Distinguishing between calpain heterodimerization and homodimerization Ravikiran Ravulapalli 1 , Robert L. Campbell 1 , Sherry Y. Gauthier 1 , Sirano Dhe-Paganon 2 and Peter L. Davies 1 1 Department of Biochemistry, Queen’s University, Kingston, Canada 2 Structural Genomics Consortium and the Department of Physiology, University of Toronto, Canada Calpains are a family of intracellular cysteine prote- ases. They are Ca 2+ -dependent and function by modu- lating the biological activities of target proteins through selective cleavage [1]. Genome sequencing projects have revealed numerous calpain isoforms in vertebrates, invertebrates, plants, microorganisms and, recently, in kinetoplastid parasites [1–7]. In the human genome, 14 different calpain isoforms have been identi- fied to date. Several calpain isoforms are ubiquitously expressed, whereas many demonstrate tissue-specific expression patterns [8]. Although their precise func- tions are poorly understood, calpains are implicated in many intracellular processes linked to calcium signal- ing, such as cell motility, apoptosis, and cell cycle reg- ulation, as well as cell-type-specific functions, such as cell fusion in myoblasts and long-term potentiation in neurons [9–12]. Several pathologic conditions (ischemic injury, Alzheimer’s disease, limb-girdle muscular dystrophy 2A, type II diabetes mellitus, gastric cancer, etc.) have been associated with disturbances of the Keywords calcium; calpain; dimerization; EF-hand protease Correspondence P. L. Davies, Department of Biochemistry, Queen’s University, Kingston, ON K7L 3N6, Canada Fax: +1 613 533 2497 Tel: +1 613 533 2983 E-mail: daviesp@queensu.ca (Received 28 August 2008, revised 13 November 2008, accepted 4 December 2008) doi:10.1111/j.1742-4658.2008.06833.x The two main mammalian calpains, 1 and 2, are heterodimers of a large 80 kDa and a small 28 kDa subunit that together bind multiple calcium ions during enzyme activation. The main contact between the two subunits of these intracellular cysteine proteases is through a pairing of the fifth EF-hand of their C-terminal penta-EF-hand (PEF) domains. From model- ing studies and observation of crystal structures, it is not obvious why these calpains form heterodimers with the small subunit rather than homodimers of the large subunit, as suggested for calpain 3 (p94). There- fore, we have used a differential tagging system to determine which of the other PEF domain-containing calpains form heterodimers and which form homodimers. His6-tagged PEF domains of calpains 1, 3, 9 and 13 were coexpressed with the PEF domain of the small subunit that had been tagged with an antifreeze protein. As predicted, the PEF domain of cal- pain 1 heterodimerized and that of calpain 3 formed a homodimer. The PEF domain of digestive tract-specific calpain 9 heterodimerized with the small subunit, and that of calpain 13, prevalent in lung and testis, was mainly found as a homodimer with a small amount of heterodimer. These results indicate whether recombinant production of a particular calpain requires coexpression of the small subunit, and whether this calpain is likely to be active in a small subunit knockout mouse. Furthermore, as the endogenous inhibitor calpastatin binds to PEF domains on the large and small subunit, it is less likely that the homodimeric calpains 3 and 13 with two active sites will bind or be silenced by calpastatin. Abbreviations AFP, antifreeze protein; PEF, penta-EF-hand. FEBS Journal 276 (2009) 973–982 ª 2008 The Authors Journal compilation ª 2008 FEBS 973 calpain system [13–18]. Therefore, elucidating the spe- cific role of calpains in these pathologies may facilitate treatment of these diseases. The ubiquitous and well-characterized members of the family, calpains 1 and 2 (l-isoform and m-isoform, respectively), are heterodimers, containing a large 80 kDa subunit (domains I–IV) and a small 28 kDa subunit (domains V and VI) [19–21]. Both enzymes share a papain-like protease core (domains I and II) characterized by the presence of the catalytic triad resi- dues cysteine, histidine and asparagine. Domains III and IV are the C2-like and penta-EF-hand (PEF) domains, respectively. The PEF domain (IV) of the large subunit pairs with the homologous PEF domain VI of the small subunit through EF-hand 5, thus forming a heterodimer. In the absence of Ca 2+ , both isoforms are catalytically inactive, and upon binding Ca 2+ , the heterodimer undergoes multiple structural changes to form the active calpain enzyme. Structural events, such as autoproteolysis, subunit dis- sociation, intradomain ⁄ interdomain rearrangement and phospholipid binding, are suggested to be involved in this complex regulation of activation [22–25]. Five of the human calpains (calpains 5, 6, 7, 10 and 15) have significantly different domain compositions from those of the conventional calpain large subunit, suggestive of distinct functions [25–29]. In particular, they lack a PEF domain with which to dimerize, and are presumed to be monomers. The other members of the calpain family (calpains 3, 8, 9, 11, 12 and 13) do have a PEF domain (domain IV). Considering their similarity in domain arrangement to the classic cal- pains 1 and 2, these isoforms have the potential to form