The PEST Domain of IκBα Is Necessary and Sufficient forin Vitro Degradation by μ-Calpain*

Polypeptide sequences enriched in proline (P), glutamate (E), serine (S), and threonine (T), dubbed PEST domains, are proposed to expedite the degradation of proteins. The proteolysis of one PEST-containing protein, IκBα, is prerequisite to the activation of the transcription factor NF-κB. Two mechanisms of IκBα degradation in vivo have been described, one well characterized through the ubiquitin-proteasome pathway, and another less characterized through calpain. In this report, a mutational analysis was done to identify any regions of IκBα that facilitate its recognition and proteolysis by calpain in vitro. These studies revealed that the PEST sequence of IκBα is critical for its calpain-dependent degradation. Furthermore, the IκBα-PEST domain binds to the calmodulin-like domain of the large subunit of μ-calpain (μCaMLD). Transfer of the IκBα-PEST domain to a protein incapable of either binding to or being degraded by μ-calpain allowed for the interaction of the chimeric protein with μCaMLD and resulted in its susceptibility to calpain proteolysis. Moreover, the μCaMLD of calpain acts as a competitive inhibitor of calpain-dependent IκBα degradation. Our data demonstrate that the IκBα-PEST sequence acts as a modular domain to promote the physical association with and subsequent degradation by μ-calpain and suggest a functional role for PEST sequences in other proteins as potential calpain-targeting units.

mal targeted proteolysis (4). An example of inducible protein degradation is seen during the activation of the transcription factor NF-B (5). Under nonstimulated conditions, NF-B is partitioned to the cytoplasm through association with a member of the IB family of inhibitory proteins, most notably IB␣ (6). As IB␣ binds to dimeric NF-B complexes, it is able to mask the nuclear localization sequence present on NF-B and thereby achieve cytoplasmic sequestration of the complex (7,8). Activation of NF-B, therefore, is typically preceded by the proteolytic inactivation of the IB inhibitory protein (5). This has been shown to occur in response to many signals, which act to positively regulate an IB kinase complex leading to sitespecific phosphorylation of IB␣ (9 -12). This in turn targets the molecule to a ubiquitin ligase enzyme and subsequent degradation through the ubiquitin-proteasome proteolytic pathway (13,14).
Although IB␣ degradation is attributed primarily to the ubiquitin-proteasome pathway, there have recently been a number of alternate proteolytic mechanisms described for IB␣, including some which specifically implicate isoforms of calpain as the direct IB␣ protease (15)(16)(17)(18). A recent study suggests that loss of calpain 3 activity results in an accumulation of IB␣ leading to an increased sensitivity to apoptosis, which contributes to the limb-girdle muscular dystrophy type 2A phenotype (15). Chen et al. (16) showed that IB␣ is degraded in a calpain-dependent manner following treatment of the mouse macrophage cell line RAW 264.7 with the toxic particulate silica. Zhang et al. (17) showed that hypoxic conditions activated calpain activity in endothelial cells, resulting in a proteolysis of IB␣ that was sensitive to the calpain inhibitor E-64d. Furthermore, Brasier and colleagues (18) demonstrated that treatment of HepG2 liver cells with the cytokine tumor necrosis factor-␣ resulted in the degradation of IB␣ through the ubiquitin-proteasome pathway as well as through a parallel pathway dependent on calpain activity. We have also recently demonstrated a novel degradation pathway of IB␣ in murine B cells that is independent of proteasome activity yet dependent on intracellular calcium and is associated with the constitutive activity of NF-B in these cells (19). The sensitivity to cysteine-protease and calpain inhibitors as well as the calcium requirement of this IB␣ proteolysis implicate a possible involvement of calpain. Together, these studies suggest that the degradation of IB␣ required for NF-B signaling is not solely dependent on the 26S proteasome but can also occur through calpain under certain conditions.
