 1 . 1 0 -7 M monensin (Fig. 4c), there was a large increase in labelling within distended Golgi vesicles, and with 1 . 1 0 -6 M this was further increased and extended into the distended cisternae of the rough endoplasmic reticulum (Fig. 4d). The results showed that the proteoglycan protein core within the chondrocytes was immunoreactive with the antibinding region antibodies and accumulated along the main pathways of secretory protein synthesis. Discussion

In cultures of normal chondrocytes significant intracellular labelling of proteoglycan protein core was only observed within secretory vesicles and the Golgi complex. This was similar to the distribution seen in chondrocytes within intact articular cartilage [14]. The failure to localise protein core within the cisternae of the rough endoplasmic reticulum of both chondrocytes in cartilage and in culture was of particular interest in view of its likely synthesis within this compartment. This may be because the concentration of protein core was too low to be detected, but might also be because it was inaccessible to the antibodies or not yet in a conformation that was antigenic. The latter seems unlikely, as although the antibodies used in this study require intact disulphide bonds in the bind-

252

E XTRACTS

a

1

Fig. 4. Identification and localisation of proteoglycan protein core in chondrocytes and the effect of monensin on its intracellular accumulation. (a) Fluorograph of a SDS-polyacrylamide gel showing [3H]serine-labelled extracts and immunoprecipitates of control (1) and monensin treated chondrocytes (1.10-7 M (2), 1.10-6 M (3) and 1.10-5 M (4)), prepared as described in Methods. Arrows denote the position of proteoglycan core protein identified by immunoprecipitates with anti-binding region serum. (b) Electron micrograph of pig laryngeal chondrocyte with immunolabelling using protein Agold of proteoglycan core protein, which shows low density labelling over Golgi elements ( ~ ). (c) As in (b) after incubation of the cells for 4 h in the presence of 1.10-7 M monensin. There is increased labelling within Golgi elements ( --, ). (d) As in (b) after incubation of the cells for 4 h in the presence of 1.10 -6 M monensin. Labelling is even more increased and extends into the cisternae of the rough endoplasmic reticulum ( ~ ). (Bars denote 1 p,m.)

2

3

IMMUNOPRECIPITAT ES

4

1

2

3

4

m

m

B

i w

m

k

I

q~t

254 ing region to maintain antigenic sites [14], results with chondrosarcoma cells showed that protein core was able to bind to hyaluronate soon after synthesis [5] and other in vitro studies showed that this also requires intact disulphide bridges [17]. At an early stage of biosynthesis the protein core was also immunoreactive [5]. Formation of the tertiary structure of binding region was thus suggested to take place soon after synthesis. Assuming a normal pathway of protein synthesis, the results show in control chondrocytes no obvious site of protein core accumulation within the cisternae of the rough endoplasmic reticulum prior to its translocation into elements of the Golgi. However, this may not imply that translocation of newly synthesised protein core is a particularly rapid event, as the convergence of protein core from the many cisternae of the rough endoplasmic reticulum into the Golgi elements may alone be responsible for the increase in concentration. The effect of monensin on proteoglycan biosynthesis in pig laryngeal chondrocytes was to inhibit chondroitin sulphate biosynthesis and proteoglycan secretion, resulting in the intracellular accumulation of proteoglycan protein core. These results were similar to those obtained with chick chondrocytes [6,11] and rat chondrosarcoma cells [10]. Immunolocalisation showed a large increase in intracellular abundance of protein core mainly present within distended Golgi vesicles with 1. 10 -7 M monensin, but also in distended rough endoplasmic reticulum with 1 . 1 0 -6 M monensin. Although the immunolocalisation cannot discriminate between protein core that does or does not bear chondroitin sulphate chains, the results of immunoprecipitation showed that protein core devoid of chondroitin sulphate accumulated intracellularly in the presence of monensin and this was also accompanied by greatly reduced chondroitin sulphate synthesis (as measured by [35S]sulphate incorporation). Most of the proteoglycan localised intracellularly in the presence of monensin was thus likely to be devoid of chondroitin sulphate chains. The change in intracellular distribution with increased monensin concentration suggests a 'backing up' effect, whereby the initial accumulation of core protein is within Golgi elements, but

