Calmodulin Binds and Stabilizes the Regulatory Enzyme, CTP:Phosphocholine Cytidylyltransferase*

CTP:phosphocholine cytidylyltransferase (CCTα) is a proteolytically sensitive enzyme essential for production of phosphatidylcholine, the major phospholipid of animal cell membranes. The molecular signals that govern CCTα protein stability are unknown. An NH2-terminal PEST sequence within CCTα did not serve as a degradation signal for the proteinase, calpain. Calmodulin (CaM) stabilized CCTα from calpain proteolysis. Adenoviral gene transfer of CaM in cells protected CCTα, whereas CaM small interfering RNA accentuated CCTα degradation by calpains. CaM bound CCTα as revealed by fluorescence resonance energy transfer and two-hybrid analysis. Mapping and site-directed mutagenesis of CCTα uncovered a motif (LQERVDKVK) harboring a vital recognition site, Gln243, whereby CaM directly binds to the enzyme. Mutagenesis of CCTα Gln243 not only resulted in loss of CaM binding but also led to complete calpain resistance in vitro and in vivo. Thus, calpains and CaM both access CCTα using a structurally similar molecular signature that profoundly affects CCTα levels. These data suggest that CaM, by antagonizing calpain, serves as a novel binding partner for CCTα that stabilizes the enzyme under proinflammatory stress.

Tight homeostatic control of mammalian phosphatidylcholine (PtdCho) 2 biosynthesis is critical to maintain cell membrane integrity and to ensure optimal secretion of key products, such as serum lipoproteins, bile, and pulmonary surfactant. In eukaryotes, the enzyme CTP:phosphocholine cytidylyltransferase (CCT) catalyzes the pivotal step for PtdCho synthesis (1). CCT is both rate-limiting and rate-regulatory for PtdCho syn-thesis, and its deficiency is linked to apoptosis, impaired cell growth, and embryonic lethality (2)(3)(4). There are four isoforms of CCT in cells: CCT␣, CCT␤1, CCT␤2, and CCT␤3 (5). The primary structure of the dominant species, CCT␣, consists of 367 amino acids with four functional domains: an NH 2 -terminal nuclear localization signal, a catalytic core (C), a membranebinding domain containing a hydrophobic ␣-helix (M), and a carboxyl-terminal phosphorylation domain (P). CCT␣ activity is regulated by activating and inhibitory lipids, by reversible phosphorylation, and at the level of mRNA (5,6). Little is known regarding other mechanisms that control its enzymatic behavior.
CCT␣ protein turnover might be a physiologically important control mechanism, since the enzyme is degraded by calciumactivated neutral proteinases (calpains) and death effector caspases (7)(8)(9). Calpain severs CCT␣ at the amino terminus and within the catalytic domain-membrane binding domain boundary (7). Calpains exist in cells as two major isoforms, depending on calcium requirements: M-calpain and -calpain. Each isoform consists of two distinct subunits, a larger 80-kDa catalytic subunit and a smaller 30-kDa regulatory subunit, forming a heterodimeric structure. The large and small subunits consist of four (I-IV) and two (I-II) domains, respectively (10). EF-hand motifs within each subunit allow for heterodomain interactions and calcium binding. Calpain cleaves its substrates, in part, by docking to two major motifs within its substrates: PEST (proline-glutamate-serine-threonine) sequences and calmodulin (CaM) binding domains. PEST sequences within IkB␣ and the ATP-binding cassette transporter 1 serve as proteolytic signatures for calpain degradation (11,12). Likewise, CaM binding domains within a calcium-ATPase pump and inducible nitric-oxide synthase impact their sensitivity to calpain hydrolysis (13,14). A "calmodulin-like" domain also exists within the catalytic subunit of calpain that facilitates interaction with some PEST sequences or CaM binding motifs (15,16). Indeed, data base analysis (available on the World Wide Web) identified a consensus PEST sequence within the CCT␣ NH 2 -terminal domain. Thus, this PEST sequence or other structural motifs that recognize CaM might brand CCT␣ for its elimination within cells.
Despite being a calpain substrate, CCT␣ is a relatively stabile enzyme. CCT␣ is highly abundant in cells, is cytosolic in pneumocytes, and has an extended half-life (17). Typically, larger, hydrophobic, cytosolic proteins and regulatory enzymes exhibit faster turnover rates (18). The half-life of CCT␣ (ϳ8 h) also exceeds that of other metabolic enzymes, including hydroxymethylglutaryl-CoA reductase and phosphoenolpyru-vate carboxykinase (19). These observations strongly suggest existence of covalent modifications, stabilizing ligands, or binding partners that enhance the life span of CCT␣ in vivo.
CaM (16.7 kDa) is a highly conserved calcium-sensing protein that binds and modulates stability of some cytoskeletal and ion transport proteins (20 -22). CaM binds proteins in its calcium-bound (holo-CaM) or calcium-free form (apo-CaM). CaM binding proteins are thus classified into Ca 2ϩ -dependent, Ca 2ϩ -independent, and Ca 2ϩ -inhibited proteins (20). Many CaM binding proteins harbor recognition motifs characterized by a basic amphipathic helix, moderate to high helical hydrophobic moment, and a net positive charge (20). Other motifs described include an IQ motif ((I/L)QXXXRGXXXR), a 1-8-14, and a 1-5-10 CaM binding motif (20). Although CaM protects some substrates from proteolysis, the molecular basis for these observations has not been fully elucidated.