heterodimers with the small subunit. However, recent biophysical studies on the recombinant PEF domain of calpain 3 showed that it forms a very stable homodimer [30]. Molecular modeling demonstrated that this interaction could be the basis for homodimer- ization of the whole enzyme. A 180 kDa protein was formed by recombinant expression of inactive cal- pain 3 in the absence of the small subunit, which is consistent with homodimerization [31]. The situation with native calpain 3 (p94) is unclear, because the enzyme is unstable and rapidly autoproteolysed during purification, but the small subunit does not seem to copurify with the 94 kDa large subunit. Thus, it can- not be assumed that the presence of a C-terminal PEF domain in other calpain isoforms will lead to heterodi- merization with the small subunit. One of the reasons why it is important to establish which calpains form heterodimers is that calpastatin, the natural inhibitor of calpains 1 and 2 [32], binds to sites on the PEF domains of both the large and small subunits [33,34]. In the presence of Ca 2+ , subdomains A and C of calpastatin tightly associate with PEF domain IV of the large subunit and domain VI of the small subunit, respectively. This binding ensures a high local concen- tration of subdomain B that binds and blocks the active site cleft of the enzyme. In the absence of the small subunit, calpastatin would lose one of its binding sites and might not associate tightly enough with the large subunit to inhibit it. More to the point, a homodimer of the large subunit would have two active sites at opposite ends of the molecule, and these certainly could not both be inhibited by one calpastatin inhibitory domain. In this context, we sought to examine all known PEF domains from human calpain isoforms, including calpain 3, to establish whether they exist as heterodimers or homodimers. In order to screen these PEF domains, a coexpres- sion system with differential tags on the recombinant proteins was established. The human small subunit lacking the glycine-rich domain (21 kDa) was tagged with type III antifreeze protein (AFP) (7 kDa) [35] in its place on the N-terminus, whereas the recombinant domain IVs of other calpain isoforms had a His6-tag on the N-terminus (Fig. 1). This approach gave us the opportunity to exploit two distinct purification methods, ice affinity purification [36] and Ni 2+ –nitri- lotriacetate–agarose chromatography, to characterize these recombinant proteins. Results Multiple constructs representing the domain IV region of human calpain isoforms 1, 3, 8, 9, 11, 12 and 13 were designed in an effort to improve the likelihood of expressing these recombinant isoforms. Recombinant calpains 1, 3, 9 and 13 domain IV constructs produced high yields when expressed alone or when coexpressed with human small subunit (Table 1). Constructs of cal- pains 8, 11 and 12 failed to express. Further trials to stabilize their expression by coexpression with human small subunit did not influence the yield. Establishing the validity of the screening method using calpain 1 domain IV To test the functionality of the N-terminally AFP- tagged small subunit in forming a natural heterodimer [20,21], we coexpressed it with inactive rat calpain 2 (C105S-m-80 kDa) large subunit and with human cal- pain 1 domain IV. The rat large subunit was chosen for this purpose because the human ortholog is poorly expressed in Escherichia coli and the residues involved Calpain heterodimerization and homodimerization R. Ravulapalli et al. 974 FEBS Journal 276 (2009) 973–982 ª 2008 The Authors Journal compilation ª 2008 FEBS in heterodimer formation are highly conserved in the two mammals. As expected, both calpain 2 large sub- unit and the isolated domain, calpain 1 domain IV (21 kDa), formed heterodimers with recombinant type III AFP-tagged human small subunit (28 kDa). This was established by Ni 2+ –nitrilotriacetate–agarose column purification, where both the coexpressed con- structs were detected in the imidazole-eluted fractions (Fig. 2A, lane 4; Fig. 2C, lane 3). In Fig. 2A lane 4, the relative staining of large (80 kDa) and small (21 kDa) subunits is consistent with their 1 : 1 stoichi- ometry. When an immunoblot of the gel, shown in Fig. 2A, was probed with antibody against AFP, the only protein band detected was 28 kDa AFP-tagged small subunit (Fig. 2B, lane 1). Similarly, when a duplicate immunoblot was probed with antibody against His-tag, the only protein band detected was 80 kDa His-tagged large subunit (Fig. 2B, lane 2). One immediate advantage of the type III AFP- tagged small subunit construct is the increase in its molecular mass from 21 to 28 kDa, which readily distinguishes it from domain IV constructs. Thus, in lane 3 of Fig. 2C, the upper 28 kDa band of the small subunit is well separated from the lower, more abun- dant His6-tagged calpain 1 domain IV. Although the presence of AFP-tagged small subunit in the affinity- purified His6-tagged calpain 1 domain IV shows that the two different PEF domains form heterodimers, the relative staining of these two bands suggests that calpain 1 domain IV is present in excess. ABC Fig. 1. Three possible scenarios derived from coexpression of recombinant PEF fusion proteins. (A) Homodimer model of