The calpain family of proteolytic enzymes is comprised of both ubiquitous and tissue-specific isoforms of calcium-dependent thiol proteases (20,21). The ubiquitousand m-calpains share structural similarity yet differ markedly in their requirements for calcium. In vitro, -calpain is active at calcium concentrations between 5 and 50 M, whereas m-calpain is active at concentrations of 0.2-1.0 mM (22). Although -calpain re-quires far less calcium than m-calpain for its half-maximal activation in vitro, intracellular calcium concentrations are nevertheless orders of magnitude lower (0.1-0.4 M). This poses an enigma surrounding the in vivo activation of calpains and has led to several proposed models of calpain regulation (22). These include interaction of calpain with membrane phospholipids, an autolytic self-activation, and its regulation by endogenous "activator" protein. Although their activation in cells remains unclear, several physiologic roles for the calpains have been suggested spanning such processes as cell-cycle regulation, apoptosis, and long term potentiation (20). Calpains are heterodimers consisting of a large 80-kDa subunit and a smaller 30-kDa subunit. The large subunit contains the catalytic function of the enzyme as well as a calmodulin-like domain (CaMLD), 1 so named for its high sequence homology to other EF-hand containing proteins such as calmodulin (20). Several known substrates ofand m-calpain have been identified and many more proposed (23), and yet structural determinants of calpain substrates remain ill-defined. Interestingly, several calpain substrates contain either calmodulin-binding domains or PEST domains (24). The latter have been associated with proteolysis, because proteins containing high PEST scores often undergo rapid degradation in vivo (3).
IB␣ contains a C-terminal PEST sequence and is reported to be inducibly degraded by calpains in vivo. Therefore, to better understand the mechanism allowing IB␣ to be degraded by calpain, and more specifically a possible role for the IB␣-PEST domain in calpain-dependent proteolysis, we analyzed the IB␣ structural requirements for this reaction. Here, we report the fundamental role of the IB␣-PEST sequence in determining substrate susceptibility to calpain in vitro. Our data suggest a mechanism for calpain-specific proteolysis of IB␣ in which first the PEST domain of IB␣ binds in a calciumdependent manner to the CaMLD of the large subunit of calpain, followed by N-terminal cleavage of IB␣ and further proteolysis.
Protein and Purification-Calpain I (-calpain) purified from porcine erythrocytes was purchased (Calbiochem). All [ 35 S]methionine-labeled proteins were transcribed (T3 RNA polymerase) and translated over 2 h at 30°C in TNT rabbit reticulocyte lysate or TNT wheat germ extract where specified, using 1 g of pBS constructs per 50 l of reaction volume according to the manufacturer's recommendation (Promega). GST and GST-CaMLD were purified from exponentially growing Escherichia coli (strain BL-21) 2 h following induction with 0.1 mM isopropyl-1-thio-␤-D-galactopyranoside. Briefly, cell pellets were resuspended in ice-cold phosphate-buffered saline with protease inhibitors, lysed by sonication, and brought to 1% Triton X-100. The lysate was cleared at 10,000 ϫ g for 5 min and 50% slurry glutathione-Sepharose 4B beads (Amersham Pharmacia Biotech) were added to supernatant and mixed gently for 30 min. Beads were washed 4 times in ice-cold phosphate-buffered saline, and purity of the protein was confirmed by SDS-polyacrylamide gel electrophoresis (PAGE) and Coomassie staining.
In Vitro Degradation Reactions-All reaction samples included 0.5-1.0 l of [ 35 S]methionine-labeled substrate protein (approximately 2 ng, see Fig. 1B), 750 M CaCl 2 , the indicated concentration of -calpain, and were incubated in calpain reaction buffer (30 mM Tris-HCl (pH 7.5) and 1.5 mM dithiothreitol). Reactions were brought to a final volume of 10 l and following incubation at 30°C for 15 min were terminated on ice by addition of 2ϫ SDS sample buffer. For competition experiments, GST proteins coupled to Sepharose (as described below) were rinsed 4 times with calpain reaction buffer and 10 l added to the reaction mixture for a final reaction volume of 20 l.