In the present study, we investigated the hypothesis that specific molecular sequence signatures confer stability to CCT␣. We show for the first time that CCT␣ is a CaM-binding enzyme and that CaM protects CCT␣ from calpain degradation. A conceptually unique finding of this study is that a highly conserved residue (Gln 243 ) (rather than an NH 2 -terminal PEST sequence) serves as an essential molecular recognition site for competition between CaM and calpain for access to CCT␣. Mutagenesis of Gln 243 within CCT␣ totally blocked CaM binding but also the ability of calpain to degrade CCT␣ in vitro and in vivo. The intermolecular competition between a proteinase and a stabilizing protein for access to a single recognition site within CCT␣ represents a novel mechanism regulating an enzyme's availability.

EXPERIMENTAL PROCEDURES
Materials-The sources of murine lung epithelial (MLE) cells, LDL, CCT␣, extracellular signal-regulated kinase (ERK) antibodies, TrueBlot IgG, FuGENE6 transfection kits, and transcription and translation (TNT) coupled reticulocyte lysate were described previously (23). Rabbit polyclonal antibodies to M-and -calpain were from ABR-Affinity BioReagents (Golden, CO). Rabbit monoclonal calmodulin antibody was purchased from Upstate (Billerica, MA). Rabbit polyclonal calmodulin kinase II antibody was purchased from Epitomics (Burlingame, CA). Purified -calpain, recombinant CaM, and the calpain substrate peptide, LLVY, were purchased from Calbiochem. The pCR-TOPO4 cloning kit, Escherichia coli One Shot competent cells, pENTR Directional TOPO cloning kits, and the Gateway mammalian expression system were purchased from Invitrogen. The QuikChange site-directed mutagenesis kit and the X-blue cells were purchased from Stratagene (La Jolla, CA). The gel extraction kit and QIAprep Spin Miniprep Kit were from Qiagen (Valencia, CA). Nucleofector transfection kits were from Amaxa (Gaithersburg, MD). Calmodulin-Sepharose 4B beads were purchased from Amersham Biosciences. Immobilized glutathione-agarose beads were purchased from Pierce. BD TALON purification and buffer kits were purchased from BD Biosciences. Calmodulin siRNAs were purchased from Dharmacon (Chicago, IL). A mammalian calmodulin cDNA, pEx1-CaM, was kindly provided by Dr. Madeline Shea (University of Iowa, Iowa City, IA) (24). All DNA sequencing was performed by the University of Iowa DNA core facility.
Construction of CCT␣ PEST Mutants-A CCT␣ variant harboring point mutations in the PEST domain (CCTPEST SDM ) was constructed where Thr 25 and Ser 32 were mutated to Ala using the QuikChange site-directed mutagenesis kit. A fulllength CCT␣ template (pCMV-CCT FL ) plasmid DNA was used as a template. The primers used to mutate Thr 25 were 5Ј-GCC-CTAATGGAGCAGCAGAGGAAGATGG-3Ј (forward) and 5Ј-CCATCTTCCTCTGCTGCTCCATTAGGGC-3Ј (reverse). The primers used to mutate Ser 32 were 5Ј-GAAGATGGAAT-TCCTGCCAAAGTGCAGCGC-3Ј (forward) and 5Ј-GCGCT-GCACTTTGGCAGGAATTCCATCTTC-3Ј (reverse). PCR conditions were as follows: initial denaturation at 95°C for 2 min and then denaturation at 95°C for 15 s, annealing at 55°C for 30 s, elongation at 68°C for 6 min, 18 cycles for three steps.
An internal deletion mutant lacking the PEST sequence (CCTPEST SOE ) was constructed using splicing by overlapping extension PCR. Full-length CCT␣ cloned into TOPO4 (TOPO-CCT FL plasmid) was used as a template. Four primers, CCTPEST1 to -4, were designed (CCTPEST1, 5Ј-CAC-CATGGATGCACAGAGTTCAGC3Ј; CCTPEST2, 5Ј-ACT-GCACAGCGCTGCACTTTCCTCCTCTTCCTTGAATT-GACTTTA-3Ј; CCTPEST3, 5Ј-TAAAGTCAATTCAAGGAA-GAGGAGGAAAGTGCAGCGCTGTGCAGT-3Ј; CCTPEST4, 5Ј-TCAGTCCTCTTCATCCTCGCTG-3Ј). In the first step, primers CCTPEST1 and CCTPEST2 were used to amplify an NH 2 -terminal fragment of CCT␣. In the second step, primers CCTPEST3 and CCTPEST4 were used to amplify a COOHterminal CCT␣ fragment. Each fragment flanked the PEST. In the last step, the two gel-purified fragments from the steps above were used as a template in the final PCR using primers CCTPEST1 and CCTPEST4 to amplify a desired 1050-bp product lacking the PEST. This fragment was purified and cloned into pPCR4-TOPO. The PCR conditions were as follows: 95°C for 30 s and 18 cycles at 95°C for 30 s, 55°C for 60 s, and 68°C for 5 min. After DNA sequence confirmation, TOPO-CCTPEST SOE was used as the template using the primers 5Ј-AAGCTTATGGATGCACAGAGTTCAGC-3Ј (forward) and 5Ј-CTCGAGGTCCTCTTCATCCTCGCTG-3Ј (reverse) to amplify the 1050-bp CCT␣ fragment. The forward primer contains a recognition site for HindIII, the reverse primer contains a recognition site for XhoI. The PCR product was cloned into pPCR4-TOPO, followed by digestion with the same enzymes prior to directional cloning into pcDNA3.1/V5-his.