His6-tagged PEF domain. (B) Homodimer model of type III AFP-tagged PEF domain (calpain small subunit domain VI). (C) Heterodimer model of fusion protein containing type III AFP-tagged (blue) small subunit (cyan) forming a dimer with His6-tagged (light brown) PEF domain (orange). Rat small subunit (1AJ5) was used for modeling. All structures were drawn with PYMOL [51]. Table 1. Screening results of domain IV constructs from calpains 1, 3, 8, 9, 11, 12, and 13. Column 1: calpains used for screening. Col- umn 2: number of constructs designed and cloned. Column 3: number of constructs expressed. Column 4: yields of constructs when expressed alone in the absence of human small subunit. Column 5: results from coexpression of the domain IV construct with human small subunit. Column 6: results obtained by biophysical analysis of these constructs. NA, data not available; +++, very high expression; ++, high expression; +, low expression. Construct Cloned Expressed Yield Coexpression yield Dimerization Calpain 1 domain IV 1 1 +++ +++ Heterodimer Calpain 3 domain IV 5 5 +++ ++ +++ Homodimer Calpain 8 domain IV 5 Nil No expression No expression NA Calpain 9 domain IV 14 11 +++ +++ Heterodimer Calpain 11 domain IV 11 Nil No expression No expression NA Calpain 12 domain IV 6 Nil No expression No expression NA Calpain 13 domain IV 12 12 +++ ++ +++ Homodimer a a Predominant form found as homodimer. R. Ravulapalli et al. Calpain heterodimerization and homodimerization FEBS Journal 276 (2009) 973–982 ª 2008 The Authors Journal compilation ª 2008 FEBS 975 Calpain 3 domain IV is a homodimer Calpain 3 domain IV is suggested to favor homodi- merization, even though small subunit-containing calpains are produced in muscle cells. Earlier studies showed that recombinant calpain 3 domain IV, when expressed in isolation, formed a homodimer [30]. In further support of this argument, we show below that His6-tagged recombinant calpain 3 domain IV coex- pressed with type III AFP-tagged human small subunit (28 kDa) exclusively forms a homodimer. Upon purifi- cation by Ni 2+ –nitrilotriacetate–agarose chromato- graphy, the 28 kDa subunit was not detected in the imidazole-eluted fraction along with calpain 3 domain IV (Fig. 3A, lane 4). The 28 kDa subunit was present in the fractions that did not bind to the Ni 2+ –nitrilo- triacetate–agarose column (Fig. 3A, lane 2). Indeed, it was the most abundant protein in the flow-through fraction from that column. Calpain 9 domain IV forms a heterodimer with the small subunit The recombinant calpain 9 domain IV construct has 200 amino acids, including its His6 N-terminal tag. It has a theoretical pI of 5.71 and a calculated molecular mass of 23 130 Da. The amino acid sequence is 43% identical with domain IV of calpain 1, and 40% identi- cal with the small subunit (28 kDa). When calpain 9 domain IV was coexpressed with the 28 kDa small subunit fusion protein, it formed a heterodimer. Both subunits were detected in the imidazole-eluted fraction (Fig. 3B, lane 3). Their stoichiometry was close to 1 : 1. To confirm the identity of the two subunits, the gel was immunoblotted and probed with the two anti- bodies used in Fig. 2B. The antibody against AFP detected the upper band as a 28 kDa AFP-tagged small subunit (Fig. 3C, lane 1). Similarly, the antibody against His-tag reacted with the N-terminally His6- tagged calpain 9 domain IV (Fig. 3C, lane 2). In the converse approach using ice affinity purifica- tion, His6-tagged calpain 9 domain IV was included in the ice because of its heterodimerization with the AFP- tagged small subunit (Fig. 4, lane 2). Here, the amount of the His6-tagged calpain 9 domain IV in the ice frac- tion was slightly lower than would be predicted from the expected 1 : 1 stoichiometry with the small subunit as seen in the liquid fraction (Fig. 4, lane 3). This seems to be due to a small amount of subunit dissocia- tion that occurs as the ice grows over and pushes past the adsorbed AFP-tagged subunit. The shearing forces of the ice are apparently sufficient to disrupt quater- A B C Fig. 2. SDS ⁄ PAGE and immunoblot analysis of differentially tagged calpain 1 and 2 heterodimers. (A) Lane 1: molecular mass standards indi- cated at the side of the gel. Lanes 2, 3 and 4: flow-through, wash and eluate samples, respectively, from the Ni 2+ –nitrilotriacetate–agarose column chromatography of 80 kDa subunit (C105S-m-80 kDa) (triangle) coexpressed with 28 kDa AFP-tagged small subunit (dot). (B) Lanes 1 and 2: immunoblots of lane 4 from (A) probed with antibody against AFP and antibody against His-tag, respectively. (C) Lanes 1, 2 and 3: flow-through, wash and eluate samples from the Ni 2+ –nitrilotriacetate–agarosecolumn chromatography of calpain 1 domain IV (C1DIV) (square) coexpressed with 28 kDa AFP-tagged small subunit (dot). Both coexpressed constructs are predominantly detected in fractions eluted with imidazole. Calpain heterodimerization and homodimerization R. Ravulapalli et al. 976 FEBS Journal 276 (2009) 973–982 ª 2008 The Authors Journal compilation ª 2008 FEBS nary structure in a portion of the dimers, but do not break covalent bonds between the AFP moiety and a fusion partner [36]. A similar partial dissociation of subunits was seen during ice affinity purification of