OSTEOARTHRITIS and CARTILAGE

Transcription

1 Osteoarthritis and Cartilage (1994) 2, Osteoarthritis Research Society /94/ $08.00/0 OSTEOARTHRITIS and CARTILAGE Analysis of changes in proteoglycan content in murine articular cartilage using image analysis BY PETER M. VAN DER KRAAN*, Jos DE LANGE~, ELLY L. VITTERS*, HENK M. VAN BEUNINGEN*, GERJO J.V.M. VAN OSCH*, PETER L.E.M. VAN LENT*, WIM B. VAN DEN BERG* *Department of Rheumatology, University Hospital Nijmegen, Geert Grooteplein zuid 8, 6525 GA, Nijmegen, The Netherlands; ~Carl Zeiss B.V., Van Houten Industriepark 21, Weesp, The Netherlands Summary The extracellular matrix of articular cartilage consists mainly of type II collagen and large aggregating proteoglycan (aggrecan). During arthritis and other joint diseases, the proteoglycan (PG) level of cartilage matrix is diminished, leading to impairment of normal joint function. A new method is described for measuring the changes in PG content of murine articular cartilage. The method is based on the automated densitometric analysis of patellar cartilage of standard, safranin O-stained sections of whole murine knee joints. It appeared to be possible to measure optical density in parallel layers of articular cartilage with high reproducibility. Approximately 25 sections can be evaluated within 1 h. Measuring a single section 10 times resulted in a coefficient of variation (CV) of %. A mean CV of 5-14~/o was calculated when a group of 18 sections was analyzed in quintuplicate. To validate the procedure, changes in PG content induced by arthritis or by intra-articular injection of TGFfl-1 were analyzed by the image analysis method, the dimethylmethylene blue (DMB) assay and by visual grading. Although not a quantitive method, the newly developed image analysis method appeared to be more sensitive in detecting significant change in PG content of murine articular cartilage than the DMB method or visual grading. The image analysis method makes it possible to measure changes in PG content of specific areas of articular cartilage with higher sensitivity than the DMB method and eliminating the bias inherent to visual grading by human observers. Key words: Image analysis, Proteoglycans, Cartilage, Murine. Introduction ARTICULAR cartilage is a specialized form of connective tissue consisting of chondrocytes which deposit an extracellular matrix of giant molecules. The two main matrix macromolecules of articular cartilage are type II collagen and aggrecan, the large aggregating proteoglycan (PG). The collagen provides the tensile strength of articular cartilage while aggrecan is responsible for the unique resiliency and water binding capacity of this tissue [1]. Under pathological conditions such as arthritis or osteoarthritis, in both humans and laboratory animals, the PG Content of articular cartilage is often reduced due to the disease process [2-8]. This will result in impaired functioning of the affected joint. In an experimental setting, the measurement of PG content of diseased articular cartilage can provide information about the progression of the Submitted 14 December 1993; accepted 29 March Supported by the Dutch League against Rheumatism. Address correspondence and requests for reprints to: Peter M. van der Kraan, Department of Rheumatology, Geert Grooteplein zuid 8, 6525 GA Nijmegen, The Netherlands. disease process and the effect of pharmacological intervention. Several methods have been developed for assaying the PG content of articular cartilage. The amount of PGs in articular cartilage can be quantified after total digestion or hydrolysis of articular cartilage by colorimetric methods. The carbazole method developed by Bitter and Muir [9] for determining uronic acid content (a component of aggrecan and other PGs) and the dimethylmethylene blue method (measuring predominantly sulfated glycosaminoglycans) according to Farndale et al. [10] are widely used. Alternatively, changes in PG content of articular cartilage can be examined in histological sections using PG-specific dyes, such as safranin O, alcian blue, or toluidine blue [3, 11-14]. We studied the possibility of analyzing changes in PG content of murine patellar cartilage using standard histological sections stained with the stoichiometric binding dye safranin O in combination with automated image analysis. It appeared to be possible to measure changes in murine articular cartilage PG content with high sensitivity and reproducibility, omitting the observer-related bias inherent to visual estimation of PG content. 207

