key: cord-0005807-bi1tqa7e authors: Saklatvala, J. title: Tumour necrosis factor α stimulates resorption and inhibits synthesis of proteoglycan in cartilage date: 1986 journal: Nature DOI: 10.1038/322547a0 sha: 95803b520dd8a2f7d963e89828f3fb56160736a5 doc_id: 5807 cord_uid: bi1tqa7e During inflammatory reactions, activated leukocytes are thought to produce a variety of small proteins (cytokines) that influence the behaviour of other cells (including other leukocytes). Of these substances, which include the interleukins, interferons and tumour necrosis factors (TNFs), interleukin-1 (IL-1) has been considered potentially a most important inflammatory mediator because of its wide range of effects (reviewed in refs 1, 2). In vivo it is pyrogenic and promotes the acute phase response; in vitro it activates lymphocytes(3) and stimulates resorption of cartilage(4) and bone(5,6). Cartilage resorption is a major feature of inflammatory diseases such as rheumatoid arthritis, and IL-1 is the only cytokine hitherto known to promote it. TNFs are characterized by their effects on tumours and cytotoxicity to transformed cells(7–9), but share some actions with IL-1.1 report here that recombinant human TNFα stimulates resorption and inhibits synthesis of proteoglycan in explants of cartilage. Its action is similar to and additive with IL-1, and it is a second macrophage-derived cytokine whose production in rheumatoid arthritis, or inflammation generally, could contribute to tissue destruction. During inflammatory reactions, activated leukocytes are thought to produce a variety of small proteins (cytokines) that influence the behaviour of other cells (including other leukocytes). Of these substances, which include the interleukins, interferons and tumour necrosis factors (TNFs), interleukin-l (IL-l) has been considered potentially a most important inflammatory mediator because of its wide range of effects (reviewed in refs I, 2). In vivo it is pyrogenic and promotes the acute phase response; in vitro it activates lymphocytes 3 and stimulates resorption of cartilage 4 and bone M • Cartilage resorption is a major feature of inflammatory diseases such as rheumatoid arthritis, and IL-l is the only cytokine hitherto known to promote it. TNFs are characterized by their effects on tumours and cytotoxicity to transformed cells 7 -9 , but share some actions with IL-1. I report here that recombinant human TNF a stimulates resorption and inbibits synthesis of proteoglycan in explants of cartilage. Its action is similar to and additive with IL-l, and it is a second macropbage-derived cytokine wbose production ' in rheumatoid artbritis, or inflammation generally, could contribute to tissue destruction. Two human TNFs have been isolated: TNFa (refs 8, 9) is a product of activated mononuclear phagocytes, TNFf3 (ref. 7) of activated lymphocytes. The proteins show about 50% homology in nucleotide sequence and compete for a common class of receptors on the cervical carcinoma line ME-180 (ref. 10) . TNFa is probably identical with cachectin ll , a factor that suppresses production of lipoprotein lipase in cultured adipocytes, and may playa part in the development of cachexia during infection l2 . IL-I shows some biological similarity to TNF: it is cytotoxic to certain transfonned cells l3 and it suppresses production oflipoprotein lipase in adipocytes l4 . Furthermore, cachectin has recently been shown to stimulate production of prostaglandins and latent collagenase by human synovial and dennal fibroblasts in a manner similar to IL-lls. In view of these findings, I have investigated the effect of TNFa on both the resorption and synthesis of proteoglycan by explants of cartilage, Chondroitin-sulphate-rich proteoglycan is an essential component of the matrix of cartilage since it enables the tissue to resist compression during load-bearing. Loss of proteoglycan, such as occurs in rheumatoid arthritis, osteoarthritis and other joint diseases, results in severe impairment of the function of cartilage. IL-l is the only purified cytokine known to cause cartilage to degrade its proteoglycan 4 ,16, and to inhibit resynthesis 17 • Figure la shows the amount of proteoglycan (measured as percentage of total chondroitin sulphate) released from porcine articular cartilage during 6 days of culture in the presence of human recombinant TNFa or pure porcine IL-l. The TNFa caused up to 75% of the proteoglycan to be released, although it was less potent than the IL-l, which was significantly active at a 20-fold lower dose (0.5 pM). Figure 1 b shows a similar experiment carried out on cartilage from bovine nasal septum which was cultured for a shorter period (48 h): again, the IL-l was more potent. The time dependence of the release of proteoglycan from bovine cartilage caused by sub-maximal concentrations of the two agents revealed that their effects were additive. Figure Ie shows that 50 pM IL-l or 290 pM TNFa caused a similar rate of release, and that this was approximately doubled when the agents were combined. Maximal stimulation of cartilage by IL-I caused more rapid release of proteoglycan than did TNFa (Fig.ld) : results for two concentrations of each cytokine demonstrate that responses were maximal. Supramaximal doses of the two agents in combination caused a rate of release that was considerably faster than that due to TNFa alone, but was not significantly greater than that seen with lL-1 alone. The failure of TNFa to augment the maximal response to IL-I may be because the limit of the chondrocytes' ability to degrade their matrix in vitro was being approached. The enzymatic mechanism by which the proteoglycan is degraded in cartilage is not understood. Normally, cartilage proteoglycans aggregate in a specific manner with hyaluronic acid, and it is thought that the large size of these aggregates causes them to be trapped in the matrix. Cartilage stimulated by IL-I releases fragments of proteoglycan which, as judged by gel filtration, are smaller than nonnal proteoglycan monomers and are unable to aggregate with hyaluronic acid ls . There is no evidence of degradation of their chondroitin sulphate chains. These changes suggest that degradation is by limited proteolysis of the protein core. The fragments of