Construction of Glutathione S-Transferase (GST)-tagged CCT␣ Domain and Carboxyl-terminal Mutants-A series of internal CCT␣ domain deletion mutants were constructed using TOPO-CCT MEM and TOPO-CCT CAT , TOPO-CCT PEST that were first generated as described previously or by using splicing by overlapping extension PCR (23). These constructs were used as a template in PCR to generate GST-CCT MEM and GST-CCT CAT , two constructs devoid of the membrane binding domain or catalytic domain, respectively. The forward primer, 5Ј-CACCATGGATGCACAGAGTTCAGCT-3Ј, and reverse primer, 5Ј-GTCCTCTTCATCCTCGCTGA-3Ј, were used in the PCR conditions as follows: 95°C for 30 s and 25 cycles at 95°C for 30 s, 55°C for 30 s, and 68°C for 1 min. The PCR products were gel-purified and cloned into pENTR-TOPO.
A CCT␣ variant harboring a point mutation (CCT Q243A ) where Gln 243 was mutated to Ala was generated using the QuikChange site-directed mutagenesis kit. GST-CCT FL plasmid was used as a template for PCR using forward primer 5Ј-GAAAAGAAATACCACTTGGCAGAACGAGTTGATA-AGG-3Ј and reverse primer 5Ј-CCTTATCAACTCGTTCTG-CCAAGTGGTATTTCTTTTC-3Ј. The thermal cycling program was as follows: initial denaturation at 95°C for 2 min and then denaturation at 95°C for 30 s, annealing at 55°C for 30 s, elongation at 68°C for 10 min, 18 cycles for three steps. The desired PCR product was gel-purified and fused to GST as above. A CCT␣ carboxyl-terminal deletion mutant harboring a similar point mutation (CCT267 Q243A ) was generated by PCR using GST-CCT 267 as the template using methods described above. All of the above PCR products in pENTR-TOPO were verified by DNA sequencing.
In Vitro TNT-CCT␣ cDNA constructs cloned into pCR4-TOPO4 (2 g of plasmid/reaction) were added directly to the rabbit reticulocyte lysate, incubated in 50 l of reaction solution containing 2 g of plasmids, 25 l of rabbit reticulocyte lysate, 2.5 l of RNase inhibitor, 1.2 l of 1 mM amino acids (minus methionine), 2 l of T7 RNA polymerase, and 5 l of [ 35 S]methionine (40 Ci/reaction). The reaction mixture was incubated at 30°C for 90 min as described (23).
Calpain Proteolysis Assay-A 25-l reaction volume containing 10 l of TNT reaction products, 10 l of calpain buffer (20 mM Tris, pH 7.5, 2 mM dithiothreitol, 1% Tween 20, and 0.015% Triton X-100), and 0.25-1 g of purified -calpain was incubated at 37°C for 0 -1 h after adding CaCl 2 to a final concentration of 200 M. The reaction was terminated by adding 2ϫ SDS protein loading buffer and heating to 95°C for 5 min. Effects of calpain hydrolysis of CCT␣ were further tested in separate studies by inclusion of varying concentrations of CCT␣, CaM, or the calpain substrate, LLVY, in the reaction mixture. In these studies, calpain was present at a fixed concentration of 0.6 pmol/reaction. The digestion products were resolved by SDS-PAGE, and gels were processed for autoradiography or immunoblotting. In other experiments, purified recombinant GST-CCT␣ and GST-CCT␣ mutants were used as substrates for calpain (0.005-0.05 g) digestion.
Cell Culture-MLE cells were cultured in Dulbecco's minimum essential medium containing 0% fetal bovine serum for up to 48 h with or without Ox-LDL (100 g/ml). Cell lysates were prepared by brief sonication in 150 mM NaCl, 50 mM Tris, 1.0 mM EDTA, 2 mM dithiothreitol, 0.025% sodium azide, and 1 mM phenylmethylsulfonyl fluoride (Buffer A) at 4°C prior to analysis. Cytosolic and microsomal preparations were isolated as described (25).
Lipoprotein Oxidation-Lipoproteins were dialyzed in phosphate-buffered saline at 4°C for 24 h followed by oxidation in 5 M CuSO 4 /phosphate-buffered saline for 24 h at 37°C. Confirmation of lipoprotein oxidization was by the malonaldehyde assay and by detection of apoprotein B-100 degradation as described (7).
CCT Activity-Enzyme activity was determined by measuring the rate of incorporation of [methyl-14 C]phosphocholine into CDP-choline using a charcoal extraction method (7). Assays were conducted with and without exogenous PtdCho/ oleic acid lipid activator in the reaction mixtures.
PtdCho Analysis-Cells were pulsed-labeled with 1 Ci of [methyl-3 H]choline chloride during the final 3 h of incubation, lipids were extracted and resolved, and activity within PtdCho was analyzed as described (7).
Immunoblot Analysis-Equal amounts of total protein (5-20 g) in sample buffer were resolved using 10% SDS-PAGE and transferred to nitrocellulose, and immunoreactive CCT␣ or calmodulin was detected as described (23). The dilution factor for primary and secondary antibodies was 1:2000. CCT␣ was purified to homogeneity as described (26).
Co-immunoprecipitation-200 g of total protein from MLE cell lysates were precleared with 20 l of Trueblot anti-Ig beads for 1 h at 4°C. 5 g of CCT␣, ERK, CaM, rabbit IgG, or calmodulin kinase II antibodies were added for a 2-h incubation at 4°C. 20 l of Trueblot anti-Ig beads were added for an additional 2-h incubation. Beads were spun down and washed five times using 50 mM HEPES, 150 mM NaCl, 0.5 mM EGTA, 50 mM NaF, 10 mM Na 3 VO 4 , 1 mM phenylmethylsulfonyl fluoride, 20 M leupeptin, and 1% (v/v) Triton X-100 (radioimmune precipitation) buffer as described (23). Beads were heated at 100°C for 5 min with 80 l of protein sample buffer prior to SDS-PAGE and immunoblotting.