2 208 van der Kraan et al.: Analysis of changes in PG content Materials and methods TISSUES C57B1 (20-25 g, weeks) were used in all experiments. The animals were fed a standard laboratory diet (RHM, Hope Farms, Linschoten, The Netherlands) and supplied with tap water ad libitum. Mice were kept in boxes in an air conditioned room with a sawdust bedding. PG depletion in patellar articular cartilage was induced in two ways: firstly by intra-articular injection of papain and secondly by induction of unilateral arthritis with methylated bovine serum albumin (mbsa) or zymozan. These methods are routinely used in our laboratory [3, 8, 15-17]. In short, the right knee joints of mice were injected once, through the patellar ligament, with 6 #l of 1, 2.5 and 5 mg/ml papain (type IV, double crystallized, Sigma, St Louis, MO). The papain was supplemented with 0.03 M L-cysteine hydrochloride (Sigma) to activate the papain. After one day, the papain-injected and contralateral knee joints were dissected for histology. Arthritis was induced by immunization of mice with 100 pg mbsa (Sigma) in Freund's complete adjuvant (Difco, Detroit, MI). Bordetella pertussis (heat killed, National Institute of Public Health, Bilthoven, The Netherlands) was used as an additional adjuvant. Seven days after the initial immunization, mice were boosted with 100 pg mbsa. At day 21, inflammation was induced by intra-articular injection of 60 pg mbsa. Three days after arthritis induction, inflamed and contralateral knee joints were dissected. In addition, arthritis was induced by intra-articular injection of 180 ttg zymozan (yeast-particles) in the right knee joint of nonimmunized mice. Patellar cartilage with supranormal PG levels was established by intra-articular injection of 100 ng transforming growth factor-fl-1 (TGFfl-1; Oncogene, Seattle, WA) in the joint space of the right,knee joint on three alternate days. One day after the last injection, knee joints were used for histology. HISTOLOGICAL PROCESSING Knee joints were dissected in toto and fixed for 7 days in phosphate-buffered formalin. The tissues were decalcified (50//0 formic acid, Merck, Darmstadt) and dehydrated with the use of an automated tissue processing apparatus (VIP, Miles Scientific, Naperville, IL). Neutral buffered formalin fixation preserves the articular cartilage PG content well [18]. Addition of quaternary ammonium salts in the fixative to prevent PG loss was omitted since this suppresses the staining of cartilage matrix with safranin 0 [18]. Decalcification was followed by embedding of the knee joints in paraffin wax. Frontal knee joint sections (6 #m) were prepared and mounted on gelatin-coated slides. After removal of paraffin with xylol and ethanol, sections were stained with fast green (10 min, 0.1%, Merck) and safranin O (10min, 0.1%, BDH, Poole, U.K.). After the safranin O staining, the sections were rinsed for 5 min with ethanol (100%) and twice (5 min) with xylol. Experimental and control groups of a single experiment were all processed and stained in parallel. EQUIPMENT FOR IMAGE ANALYSIS The sections were examined with a Leitz microscope (magnification x100, Leitz Dialux 20, Wetzlar, Germany) using bright field illumination. The microscopic image was recorded and digitized with a CCD video camera (DXC-151, Sony, Japan) in combination with a VIDAS image analysis system (Kontron Electronics, Germany). Images were displayed on a monitor (Multisync NEC 3D, Japan) with a resolution of pixels and 256 gray levels. Recorded images were stored on Bernouilli disks. Fast green staining, in the images, was neutralized by the use of a interference green filter (546 nm, Carl Zeiss, Germany). PROCEDURE All measurements were done with a standard camera setting. The light intensity used during the measurements was optimized. An empty image field was depicted and the light intensity was adjusted maximally, preventing over-exposure of the system. Light intensity was kept constant during the measurements. The empty image field was recorded and the image was smoothed by use of a lowpass filter. This image was used as a reference for the shading correction procedure applied later on in the measurement program. The patella was positioned in the middle of the video image marked by a circle. The patellar cartilage was placed in the video image with the articular cartilage at the lower part of the image. Approximately 90% of the articular cartilage, excluding the margins, is present in the image (Fig. 1). After focusing, the image was recorded by averaging four black and white images. The gray level of the picture was corrected for uneven light distribution by use of additive shading correction.

