Protein-digesting enzymes are kept on a tight leash to stop them from wantonly attacking targets. Two crystal structures show how an inhibitory protein domain gags one such enzyme without being chewed up itself.
All organisms have enzymes, known as proteinases, that break down other proteins, and have vital roles in such disparate processes as food digestion and blood clotting. These enzymes can't be allowed to run amok — if a protein has been mistakenly broken down, then a replacement has to be painstakingly re-synthesized from scratch. Several strategies have therefore evolved to ensure that proteinases are allowed to act only in appropriate circumstances. For example, inhibitor proteins block proteinase activity using various mechanisms1,2. In this issue, two crystallographic studies3,4 show that the protein inhibitor calpastatin uses a previously undiscovered mechanism to block the activity of the calpain proteinase. Two helical regions of calpastatin bind to widely separated sites on calpain; this reinforces the wrapping of an intervening, unstructured region of the inhibitor around calpain's active site.
Calpains are calcium-dependent proteinases found inside cells5. In most mammalian cells, the two main forms are µ-calpain (calpain 1) and m-calpain (calpain 2), each of which consists of two protein chains: a large catalytic subunit and a smaller regulatory subunit. It has been shown that m-calpain is required for embryonic development in mice6. Calpains also cut several structural proteins that regulate cell shape, disrupting their association with other proteins and so affecting the organization of the cell's protein 'skeleton' (the cytoskeleton). Studies on calpain-deficient cells show that calpains are important for remodelling the cytoskeletal structures that abut the cell membrane and regulate cell movement7. Furthermore, calpains contribute to several calcium-initiated cell-death programs.
Calpastatin inhibits both µ- and m-calpains, and is approximately the same size as the large calpain subunits. But unlike the proteinases, it is an intrinsically unstructured protein, adopting a defined structure only on binding to active calpain. Each calpastatin molecule contains four regions — the calpastatin inhibitory domains, CIDs I–IV — that bind calpains and block their activity (Fig. 1a). Each CID is, in turn, subdivided into three regions (A to C) that are predicted to interact with calpain. Because calpastatin molecules are largely unravelled in solution, they can potentially interact with four calpain molecules simultaneously, with one calpain at each CID.
The current papers3,4 reveal for the first time a complete picture of how calpastatin shuts down calpain activity. Hanna et al.3 (page 409) present the structure of calcium-bound m-calpain in complex with the CID-IV fragment of calpastatin, whereas Moldoveanu et al.4 (page 404) report the structure of calcium-bound m-calpain in complex with the CID-I fragment. The two studies show that the binding of calcium ions to m-calpain causes changes in the positioning of the four catalytic subunit domains DI–DIV, and of a regulatory subunit domain, DVI (Fig. 1b). Calcium binding thus generates interaction sites in the proteinase to which CIDs bind; specifically, a site created in DIV is bound by CID region A, and a site in DVI is bound by CID region C. In this way, CIDs bind simultaneously to two widely separated domains of calpain. This strong binding allows CID region B — a largely unstructured area between the helices of regions A and C — to make several contacts with the DI–DIII regions of calpain, thus blocking the proteinase's active site. These observations3,4 confirm and refine the findings of recent nuclear magnetic resonance studies8, which also indicated that calpastatin wraps around calpain as described above.
This inhibitory mode of action by calpastatin seems like a dangerous trick — deliberately inserting an unfolded protein into the active site of a proteinase is a bit like putting one's head in a lion's mouth and hoping that it won't be bitten off. But the crystal structures3,4 show that calpastatin has evolved a neat trick to avoid being cleaved by calpain: the short stretch of protein that would be expected to enter the active site actually loops away from the proteinase (Fig. 1b). Thus, calpastatin, although largely disordered on its own, is fine-tuned to generate a local structural motif that protects it from attack when attached to its calcium-bound substrate.
The new studies3,4 tell us as much about calpains as they do about calpastatin. Although the crystal structure of calcium-free (and thus inactive) m-calpain was solved several years ago9, crystallization of calcium-bound calpain has been problematic. Structural studies have thus relied on the extrapolation of results from engineered calpain fragments that contain only DI and DII (ref. 10). Calpastatin does not bind calcium, and has little affinity for calpains in the absence of calcium. It therefore seems highly likely that, as the authors of the studies suggest3,4, the structure of calcium-bound calpain in their calpastatin co-crystals is at least a close approximation of the active conformation of m-calpain.
The structures3,4 show that, on calcium binding, four specific arginine amino-acid side chains in the DIII domain of m-calpain contact the DII domain, stabilizing the catalytically active conformation. Moldoveanu et al.4 observe that mutations in m-calpain, in which the arginines are replaced with other amino acids, lower m-calpain's activity. These arginines are also found in a muscle-specific isoform of calpain known as calpain-3; mutation of the arginines in calpain-3 leads to loss of the proteinase's activity, an effect that has been associated with some cases of limb-girdle muscular dystrophy11.
More broadly, the studies3,4 provide further evidence for the involvement of m- and µ-calpains in continuous cellular processes. Calpastatin CIDs apparently evolved to remain intact as they inhibit calpain activity, even though resistance to cleavage is not necessary for this purpose3. But if cleaved and reused, CIDs would probably be less effective at blocking calpain activity than the intact inhibitors. It therefore seems that there is an evolutionary advantage in salvaging CIDs for future inhibition cycles, to avoid organisms using vital resources to make new molecules whenever the need arises. It is thus reasonable to conclude that calpains are activated within cells often enough to warrant this energy-saving measure.
Previous studies have shown that calpains can cleave calpastatin in the disordered regions between CIDs, even at relatively high concentrations of calpastatin12. The resulting calpastatin fragments are themselves calpain inhibitors. Some calpastatin is anchored to membranes through a terminal region (the L-domain, Fig. 1a), but the fragmentation mechanism could provide water-soluble calpain inhibitors that access different subcellular regions. Moldoveanu et al.4 show that region B of CID-I does not seem to inhibit µ-calpain well — could this explain how apparently inhibited calpain can still find a way to cleave the calpastatin to which it is bound? By damaging calpastatin, calpain could paradoxically sow the seeds of its own defeat, because, once formed, the calpastatin fragments have the potential to effectively shut down the proteinase's activity throughout the cell. With such fascinating nuances still to be explained, the calpain–calpastatin system certainly deserves continued exploration.
Borth, W. FASEB J. 6, 3345–3353 (1992).
Estrada, S., Olson, S. T., Raub-Segall, E. & Bjork, I. Protein Sci. 9, 2218–2224 (2000).
Hanna, R. A., Campbell, R. L. & Davies, P. L. Nature 456, 409–412 (2008).
Moldoveanu, T., Gehring, K. & Green, D. R. Nature 456, 404–408 (2008).
Goll, D. E., Thompson, V. F., Li, H., Wei, W. & Cong, J. Physiol. Rev. 83, 731–801 (2003).
Dutt, P. et al. BMC Dev. Biol. 6, doi:10.1186/1471-213x/6/3 (2006).
Franco, S. J. & Huttenlocher, A. J. Cell Sci. 118, 3829–3838 (2005).
Kiss, R. et al. FEBS Lett. 582, 2149–2154 (2008).
Hosfield, C. M., Elce, J. S., Davies, P. L. & Jia, Z. EMBO J. 18, 6880–6889 (1999).
Moldoveanu, T. et al. Cell 108, 649–660 (2002).
Jia, Z. et al. Biophys. J. 80, 2590–2596 (2001).
Mellgren, R. L., Mericle, M. T. & Lane, R. D. Arch. Biochem. Biophys. 246, 233–239 (1986).