full length l-calpain heterodimerized to the AFP-tagged subunit (results not shown). The control experiment in this series showed that His6-tagged calpain 9 domain IV, when expressed alone, was not included in the ice but remained in the liquid fraction (Fig. 4, lanes 4 and 5, respectively). Calpain 13 domain IV The recombinant calpain 13 domain IV construct con- tains 174 amino acids, including the His6 N-terminal tag. It has a theoretical pI of 6.75 and a calculated molecular mass of 19 901 Da. Unlike other calpain PEF domains, it has low sequence identity with domai- n IV of calpain 1 (28%) and the small subunit (29%). When the recombinant calpain 13 domain IV construct was coexpressed with type III AFP-tagged human small subunit (28 kDa), calpain 13 domain IV was pre- dominantly seen in the eluant. The 28 kDa small sub- unit was mainly observed in the flow-through, although a faint band was seen in the wash and eluant (Fig. 5). On the basis of these SDS ⁄ PAGE results, a small amount of heterodimer is produced but cal- pain 13 domain IV is predominantly a homodimer. Discussion The PEF domain was first described in calpain [37– 39], and has since been found in other proteins such as ALG-2, grancalcin, sorcin and peflin [40,41]. It is char- acterized by having a fifth EF-hand available to pair with that of another PEF domain to form heterodi- mers or homodimers. More than half of the human calpain isoforms (1, 2, 3, 8, 9, 11, 12 and 13) have a PEF domain. Of these, the ubiquitous well-studied cal- pains 1 and 2 are known to form heterodimers with the small subunit PEF domains. However, previous investigations on calpain 3 suggest that PEF domain- containing calpain isoforms need not necessarily form a heterodimer, like calpains 1 and 2. In this study, we set out to determine what kind of dimers the different calpain isoforms make. Modeling studies using shape complementarity as a tool to measure the likelihood of forming a hetero- dimer or homodimer were performed using calpain 2, the previously generated model of calpain 3 domain IV A B C Fig. 3. SDS ⁄ PAGE and immunoblot analysis of calpain 3 domain IV (C3DIV) and calpain 9 domain IV (C9DIV) samples coexpressed with small subunit. (A) Lane 1: molecular mass standards indicated at the side of the gel. Lanes 2, 3 and 4: flow-through, wash and eluate sam- ples, respectively, from the Ni 2+ –nitrilotriacetate–agarose column chromatography of His-tagged (C3DIV) (triangle) coexpressed with 28 kDa AFP-tagged small subunit (dot). Only the C3DIV domain is detected in the eluant. (B) Lanes 1–3: flow-through, eluate and wash samples from the Ni 2+ –nitrilotriacetate–agarose column of His-tagged C9DIV (square) coexpressed with 28 kDa AFP-tagged small subunit (dot). Both the human small subunit and C9DIV are present in the eluant. (C) Lanes 1 and 2: immunoblots of lane 3 from (B) probed with antibody against AFP and antibody against His-tag, respectively. R. Ravulapalli et al. Calpain heterodimerization and homodimerization FEBS Journal 276 (2009) 973–982 ª 2008 The Authors Journal compilation ª 2008 FEBS 977 [30], and the small subunit structures as a guide. In addition, models were generated for artificial structures of the calpain 3 domain IV–small subunit heterodimer and of the calpain 2 domain IV homodimer. Shape complementarity values differed only slightly between the different dimers. In order of best to worst, the complementarity values were calpain 3 domain IV homodimer (0.751), small subunit homodimer (0.751), calpain 3 domain IV–small subunit heterodimer (0.734), calpain 2 domain IV–small subunit hetero- dimer (0.734) and calpain 2 domain IV homodimer (0.715). These values are not significantly different from each other, and therefore do not appear to pro- vide a method for distinguishing correct from incorrect dimers. Comparison of the buried surface areas for the various complexes also shows little variation, with the calpain 2 domain IV homodimer displaying the small- est surface area (average value of 1182 A ˚ 2 ) as com- pared to the others (average values ranging from 1311 to 1391 A ˚ 2 ). As tight packing of residues involved in the dimerization interfaces might not be the only factor influencing dimer formation, we used experimen- tation to distinguish which isoforms form heterodimers or homodimers. The recombinant small subunit domain VI has a molecular mass of 21 264 Da, and forms a homo- dimer when expressed alone [42]. Its molecular mass is close in value to those of isolated calpain PEF domains (domain IV), making it hard to distinguish whether they formed homodimers or heterodimers when coexpressed. In order to overcome this uncer- tainty, we devised a differential tag approach whereby all the calpain PEF domains contain a His6 N-termi- nal tag and the small subunit has an N-terminal type III AFP tag (7 kDa), allowing us to distinguish these two domains by size. Like the rat small subunit, the recombinant 28 kDa human small subunit fusion protein formed a homodimer when expressed alone (results not shown). Fig. 4. Ice affinity purification of type III AFP-tagged small subunit and calpain 9 domain IV (C9DIV). Lane 1: molecular mass standards indicated at the side of the gel. Lanes 2 and 3: equal volumes of the ice and liquid