3 Osteoarthritis and Cartilage Vol. 2 No was measured while in the calcified cartilage at a depth of 55 #m a zone of 11 #m was analyzed (Fig. 1). The optical density is expressed as the mean optical density in the measurement area (total O.D./measurement area). Since safranin O binds stoichiometrically to cartilage glycosaminoglycans [12, 13], the best approximation of changes in glycosaminoglycan content is comparison of the measured optical density values, assuming that O.D. = -1 log T (T = transmission). FIG. 1. Microphotograph of murine knee joint depicting the zones of patellar cartilage measured by the image analysis method. In the noncalcified cartilage a zone of 20 pixels (22 pm) adjacent to the cartilage surface was measured while in the calcified cartilage at a depth of 55 #m a zone of 11 pm was analyzed. The analyzed zones are indicated. The figure is similar to the video image recorded by the CCD camera (magnification x65, safranin O and fast green stained). SEGMENTATION A gray value threshold was used to delete objects not stained by safranin O from the image. An interactive dynamic threshold was used in pilot experiments to determine the optimal setting of this threshold. In the final measurement program a fixed threshold was used for segmentation. During this procedure, bone and chondrocyte lacunae were omitted from the image. However, the deletion from the image of highly PG-depleted areas in the cartilage matrix by this procedure could be restored by hand. If necessary, inflammatory cells and debris aligning the cartilage surface (arthritic knees) still present in the image could be eliminated from the image by use of the cursor. Finally, all image objects not in contact with the circle defined earlier were omitted from the image. MEASUREMENT The optical density in zones of cartilage parallel to the patellar cartilage surface was measured. Due to the above described segmentation procedure, the chondrocyte lacunae do not contribute to the measured values. The width of the zone (in pixels) and the number of zones measured can be set interactively at the start of the measurement session. In initial measurements, 10 zones with a width of 10 pixels (11 #m) were measured. In later sessions, in the noncalcified cartilage a zone of 20 pixels (22 #m) adjacent to the cartilage surface SPECTROPHOTOMETRIC ANALYSIS OF CARTILAGE GLYCOSAMINOGLYCAN CONTENT In addition to the densitometric analysis of glycosaminoglycan content, the glycosaminoglycan quantity of patellar cartilage was measured with the dimethylmethylene blue (DMB) method according to Farndale et al. [10]. Patellae were dissected from the knee joints as described before [19], fixed in ethanol (70%) and subsequently decalcified in 5% formic acid. Cartilage was stripped from the underlying bone with the use of forceps. Using this method, both the noncalcified and the calcified layer of the articular cartilage are obtained. The isolated cartilage was digested overnight with papain at 60 C (100 #1/patella). The papain mixture (ph 6.0) consisted of I mg/ml papain, 0.1 M sodium acetate, 10 mm L-cysteine hydrochloride and 50 mm ethylenediaminetetraacetic acid sodium salt (all from Sigma). Immediately after mixture of the digests with DMB the glycosaminoglycan content was assayed by spectrophotometric analysis. Chondroitin sulfate type A (Sigma) from whale cartilage was used as a standard. VISUAL ESTIMATION OF PG CONTENT In addition to examination of the histological sections by image analysis, PG content of patellar cartilage was visually scored by a blinded observer. The same sections were used for densitomet~c analysis with the image analyzer and visual grading. Changes in PG content of the noncalcifled zone were scored. PG depletion was graded on a scale of 0 to -4 (total depletion), while an elevated PG content was indicated by a positive score. STATISTICAL EVALUATION Statistical differences were evaluated with the Wilcoxon rank sum test. P < 0.05 was considered as statistically significant.

