阿斯匹林抗癌机理被揭示

People with cylindromatosis (also known as turban tumour syndrome) are predisposed to benign tumours that arise in hair follicles and in cells of the sweat and scent glands. The syndrome is caused by a variety of mutations in the CYLD gene, found on chromosome 16 (refs 4, 5). As with all classical tumour suppressors, both copies of the gene must be inactivated to produce cylindromatosis, usually by a mutation in one copy and a deletion of the region of the chromosome carrying the other copy.

But how does the loss of CYLD cause this tumour syndrome? Many cancers exhibit disturbances in signalling pathways that activate the NF-B protein, probably related to this protein's role in regulating genes that control inflammation, the immune response, or cell death⁶. These pathways are themselves regulated at several steps by polyubiquitination — the covalent attachment of a target protein to chains of the small protein ubiquitin⁷. Such chains can have different effects, depending on how they are formed. When the ubiquitins are linked to each other through the lysine amino acid found at position 48 of each ubiquitin, the target protein is directed to the cellular waste-disposal unit, the proteasome⁸. If lysine 63 is used instead, it can serve as a signal for the target to assemble with other proteins^{9, 10}.

For various reasons^{5, 11}, CYLD has been suspected of being a deubiquitinating enzyme — an enzyme that removes ubiquitin from target proteins. Might it therefore contribute to the NF-B pathway? The new papers^1-3 show that it does. Brummelkamp et al.¹ used a panel of small interfering RNAs to reduce the expression of CYLD and other deubiquitinating enzymes in cultured cells. They found that blocking CYLD expression resulted in the activation of signalling from tumour-necrosis factor- (TNF-) through NF-B. They also demonstrated a direct interaction between CYLD and NEMO, a component of the NF-B pathway. In independent studies, Kovalenko et al.² and Trompouki et al.³ also identified CYLD as a protein that interacts with NEMO.

So how does NEMO fit into the picture? NF-B is normally found in the cell cytoplasm, where it is kept inactive by association with inhibitor (IB) proteins. Upon stimulation by proteins such as TNF- or interleukins, a family of TNF receptors signals to intracellular adapter proteins known as TRAFs (Fig. 1). These proteins have a variety of effects, one of which is to activate the IB kinase (IKK) complex. IKK in turn adds phosphate groups to (phosphorylates) IB, leading to its polyubiquitination and degradation. NF-B is then released from its captor and is free to move into the nucleus, where it regulates a variety of genes. NEMO is one component of the IKK complex, and Kovalenko et al.² and Trompouki et al.³ show that it binds to the middle of CYLD. Kovalenko et al. also found that another part of CYLD can bind to TRAF2. So, it seems likely that TRAF2, NEMO and CYLD are all components of a signalling complex formed in response to TNF- (although it is not clear whether they can all be part of the complex at the same time).

Figure 1 The well-characterized NF-B signalling pathway, and a new contributor.   Full legend   High resolution image and legend (70k)

What does CYLD do there? The new papers^1-3 show that signalling through the NF-B pathway increases when CYLD activity is lost, whether as a result of reducing the protein's concentration with small interfering RNAs, transfecting cells with forms of the protein that bear mutations seen in cylindromatosis patients, or transfecting cells with engineered inactive CYLD. Overexpressing the protein has the converse effect. The findings imply that this protein usually downregulates signalling, at a step upstream of IKK activation — possibly at the point where TRAFs and NEMO interact.

So what are CYLD's substrates? The papers show that CYLD can deubiquitinate TRAF2 in vivo^1-3 and in vitro^{2, 3}. It is known that TRAF2 can be connected to ubiquitin chains linked through either lysine 48 or lysine 63; the former leads to its degradation¹² and the latter (which occurs by self-ubiquitination) seems to lead to the assembly of a multi-protein complex that activates IKK, and hence NF-B. Kovalenko et al. now show that the substrates for CYLD in vivo are not lysine-48-linked chains, while Trompouki et al. find that the formation of most of the ubiquitinated TRAF2 in the cells studied requires an enzyme of the lysine-63-ubiquitination pathway.

The likely model, then, is that — by removing lysine-63-linked ubiquitin chains from TRAF2 — CYLD causes the dissociation of a complex that would otherwise activate NF-B signalling. In the absence of CYLD, as in people with cylindromatosis, NF-B signalling is more persistent than normal, contributing to aberrant cell growth. Many other related proteins have deubiquitinating activity^{11, 13}, including a protein that is already known to inhibit NF-B signalling¹⁴. It is plausible that all these proteins perform the same task in different tissues.

But how could persistent activation of the NF-B pathway, which is best known for its role in inflammation, lead to tumours? One hint comes from the observation that activation of the pathway prevents cell death by apoptosis¹⁵. Apoptosis kills cells that have accumulated mutations or other damage — damage that might otherwise lead to cancer. Accordingly, Brummelkamp et al. found that a loss of CYLD decreased apoptosis, suggesting that cells lacking this protein could be more susceptible to the accumulation of potentially cancer-predisposing genetic alterations.

Of course questions remain, about both the NF-B pathway and the new findings. How, for instance, does the lysine-63-linked ubiquitination of TRAFs lead to the activation of IKK? What is the interplay between the direct and indirect effects of NF-B on cell growth and apoptosis? And can we use this knowledge to develop drugs that ameliorate cylindromatosis and other cancers? Velcade, a proteasome inhibitor that blocks NF-B activity, has been approved by the US Food and Drug Administration for the treatment of multiple myeloma¹⁵. Moreover, Brummelkamp et al.¹ have shown that aspirin and prostaglandin A1 — anti-inflammatory drugs with effects on the NF-B pathway — prevent the activation of NF-B-responsive genes and the inhibition of apoptosis caused by the loss of CYLD. Much remains to be investigated, but this finding points to the possibility of a simple, yet potentially important, treatment for cylindromatosis.