fractions obtained from the distribution of coex- pressed 28 kDa AFP-tagged small subunit (dot) with His-tagged C9DIV (square). Lanes 4 and 5: equal volumes of the ice and liquid fractions obtained from the distribution of His-tagged C9DIV (square) in the absence of 28 kDa AFP-tagged small subunit. Fig. 5. SDS ⁄ PAGE analysis of calpain 13 domain IV (C13DIV) sam- ples from the Ni 2+ –nitrilotriacetate–agarose column. Lane 1: molec- ular mass standards indicated at the side of the gel. Lanes 2, 3 and 4: flow-through, wash and eluate fractions from the column, respectively. The 28 kDa subunit and C13DIV proteins are indicated by dot and square symbols, respectively. Calpain heterodimerization and homodimerization R. Ravulapalli et al. 978 FEBS Journal 276 (2009) 973–982 ª 2008 The Authors Journal compilation ª 2008 FEBS Calpain 1, 3, 9 and 13 PEF domains were success- fully cloned and coexpressed as soluble recombinant products. However, numerous attempts to express cal- pain 8, 11 and 12 PEF domain constructs in E. coli were unsuccessful, and thus the dimerization potential of these PEF domains could not be analyzed. As the wild-type calpains 1 and 2 are both known to form heterodimers, we used calpain 2 large subunit and cal- pain 1 domain IV as controls in our experiments. Even in the absence of its adjacent domains, calpain 1 domain IV formed a heterodimer with the small sub- unit, rather than a homodimer. It should be noted that this construct lacks the N-terminal anchor peptide, which, on the basis of the structure of calpain 2 [19,21], should make additional heterodimerization contacts between the large and small subunits. Recombinant calpain 3 domain IV was previously shown to form a homodimer when expressed alone [30]. In this study, it was coexpressed with small subunit (28 kDa) but still formed a homodimer, further support- ing the argument that calpain 3 is a natural homodimer. Calpain 9 has been previously suggested to form a hete- rodimer when coexpressed with small subunit in the baculovirus expression system [43]. Coexpression of recombinant proteins calpain 9 domain IV and small subunit fusion product (28 kDa) led these proteins to associate as a heterodimer, in agreement with these pre- vious studies. As with calpain 1, the absence of the other domains in the large subunit did not alter the propensity of calpain 9 domain IV to heterodimerize. When expressed alone, calpain 9 domain IV formed an oligo- mer, unlike other PEF domains (results not shown). Calpain 13 is a tissue-specific calpain expressed predom- inantly in testis and lung. Its physiological role is not well understood, and its dimerization state is unknown [8]. Calpain 13 domain IV appeared as a predominant homodimer when coexpressed with small subunit fusion protein (28 kDa), although there were small amounts of heterodimer present in the eluate from the Ni 2+ –nitri- lotriacetate–agarose column. Most of the PEF domains in calpain isoforms share a high degree of sequence identity; however, it is not clear why they prefer one form of dimerization over the other. Further analysis of these constructs by determining their structure through crystallography may help us to gain more insight into the preference for homodimerization versus heterodimerization. Meanwhile, on the basis of these results, we predict that calpain 9 can be bound and silenced by calpasta- tin. Silencing of calpains 3 and 13 would require the simultaneous binding of two calpastatin inhibitory domains. Although this is a theoretical possibility, especially as calpastatin has four inhibitory domains and is an intrinsically unstructured protein, the absence of a small subunit in these two calpains would deprive calpastatin of one of its three calpain-binding sequences. The loss of this binding site would signifi- cantly weaken the overall binding interaction. Experimental procedures High-throughput cloning The cDNA fragments encoding the domain IV regions of cal- pains 1, 9, 11, 12 and 13 were obtained by PCR amplification of full-length cDNA templates of human calpains 1, 9, 11, 12 and 13 obtained from the Mammalian Gene Collection, using Expand high-fidelity DNA polymerase (Roche, India- napolis, IN, USA). Human calpain 8 domain IV was obtained by PCR amplification of reverse transcripts (RT-PCR) of total RNA from human stomach (Stratagene, La Jolla, CA, USA), using an RT-Thermoscript kit (Invitro- gen, Carlsbad, CA, USA) and Expand high-fidelity DNA polymerase. Human calpain 3 domain IV was obtained as previously described [30]. Multiple constructs were designed for each of these domains. The amplified fragments encoding domain IV regions of calpains 1, 8, 9, 11, 12 and 13 were inserted using the infusion ligation independent cloning system (BD Biosciences, Mountainview, CA, USA) into a modified pET28-LIC expression vector (EMD-Novagen, Gibbstown, NJ, USA) using a 96-well format high-through- put approach [44], downstream of the nucleotide sequence encoding MGSSHHHHHHSSGLVPRLGS. This 20 amino acid sequence contains a hexahistidine tag (His6-tag) and a thrombin cleavage site. Type III AFP-tagged human small subunit The cDNA fragment encoding domain VI of the human small subunit was obtained by PCR amplification of reverse transcripts (RT-PCR) of