at higher concentrations it also accumulates within rough endoplasmic reticulum. This would imply that the primary site of monensin inhibition of translocation is within the Golgi complex, preventing movement between different Golgi elements. This is in agreement with the results of other investigations with glycoproteins which showed that monensin prevented completion of the late events in glycoprotein biosynthesis [18-20]. These include mannose trimming and galactose and sialic acid transfer on complex-type N-linked oligosaccharides which are suggested to occur in trans elements of the Golgi [21-23]. The results with proteoglycans suggest that chondroitin sulphate synthesis occurs after the entry of proteoglycan protein core into elements of the Golgi system at a site comparable to these late events of complex oligosaccharide synthesis. These observations are also compatible with others on chondroitin sulphate synthesis, which have shown by autoradiography that [35S]sulphate incorporation was localised in perinuclear regions similar to the location of the Golgi complex [24], and the results of localisation by immunofluorescence which showed proteoglycan within vesicles in the region of the Golgi [25,26]. Together these results suggest that the synthesis of chondroitin sulphate (which comprises 90% (w/w) of the completed proteoglycan) only occurs on the protein core when it passes into distal Golgi elements and this is followed rapidly by its secretion into the extracellular matrix.

Acknowledgements We thank Miss Fatemah Saed-Nejad, Miss Clare Hughes, Mrs. Rita Manhas and Mr. C. Wells for excellent technical assistance and the Arthritis and Rheumatism Council and M.R.C. for their support.

References 1 Hardingham, T.E. (1981) Biochem. Soc. Trans. 9, 489-497 2 Hardingham,T.E., Burditt, L. and Ratcliffe,A. (1984) Prog. Clin. Biol. Res. 151, 17-29. 3 Kimura, J.H., Caputo, C.B. and Hascall, V.C. (1981a) J. Biol. Chem. 256, 4368-4376 4 Mitchell, D. and Hardingham, T.E. (1981) Biochem.J. 196, 521-529

255 5 Kimura, J.H., Thonar, E.J., Hascall, V.C., Reiner, A. and Poole, A.R. (1981) J. Biol. Chem. 256, 7890-7897 6 Nishimoto, S.K., Kajiwara, T. and Tanzer, M.L. (1982) J. Biol. Chem. 257, 10558-10561 7 Tartakoff, A.M. and Vassalli, P. (1977) J. Exp. Med. 146, 1332-1345 8 Tartakoff, A.M. and Vassalli, P. (1978) J. Cell. Biol. 79, 694-707 9 Tajiri, K., Uchida, N. and Tanzer, M.L. (1980) J. Biol. Chem. 255, 6036-6039 10 Mitchell, D. and Hardingham, T.E. (1982) Biochem. J. 202, 249-254 11 Nishimoto, S.K., Kajiwara, T., Ledger, P.W. and Tanzer, M.L. (1982) J. Biol. Chem. 257, 11712-11716 12 Ratcliffe, A. and Hardingham, T.E. (1983) Biochem. J. 213, 371-378 13 Skinner, M.K. and Griswold, M.D. (1983) Biochem. J. 209, 281-284 14 Ratcliffe, A., Fryer, P.R. and Hardingham, T.E. (1984) J. Histochem. Cytochem. 32, 193-201 15 Fryer, P.R., Wells, C. and Ratcliffe, A. (1983) Histochemistry 77, 141-143

16 Roth, J., Bendayan, M., Carlemalm, E., Villiger, W. and Garavito, M. (1981) J. Histochem. Cytochem. 29, 663-671 17 Hardingham, T.E., Ewins, R.J.F. and Muir, H. (1976) Biochem. J. 157, 127-143 18 Tartakoff, A.M. and Vassalli, P. (1979) J. Cell Biol. 83, 284-299 19 K ~ i a i n e n , L., Hashimoto, K., Saraste, J., Virtanen, I. and Penttinnen, K. (1980) J. Cell. Biol. 87, 783-791 20 Griffiths, G., Quinn, P. and Warren, G. (1983) J. Cell. Biol. 96, 835-850 21 Roth, J. and Berger, E.G. (1982) J. Cell. Biol. 92, 223-229 22 Quinn, P., Griffiths, G. and Warren, G. (1983) J. Cell. Biol. 96, 851-856 23 Strous, G.J., Kerkhof, P.V., Willemsen, R., Geuze, H.J. and Berger, E.G. (1983) J. Cell. Biol. 97, 723-727 24 Horwitz, A.L. and Dorfman, A. (1968) J. Cell. Biol. 38, 358-368 25 Vertel, B. and Dorfman, A. (1979) Proc. Natl. Acad. Sci. USA 76, 1261-1264 26 Pacifici, M., Soltesz, R., Thai, G., Shanley, D., Boettiger, M. and Holtzer, H. (1983) J. Cell Biol. 97, 1724-1736