Drug Insight: Tumor Necrosis Factor Converting Enzyme as a Pharmaceutical Target for Rheumatoid Arthritis

Marcia L Moss; Liora Sklair-Tavron; Raphael Nudelman


Nat Clin Pract Rheumatol. 2008;4(6):300 

In This Article

Tace as a Therapeutic Target For Rheumatoid Arthritis

Inhibitors (rather than antagonists) that are orally available and are considered as anti-TNF therapies fall into two main classes: those that prevent the production of TNF and those that stop the release of soluble TNF from immune-competent cells. TACE, the enzyme that processes precursor TNF in order to release soluble TNF, is a member of the ADAM (a disintegrin and metalloproteinase) family and is known as ADAM17.[14,15] TACE was purified on the basis of its capacity to process precursor TNF, and cells taken from TACE-deficient mice in which the zinc-binding motif of TACE is mutated are unable to process precursor TNF.[14] TACE is also required for epidermal growth factor signaling, as TACE-deficient mice have a phenotype similar to that of mice that lack transforming growth factor α and of mice that lack epidermal growth factor receptor.[16] TACE-deficient mice exhibit open eyes and wavy hair at birth, which is reminiscent of the phenotype of transforming growth factor α knockout mice. Mice lacking TACE also show epithelial defects in the lung bronchioli that are similar to those in mice that lack the epidermal growth factor receptor. Given the characteristics of the TACE knockout phenotype, investigators are actively seeking inhibitors of TACE that could potentially treat cancer ( Box 1 ). These inhibitors might be more beneficial for the treatment of cancer than for RA because of possible adverse effects -- since its discovery, TACE has been reported as a processing enzyme or 'sheddase' for more than 30 type I and II membrane-spanning proteins ( Table 1 ).

Early preclinical studies indicated that inhibition of TACE is beneficial for patients with arthritis, even though blocking TACE activity does not neutralize the effects of membrane-bound precursor TNF. Two dual MMP-TACE inhibitors, GW3333 (compound 1 in Table 2 , Figure 2) from GlaxoSmithKline (GSK)[17] and TMI-1 (compound 2 in Table 2 , Figure 2) from Wyeth,[18] proved to be efficacious in various animal models ranging from arthritis induced by collagen (collagen-induced arthritis [CIA]) and peptidoglycan-polysaccharide polymer to the more severe adjuvant-induced arthritis.

Structural Representation of Preclinical and Clinical Lead Compounds For TACE Inhibition.
The available structures for compounds detailed in Table 2 are outlined. The structures for compounds 12-16 in Table 2 are unavailable. Abbreviation: TACE, TNF-converting enzyme.

More-selective inhibitors by Bristol-Myers Squibb (BMS, previously Dupont-Merck; BMS-561392; compound 3 in Table 2 , Figure 2)[19] and Wyeth (TMI-2; compound 4 in Table 2 , Figure 2)[20] were also effective in the mouse CIA model, providing preclinical validation of TACE as a target for RA therapies. In the CIA model, paw swelling and erythema are measured on a scale of 0-4. In the BMS study, mice with CIA were given four monoclonal antibodies against chicken type II collagen on day 1 followed by lipopolysaccharide (LPS) on day 2. A dose of 10.5 mg/kg BMS-561392 twice daily worked as well as etanercept, reducing the clinical score per paw after 12 days from 1.5 to 0.4. BMS-561392 at 2.8 mg/kg was also more effective than infliximab, supporting the hypothesis that TACE inhibition would be beneficial for patients with RA. In a study of the efficacy of TMI-2 in the CIA model, type II collagen in incomplete Freund's adjuvant was administered on days 0 and 21, and compound was given twice daily at 100 mg/kg when 10% of the mice developed signs of arthritis. Paw swelling was reduced from 1.28 to 0.5.

Recent clinical studies, however, have failed to show that TACE inhibition is an effective way of treating RA. For example, in the first phase II study from BMS, there were indications that TACE inhibition induced mechanism-based toxic effects in the liver.[21] The rationale for designating this toxicity as mechanism-based stems from the hypothesis that TACE acts not only on precursor TNF, but also on other membrane-bound proteins, such as TNF receptor I (TNFRI) and TNFRII (Figure 3).[16] A consequent build-up on the cell surface of precursor TNF and of TNFRI and TNFRII would make cells supersensitive to the effects of TNF. In a proof-of-concept experiment, rats were treated with BMS-561392 (a partially selective inhibitor)[19,21] or BMS-566394 (a selective TACE inhibitor; compound 5 in Table 2 , Figure 2),[21] both of which demonstrated hepatotoxicity. Livers from rats treated with BMS-561392 showed a 10-fold greater accumulation of precursor TNF than those of controls and a 3-fold greater increase in TNFRII. No data were presented on the effects of the selective TACE inhibitor BMS-566394.