Expression and Knockdown of Recombinant Proteins-CCT␣ PEST mutants were expressed in cells using the Amaxa nucleofector system per the manufacturers' instructions. Cellular expression of green fluorescent tagged plasmids using this device was achieved at Ͼ90% in MLE cells. Transfection of GST-CCT␣ fusion constructs was also conducted for 24 h in Dulbecco's minimum essential medium/F-12 medium containing 0% fetal bovine serum using 18 l of FuGene6 reagent and 6 -10 g/dish of the desired plasmid. After 24 h, the cells were harvested in Buffer A followed by brief sonication. In some studies, 24 h after transfection, 100 g/ml Ox-LDL was added for an additional 24 h. For overexpression of CaM, 1 ϫ 10 6 cells were infected with Adv-CaM or an empty vector (Adv-Con) at MOI ϭ 40 for 6 h prior to an additional 24-h incubation with Ox-LDL. For CaM knockdown, 1 ϫ 10 6 cells were electroporated with 1 nmol of siRNA. Cells were either transfected with siRNA (sense sequence, 5Ј-UGACAAACCUUGGAGA-GAAUU-3Ј; antisense sequence, 5Ј-PUUCUCUCCAAGGU-UUGUCAUU-3Ј) against CaM or with a control siRNA. 72 h after transfection, cells were exposed to Ox-LDL for an additional 24 h prior to harvest. In rescue studies, cells were either transfected with CaM siRNA or control siRNA; 48 h after transfection, cells were infected with Adv-CaM for another 24 h. Finally, Ox-LDL was added to the medium for an additional 24 h prior to harvest.
CCT␣ Degradation-CCT␣ degradation was determined by nucleofecting CaM siRNA or control siRNA in MLE cells as above, and 72 h later, cells were preincubated for 1 h in methionine-deficient medium and then pulsed with [ 35 S]methionine (60 Ci/ml) for 4 h at 37°C as described (25). Cells were rinsed, chased in medium replete with methionine and cysteine for 0 -8 h, and processed for CCT␣ immunoprecipitation, SDS-PAGE, and autoradiography as described (7).
GST Pull-down Assays-After plasmid transfection, cellular lysates were prepared as described above, followed by incubation with 20 l of immobilized glutathione-agarose beads at 4°C for 1 h. After incubation, the beads were spun down and rinsed three times using buffer containing 250 mM NaCl and 0.2% Nonidet P-40. Beads were heated at 100°C for 5 min with 80 l of protein sample buffer for subsequent immunoblotting. For purification of recombinant GST-CCT␣, 100 l of immobilized glutathione-agarose beads were incubated with cell lysates (prepared from two 100-mm dishes) at 4°C for 4 h. After incubation, beads were washed as described above. Recombinant GST-CCT␣ proteins were eluted using a 5 mM glutathione Buffer A solution, followed by concentration using YM-30 spin columns.
Calmodulin-Sepharose Binding Assay-20 l of CaM-Sepharose beads were incubated with cell lysates (30 g) or purified rat liver CCT␣ (1-2 g) with or without calcium at 4°C for 2 h. After incubation, beads were spun down and washed three times using buffer containing 300 mM NaCl and 0.1% Nonidet P-40. Beads were heated at 100°C for 5 min with 40 l of protein sample buffer. Released products were resolved using SDS-PAGE prior to immunoblotting.
Mammalian Two-hybrid Binding Assay-CCT␣ was PCRamplified using GST-CCT␣ as a template and cloned into the pM vector (Clontech) that expresses the CCT␣-Gal4BD fusion protein. CaM was also PCR-amplified using the Adv-CaM plasmid and cloned into a pVP16 Gal4AD vector that expresses a CaM-Gal4AD fusion protein. After sequence confirmation, CCT␣-Gal4BD, CaM-Gal4AD, and pG5CAT reporter vector were co-electroporated into cells per the manufacturers' instructions. 48 h after transfection, cells were lysed and assayed for ␤-galactosidase activities. pM-53 and pVP16-T plasmids served as positive controls. pM3-VP16 and pVP16-CP plasmids served as negative controls.
Fluorescence Resonance Energy Transfer (FRET) Analysis-PCR-based strategies were used to create compatible restriction enzyme sites that would allow construction of chimeric cDNAs for FRET. A 1050-bp CCT␣ fragment was amplified in PCR using the primers 5Ј-ACAAGATCTATGGATGCACAG-AGTTCAGCT-3Ј (forward) and 5Ј-ACTGTCGACTTAGTC-CTCTTCATCCTCGCTG-3Ј (reverse) using pCMV5-CCT as a template. The forward primer contains a recognition site for BglII, and the reverse primer contains a recognition site for SalI. The PCR product was gel-purified and digested with the same enzymes prior to cloning into pAmCyan-C1 (Clontech), generating CFP-CCT␣.
An Adv-CaM vector (below) was used as template in PCR using the primers 5Ј-ACTAGATCTATGGCTGATCAGCTG-ACCG-3Ј (forward) and 5Ј-ACTGTCGACTCATTTTGCAG-TCATCATCTGTAC-3Ј (reverse) to amplify a 450-bp CaM fragment. The forward primer contains a recognition site for BglII, and the reverse primer contains a recognition site for SalI. This PCR product was gel-purified and digested with BglII and SalI prior to cloning into a linearized pZsYellow1-C1 vector (Clontech), generating YFP-CaM.