GST Pull-down Assays-Association reactions were performed in the presence of 10 mg/ml E. coli protein extract in 1ϫ binding buffer (50 mM KPO 4 (pH 7.5), 150 mM KCl, 10% (v/v) glycerol, and 1% (v/v) Triton X-100) and in the presence of 1 mM phenylmethylsulfonyl fluoride and 20 ng/l aprotonin A (Sigma). For each sample 1 l of [ 35 S]methioninelabeled protein was incubated with 200 l of E. coli extract on ice for 15 min, and cleared at 14,000 rpm for 15 min. 20 l of GST or GST-CaMLD bound to Sepharose beads (1:1 slurry) was added to this, and the final volume brought up to 0.5 ml with E. coli protein extract. Reactions were tumbled at 4°C for 2 h, washed 5 times in ice-cold 1ϫ binding buffer, boiled in 1ϫ SDS sample buffer, and separated by SDS-PAGE.
Immunoprecipitations and Immunoblotting-Antibodies used were rabbit polyclonal raised against either the N-terminal 56 amino acids of murine IB␣ conjugated to GST (5432 antibody, see Ref. 28) or the C-terminal 21 amino acids of IB␣ (C21 antibody, Santa Cruz Biotechnology, Inc.). Protein A-Sepharose and the appropriate antibodies were added to each reaction. IP buffer and protocol have been described previously (19). All immunoblotting was done with the C21 antibody as described previously (19).
In Vitro Phosphorylation Reaction-In vitro phosphorylation of IB␣ translated in rabbit reticulocyte lysate was performed according to Hunter et al. (29). Briefly, following translation the extract was passed through G-50 Sephadex (Amersham Pharmacia Biotech) equilibrated with 10 mM NaCl, 25 mM KCl, 1 mM MgCl 2 , 10 mM Tris (pH 7.5), 0.25 mM dithiothreitol, 0.1 mg/ml bovine serum albumin). [␥-32 P]ATP was added, and the reaction incubated at 30°C for 60 min. Phosphatase inhibitors were then added to all but one wild-type IB␣ reaction, and the remaining [␥-32 P]ATP removed by G-50 chromatography. The IB␣ sample lacking phosphatase inhibitors was incubated with 10 units of calf intestinal alkaline phosphatase (Roche Molecular Biochemicals) at 37°C for 30 min. All IB␣ proteins were then immunoprecipitated with N-terminal-specific 5432 antibody and transferred to Immobilon-P following SDS-PAGE. Western blotting was done to confirm equal amounts of in vitro translated proteins, and phosphoproteins were visualized by autoradiography.

IB␣ Is Degraded by -Calpain in Vitro-
To study the substrate characteristics of IB␣ for calpain it was important to first determine the calpain-specific cleavage and/or degradation of IB␣ in vitro. We analyzed the purity of commercially available -calpain extracted from porcine erythrocytes by SDS-PAGE. A Coomassie Blue stain of the purified calpain used (Fig. 1A) demonstrates that the major band migrates above the 68-kDa marker, representative of the large 80-kDa subunit. The small subunit of calpain, migrating at ϳ29 kDa, is also visible. Composition among different lots of enzyme differed (not shown). In addition, the sensitivity of IB␣ to degradation among the different -calpain lots also varied (not shown). The amount of in vitro translated IB␣ substrate added to reaction mixtures is estimated to be approximately 2 ng (ϳ5 nM concentration), as judged by relative band intensities of known amounts of recombinant GST-IB␣ protein by Western blot analysis (Fig. 1B). To test the protease activity of the purified -calpain on IB␣, we incubated increasing concentrations of -calpain with in vitro translated IB␣. Addition of -calpain to [ 35 S]methionine-labeled IB␣ translated in rabbit reticulocyte lysate resulted in the complete degradation of IB␣ in a dose-dependent fashion (Fig. 1C, upper panel). Furthermore, the degradation of IB␣ was calcium dependent, because the addition of the calcium chelator EGTA (30) to the reaction mixture completely blocked IB␣ degradation at the highest calpain concentration tested (lane 3), consistent with the calcium requirement of calpain. A similar degradation pattern can be observed using IB␣ translated and [ 35 S]methionine-labeled in wheat germ extract (Fig. 1C, lower panel).