total RNA from human stomach (Stratagene, La Jolla, CA, USA), using an RT-Thermo- script kit (Invitrogen, Carlsbad, CA, USA) and Expand high-fidelity DNA polymerase (Roche, Indianapolis, IN, USA). The amplified product was cloned into the modified pET vector (pAC-pET) as previously described [45]. The type III AFP sequence was previously prepared by gene synthesis [46]. It was cloned into the pAC-pET vector 5¢ of the truncated 21 kDa subunit sequence. At the protein level, the two domains are joined by a linker of three alanine residues. Protein expression and purification by Ni 2+ –nitrilotriacetate–agarose The pET28-LIC vectors encoding the domain IV regions were transformed along with the pAC-pET plasmid R. Ravulapalli et al. Calpain heterodimerization and homodimerization FEBS Journal 276 (2009) 973–982 ª 2008 The Authors Journal compilation ª 2008 FEBS 979 containing the small subunit fusion construct into E. coli BL21(DE3) cells (Novagen) by electroporation. The trans- formed cells were grown in 1 L of LB medium under kana- mycin and ampicillin selection. The cells were grown to a D 600 nm of 0.8–1.0 at 37 °C. Protein expression was induced at 16 °C using 0.4 mm isopropyl thio-b-d-galactoside for 16 h. The cells were collected by centrifugation, resuspended in lysis buffer [25 mm Tris ⁄ HCl, pH 7.6, 5 mm EDTA, 5% (v ⁄ v) glycerol, 10 m m 2-mercaptoethanol, and 0.1 mm phen- ylmethanesulfonyl fluoride], and lysed by sonication. The resulting lysate was clarified by centrifugation at 27 000 g for 45 min. The supernatant obtained was incubated with 5mL Ni 2+ –nitrilotriacetate–agarose resin (Qiagen, Chats- worth, CA, USA) for 30 min at 4 °C with constant stirring. The Ni 2+ –nitrilotriacetate–agarose resin was later trans- ferred to a column and washed with N-buffer (50 mm Tris ⁄ HCl, pH 7.6, 100 mm NaCl, 5 mm imidazole, and 0.01% sodium azide). His6-tagged proteins were eluted with the lysis buffer containing 250 mm imidazole. The samples collected were later analyzed by SDS ⁄ PAGE. The inactive recombinant rat calpain 2 large subunit (C105S-m-80 kDa) was also coexpressed with the AFP-tagged small subunit and purified as described previously [45]. Ice affinity purification Ice affinity purification [36] was explored as a way of isolat- ing and identifying products containing the type III AFP fusion. In this method, the AFP fusion protein was adsorbed from solution (50 mL) into growing polycrystal- line ice frozen onto a cooled brass cold finger. The growth of the ice was controlled by circulating cold ethylene glycol solution through the hollow cold finger. After a thin layer of ice ( 1 mm) had initially formed on the cold finger, it was immersed in the AFP-containing solution prechilled to 1 °C in an insulated beaker. The solution was gently mixed using a stir bar, and the temperature of the cold finger was gradually reduced at a linear rate ()0.5 to )2.5 °C over 36 h), using a temperature-programmable water bath (Ne- slab), until approximately half to two-thirds of the volume was incorporated into the ice hemisphere. The ice hemi- sphere was then removed from the liquid and allowed to melt for  10 min to remove any protein that was nonspe- cifically bound to the surface of the ice. The ice hemisphere was melted to release the AFP. Samples (2 and 5 lL) from both melted ice (ice fraction) and leftover liquid (liquid fraction) were analyzed by SDS ⁄ PAGE [47]. Modeling studies Shape complementarity of various dimer structures and models was calculated using the program sc from the ccp4 program suite [48]. Crystal structures of the rat small sub- unit homodimer (Protein Data Bank code: 1dvi) and of the human calpain 2 heterodimer (Protein Data Bank code: 1kfu) were used as references. Homology models of the cal- pain 3 domain IV homodimer, the heterodimer of calpain 3 domain IV with the small subunit and of a calpain 2 domain IV homodimer were generated using the program modeller 9v3 [49]. The best of 100 models were then used in an energy minimization and molecular dynamics proto- col using the program gromacs 3.3 [50]. The protein was solvated, energy minimized using the steepest descents pro- tocol, and subjected to position-restrained molecular dynamics to relax the solvent. This was followed by a 2 ns molecular dynamics simulation. Structures were extracted from the trajectory every 20 ps, and the surface comple- mentarity at the dimer interface was calculated with the program sc from the ccp4 program suite [48]. The average sc value from these 100 structures is reported. For compar- ison, the same molecular dynamics protocol was used on the crystal structures of the rat small subunit homodimer (Protein Data Bank code: 1dvi) and of the human calpain 2 heterodimer (Protein Data Bank code: 1kfu). Immunoblotting Immunoblotting was performed using 10% Tris ⁄ Tricine SDS ⁄ PAGE gels transferred onto poly(vinylidene difluoride) membranes. Polyclonal antibodies against the His-tag and against type III AFP were raised in rabbits. The secondary antibody was anti-(rabbit IgG) conjugated to horseradish peroxidase (Promega, Madison, WI, USA), which was detected by ECL (Perkin-Elmer, Fremont, CA, USA). Acknowledgements This research was funded by a grant to P. L. Davies from the Canadian Institutes for Health Research. P. L. Davies holds a Canada Research Chair in Pro- tein Engineering. The Structural Genomics Consortium is a registered charity (number 1097737) that receives funds from the Canadian Institutes for Health Research, the Canadian Foundation for Innovation, Genome Canada through the Ontario Genomics Insti- tute, GlaxoSmithKline, Karolinska Institute, the Knut and Alice Wallenberg Foundation, the Ontario Inno- vation Trust, the Ontario Ministry for Research and Innovation, Merck & Co., Inc., the Novartis Research Foundation, the Swedish Agency for Innovation Sys- tems, the Swedish Foundation for Strategic Research, and the Wellcome Trust. References 1 Goll DE, Thompson VF, Li H, Wei W & Cong J (2003) The calpain system. Physiol Rev 83, 731–801. 