For analysis of CCT␣ and CaM interaction by FRET, cells were first plated at 0.12 ϫ 10 6 cells/well in a two-chamber cover glass system. Cells were co-transfected with YFP-CaM and CFP-CCT␣ (2 g of plasmid/chamber) with Fugene 6 (6 l). CaM-CCT␣ interaction was detected at the single cell level using a combination laser-scanning microscope system (LSM510/ConfoCor2; Zeiss, Jena, Germany). To achieve excitation, the 458-nm line of an argon ion laser was focused through the Zeiss ϫ60 oil differential interference contrast objective lens onto the cell. Emissions of YFP (the FRET acceptor) and CFP were collected through 535-595-nm and 470 -500-nm barrier filters, respectively. Photobleaching was performed with 50 iterations and 100% intensity of a 514-nm laser. FRET quantitation of fluorescence images were generated using Zeiss Rel3.2 image software. The average fluorescence intensities/per pixel were calculated following background subtraction.
Construction of an Adenoviral CaM Vector-A pEx1-CaM plasmid was used as a PCR template using the primers 5Ј-CTC-GAGATGGCTGACCAACTGACTGA-3Ј (forward) and 5Ј-GGATCCCTTTGCTGTCATTTGTACAAAC-3Ј (reverse) to amplify a 450-bp CaM fragment. The forward primer has an engineered XhoI site, and the reverse primer has an engineered BamHI site. PCR conditions were as follows: 95°C for 30 s and 35 cycles at 95°C for 30 s, 55°C for 30 s, and 72°C for 1 min. The PCR products were cloned into pPCR4-TOPO followed by digestion with XhoI and BamHI. The pacAD5 cytomegalovirus IRES eGFP pA vector was also digested with XhoI and BamHI. These digestion products were fractionated by gel electrophoresis, and the desired 450-bp and 7.5-kb fragments were then purified and ligated using T4 DNA ligase at 25°C for 1 h. The Gene Transfer Vector Core (University of Iowa) used the newly constructed adenovirus-CaM shuttle plasmid to generate a first generation adenovirus-CaM expression vector (27). Adenovirus vectors expressing the CaM transgene driven by the cytomegalovirus promoter (Adv-CaM) or an empty control vector containing the cytomegalovirus promoter but no trans-gene (Adv-Con) were used in experiments. Adenoviral vectors were replication-deficient (deletion of the E1 gene) and free of wild type contamination as determined by plaque assay and by PCR for E1 sequences. The particle titers of adenoviral stocks were ϳ10 12 particles/ml that were used in studies.
Statistical Analysis-Statistical analysis was performed using one-way analysis of variance with a Bonferroni adjustment or Student's unpaired t test. Data are expressed as mean Ϯ S.E.

CCT␣ PEST Mutants Are Not Resistant to Calpain-A strong
PEST sequence was identified in CCT␣ using the PEST-FIND algorithm (PEST find Analysis webtool) (Fig. 1A). The calculated PEST score was ϩ8.56 (PEST scores greater than ϩ0 are considered highly significant). There are four less conserved CCT␣ PEST sequences, with scores ranging from Ϫ1.37 to Ϫ18.37. Mutagenesis was performed to disrupt this signature motif and potentially reduce the ability of calpain to degrade CCT␣. Mutagenesis of highly conserved Thr 25 and Ser 32 to Ala resulted in loss of the PEST sequence as identified by this algorithm. First, three constructs, CCT␣ full-length (CCT␣ FL ) and CCT SDM and CCT SOE (mutants with disrupted PEST motifs), were synthesized in vitro and incubated with calpain, and reaction products were resolved and visualized by autoradiography.
Calpain produced a dose-dependent decrease in the levels of the 42-kDa CCT␣ FL product (Fig. 1B, upper left). CCT␣ FL was degraded within 15 min of calpain exposure (Fig. 1B, lower left) (6). Calpain (0.5 g) effectively hydrolyzed CCT SDM and CCT SOE within 60 min (Fig. 1B, upper right), and at higher calpain concentrations (1 g) levels of PEST mutants were undetectable within 30 min (Fig. 1B, lower  right). Thus, disruption of the PEST sequence in CCT␣ does not protect against calpain in vitro. To assess the physiologic role of the CCT␣ PEST motif, functional CCT␣ FL or CCTPEST SDM plasmids were transiently expressed in MLE cells prior to exposure to Ox-LDLs that activate calpains (7). In MLE cells, CCT␣ is detected in cytosol, and the remainder is membrane-associated (data not shown). Ox-LDL reduced immunoreactive CCT␣ levels in both the cytosolic and membrane fractions (data not shown). Ox-LDL uniformly produced a significant decrease in CCT activity to ϳ50 -60% of control values in untransfected cells when assayed in the absence of exogenous lipid (reflecting membrane-bound enzyme) (Fig. 1C) and from 5.28 Ϯ 0.2 to 3.12 Ϯ 0.60 nmol/ min/mg protein when assayed with exogenous lipid (total enzyme). Thus, Ox-LDL did not inactivate CCT␣ by simply shifting CCT␣ off of the membrane into the cytosol but rather by degrading the enzyme in both cellular compart- ments. Further, consistent with other calpain substrates, CCT␣ degradation products are not easily identified, since these fragments are probably cleared rapidly in cells by endopeptidases or the proteasome thereby often evading detection (11,28). The sizes of products of the 20 and 26 S proteasome are only 500 Da (5-6 residues) and would run near the dye front of the PAGE, requiring more sensitive approaches for their detection. Oxidized lipids also impaired CCT activity in cells transfected with either CCT␣ FL or CCTPEST (Fig. 1C). In these experiments, high level expression of CCT␣ FL or CCTPEST plasmids by nucleofection was observed; reductions of the levels of both overexpressed and endogenous CCT␣ proteins were observed after Ox-LDL treatment (Fig. 1D). These data indicate that the CCT␣ NH 2terminal PEST sequence within does not confer resistance to actions of calpains.
