Calpain Mutants with Increased Ca2+ Sensitivity and Implications for the Role of the C2-like Domain*

The ubiquitous calpain isoforms (μ- and m-calpain) are Ca2+-dependent cysteine proteases that require surprisingly high Ca2+concentrations for activation in vitro (∼50 and ∼300 μm, respectively). The molecular basis of such a high requirement for Ca2+ in vitro is not known. In this study, we substantially reduced the concentration of Ca2+ required for the activation of m-calpain in vitro through the specific disruption of interdomain interactions by structure-guided site-directed mutagenesis. Several interdomain electrostatic interactions involving lysine residues in domain II and acidic residues in the C2-like domain III were disrupted, and the effects of these mutations on activity and Ca2+sensitivity were analyzed. The mutation to serine of Glu-504, a residue that is conserved in both μ- and m-calpain and interacts most notably with Lys-234, reduced the in vitro Ca2+requirement for activity by almost 50%. The mutation of Lys-234 to serine or glutamic acid resulted in a similar reduction. These are the first reported cases in which point mutations have been able to reduce the Ca2+ requirement of calpain. The structures of the mutants in the absence of Ca2+ were shown by x-ray crystallography to be unchanged from the wild type, demonstrating that the increase in Ca2+ sensitivity was not attributable to conformational change prior to activation. The conservation of sequence between μ-calpain, m-calpain, and calpain 3 in this region suggests that the results can be extended to all of these isoforms. Whereas the primary Ca2+ binding is assumed to occur at EF-hands in domains IV and VI, these results show that domain II–domain III salt bridges are important in the process of the Ca2+-induced activation of calpain and that they influence the overall Ca2+ requirement of the enzyme.

The two ubiquitous calpains,and m-calpain, are cytosolic thiol proteases entirely dependent on Ca 2ϩ for their activity. These two calpains consist of a large or catalytic subunit (80 kDa) (from the genes capn1 and capn2, respectively) and a common small or regulatory subunit (28 kDa) (from capn4). Theand m-isoforms differ also in the concentration of Ca 2ϩ required for half-maximal activation in vitro. Both enzymes require a Ca 2ϩ concentration (ϳ50 and ϳ300 M forand m-calpain, respectively) that is significantly higher than that available in vivo (Ͻ1 M). Autolysis causes a drop in Ca 2ϩ requirement, but calpain activation in vivo must involve additional factors, such as membrane-binding and activator proteins. Although the physiological roles of these calpains remain unclear, there is much evidence suggesting that they contribute to many cellular processes, including signal transduction, apoptosis, cell cycle regulation, and cytoskeletal reorganization (1)(2)(3)(4)(5). Their physiological importance is exemplified by the recent demonstration that transgenic mice lacking the classical calpain isoforms die during embryonic development (6). Excessive proteolysis by these enzymes in response to altered Ca 2ϩ homeostasis has been observed in several neuropathological states, including Alzheimer's disease (7,8). With regard to other forms of calpain, defects in calpain-3 lead to the development of limb-girdle muscular dystrophy 2A (9), calpain-10 is linked to type II diabetes (10), and in Caenorhabditis elegans, the protease activity of the calpain homologue TRA-3 is required for sex determination (11,12).
The x-ray structures of rat (13) and human (14) m-calpain in the absence of Ca 2ϩ show that the large subunit consists of a short ␣-helix at the N terminus; domains I and II, which constitute the protease function; domain III, which resembles a C 2 -domain; and domain IV, which contains several EF-hands. The small subunit contains domain V, which is glycine-and proline-rich and was either not present (13) or not detectable (14) in the x-ray structure, and domain VI, which contains several EF-hands and is structurally very similar to domain IV. Importantly, the structure revealed that the protease active site is not assembled in the absence of Ca 2ϩ , a feature not previously observed in any other cysteine protease. The protease domains (domains I and II) are held apart in the Ca 2ϩ -free conformation by interactions within the molecule, which prevent the catalytic triad residues (Cys-105 in domain I and His-262 and Asn-286 in domain II) from assuming the correct conformation to hydrolyze substrates. Activation by Ca 2ϩ must somehow relieve these constraints, permitting domains I and II to move toward each other and form a competent active site. A Ca 2ϩ -bound structure of calpain might answer several questions concerning the Ca 2ϩ -induced activation mechanism, but this structure has not yet been determined because of significant technical difficulties.