2 Pinter M, Stierandova A & Friedrich P (1992) Purifica- tion and characterization of a Ca(2+)-activated thiol Calpain heterodimerization and homodimerization R. Ravulapalli et al. 980 FEBS Journal 276 (2009) 973–982 ª 2008 The Authors Journal compilation ª 2008 FEBS protease from Drosophila melanogaster. Biochemistry 31, 8201–8206. 3 Barnes TM & Hodgkin J (1996) The tra-3 sex determi- nation gene of Caenorhabditis elegans encodes a mem- ber of the calpain regulatory protease family. EMBO J 15, 4477–4484. 4 Huang X, Czerwinski E & Mellgren RL (2003) Purifica- tion and properties of the Dictyostelium calpain-like protein, Cpl. Biochemistry 42, 1789–1795. 5 Margis R & Margis-Pinheiro M (2003) Phytocalpains: orthologous calcium-dependent cysteine proteinases. Trends Plant Sci 8, 58–62. 6 Dear N, Matena K, Vingron M & Boehm T (1997) A new subfamily of vertebrate calpains lacking a calmodulin-like domain: implications for calpain regulation and evolution. Genomics 45, 175–184. 7 Ersfeld K, Barraclough H & Gull K (2005) Evolution- ary relationships and protein domain architecture in an expanded calpain superfamily in kinetoplastid parasites. J Mol Evol 61, 742–757. 8 Farkas A, Tompa P & Friedrich P (2003) Revisiting ubiquity and tissue specificity of human calpains. Biol Chem 384, 945–949. 9 Glading A, Lauffenburger DA & Wells A (2002) Cutting to the chase: calpain proteases in cell motility. Trends Cell Biol 12, 46–54. 10 Neumar RW, Xu YA, Gada H, Guttmann RP & Siman R (2003) Cross-talk between calpain and caspase prote- olytic systems during neuronal apoptosis. J Biol Chem 278, 14162–14167. 11 Janossy J, Ubezio P, Apati A, Magocsi M, Tompa P & Friedrich P (2004) Calpain as a multi-site regulator of cell cycle. Biochem Pharmacol 67, 1513–1521. 12 Suzuki T, Okumura-Noji K, Ogura A, Tanaka R, Nakamura K & Kudo Y (1992) Calpain may produce a Ca(2+)-independent form of kinase C in long-term potentiation. Biochem Biophys Res Commun 189, 1515– 1520. 13 Cox NJ, Hayes MG, Roe CA, Tsuchiya T & Bell GI (2004) Linkage of calpain 10 to type 2 diabetes: the biological rationale. Diabetes 53(Suppl. 1), S19–S25. 14 Kuwako K, Nishimura I, Uetsuki T, Saido TC & Yoshikawa K (2002) Activation of calpain in cultured neurons overexpressing Alzheimer amyloid precursor protein. Brain Res Mol Brain Res 107, 166–175. 15 Enns D, Karmazyn M, Mair J, Lercher A, Kountchev J & Belcastro A (2002) Calpain, calpastatin activities and ratios during myocardial ischemia–reperfusion. Mol Cell Biochem 241, 29–35. 16 Richard I, Broux O, Allamand V, Fougerousse F, Chiannilkulchai N, Bourg N, Brenguier L, Devaud C, Pasturaud P, Roudaut C et al. (1995) Mutations in the proteolytic enzyme calpain 3 cause limb-girdle muscular dystrophy type 2A. Cell 81, 27–40. 17 Yoshikawa Y, Mukai H, Hino F, Asada K & Kato I (2000) Isolation of two novel genes, down-regulated in gastric cancer. Jpn J Cancer Res 91, 459–463. 18 Wang KK & Yuen PW (1994) Calpain inhibition: an overview of its therapeutic potential. Trends Pharmacol Sci 15, 412–419. 19 Hosfield CM, Elce JS, Davies PL & Jia Z (1999) Crystal structure of calpain reveals the structural basis for Ca(2+)-dependent protease activity and a novel mode of enzyme activation. EMBO J 18, 6880–6889. 20 Hosfield CM, Ye Q, Arthur JS, Hegadorn C, Croall DE, Elce JS & Jia Z (1999) Crystallization and X-ray crystallographic analysis of m-calpain, a Ca2+-depen- dent protease. Acta Crystallogr D Biol Crystallogr 55, 1484–1486. 21 Strobl S, Fernandez-Catalan C, Braun M, Huber R, Masumoto H, Nakagawa K, Irie A, Sorimachi H, Bourenkow G, Bartunik H et al. (2000) The crystal structure of calcium-free human m-calpain suggests an electrostatic switch mechanism for activation by calcium. Proc Natl Acad Sci USA 97, 588–592. 22 Moldoveanu T, Hosfield CM, Lim D, Elce JS, Jia Z & Davies PL (2002) A Ca(2+) switch aligns the active site of calpain. Cell 108, 649–660. 23 Reverter D, Strobl S, Fernandez-Catalan C, Sorimachi H, Suzuki K & Bode W (2001) Structural basis for possible calcium-induced activation mechanisms of calpains. Biol Chem 382, 753–766. 24 Tompa P, Emori Y, Sorimachi H, Suzuki K & Fried- rich P (2001) Domain III of calpain is a Ca2+-regu- lated phospholipid-binding domain. Biochem Biophys Res Commun 280, 1333–1339. 25 Suzuki K, Hata S, Kawabata Y & Sorimachi H (2004) Structure, activation, and biology of calpain. Diabetes 53(Suppl. 1), S12–S18. 26 Waghray A, Wang DS, McKinsey D, Hayes RL & Wang KK (2004) Molecular cloning and characteriza- tion of rat and human calpain-5. Biochem Biophys Res Commun 324, 46–51. 