CCT␣ Is a CaM-binding Protein-An alternative recognition signal for calpain in substrates is a CaM-binding domain (13,14). We first investigated whether CCT␣ interacts with CaM. Thus, CaM or ERK were immunoprecipitated, followed by immunoblotting with anti-CCT␣ antibodies. CCT␣ was detected in association with immunoprecipitated CaM from cell lysates ( Fig. 2A, top); consistent with our prior studies, CCT␣ also bound to ERK, whereas this association was not detected using negative controls (rabbit IgG or beads alone). Conversely, immunoprecipitation of CCT␣ or calmodulin kinase II followed by immunoblotting with CaM antibody revealed that CaM was detected in association with CCT␣ and calmodulin kinase II (positive control) but not with preimmune serum (negative control) ( Fig. 2A, bottom). Next, cells were transfected with GST-CCT␣ fusion proteins, and GST pulldown products were eluted and processed for CCT␣ and CaM immunoblotting (Fig. 2B). CaM was detected in association with overexpressed, purified GST-CCT␣, whereas this association was not demonstrated in untransfected cells or by using agarose beads. Further, cell lysates were run over CaM-agarose beads, the beads were extensively rinsed using buffer containing 0.1% Nonidet P-40, and products were eluted. The elution products were then resolved by SDS-PAGE prior to CCT␣ immunoblotting. Different calcium concentrations ranging from 0 to 2000 M were used in the binding buffer. Indeed, CaM interacts with CCT␣ in a calcium-independent manner (Fig.  2C). Preliminary studies also showed that varying calcium concentrations after the proteins were bound using the CaM-agarose assay did not dissociate CaM from CCT␣ (data not shown). To confirm a more direct interaction between CCT␣ and CaM, purified CCT␣ was run over CaM-agarose beads, and the beads were processed as above for CCT␣ binding (Fig. 2D). CaM was detected in association with incubation of 1-2 g of purified CCT␣.
To assess in vivo binding between CaM and CCT␣, we used mammalian two-hybrid assays and FRET (Fig. 3). Cells were co-transfected with CCT␣-Gal4BD and CaM-Gal4AD plasmids as fusion proteins together with a plasmid construct encoding a ␤-galactosidase reporter gene (pG5CAT). CCT␣-Gal4BD and CaM-Gal4AD transfected separately did not increase reporter activity (Fig. 3A, inset). Co-transfection of the CCT␣-Gal4BD and CaM-Gal4AD plasmids together stimu-lated reporter activity comparable with the positive control plasmid indicative of CaM-CCT␣ binding. Finally, we also employed FRET analysis using an acceptor photobleaching technique (29) (Fig. 3B). In FRET, energy is transferred from a donor fluorophore molecule to an acceptor fluorophore molecule when proteins are in close (nanometer range) proximity. Thus, FRET is a powerful tool providing more direct visual evidence of protein-protein interaction in vivo. If FRET is observed using the acceptor photobleaching method, the donor emission (CFP) signal increases after a nearby acceptor fluorophore (YFP) is inactivated by irreversible photobleaching. Cellular transfection with YFP-CaM chimera led to diffuse cellular fluorescence of YFP-CaM in line with CaM localization in both FIGURE 2. CCT␣ is a CaM-binding enzyme. A, co-immunoprecipitation. MLE cells were lysed and incubated with TrueBlot beads alone or with rabbit IgG, CaM, or ERK1 polyclonal antibodies (top). Cells were also lysed and incubated with beads alone or with rabbit IgG, CCT␣, or calmodulin kinase II (CaMKII) polyclonal antibodies (bottom). Immunoprecipitants were resolved by SDS-PAGE prior to CCT␣ (top) or CaM (bottom) immunoblotting. CCT␣ migrates at ϳ42 kDa, whereas CaM is detected at ϳ17 kDa; the band (bottom) at ϳ23 kDa is Ig light chain. B, GST pull-downs. A GST-CCT␣ fusion construct (10 g of plasmid/dish) was transfected in cells using FuGene6. After transfection, cell lysates were incubated with glutathione-agarose beads and rinsed, and products eluted and resolved by SDS-PAGE prior to CaM immunoblotting. Nontransfected (NT) cell lysates or agarose beads alone processed similarly served as a negative control. C and D, CaM-Sepharose binding assay. Cell lysates (30 g) (C) were incubated with CaM-Sepharose beads at different Ca 2ϩ concentrations. After extensive rinsing, elusion products were resolved by SDS-PAGE prior to CCT␣ immunoblotting. Cell lysates were incubated with agarose beads as a negative control. D, purified CCT␣ (1-2 g) was incubated with CaM-Sepharose beads and processed as above for CCT␣ immunoblotting. Data in each panel represent at least n ϭ 3 experiments except for D (n ϭ 2). cytosol and nucleus as described previously (30) (data not shown). Consistent with the ability of CaM to recruit binding partners to the nucleus (31), co-transfection of these fluorescent plasmids led to detection of a robust CFP-CCT␣ signal within the nucleus, making CCT␣ accessible to CaM (Fig. 3B,  top). More importantly, the emission fluorescence values of both the donor CFP-CCT␣ and acceptor YFP-CaM before and after acceptor photobleaching in the region of interest are shown (Fig. 3B, upper arrow and lower plots). These data show that upon bleaching, there was decreased acceptor fluorescence (YPF) coupled with an increase in donor emission fluorescence (CFP), because the acceptor cannot take in energy after its photobleaching. As a whole, these data complement the physical interaction data in Fig. 2, demonstrating that CCT␣ binds CaM in vivo and in vitro.