4 210 van der Kraan et al.: Analysis of changes in PG content Results In Fig. 2. a microphotograph of an arthritic, PG-depleted knee joint and a control knee joint are shown. Fig. 3 represents the optical density measured in 10 parallel zones of patellar cartilage from a normal knee and a knee with antigeninduced arthritis. It can be seen that the arthritic process mainly leads to PG depletion in the noncalcified cartilage, approximately zones 1-4. The calcified cartilage appeared relatively insensitive to PG depletion. The surface zone of the articular cartilage was depleted most extensively [Fig. 3(b)]. However, in normal knee joints also, the surface of murine articular cartilage stains less than the deeper zones [Fig. 3(a)]. In later experiments only zones 1 and 2 (combined) and zone 6 were measured in each knee. For each section it was visually checked that zones 1 and 2 were confined to the noncalcified articular cartilage while zone 6 was exclusively situated in the calcified cartilage below the tidemark. FIG. 2. Micrograph of a normal knee joint and a knee joint with antigen-induced arthritis. Sections were stained by safranin O and fast green. (a) Normal knee joint; (b) arthritic knee joint. N: non-calcified cartilage; C: calcified cartilage; T: tidemark (magnification x 65, safranin O and fast green stained). To determine the reproducibility of the measurements, optical density in a representative control and a representative depleted knee joint section was analyzed 10 times (Table I). In addition, the reproducibility of the measurements was checked by five subsequent measurements in six control knees and six PG-depleted knees (papain-injected). Three sections of each knee joint were measured. In between the measurements all systems were turned off. As can be seen in Table I, the measurement of the optical density of a particular section is highly reproducible. The coefficient of variation (CV) varied from %. Also, the measurements of a group of knee joints appeared to be very reproducible (Table II). The mean CV between the measurements ranged from 5-14 //o. The variability in the PG-depleted patellar cartilage was, although still very small, higher than in the control knee joints. Statistical differences between subsequent measurements were never observed. These results indicate that adjustment of the equipment has little effect on the measurements and that optical density measurements are highly reproducible. Whether the calcified cartilage (zone 6) could be used as a reference value applied for correction of variation in section thickness or staining variability was examined. However, it appeared that under certain conditions (e.g. injection of 5 mg/ml papain) the optical density in the calcified layer changed. The optical density in the calcified cartilage area can only be used as an internal reference value, as no significant changes in optical density are induced by the experimental procedure. Nevertheless it appeared that the relative standard deviations are corrected values and uncorrected values were nearly identical. This indicates that the reliability of the measurements is not improved significantly by the use of corrected values, and that variability within one experiment due to variations in section thickness or staining variability is limited. The effectiveness of detecting changes in PG content by image analysis on standard histological sections were compared with the colorimetric method using DMB and cartilage digests and with visual examination. It must be kept in mind that the measurement by the DMB method measures the integral PG content of both the calcified and noncalcified cartilage while the image analysis method makes it possible to measure specific areas of articular cartilage. The DMB method will therefore underestimate the PG depletion in the noncalcified cartilage. A dose-response relation between papain dose and cartilage depletion could only be seen using image analysis (Table III). The CV

5 Osteoarthritis and Cartilage Vol. 2 No i(a) 0"8 I "~ 0.6 e~ 0.5 o.7- I T T T T T T T I I f I i I I t I l Cartilage zone 0.9 (b) 0.8 ~0.7 T T T T T "~ T T I 0.3 i i i i I i I i I Cartilage zone FIG. 3. Optical density of patellar cartilage of 10 perpendicular zones of 11 pm each. (a) Control knee joint; (b) knee joint with antigen-induced arthritis. Values shown are mean_+ S.D. of 10 normal and 10 arthritic knee joints, measured once. ranged from 5-7% with the image analysis method while with the DMB method it ranged from 5-20%. Although we are not able to measure the absolute cartilage PG content using image analysis, in contrast to the DMB method, the image analysis method appears to be more sensitive in measuring changes in PG content in the noncalcified cartilage then the DMB method. Also, visual scoring of Table I Reproducibility of optical density measurements in a control and papain-injected routine knees:joint Control knee Papain-injected knee Noncalcified Calcified Noncalcified Calcified layer layer layer layer Mean optical density t- S.D. -} ± ± ± Ten measurements were carried out. After each measurement, the VIDAS equipment and the microscope light were turned off and the microscope was defocused and refocused. In the noncalcified cartilage a zone of 20 pixels (22 #m) adjacent to the cartilage surface was measured while in the calcified cartilage at a depth of 55 pm a zone of 11/~m was analyzed.