27 Tonami K, Kurihara Y, Aburatani H, Uchijima Y, Asano T & Kurihara H (2007) Calpain 6 is involved in microtubule stabilization and cytoskeletal organization. Mol Cell Biol 27, 2548–2561. 28 Yorikawa C, Takaya E, Osako Y, Tanaka R, Terasawa Y, Hamakubo T, Mochizuki Y, Iwanari H, Kodama T, Maeda T et al. (2008) Human calpain 7 ⁄ PalBH associ- ates with a subset of ESCRT-III-related proteins in its N-terminal region and partly localizes to endocytic membrane compartments. J Biochem 143, 731–745. 29 Dong B & Liu R (2008) Characterization of endoge- nous and recombinant human calpain-10. Biochimie 90, 1362–1371. 30 Ravulapalli R, Diaz BG, Campbell RL & Davies PL (2005) Homodimerization of calpain 3 penta-EF-hand domain. Biochem J 388, 585–591. R. Ravulapalli et al. Calpain heterodimerization and homodimerization FEBS Journal 276 (2009) 973–982 ª 2008 The Authors Journal compilation ª 2008 FEBS 981 31 Kinbara K, Ishiura S, Tomioka S, Sorimachi H, Jeong SY, Amano S, Kawasaki H, Kolmerer B, Kimura S, Labeit S et al. (1998) Purification of native p94, a mus- cle-specific calpain, and characterization of its autolysis. Biochem J 335, 589–596. 32 Wendt A, Thompson VF & Goll DE (2004) Interaction of calpastatin with calpain: a review. Biol Chem 385, 465–472. 33 Todd B, Moore D, Deivanayagam CC, Lin GD, Chattopadhyay D, Maki M, Wang KK & Narayana SV (2003) A structural model for the inhibition of calpain by calpastatin: crystal structures of the native domain VI of calpain and its complexes with calpastatin peptide and a small molecule inhibitor. J Mol Biol 328, 131–146. 34 Hanna RA, Campbell RL & Davies PL (2008) Cal- cium-bound structure of calpain and its mechanism of inhibition by calpastatin. Nature 456, 409–412. 35 Jia Z, DeLuca CI, Chao H & Davies PL (1996) Struc- tural basis for the binding of a globular antifreeze protein to ice. Nature 384, 285–288. 36 Kuiper MJ, Lankin C, Gauthier SY, Walker VK & Davies PL (2003) Purification of antifreeze proteins by adsorption to ice. Biochem Biophys Res Commun 300, 645–648. 37 Lin GD, Chattopadhyay D, Maki M, Wang KK, Carson M, Jin L, Yuen PW, Takano E, Hatanaka M, DeLucas LJ et al. (1997) Crystal structure of calcium bound domain VI of calpain at 1.9 A resolution and its role in enzyme assembly, regulation, and inhibitor bind- ing. Nat Struct Biol 4, 539–547. 38 Blanchard H, Grochulski P, Li Y, Arthur JS, Davies PL, Elce JS & Cygler M (1997) Structure of a calpain Ca(2+)-binding domain reveals a novel EF-hand and Ca(2+)-induced conformational changes. Nat Struct Biol 4, 532–538. 39 Kretsinger RH (1997) EF-hands embrace. Nat Struct Biol 4, 514–516. 40 Maki M, Kitaura Y, Satoh H, Ohkouchi S & Shibata H (2002) Structures, functions and molecular evolution of the penta-EF-hand Ca2+-binding proteins. Biochim Biophys Acta 1600, 51–60. 41 Jia J, Han Q, Borregaard N, Lollike K & Cygler M (2000) Crystal structure of human grancalcin, a member of the penta-EF-hand protein family. J Mol Biol 300, 1271–1281. 42 Blanchard H, Li Y, Cygler M, Kay CM, Simon J, Arthur C, Davies PL & Elce JS (1996) Ca(2+)-binding domain VI of rat calpain is a homodimer in solution: hydrodynamic, crystallization and preliminary X-ray diffraction studies. Protein Sci 5, 535–537. 43 Lee HJ, Tomioka S, Kinbara K, Masumoto H, Jeong SY, Sorimachi H, Ishiura S & Suzuki K (1999) Charac- terization of a human digestive tract-specific calpain, nCL-4, expressed in the baculovirus system. Arch Biochem Biophys 362, 22–31. 44 Benoit RM, Wilhelm RN, Scherer-Becker D & Ostermeier C (2006) An improved method for fast, robust, and seamless integration of DNA fragments into multiple plasmids. Protein Expr Purif 45, 66–71. 45 Elce JS, Hegadorn C, Gauthier S, Vince JW & Davies PL (1995) Recombinant calpain II: improved expression systems and production of a C105A active-site mutant for crystallography. Protein Eng 8, 843–848. 46 Chao H, Davies PL, Sykes BD & Sonnichsen FD (1993) Use of proline mutants to help solve the NMR solution structure of type III antifreeze protein. Protein Sci 2, 1411–1428. 47 Marshall CB, Tomczak MM, Gauthier SY, Kuiper MJ, Lankin C, Walker VK & Davies PL (2004) Partitioning of fish and insect antifreeze proteins into ice suggests they bind with comparable affinity. Biochemistry 43, 148–154. 48 Collaborative Computational Project, Number 4 (1994) The CCP4 suite: programs for protein crystallography. Acta Crystallogr D Biol Crystallogr 50, 760–763. 49 Sali A & Blundell TL (1993) Comparative protein mod- elling by satisfaction of spatial restraints. J Mol Biol 234, 779–815. 50 Van Der Spoel D, Lindahl E, Hess B, Groenhof G, Mark AE & Berendsen HJ (2005) GROMACS: fast, flexible, and free. J Comput Chem 26, 1701–1718. 51 DeLano WL (2003) The PyMOL Molecular Graphics System. DeLano Scientific, San Carlos, CA. Calpain heterodimerization and homodimerization R. Ravulapalli et al. 982 FEBS Journal 276 (2009) 973–982 ª 2008 The Authors Journal compilation ª 2008 FEBS . Distinguishing between calpain heterodimerization and homodimerization Ravikiran Ravulapalli 1 , Robert. Pinter M, Stierandova A & Friedrich P (1992) Purifica- tion and characterization of a Ca(2+)-activated thiol Calpain heterodimerization and homodimerization