A place for TACE

Vertebrates, and humans are no exception, have developed potent innate and adaptive immune systems that deal with potential lethal encounters with microorganisms, which are critically dependent on the biological effects of proinflammatory cytokines such as interleukin 12, interleukin 18, interferon γ, and tumour necrosis factor α (TNF-α). The pivotal role of these cytokines is underscored by the observation that neutralisation of TNF-α in patients with Crohn's disease may convert latent tuberculosis into a potentially lethal disseminated form.¹ Thus proinflammatory cytokines are necessary for survival of the human species, but as they may cause significant inflammatory damage, their production and secretion need to be tightly regulated.

Indeed, the production and secretion of TNF-α are controlled at multiple checkpoints, no doubt in order to prevent unrestrained inflammation and tissue damage. TNF-α is translated as a precursor protein which contains an unusually long signal peptide that anchors the protein to the outer membrane. During local and systemic inflammatory reactions, membrane bound TNF-α is cleaved extracellularly by a specific zinc dependent metalloprotease that has been designated TNF-α converting enzyme (TACE), yielding the soluble homotrimeric form of TNF-α that can act either as a compartmentalised or circulating cytokine. It was initially thought that cleavage of TNF-α constituted a major final road block for TNF-α production but subsequently it was recognised that membrane expressed TNF-α is biologically active as a homotrimer during cell-cell contact by interacting with both p55 and p75 TNF receptors.²

TACE is member of the ADAM (a disintegrin and metalloproteinase) family of cell surface proteases, which are induced during inflammation,^3,4 and has attracted a lot of attention because small molecules that can be relatively easily manufactured are known to effectively block the function of metalloproteases. In this issue of Gut, Brynskov and colleagues⁵ report that in the normal colon mucosa, mononuclear as well as epithelial cells express bioactive TACE, and that TACE activity is increased in mucosal biopsies from patients with active ulcerative colitis but not in Crohn's disease [see page 37]. TACE activity was blocked ex vivo by metalloprotease/TACE inhibitors or zinc chelating agents, but not by trocade, a broad spectrum metalloprotease that is known not to affect TACE activity.

Is TACE an attractive therapeutic target in inflammatory bowel disease? Clearly, inhibition of TACE will lead to a reduction in the amount of secreted TNF-α but the number of membrane bound TNF-α molecules is not significantly altered.⁶ Soluble and membrane bound forms have different biological functions and this has been clearly demonstrated by studying the phenotype of mice that were engineered to express a form of T lymphocyte targeted TNF-α that cannot be cleaved by TACE or other metalloproteases.⁷ As a consequence, these mice exclusively express membrane bound TNF-α. Such mice are still susceptible to the development of a wide range of T lymphocyte mediated inflammatory diseases, including arthritis, hepatitis, and encephalitis, clearly indicating an important role of membrane bound TNF-α in the pathogenesis of this diseases.

It should also be noted that none of the currently available TACE inhibitors is entirely specific, and they also affect the function of other metalloproteases as well as cleavage of multiple human membrane expressed molecules, including both TNF receptors.^6,8 Normally, TNF receptors are cleaved during inflammation, and this has a TNF-α regulatory function because further TNF-α signalling is inhibited and because soluble TNF receptors retain the ability to bind and neutralise soluble TNF-α. In humans, TACE inhibitors indeed interfere not only with TNF-α production but also with the shedding of TNF-α receptors.⁶ This mechanism has been postulated to explain the paradoxical finding that treatment of experimental arthritis with a TACE inhibitor increased rather than inhibited the inflammatory response.⁹ Another study reported that a TACE inhibitor reduced the severity of TNBS induced colitis in mice but this was not associated with a reduction in TNF-α production, strongly suggesting that inhibition of metalloproteases other than TACE was the mechanism of action.¹⁰ Because several non-TACE metalloproteases have been strongly implicated in tissue damage and remodelling during inflammatory bowel disease,¹¹ this would still be an interesting therapeutic intervention but it should be noted that long term non-specific inhibition of metalloproteases could result in (ectopic) collagen deposition and fibrosis, and this may lead to stenosis.

In conclusion, TACE inhibitors block the release of TNF-α by mononuclear cells but not the membrane expressed form. Therefore, these compounds act in a very different manner than TNF-α binding molecules that also target membrane bound TNF-α, resulting in apoptosis of TNF-α producing cells.¹² Inhibition of TNF-α release by TACE inhibitors is expected to have anti-inflammatory effects, but in experimental models of T lymphocyte mediated diseases, expression of membrane bound TNF-α is sufficient for induction of disease. Finally, none of the currently available TACE inhibitors is entirely specific and inhibition of non-TACE metalloproteases may cause unexpected side effects. Gut 2002;51:5–6