Because the proteasome is abundant in cell extract and known to degrade IB␣, we tested whether lactacystin, which specifically inhibits the proteasome by modifying the active site Nterminal threonine residue of the catalytic ␤ subunit (31), could block the degradation of IB␣ under these conditions. As shown in Fig. 1D, we found that a concentration of lactacystin as high as 100 M had no effect on the reaction whereas calpeptin, a synthetic calpain inhibitor, showed a dose-dependent inhibition of IB␣ proteolysis. These data indicate that IB␣ translated in either a mammalian or a plant cell-free system is degraded by calpain in vitro.
The C-terminal PEST Domain of IB␣ Facilitates Its in Vitro Degradation by Calpain-IB␣ can be divided into three general domains. These include (i) an N-terminal domain that contains amino acid constituents required for the inducible degradation of IB␣ (32-34); (ii) an internal region made up of six repeats of a 30 -34 amino acid ankyrin-like domain that facilitates association with NF-B (35,36); and (iii) a C-terminal acidic domain that includes a PEST sequence and appears to be involved in regulating some cases of IB␣ protein turnover (26,(37)(38)(39), as well as the inhibition of DNA binding by NF-B (40,41). We questioned whether either the N-terminal signal responsive domain or the C-terminal PEST domain of IB␣ are determinants for the efficiency of in vitro degradation by calpain. This was tested through an analysis of calpainmediated proteolysis using either N-terminally or C-terminally truncated IB␣ proteins in comparison to the full-length IB␣ protein as substrates. [ 35 S]methionine-labeled IB␣ proteins lacking either amino acids 1-36 (IB⌬N) or amino acids 278 -314 (IB⌬C) were incubated with increasing concentrations of -calpain, terminated, and separated by SDS-PAGE ( Fig. 2A). At a calpain concentration of 100 nM the wild-type IB␣ protein was almost completely degraded within 15 min (Fig. 2B). The IB⌬N mutant was proteolyzed in a similar fashion. Proteolytic breakdown of IB⌬C, however, was reduced in comparison to full-length IB␣ at equal concentrations of calpain. Following incubation with 100 nM calpain the majority of the input protein still remained. These data suggest that a sequence, or sequences, within the IB␣ C-terminal domain promotes the in vitro degradation of IB␣ by -calpain.
As a more direct test of the contributing role of the Cterminal PEST domain to IB␣ as a calpain substrate, we examined the degradation profile of an IB␣ mutant in which the PEST score has been lowered. MutF contains five alanine substitutions for serine and threonine residues within the PEST domain (25), lowering the PEST score of the C terminus from ϩ4.6 to Ϫ0.7 (determined using the PEST-FIND algorithm developed by Rogers et al. (3)). When MutF was incubated with increasing calpain concentrations we detected a significant decline in proteolysis relative to that of the wildtype protein.
At the highest calpain concentration tested MutF was only degraded to ϳ60% of its original amount, which represents a reduction in the proteolytic efficiency of calpain in vitro. MutF can bind to NF-B when expressed in cells and is sensitive to the proteasome-dependent degradation pathway, and therefore gross structural alterations in the protein caused by amino acid substitution that result in proteolytic insensitivity are unlikely (25).