6 212 van der Kraan et al.: Analysis of changes in PG content Table II Reproducibility of optical density measurements of 6 control knees (18 sections) and 6 papa±n-injected knee joints Control knee Papa±n-injected knee Noncalcified Noncalcified Measurement layer Calcified layer layer Calcified layer ±_ ± ± _ _ ± _ ± ± ± ± ± ± _ ± ± ± ± ± 0.07 Three sections of each knee were measured and the mean value of the three sections was calculated. The values shown are mean + S.D. for six knees. In the noncalcified cartilage a zone of 20 pixels (22 pm) adjacent to the cartilage surface was measured while in the calcified cartilage at; a depth of 55 #m a zone of 11 gm was analyzed. PG depletion appeared to be less sensitive in detecting significant changes in PG content than examination by image analysis. The effects of triple TGFfl injection and both antigen-induced and zymozan-induced arthritis were measured using image analysis (Table IV). From earlier histological and DMB measurements it was known that triple TGFfl injections lead to an elevated PG content while arthritis results in depletion of the cartilage matrix. Elevated (TGFfl) Table III The effect of papain injection on patellar cartilage PG content as measured by the DMB method and image analysis. DMB method -- glyco- Papain saminoglycan content (#g/ (mg/ml) patella) ~o Image analysis method -- optical density % Visual scoring method 0 (control) 3.91 ± ± ± 0.22* ± 0.03* ± 0.30* _ 0.04* ± 0.54* ± 0.04"~ PG content was measured using the DMB method after isolation and digestion of the patellar cartilage. With image analysis, in the noncalcified cartilage a zone of 22 gm adjacent to the cartilage surface was measured. In addition, PG content was visually estimated (PG depletion on a scale of 0 to -4, PG increase > 0, 0 = control). The percentage PG depletion was, in the case of image analysis, estimated with the premise that safranin O binds stoichiometrically to PGs. The values shown are meanis.d, for six knees (18 sections). *Significantly different from control; tsignificantly different from 1 mg/ml. A reduction of glycosaminoglycan content of approximately 50% (image analysis or DMB) results in a total absence of safranin O staining as noted by visual examination. Table IV Effect of antigen-induced arthritis and zymozan-induced arthritis and of triple TGFfl-1 injections on optical density of articular cartilage measured by image analysis Image analysis method -- optical density DMB method -- glycosaminoglycan Visual scoring Treatment Noncalcified layer Calcified layer content (gg/patella) method None 0.82 ± ± ± x TGFfl 0.97 ± 0.06* 0.98 ± _ /+ 1 Antigen-induced arthritis 0.65 ± 0.10" 0.83 ± ± 0.31" - 3* Zymozan-induced arthritis 0.73 ± 0.05* 0.84 ± /- 2* In the noncalcified cartilage a zone of 20 pixels (22 pm) adjacent to the cartilage surface was measured while in the calcified cartilage at a depth of 55 pm a zone of 11 #m was analyze&in addition, in parallel experiments PG content was measured using the DMB method after isolation and digestion of the patellar cartilage and by visual estimation (PG depletion on a scale of 0 to -4, PG increase > 0, 0 = control). The values shown are mean ± S.D. of six knees. *Statistically different from control.