Based on the primary sequence alone, it was for a long time assumed that the EF-hand-containing domains IV and VI were the major determinants of the Ca 2ϩ requirement of calpain. Recent studies have shown that Ca 2ϩ binding at EF-hand 3 makes the largest contribution to calpain activation, EF-hand 2 may make some contribution, and the other EF-hands apparently have no direct role in Ca 2ϩ regulation (15). Additionally, it has recently been demonstrated that TRA-3 has Ca 2ϩ -dependent protease activity although it lacks the EF-hand domain (12). Thus it is likely that the observed Ca 2ϩ requirement * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  of the whole enzyme is not determined solely by the EF-hands and is greatly affected by interactions elsewhere in the molecule.
The x-ray structures revealed two interesting features in domain III that may be very important in this regard (Fig. 1A). First, domain III was found to be structurally similar to the Ca 2ϩ -dependent C 2 domains, which are known to promote the phospholipid binding of C 2 -containing proteins (16,17). This is consistent with the following two considerations: 1) calpains are thought to translocate to the cell membrane when activated by Ca 2ϩ and 2) the Ca 2ϩ requirement of calpain is greatly reduced in the presence of phospholipids, which appears to be a part of the in vivo activation mechanism of calpain. Second, at the interface between domains II and III, there is a set of electrostatic interactions involving a remarkably acidic loop composed of Glu-392-Asp-400 and Glu-504 in domain III and clustered lysine residues Lys-226, Lys-230, and Lys-234 in domain II (Fig. 1B). All these residues are highly conserved within the known m-calpain sequences. We hypothesize therefore that this salt bridge region exerts a conformational constraint on the movement of domain II and therefore on the assembly of the active site, which will tend to elevate the Ca 2ϩ requirement of the enzyme. By disrupting these salt bridges, the mutations described here were designed to investigate the effects on the Ca 2ϩ requirement of calpain. Our experiments would further address the question of whether these electro-static interactions directly contribute to maintaining the inactive conformation of calpain.
The expression and purification of wild-type rat m-calpain and mutant enzymes were performed as previously described (18). Coexpression of the constructs pET-24-m-80k-CHis 6 (encoding the m-calpain 80-kDa subunit with a C-terminal histidine tag) and pACpET-21k (encoding the C-terminal 184 residues of the regulatory subunit, referred to as domain VI or the 21-kDa subunit) in Escherichia coli strain BL21(DE3) gives rise to an active heterodimer. Enzymes were purified through several columns and finally concentrated by centrifugation using a BioMax 30-kDa molecular mass exclusion device to ϳ10 mg/ml in a buffer containing 10 mM dithiothreitol, 50 mM Tris-HCl, pH 7.6, 100 mM NaCl, and 200 M EDTA. Small aliquots of the enzyme samples (typically 50 l) were flash frozen in liquid nitrogen and stored at Ϫ70°C.