Mapping the CaM Binding Site within CCT␣-CaM has a propensity to bind amphipathic ␣-helices (20). There are two major helices within the CCT␣ membrane binding domain (Fig. 4A) (32). We hypothesized that these helixes might harbor a potential CaM binding domain. We employed a reductionist approach by expressing GST-tagged CCT␣ constructs lacking functional domains (Fig. 4B). Following cellular plasmid trans-fection, lysates were resolved by SDS-PAGE followed by immunoblotting using GST antibodies to confirm expression of these mutants (Fig. 4C); cellular lysates were also run over glutathione-agarose beads, and after stringent washing, GST pull-down products were eluted and resolved by SDS-PAGE prior to CCT␣ and CaM immunoblotting. As shown in Fig. 4C, each of these constructs was sufficiently expressed and exhibited appropriate mobilities as fusion proteins on immunoblots. CCT␣ immunoblotting, as expected, revealed a missing band after expression of GST-CCT CAT as the antiserum is directed against the catalytic domain (Fig. 4D, top). Full-length CCT␣ and GST-CCT␣ mutants devoid of the catalytic core (CCT CAT ), PEST sequence (residues 16 -32) (CCT PEST ), NH 2 -terminal sequence (residues 1-40) (CCT N40 ), and carboxyl terminus (residues 315-367) (CCT 315 ) all bound CaM (Fig. 4D, bottom). However, deletion of the CCT␣ membrane binding domain (residues 236 -315) (CCT MEM ) totally disrupted CaM-CCT␣ association. Thus, CaM binds CCT␣ within the membrane domain.
We next tested several GST-CCT␣ mutants that were progressively truncated within the membrane-binding domain (at the carboxyl terminus) to further localize a CaM binding motif (Fig. 5A, upper map). C288 retains helix 1 and helix 2, C267 contains helix 1, and both mutants bound CaM (Fig. 5A, bottom). However, binding of CaM was not observed with mutants C243 and C210. Thus, CaM binds CCT␣ in a span of residues from 243 and 267 in helix 1. Additional mapping studies (Fig.  5B) revealed that C260 and C250 also bound CaM, thus localizing CaM interactions with CCT␣ to a motif, LQERVDKVK. This sequence displays some similarity to CaM IQ-binding motifs with regard to conservation at Gln 243 (20). To evaluate the significance of this highly conserved site, we constructed a full-length CCT␣ and a truncated CCT␣ each with a single amino acid substitution at Gln 243 (FL Q243A and C267 Q243 ). These mutants were tested in the GST pull-down interaction assay. The results (Fig. 5C) demonstrate that like the C243 construct, both FL Q243A and C267 Q243 lose the ability to interact with CaM. Thus, CCT␣ sequence 243 LQERVDKD 250 is required for CaM binding, and Gln 243 is an important recognition site mediating this interaction.
CaM Modulates CCT␣ Stability and Function-The above data suggest that CaM binds CCT␣ within a distinct recognition motif. Since CaM stabilizes proteins, we next executed gain-of-function and loss-of-function analysis by manipulating its expression in vivo. First, purified CCT␣ was used in the calpain digestion assay in the absence or presence of exogenous CCT␣, CaM, or the calpain peptide substrate, LLVY. Each reaction contained 0.6 pmol of calpain. After hydrolysis, reaction products were processed for levels of immunoreactive CCT␣ and quantified by densitometry. As shown in Fig. 6A, 0.6 pmol of calpain effectively cleaved 80% of purified CCT␣ (0.7 pmol) in the absence of CaM; only at a 9-fold excess of recombinant CCT␣ (6.3 pmol) was calpain hydrolysis of the enzyme significantly inhibited (Fig. 6A, left). Importantly, calpain-mediated CCT␣ hydrolysis was significantly reduced by increasing either the CaM/CCT␣ molar ratio or the amounts of a calpain substrate in the reaction mixture (Fig. 6A, middle and right, respectively). In these studies, CCT␣ was present at 0.7 pmol/reaction (Fig. 6A, middle and right). Of note, even at very low molar

. CaM binds CCT␣ in vivo.