However, the amino acid replacements in MutF disrupt the basal phosphorylation of IB␣ by casein kinase II (25,42). Thus, we compared the phosphorylation status of the fulllength protein to that of MutF following translation in rabbit reticulocyte lysate. Additionally, a mutant of the inducible phosphorylation sites, serines 32 and 36 within the N terminus of IB␣ (32-34), was examined. Following depletion of cold ATP from the reaction mixture by G-50 chromatography, [␥-32 P]ATP was added to the translation reaction mixtures, and IB␣ immunoprecipitated following the reaction (29). Fig. 3A shows that IB␣ was phosphorylated in the reticulocyte lysate, and treatment with calf intestinal alkaline phosphatase (CIP) reduced the incorporated radiolabeled phosphate. Replacing serines 32 and 36 with alanine residues had no effect on the phosphorylation of IB␣ in vitro, whereas phosphorylation of the MutF protein was barely detectable by this assay. [ 35 S]Methionine-labeled IB␣ was either left untreated or was treated with CIP. Both reactions were incubated at 65°C to inactivate CIP and then subjected to calpain proteolysis in both the presence and absence of EGTA. CIP-treated IB␣ was a poorer substrate of calpain in vitro when compared with untreated IB␣ (Fig. 3B), illustrating that the phosphorylation of IB␣ facilitates its in vitro degradation by calpain (compare lanes 3  and 7). Degradation was blocked by EGTA, which again demonstrates the calcium requirement of the calpain-specific reaction (compare lanes 3 and 4). Furthermore, IB␣ that had been pretreated with CIP was less sensitive to calpain-specific degradation than untreated IB␣ over time (Fig. 3C). These findings suggest that the integrity and/or posttranslational modifications of the IB␣ PEST sequence can alter the efficacy of calpain to degrade IB␣ in vitro.
The PEST Domain of IB␣ Associates with the Calmodulinlike Domain of the Large Subunit of -Calpain-Since their first description in 1986, it has been speculated that PEST domains are able to increase a protein's turnover rate through association with calpain (3). However, to our knowledge, neither calpain nor any other protease has been observed to di-rectly interact with a protein's PEST domain. The large subunit of -calpain contains a calmodulin-like domain (CaMLD) that has been shown to mediate a calcium-dependent association with the endogenous calpain inhibitor calpastatin (27,43). Thus, to test whether CaMLD directly associates with IB␣, possibly through the IB␣-PEST domain, we used a fusion protein consisting of GST and the CaMLD (GST-CaMLD) and tested for its association with IB␣ and deletion mutant proteins in a GST pull-down assay. A representative gel is shown in Fig. 4A. Following incubation, GST-CaMLD, but not GST, was able to bind to full-length IB␣ in the presence of calcium but only weakly when calcium was not added to the binding buffer. Alternatively, IB⌬C was unable to associate with GST-CaMLD or GST in either the presence or the absence of added calcium. To extend these findings we tested for association between GST-CaMLD and MutF both with and without added calcium. We observed that while MutF was pulled down in the presence of added calcium by GST-CaMLD, this interaction represents a weaker one than that between GST-CaMLD and wild-type IB␣. Additionally, when IB␣ was dephosphorylated by CIP before the association reaction the binding of IB␣ to GST-CaMLD was also weak, similar to that seen for MutF (Fig. 4B). These findings are consistent with and correlate well with the inability of IB⌬C to be degraded by -calpain and the reduced ability of MutF or CIP-treated IB␣ to be degraded by -calpain (Figs. 2 and 3). These results suggest that the PEST-containing C-terminal domain of IB␣ facilitates the association of IB␣ with the CaMLD of calpain in vitro and that the affinity of the protein substrate for CaMLD affects its efficiency of degradation by calpain in vitro.
IB␣-PEST Acts as a Modular Domain to Allow Association with CaMLD to an Otherwise Nonbinding Protein-Data from the GST pull-down experiments demonstrated that the PEST domain of IB␣ is indeed necessary for association of the protein with CaMLD. To ascertain whether the PEST domain itself is sufficient for binding the CaMLD, we generated chimeric proteins in which bacterial CAT was fused to either N-or C-terminal portions of IB␣. The N-terminal 66 amino acids of IB␣ were fused to the N terminus of CAT to create N-CAT, whereas the C-terminal 54 amino acids of IB␣, inclusive of the PEST domain, were fused to the C terminus of CAT to create CAT-PEST (Fig. 5A). Following translation and incorporation of [ 35 S]methionine in vitro, these proteins were incubated with either GST or GST-CaMLD in both the presence and absence of added calcium, as previously for the IB␣ proteins. CAT-PEST showed a strong association with GST-CaMLD, which was favored by the addition of exogenous calcium to the binding reaction (Fig. 5A). However, neither CAT nor N-CAT were detectable in the fraction pulled down by GST-CaMLD even on exposures up to 10 times longer than that required to visualize CAT-PEST. Immunoprecipitations demonstrated that CAT-PEST, but not CAT, is phosphorylated at similar levels as the IB␣ protein during in vitro translation in rabbit reticulocyte lysate, consistent with the phosphorylation of a residue(s) within the IB␣-PEST domain (data not shown). Strikingly, these data show that the IB␣-PEST domain acts autonomously to bind the CaMLD of -calpain and that this association is transferable to a normally nonbinding protein.