Titration of Ca 2ϩ Requirement-For the measurement of the Ca 2ϩ dependence of calpain activity, a modification of the standard casein assay was used. The duplicate assays contained 4 mg/ml casein, 0.2 M NaCl, 10 mM 2-mercaptoethanol, and 50 mM Tris-HCl, pH 7.6, in a final volume of 100 l. Net final CaCl 2 concentrations ranged from 0 M to 5.0 mM. The reaction was initiated with 4 l of enzyme sample (see below), and the mixtures were incubated at 25°C for 30 min before the reaction was terminated by the addition of 70 l of ice-cold 10% trichloroacetic acid. The resultant mixture was placed on ice for 10 min and centrifuged at 15,000 rpm for 15 min, and the absorbance values of the supernatants were recorded at 280 nm. Immediately prior to the Ca 2ϩ titration, the enzyme aliquots were freshly thawed from Ϫ70°C and diluted in 50 mM Tris-HCl, pH 7.6, to a final enzyme concentration of 0.2 mg/ml (wild type, E504S, K226S, K230S, K234S) or 2 mg/ml (K230E, K234E). The Ca 2ϩ concentration required for half-maximal activity with casein as substrate is given as [Ca 2ϩ ] 0.5 . This value was calculated by fitting the normalized activity data to the Hill equation y ϭ x n /(k n ϩ x n ) where y is the fraction of maximum activity, k ϭ [Ca 2ϩ ] 0.5 , n is the Hill constant, and x is [Ca 2ϩ ].
Crystallization, Diffraction Data Collection, and Processing-Crystals of the m-calpain mutant enzymes were grown in conditions very similar to those for wild-type m-calpain (19). X-ray data from the K230E mutant were collected at 1.54 Å at an in-house facility that consists of an MAR Research imaging plate and a Rigaku RU-200 x-ray generator operated at 50 kV and 100 mA. Data for the E504S mutant were collected at beamline A-1 at the Cornell High Energy Synchrotron Source, using monochromatic radiation at 0.925 Å and a Quantum-4 CCD camera from Area Detector Systems Corporation (San Diego, CA). All crystals were frozen in liquid propane, and data sets were collected at 100 K. Raw x-ray data were processed using the HKL program suite (20). Both mutants and wild-type m-calpain crystallized isomorphously in space group P2 1 . Cell dimensions for the K230E mutant were a ϭ 51.62 Å, b ϭ 156.13 Å, c ϭ 64.00 Å, ␤ ϭ 95.21°whereas the dimensions for the E504S mutants were a ϭ 51.82 Å, b ϭ 157.80 Å, c ϭ 64.49 Å, ␤ ϭ 95.18°.
Structure Solution and Refinement-Because the mutant crystals were virtually isomorphous to the wild-type protein, we were able to incorporate the coordinates of wild-type m-calpain (PDB code 1DF0) directly for the initial refinement of the mutant enzymes. All refinement was carried out using the CNS package (21) with 10% of the reflections excluded from the refinement for the calculation of R free . The initial R factor for the K230E mutant was 0.366, which dropped to 0.260 (R free ϭ 0.321) after refinement at 2.8 Å. The initial R factor for the E504S mutant was 0.372, which was reduced to 0.258 (R free ϭ 0.302) after refinement at 2.4 Å. The final models and the calculated ͉2F obs Ϫ F calc ͉ electron density maps were analyzed using the XFIT program (22). matographic columns to ϳ90 -99% purity based on Coomassie staining of SDS-polyacrylamide electrophoresis gel, depending on the mutant (Fig. 2).

Expression and
Effect of Mutations on Ca 2ϩ Requirement and Specific Activity-Enzymes were assayed for their ability to hydrolyze casein at varying concentrations of Ca 2ϩ . The calculated [Ca 2ϩ ] 0.5 served as the basis for the comparison of the Ca 2ϩ requirement of wild-type and mutant enzymes in this study. In the conditions employed here, the [Ca 2ϩ ] 0.5 for wild-type m-calpain was 242 Ϯ 6 M (Fig. 3), which is in general agreement with previous reports (1)(2)(3)(4)(5). Introducing the mutation E504S in domain III had the most dramatic effect, reducing the [Ca 2ϩ ] 0.5 to 129 Ϯ 1 M, corresponding to a 47% reduction in the Ca 2ϩ requirement compared with wild-type m-calpain. The effects of mutating each of the three lysine residues (Lys-226, Lys-230, and Lys-234) individually to serine had varying results. Mutating Lys-226 or Lys-230 had minor effects on the Ca 2ϩ sensitivity of the enzyme (Fig. 3, Table I), whereas the mutation of Lys-234 reduced the [Ca 2ϩ ] 0.5 to 183 Ϯ 1 M (a 24% reduction in the Ca 2ϩ requirement). Further, the mutation of Lys-234 to glutamic acid resulted in a reduction in the [Ca 2ϩ ] 0.5 to 159 Ϯ 3 M, corresponding to a 34% decrease. Interestingly, the K230E and K234E mutations decreased the specific activity of the enzyme to ϳ16% compared with the wild type, whereas the K230S and K234S mutations had no significant effect on specific activity.