7 Osteoarthritis and Cartilage Vol. 2 No as well as reduced PG content (arthritis) in the noncalcified cartilage could easily be detected using image analysis. With the DMB method only a significant depletion of PGs could be demonstrated in patellar cartilage from knees with antigen-induced arthritis. When visually scored, a significant elevation in PG content could not be detected although a trend indicating a supranormal PG quantity in the noncalcified cartilage was observed. It was obvious that changes in PG content of the noncalcified cartilage of murine patellar cartilage could be more easily detected using image analysis than with the DMB method or visual scoring. Discussion This paper has described a method for quantitation of changes in PG content of patellar cartilage of murine knee joints using standard histological sections and image analysis. The method, although here only depicted for patellar cartilage of the mouse, can easily be adapted to other cartilage areas or joint sections of other small animals, such as rats or guinea-pigs. The densitometric measurements appeared to be highly reproducible and insensitive to changes in equipment setting. The newly developed method was compared with two established methods of estimating changes in PG content in articular cartilage. Estimation of changes in PG content with the DMB method of Farndale et al. appeared to be less sensitive than the densitometric method. This is probably due to the fact that only a limited amount of cartilage is available for this method (mouse patella), leading to measurement of glycosaminoglycan concentrations near the detection limit. Moreover, since the entire cartilage layer is stripped off the patella, the DMB method measures the total PG content of both the calcified and noncalcified cartilage while the image analysis method can measure in an area restricted to the noncalcified cartilage. One of the main advantages of the newly developed method is the possibility of measuring changes in PG content in specific depths of articular cartilage (Fig. 3). In addition, in contrast to the DMB method, the cartilage does not need to be digested and standard histological sections can be used to study changes in PG content. Approximately 25 sections can be evaluated within 1 h. Visual examination of histological sections could less easily detect changes in PG content than the image analysis method (Tables III& IV). Statistical significance was reached more easily with the latter method. An elevated PG content in the noncalcified zone of patellar cartilage (TGFfl, Table IV) was especially difficult to detect visually. That the amount of PG is really raised in articular cartilage after triple TGFfl injections was indicated by the observation that in the same cartilage the PG synthesis was strongly elevated [20]. The parallel elevated optical density of the calcified cartilage probably prevents an easy detection of an increased optical density in the noncalcified cartilage for the human observer. Since depletion of cartilage takes place mainly in the noncalcified zone, comparison of the noncalcifled zone and calcified zone in one section makes it relatively simple to observe PG depletion. The image analysis method makes it possible to score histological sections independent of variations between different human observers and timedependent variations in a visual scoring system. Minimizing the role of the human observer will reduce inaccuracies inherent to human observerrelated bias. A limitation of the image analysis method is that changes in PG content can be very simply detected but absolute values for PG content cannot be given. It is also recommended to perform the staining procedure on all groups within an experiment with the same solutions and at the same time, since variation in staining procedure can introduce variation in measured optical densities. The use of standard control sections to be stained along with the experimental groups could give an indication of staining performance. Variation in section thickness between experimental groups can also influence conclusions. However, correction of section thickness and staining variation using an internal control (calcifled zone) had litte effect on the outcome of the measurements in this paper. In conclusion, an automated image analysis method is presented which makes it possible to measure changes in the PG content in murine articular cartilage with greater sensitivity and reproducibility than visual scoring or the DMB method. References 1. Hardingham TE, Fosang AJ, Dudhia J. Aggrecan, the chondroitin sulfate/keratan sulfate proteoglycan from cartilage. In: Kuettner KE, Schleyerbach R, Peyron J, Hascall VC, Eds. Articular cartilage and osteoarthritis. New York: Raven Press Ltd 1992: Mankin HJ, Dorfman H, Lippiello L, Zarins A. Biochemical and metabolical abnormalities in articular cartilage from osteoarthritic human hips. II Correlation of morphology with biochemical and metabolical data. J Bone Joint Surg [Am] 1971;63:131-9.