As mentioned, the CAT-PEST fusion protein was pulled down by GST-CaMLD even in the absence of added calcium, though slightly less efficiently than when calcium was added.
This interaction could indicate that the protein-protein association is not completely calcium-dependent. Alternatively, contaminating calcium ions co-purified with GST-CaMLD from bacterial lysate or were provided from rabbit reticulocyte lysate and facilitated the interaction. When residual calcium ions were eliminated by the addition of EGTA to the binding buffer we observed that CAT-PEST no longer associated with GST-CaMLD (Fig. 5B). Therefore, the binding of the IB␣-PEST sequence to GST-CaMLD is calcium-dependent.

Conferred Association with CaMLD by IB␣-PEST Is Sufficient to Confer Susceptibility to Proteolysis to a Nonsubstrate
Protein-We have demonstrated that the PEST domain of IB␣ is sufficient to allow the binding of a chimeric protein to CaMLD in a calcium-dependent manner. Also, the association of IB␣ with CaMLD appears to be a requirement for its calpain-mediated degradation in vitro. This evidence suggests that by fusing the IB␣-PEST domain to a nonsubstrate protein we would be able to promote the association and subsequent proteolysis of the chimeric protein by calpain. Also, however, we have found using gradient gels and Tris-Tricine gel electrophoresis that calpain preferentially cleaves IB␣ at one or both of two sites within the IB␣ N terminus (data not shown). Consequently, in transferring the N-terminal 66 amino acids of IB␣ to CAT we add to CAT one or more calpain cleavage sites. Therefore, to test whether the binding to calpain or the presence of a preferred calpain cleavage site(s) is sufficient for proteolysis, we analyzed the sensitivities to calpaindependent degradation of the CAT protein and related CAT chimeric proteins.  1, 6, and 11. IB␣ or mutant input proteins are labeled with asterisks. Exposure times for IB␣ and MutF pull-downs was 40 h, and for IB⌬C, 54 h. Molecular mass in kDa is indicated to the right. B, relative amount of IB␣ proteins associated with GST-CaMLD. Percentage of input wild-type IB␣ recovered by GST-CaMLD in the presence of added calcium was used as 100% to calculate the relative amounts of input proteins recovered by GST-CaMLD in the presence or absence of added calcium. Similar experiments (as outlined in A) were performed with IB␣ translated in vitro and CIP-treated. Data for each reaction is tabulated from two to three independent experiments.  1 and 2) were allowed to bind GST-CaMLD in the presence or absence of 1.5 mM EGTA. Gels were exposed for the same amount of time.
The CAT protein was not detectably proteolyzed by calpain even at the highest enzyme concentration tested (100 nM, Fig.  6A). On the other hand, the CAT-PEST fusion protein displayed a degradation profile similar to that of IB␣. N-CAT was only weakly susceptible to calpain proteolysis, but when the IB␣-PEST domain was fused to N-CAT (N-CAT-PEST) this chimeric protein was efficiently degraded by calpain in vitro. As expected, the 66 amino acids from IB␣ that were fused to CAT appear to contain the two preferred calpain cleavage sites, because Western blotting with antibody raised against the IB␣ C terminus following degradation of N-CAT-PEST by calpain detects two intermediates that migrate at the predicted molecular mass for degradative products generated by cleavage at these two sites (Fig. 6B). Therefore, transfer of the IB␣ N-terminal calpain cleavage sites alone to CAT is not sufficient to allow the fusion protein to be degraded by calpain, nor does it make the CAT-PEST chimera a better substrate for calpain. However, transfer of the IB␣-PEST domain to CAT does confer to the fusion protein sensitivity to calpain degradation in vitro.