Processing of TNFRI and TNFRII by TACE and Other ADAM-family Members.
TACE can process TNFRI and TNFRII to generate the soluble receptors sTNFRI and sTNFRII -- these soluble forms help to neutralize the effect of TNF. TACE seems to be the major convertase for precursor TNF and for TNFRII, whereas unidentified ADAMs also process TNFRI. When TACE is inhibited, membrane-bound forms of TNF, TNFRI and TNFRII can potentially build up on the cell surface, leading to sensitization of the cells to precursor TNF and the potential for liver toxicity. Abbreviations: ADAM, a disintegrin and metalloproteinase; proTNF, precursor TNF; sTNF, soluble TNF; sTNFRI, soluble TNF receptor I; sTNFRII, soluble TNF receptor II; TACE, TNF-converting enzyme; TNF, tumor necrosis factor; TNFRI, TNF receptor I; TNFRI, TNF receptor II.

Arguments against mechanism-based toxicity of TACE inhibitors arise from the fact that, although TACE is an important physiological metalloproteinase for processing precursor TNF, other members of the ADAM family have been implicated in the processing of TNF receptors.[22,23] In one phase I study by GSK, the levels of soluble TNF, TNFRI and TNFRII in response to LPS administration were measured before and after treatment with the metalloproteinase inhibitor GI5402.[24] GI5402 did not affect the increase in monocyte expression of precursor TNF or the decrease in granulocyte-membrane expression of TNFRI and TNFRII; however, it did decrease the plasma levels of soluble TNF, TNFRI and TNFRII. Dupont-Merck carried out a similar study using BMS-561392 with LPS in humans and found that precursor TNF and TNFRI and TNFRII transiently accumulated on the cell surface of monocytes and T lymphocytes.[22] Most of the unprocessed precursor TNF was degraded after internalization. These two studies raise some doubt as to whether TACE inhibition will always lead to toxic effects in the liver.

The results of the GSK and Dupont-Merck studies are, however, challenged by evidence that TACE is also the principal sheddase for TNFRII as well as for TNF.[16] In TACE knockout mice, the level of circulating TNFRII in the serum is over 90% lower than that in wild-type mice, whereas TNFRI levels are approximately 50% lower.[25] For many shedding processes, the ADAM-family members that seem to be active during constitutive release of membrane proteins differ from those that are active during induced release. In addition, the use of particular family members might depend on the cell type. The release of membrane proteins by ADAM proteins can, therefore, be quite complicated, and often multiple enzymes are involved.

In a phase II study of the dual MMP-TACE inhibitor TMI-005 (compound 6 in Table 2 , Figure 2), also known as apratastat, no hepatotoxicity was found, but the efficacy results were discouraging.[26] This was a 12-week, randomized placebo-controlled, phase II study, in which 390 patients with RA were treated with both methotrexate and TMI-005 (given orally three times daily at 50, 100 or 150 mg). There were seven adverse events in the treated group compared with none in the placebo group; however, none of the participants developed an increase in the levels of hepatic aminotransferases, ruling out hepatotoxicity. In addition, the treated patients showed no improvement in symptoms relative to patients in the placebo group. The lack of efficacy of TMI-005 might partly result from basal signaling via TNF receptors by membrane-bound precursor TNF on immunological cells. Dosing was probably adequate, as this trial followed protocols established in a phase I study that used LPS in human individuals to determine the levels of TMI-005 that were needed in the blood to inhibit TNF release from immune-competent cells. On the basis of the phase II results, the development of TMI-005 by Wyeth was terminated. Although this study showed that dual inhibition of TACE and MMPs was not beneficial for the treatment of RA, it did provide impetus for the use of TACE inhibitors for other indications. The development of TMI-005 for other indications is unlikely, however, as cases of tendonitis were reported. This condition is reminiscent of that occurring in cancer patients treated with MMP inhibitors such as marimastat.[27] This adverse effect probably occurs because TMI-005 is a broad-spectrum MMP inhibitor.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.