8 214 van der Kraan et al.: Analysis of changes in PG content 3. Van den Berg WB, Kruijsen MWM, van de Putte LBA, van Beusekom HJ, van der Sluis M, Zwarts WA. Antigen-induced and zymozan-induced arthritis in mice: studies on in vivo cartilage proteoglycan synthesis and chondrocyte death. Br J Exp Pathol 1981;62: Kruijsen MWM, van den Berg WB, van de Putte LBA. Sequential alterations of periarticular structures in antigen-induced arthritis in mice. Histological observations on fibrous capsule, ligaments, bone and muscles using whole joint sections. Br J Exp Pathol 1983;12: Carney SI, Billingham MEJ, Muir H, Sandy JD. Demonstration of increased proteoglycan turnover in cartilage explants from dogs with experimental osteoarthritis. J Orthop Res 1984;2: Poole AR. Changes in collagen and proteoglycan of articular cartilage in arthritis. Rheumatology 1986;10: Van Beuningen HM, van den Berg WB, Schalkwijk J, Arntz OJ, van de Putte LBA. Age- and sexrelated differences in antigen-induced arthritis in C57B1/10 mice. Arthritis Rheum 1989;32: Van der Kraan PM, Vitters EL, van de Putte LBA, van den Berg WB. Development of osteoarthritic lesions in mice by "metabolic" and "mechanical" alterations in the knee joints. Am J Pathol 1989;135: Bitter T, Muir HM. A modified uronic acid carbazole reaction. Anal Biochem 1962;4: Farndale RW, Buttle DJ, Barrett AJ. Improved quantitation of sulfated glycosaminoglycans by use of dimethylmethylene blue. Biochim Biophys Acta 1986;883: Poole AR. The relationship between toluidine staining and hexuronic acid content of cartilage matrix. Histochem J 1970;2: Rosenberg L. Chemical basis for the histological use of safranin O in the study of articular cartilage. J Bone Joint Surg [Am] 1971;53: Kiviranta I, Jurvelin J, Tammi M, Helminnen HJ. Microscopic quantitation of glycosaminoglycans in articular cartilage sections stained with safranin O. Histochemistry 1985;80: Sayers DCJ, Volpin G, Bentley G. The demonstration of bone and cartilage remodelling using alcian blue and haematoxylin. Stain Technol 1987;63: Van der Kraan PM, Vitters EL, van Beuningen HM, van den Berg WB. Proteoglycan synthesis and osteophyte formation in "metabolically" and "mechanically" induced murine degenerative joint disease: an in-vivo autoradiographic study. Int J Exp Pathol 1992;73: Van Lent PLEM, Dekker C, Mosterd J, van den Bersselaar L, van den Berg WB. Allergic arthritis induced by cationic proteins: role of molecular weight. Immunology 1989;67: Van de Loo AAJ, Arnzt OJ, Otterness IG, van den Berg WB. Protection against cartilage proteoglycan synthesis inhibition by anti-interleukin 1 antibodies in experimental arthritis. J Rheumatol 1992;19: Kiviranta I, Tammi M, Jurvelin J, S~im~inen A-M, Helminen HJ. Fixation, decalcification, and tissue processing effect on articular cartilage proteoglycans. Histochemistry 1984;86: Van den Berg WB, Kruijsen MWM, van de Putte LBA. The mouse patella assay. An easy method of quantitating articular cartilage chondrocyte function in vivo and in vitro. Rheumatol Int 1982;1: Van Beuningen HM, van der Kraan PM, Arntz OJ, van den Berg WB. Stimulation of osteophyte formation and cartilage proteoglycan synthesis by intraarticularly injected TGF-fl. Trans Orthop Res Soc 1993;18:289(abstract).