proteoglycan that were released by cartilage stimulated with TNF behaved similarly on gel filtration to those generated by stimulation with IL-I (Fig. 2) . The bulk of the fragments generated by stimulation with either agent emerged from a Sepharose 2B column at a region between the elution positions of intact proteoglycan and proteoglycan digested with papain (which consists largely of single-chain chondroitin sulphate peptides). Addition of hyaluronic acid to the prot eo glycan fragments before chromatography caused little or no fonnation of aggregates. This suggested that the hyaluronate binding region was blocked or had been lost. When the proteolgycan fragments were chromatographed under dissociative conditions (4 M guanidine-HCl in the chromatographic buffer) the position of the main peak was unchanged. These experiments showed that chondrocytes activated by TNF or IL-I caused a similar limited proteolysis of the proteoglycans. In order to study the effect of TN Fa on the synthesis of proteoglycan, cartilage was stimulated for 48 h, and 3iS04 was added to the culture medium for the last 6 h. In this procedure the isotope becomes incorporated into newly synthesized sulphated glycosaminoglycan (mainly chondroitin sulphate). At the end of the experiment the medium and · cartilage were digested with papain, and glycosaminoglycan was precipitated from the digests with cetylpyridinium chloride. The amount of radioactivity present in the precipitates was a measure of chondroitin sulphate (and, by inference, proteoglycan) synthesis. In experiments made with porcine articular (Fig.3a) or bovine nasal septal (Fig.3b) cartilages, TNFa caused a marked sup- Hours for 30 min). Each was then transferred to a well of a 96-well multititre plate and incubated under the same conditions in 0.2 ml of culture medium, either with no addition, or with human TNFa or porcine IL-l of p15. The medium was changed at 3 days and the culture was terminated after 6 days. Human TNFa was a recombinant protein expressed in Escherichia coli and purified as described previously7.22. Porcine IL-l was a natural leukocyte protein purified to homogeneity as described elsewhere!6. The pI 5 form rather than the pI 8 form was used for these experiments. After culture the medium and cartilage were separated. The cartilage was digested completely with papain (see legend to Methods. a, Pieces of pig arti~ular cartilage were dissected, precultured and then stimulated either with TNFa or porcine IL-I for 48 h exactly as described for Fig. I , except that the culture medium (DMEM) contained 1 % normal bovine serum (heat inactivated) during the stimulation period. For the last 6 h the medium was replaced with S04-free culture medium (still containing TNFa or IL-I) to which was added 2.5 !LCi ml-I of 35S04 (25-40 Ci mg-I ; Amersham). After culture, the cartilage pieces were separated from the medium, briefly blotted to remove excess medium, and then weighted. Each piece was digested at 65°C for 2 h in 0.2 ml of 0.05 M sodium phosphate buffer pH 6.5 containing 1 mM EDTA, 2 mM N-acetylcysteine, 28!Lg ml-I papain (Sigma Type III). Samples of medium (0.2 ml) were also digested with 0.1 ml of the papain solution under the same conditions. Chondroitin sulphate (0.1 ml of 2 mg ml-I ) was added to all the samples followed by 0.1 ml of cetylpyridinium chloride (10% w/v). Samples were centrifuged, the precipitates were washed twice in 3% cetylpyridinium chloride, then dissolved in 0.5 ml of formic acid and added to 5 ml of a scintillation mixture (Pico-fluor-30, Packard) and counted in a liquid scintillation counter. The radioactivity of the digests of tissue and medium were added together and the results expressed as d.p.m. per mg wet weight of cartilage. b, As for a except that disks of bovine nasal septal cartilage were used (see Fig. I ), and the culture medium contained 5% normal bovine serum throughout the experiment. pression of proteoglycan production as judged the incorporation of 35S04 into glycosaminoglycan, but was less potent than IL-l. The porcine IL-1 inhibited incorporation in porcine articular cartilage in the range 0.1-10 pM; TN Fa was 20-fold less active. A similar differential was observed on the bovine cartilage. The lower potency of the human TNFa compared with porcine IL-1 may be due to species differences. Taken together, the experiments demonstrate that TNFa has a similar action to IL-1 on chondrocytes. It causes them to degrade proteoglycan by limited proteolysis and inhibits their synthesis of new proteoglycan. Exposure of cartilage to either of these leukocyte products during inflammation could lead to loss of proteoglycan and impairment of function; furthermore, their effects may be additive. Pigs l6 , like humans l9 , have two different IL-1 proteins: for these experiments the pI 5 IL-1 was used rather than the pi 8 form. The IL-1s are equipotent on cartilage l6 and compete for the same receptors on porcine synovial fibroblasts: TNFa at 400 times excess over IL-l) did not compete for these receptors (T. A. Bird and J. S., in preparation). The augmentation of the effect of maximal doses of TNFa by IL-1 reported here is consistent with there being different receptors on chondrocytes for the two cytokines. Since they are apparently not homologous 8 ,16, IL-1 and TNFa would be expected to combine with different receptors. These considerations suggest that chondrocytes (and probably other connective tissue cells) could have two distinct types of receptor (one for IL-1 and one for TNF) whose interaction with ligand promotes resorption of matrix polymers while inhibiting their synthesis. Such a possibility could have important implications for the pharmacological control of inflammatory tissue destruction. Following submission of this manuscript, Bertolini et al. 20 have reported that human TNFs, like IL-1, stimulated Ca2+ release from fetal rat bones. I thank Dr M. Palladino (Genentech) for supplying TNF, Dr C. A. Dinarello (Tufts University) for suggesting the experiments, Dr J. Tyler (Strangeways Laboratory) for doing the gel chromatography, Mrs V. Curry and Mrs J. Mason for technical help, and the Arthritis and Rheumatism Council for financial support. Biochemistry and Biology of Corona viruses