X-ray Crystallography-To ascertain whether mutations had caused unexpected structural changes that might influence the Ca 2ϩ sensitivity, we attempted structure determination of these mutants by x-ray crystallography. We were successful in crystallizing K226S, K230E, K234S, and E504S mutants in the absence of Ca 2ϩ in conditions very similar to those for wild-type rat m-calpain (19). To date, we have solved the structures of the K230E and E504S mutants, from which larger crystals suitable for diffraction were obtained. The refined structures of these mutants are virtually indistinguishable from wild-type calpain (root mean square deviation of 0.47 and 0.49 Å on ␣-carbon atoms for the E504S mutant and the K230E mutant, respectively), even in the vicinity of the mutations. Because these two structures were almost identical to the wild type, we did not proceed with the mutants that only gave rise to weakly diffracting crystals and assumed that they would have minimal, if any, conformational changes as well. DISCUSSION The mechanism of calpain activation by Ca 2ϩ is a fundamental biochemical question that has remained poorly understood. A particularly interesting feature of these enzymes is that their in vitro requirements for Ca 2ϩ are orders of magnitude higher than typical Ca 2ϩ concentrations found in the cell. This char-acteristic is not shared by similar members of the EF-hand family, such as calmodulin, which have in vitro Ca 2ϩ requirements within the physiological range. Efforts to convert calpain EF-hand sequences to more canonical forms have been unsuccessful in reducing the Ca 2ϩ requirement (15), suggesting that other structural features contribute to calpain's requirement for Ca 2ϩ . The x-ray structure of Ca 2ϩ -free rat m-calpain revealed several features that appear to maintain the inactive conformation of the enzyme. Of particular interest is a loop of nine acidic residues (Glu-392-Asp-400) in domain III, together with Glu-504, which make several electrostatic contacts with three lysine residues (Lys-226, Lys-230, Lys-234) on one ␣-helix in domain II. The structure of human m-calpain showed an additional interaction between Glu-504 and Lys-354. It should be noted particularly that Glu-504 would have the strongest contact with Lys-234. Structures are not yet available for -calpain and calpain 3, but these residues are almost all conserved in both -calpain and calpain 3, so these domain II-domain III interactions are likely to be present in all three calpains. It therefore seemed attractive to propose that this set of salt bridges might affect the ability of domain II to approach domain I in the process of assembling the active site, leading to the activation of calpain.