We designed a competition experiment to test the model of calpain-dependent proteolysis in which association of IB␣ to the CaMLD of calpain precedes IB␣ degradation. Increasing amounts of either GST (in molar excess over IB␣ of up to 200-fold) or GST-CaMLD (in molar excess over IB␣ of up to 80-fold) were added to a reaction mixture containing [ 35 S]methionine-labeled IB␣ and calpain. Following incubation, the amount of IB␣ substrate remaining was analyzed (Fig. 6C). Addition of GST-CaMLD displayed a dose-dependent block of IB␣ proteolysis, whereas similar amounts of GST added alone had no effect. To ensure that GST-CaMLD did not directly inhibit calpain activity in these experiments, excess GST-CaMLD was incubated with calpain before addition of IB␣ substrate. The ability of the calpain protease to digest IB␣ was not diminished (data not shown). Together these studies demonstrate that the IB␣-PEST sequence is both necessary and sufficient for association with the CaMLD of calpain, which subsequently confers susceptibility to -calpain-dependent proteolysis in vitro.

DISCUSSION
Activation of the transcription factor NF-B by a wide variety of inducing agents is preceded by the proteolytic inactivation of the inhibitory molecule IB␣. Because of its central role in NF-B signaling, great effort has gone into delineating the events required to degrade IB␣ through the ubiquitin-proteasome pathway. In addition to proteasome-dependent IB␣ proteolysis, several groups have either directly or indirectly implicated calpain as an in vivo IB␣ protease (15)(16)(17)(18)(19). However, the mechanism of IB␣ degradation by calpain is uncertain. In this report we examined the structural determinants of IB␣ required for degradation by calpain in vitro and demonstrate a functional role for the C-terminal PEST domain of IB␣ as a necessary and sufficient calpain-targeting module. A necessary role for the PEST sequence is inferred because deletion of this region inhibits the degradation of IB␣ by calpain, and mutation of residues within this domain reduces the efficiency of calpain-dependent IB␣ proteolysis as well. Furthermore, the hierarchy of sensitivity to calpain degradation among the IB␣-PEST mutants is mirrored in their relative abilities to associate with the CaMLD of the large calpain subunit, i.e. IB␣ Ͼ MutF Ͼ IB⌬C. The finding that transfer of the IB␣-PEST domain to CAT yields a chimeric molecule that only then binds to and is a substrate for calpain indicates that this sequence is sufficient for calpain-dependent proteolysis.
Phosphorylation of some proteins within PEST or PEST-like sequences is necessary for their recognition by F-box or other proteins that serve to recruit the substrate to a ubiquitin-ligase complex, and degradation through the 26 proteasome ensues (44 -48). Although it has long been proposed that by sequestering calcium, the PEST motif could form a calcium-dependent interaction with calpain that would place the catalytic site of calpain in close proximity to the PEST-containing substrate (3,24) to our knowledge this is the first demonstration of direct association between a PEST sequence and calpain or any other protease.
We determined that the degradation of IB␣ by calpain is preceded by cleavage at either one or both of two N-terminal  6 -9) bound to glutathione-Sepharose were added to reactions containing [ 35 S]methionine-labeled IB␣ and 100 nM calpain. After 10 min on ice, the reactions were incubated at 30°C for 15 min and terminated samples were separated by SDS-PAGE followed by autoradiography (top panel). Four times the amounts of GST or GST-CaMLD, which were added to each reaction, were separated by SDS-PAGE followed by Coomassie Blue (CB) staining and are shown in the bottom panel.