The K234S and E504S mutations both significantly reduced the Ca 2ϩ requirement without affecting the specific activity of the enzyme. The reduction caused by the E504S mutation was  greater than that caused by the K234S mutation. This suggests that Glu-504 makes multiple electrostatic contacts and probably reflects the fact that the E504S mutation abolishes the salt links to Lys-234 and Lys-354. The K234S mutation abolishes only one of these salt links. The crystal structure of E504S is virtually identical to that of the wild-type enzyme in the inactive state, demonstrating that the reduction in the Ca 2ϩ requirement observed in E504S is not a result of the disruption of the inactive conformation but must involve the facilitation of domain movement during Ca 2ϩ activation. Another interesting result was observed with the K234E mutation, which reduced the Ca 2ϩ requirement of the enzyme even further than the K234S mutation. This mutation also reduced the specific activity of the enzyme, illustrating the sensitivity of this region to changes in the electrostatic potential. This mutant was expected to exert a repulsive interaction with Glu-504. Surprisingly, the results of this work showed that the electrostatic interactions of Lys-226 and Lys-230 with the acidic Glu-392-Asp-400 loop do not appear to be critical to the activation of calpain by Ca 2ϩ because the K226S and K230S mutations did not affect either the Ca 2ϩ requirement of the enzyme or its specific activity. The K230E mutation, which is expected to introduce a strong repulsion between this position and the acidic loop, did not affect the Ca 2ϩ requirement of the enzyme. It did, however, greatly reduce the specific activity of the enzyme, for reasons that we cannot presently explain. The crystal structure of K230E in the absence of Ca 2ϩ was identical to that of wild-type m-calpain, showing that the mutation did not affect the domain II-domain III geometry in the resting state of the enzyme. Because it has not so far been possible to crystallize calpain in the presence of Ca 2ϩ , the effects of these mutations on the activated state of the enzyme cannot be directly observed.
Of the many mutations that have been made in m-calpain, the Lys-234 and Glu-504 mutations described here are the only examples so far described that result in a lowering of the Ca 2ϩ requirement. The question therefore arises of how a loss of salt linkages around Glu-504 can have a major effect whereas alteration at Lys-226 and Lys-230 has very little effect. The residue Glu-504 is at one end of the "transducer arm" or linker peptide that connects domain IV to domain III, and this arm is assumed to communicate Ca 2ϩ -induced conformation changes in domain IV to the rest of the molecule. The evidence therefore suggests that the loss of the salt bridges at Glu-504 makes the movement of domain II toward domain I "easier," i.e. permits it to occur at a lower Ca 2ϩ concentration. As can be seen in Fig. 1, Glu-504 is "strategically" located at the top right end of the salt bridge series. The movement of domain II in the assembly of the active site, according to our modeling studies guided by the many well established thiol protease structures, would be based on a pivotal point in the Glu-504 region. Therefore, given that the Glu-504 and Lys-234 mutations weaken the interaction of domain II with domain III, we would suggest that domain II does in fact move apart from domain III during the activation process. Whether domain III undergoes additional conformational changes remains unknown in the absence of a Ca 2ϩ -bound structure. Illustrating the importance of this interaction in the calpain family, the corresponding mutation in -calpain, E515S, also lowers the Ca 2ϩ requirement (data not shown).
Recalling the resemblance of domain III to C 2 domains, it should also be noted that Glu-504 and the Glu-392-Asp-400 loop are in positions very similar to the loops in C 2 domains that participate in Ca 2ϩ -dependent phospholipid binding. The Ca 2ϩ requirement of calpain is greatly reduced in the presence of some phospholipids in vitro, and this is assumed to reflect membrane binding in vivo. We speculate therefore that Ca 2ϩ and phospholipid binding to this region of calpain could disrupt these critical salt links, thereby releasing some constraints on the movement of domain II and lowering the Ca 2ϩ requirement of the enzyme. Coordination of Ca 2ϩ in the acidic loop by the side chains of the acidic residues could break most or even all of the electrostatic interactions with the basic residues in domain II. With the exception of pivotal Glu-504, perhaps breaking one or two such salt links by point mutation may not be enough to completely relieve the conformational constraints to permit the movement of domain II. Given that a Ca 2ϩ -bound structure may not be available in the foreseeable future, experiments of the type employed here will be required to more fully understand the mechanism of calpain activation by Ca 2ϩ . In this work, we have for the first time generated calpain mutants that are responsive to significantly lower concentrations of Ca 2ϩ for activation. Further, we have identified a specific structural feature remote from the EF-hand domains that affects the Ca 2ϩ sensitivity and activation of calpains. In light of the sequence conservation displayed in this region, it is clear that these key interdomain interactions are important factors contributing to the overall Ca 2ϩ requirement of the calpains.