sites (data not shown). This is consistent with findings based both on -calpain and m-calpain (16,18,49). A strong preference for these sites by calpain is evident, because the N-CAT-PEST protein displays a similar pattern to IB␣ as it is proteolyzed. The results suggest that one requisite role for calcium in calpain proteolysis may be proximity, i.e. the recruitment of the protein substrate to the enzyme. Although the 66 N-terminal residues of IB␣ contain strong calpain cleavage sites, these alone were not sufficient to confer calpain susceptibility to CAT. However, when CAT was brought into association with calpain in the presence of calcium by means of the IB␣-PEST sequence, proteolysis ensued. Thus our data indicate that a calcium-dependent association, not catalysis, is the rate-limiting step in this reaction. Though it was not demonstrated, a similar role for the PEST sequences of the common cytokine receptor ␥ chain may exist, because this protein bound to the CaMLD of the calpain small subunit and was not degraded by calpain when the PEST sequences were removed (50).
Recently, evidence has accumulated, which suggests that the ubiquitin-proteasome proteolytic pathway may be responsible for degradation of many PEST-containing proteins (51). Indeed, recent reports demonstrate that the PEST domains of three different calpain substrates are dispensable for calpain proteolysis (52)(53)(54). In one instance it was shown that rather than the PEST domains, the calmodulin-binding domain of the calcium-dependent ATPase was critical for its degradation by calpain (52). A common theme among the vast majority of calpain substrates is that they contain either a PEST sequence, a CaM binding domain, or both. Though unlikely to apply to all calpain substrates, it is intriguing to think that in these proteins the PEST domains or CaM binding domains serve overlapping functions, i.e. by promoting the association of substrate with CaMLD these protein motifs could act interchangeably as calpain-homing sequences.
Several groups have shown that the basal turnover of IB␣ becomes retarded when the casein kinase II phosphorylation sites within the PEST sequence are deleted or replaced with alanine (25,26,39,55,56). Interestingly, our data indicate that phosphorylation of IB␣ at casein kinase II sites within the PEST domain enhances both the association with and the degradation by calpain in vitro, consistent with the possibility that calpain activity, either alone or in addition to the proteasome (56), may be responsible for the basal degradation of IB␣ under some in vivo conditions. Furthermore, we show that the N-terminal signal responsive domain of IB␣ does not affect calpain-dependent proteolysis. In keeping with this, we discovered that neither the induced phosphorylation of IB␣ at serines 32 and 36, nor the substitution of alanine at these sites, alters the rate and pattern of IB␣ degradation by calpain (data not shown). These data indicate that the S32A/S36A mutants, which have been used to identify the inducible phosphorylation-dependent NF-B activation pathway, can be effectively degraded by calpain. Thus, the S32A/S36A IB␣ protein will likely prove to be a useful reagent in distinguishing the two proteolytic mechanisms. Similarly, future investigations directed at determining the amino acid components and points of contact along the interface between IB␣-PEST and CaMLD could be helpful to generate IB␣ mutants specifically resistant to calpain-dependent proteolysis. Such studies may also help to specify the role of PEST motifs as general calpain-targeting sequences.
It will also be of interest to resolve which, if any, PEST sequences derived from other proteins, such as other IB members, are able to interact with the CaMLDs of calpain. IB␤ has a C-terminal PEST sequence similar to IB␣, and IB⑀ has putative PEST sequences in its amino-terminal domain. We have observed that in vitro, IB␤ is readily degraded by -calpain, 2 suggesting potential calpain-mediated regulation of NF-B complexes in association with IB members other than IB␣. Additionally, it may be insightful to examine the ability of calpain substrates that lack high PEST scores to associate with the CaMLD. The findings in this report suggest that association through PEST sequences may be an indication of a protein's sensitivity to calpain proteolysis, though one can imagine CaM binding domains or other uncharacterized motifs playing a similar role. Though the contribution of IB␣-PEST to calpain-dependent IB␣ proteolysis in vivo has not been investigated here, our demonstration of its calpain-targeting potential in vitro provides information on IB␣ structure that can be used to examine systems where calpain is known or suspected to be activated.