A, mammalian two-hybrid assay. Cells were co-transfected using electroporation with CCT␣-Gal4BD (CCT␣) and CaM-Gal4AD (CaM) plasmids as fusion proteins separately (inset) or in combination with a plasmid construct encoding a ␤-galactosidase reporter gene (pG5CAT) per the manufacturers' instructions. Cells were lysed and assayed for ␤-galactosidase activities. pM-53/pVP16-T and pM3-VP16/pVP16-CP plasmids served as positive and negative controls, respectively. p Ͻ 0.05 versus empty. B, FRET analysis. Cells were transfected with YFP-CaM, and CFP-CCT␣ and CaM-CCT␣ interaction at the single cell level was imaged using laserscanning microscopy before and after photobleaching. Shown in the upper sets of panels is single cell imaging showing that after acceptor photobleaching, fluorescence intensity of YFP decreased, and CFP increased, confirming protein interaction between CaM and CCT␣. Bottom, the same FRET was confirmed quantitatively by graphing of fluorescence intensities. ratios (ϳ0.5) of CaM/CCT␣ (i.e. ϳ0.35 pmol of CaM per 0.7 pmol of CCT␣), calpain hydrolysis of the substrate was inhibited by 80% (Fig. 6A, middle). Further, initial results show that CaM did not directly alter calcium dependence of calpain activity in vitro (data not shown). Next, cells were infected with a replication-deficient adenovirus that expresses CaM (Adv-CaM) or an empty virus (Adv-Con), and 6 h later, cells were exposed to Ox-LDL for 24 h. Adv-CaM, unlike the empty virus, totally blocked Ox-LDL inhibition of CCT activity (Fig. 6B) and CCT␣ degradation (Fig. 6C) while producing a robust increase in CaM levels (Fig. 6C). Next, MLE cells were either transfected with CaM siRNA or a control siRNA, and 72 h later, cells were exposed to Ox-LDL. As shown in Fig. 6D, Ox-LDL and CaM siRNA each significantly reduced CCT activity to ϳ40 -50% of control; CaM siRNA in combination with Ox-LDL accentuated this effect. These changes in enzymatic behavior were reflected in CCT␣ protein levels, since CaM siRNA (while reducing CaM levels) significantly reduced steady-state CCT␣ mass, an effect accentuated when it was applied in combination with Ox-LDL (Fig. 6E). Although Ox-LDL in some studies tended to reduce CaM levels, this was not a consistent finding, and the modified lipoproteins did not reduce CaM synthesis as determined by [ 35 S]methionine pulsechase (n ϭ 2; data not shown). Finally, to assess a causal role for CaM in regulating CCT␣ availability, we conducted rescue experiments to determine if Adv-CaM can restore CCT function after CaM siRNA (Fig. 6, F and G). Cells were transfected with either CaM siRNA or control siRNA; 48 h after transfection, cells were infected with Adv-CaM for 24 h, and Ox-LDL was added to the medium for an additional 24 h. Ox-LDL inhibited CCT activity by ϳ30% in these studies, an effect that was enhanced in combination with CaM siRNA (Fig. 6F). Moreover, Adv-CaM overexpression effectively restored CCT activity (Fig. 6F) and CCT␣ content (Fig.  6G) in cells after CaM siRNA transfection alone or in combination with Ox-LDL treatment. Together, these data indicate that CaM protects CCT␣ stability in vitro and in vivo.
To confirm that endogenous CaM regulates CCT␣ protein stability, we performed CaM knockdown using siRNA, followed by measurements of CCT␣ half-life by pulse-chase (Fig. 7, A and  B). Consistent with prior studies, in the presence of scrambled siRNA, CCT␣ exhibited a t1 ⁄ 2 of ϳ8 h (25). CCT␣ turnover was significantly accelerated in cells pretreated with CaM siRNA, since enzyme levels were almost undetectable by 2 h (Fig. 7B). These effects of CaM siRNA were associated with significant reductions in PtdCho synthesis (Fig. 7, C and D). As a whole, these observations indicate that endogenous CaM stabilizes CCT␣ in vivo.
A Q243A Mutant Confers Resistance to Calpain-Calpain efficiently cleaves its substrates, in part, by docking to CaM binding motifs within target substrates (33). Since Q243A Lysates from transfectants were resolved by SDS-PAGE prior to GST immunoblotting. D, overexpressed GST-CCT fusion proteins were incubated with glutathione-agarose beads and rinsed extensively, and elusion products were resolved by SDS-PAGE prior to CCT␣ immunoblotting (above) or CaM immunoblotting (below). CCT␣ immunoblotting was performed using a polyclonal antibody generated against epitopes within the catalytic domain. Nontransfected (NT) cell lysates incubated with glutathione-agarose beads served as a negative control. Data in each panel represent at least n ϭ 3 experiments.
totally disrupts CaM binding, we investigated whether this site was also critical for calpain recognition. GST-CCT FL and GST-CCT Q243A were purified and used as substrates for calpain digestion. Although GST-CCT FL was efficiently cleaved by calpain, resulting in the appearance of several hydrolysis products in a dose-dependent manner, the GST-CCT Q243A mutant was totally resistant to calpain cleavage (Fig. 8A). These results were confirmed with higher doses of calpain (0.5 g), where the GST-CCT Q243A mutant was still resistant to calpain cleavage (data not shown). Cells were transfected with GST-CCT Q243A to assess its physiologic role in response to Ox-LDL. GST-CCT constructs were observed to be functional and targeted appropriately to cellular compartments similar to endogenous CCT␣ (see supplemental material). Ox-LDL decreased the activity of both endogenous and overexpressed full-length CCT␣ to ϳ66% of control values. However, Ox-LDL did not significantly decrease CCT activity in cells expressing GST-CCT Q243A (Fig.  8B). Immunoblotting experiments revealed that both endogenous CCT␣ and GST-CCT FL were degraded by Ox-LDL, whereas levels of the 68-kDa GST-CCT Q243A fusion product displayed considerable stability comparable with untreated controls. Thus, Gln 243 may serve as a key dock site for competitive access of CaM and calpain within the CCT␣ enzyme, and its mutagenesis significantly blocks proteinase activity in vitro and in vivo.

DISCUSSION
To the best of our knowledge, these studies demonstrate a novel regulatory model whereby competition between a proteinase and CaM for access to a single, highly conserved residue (Gln 243 ) within a CaM binding motif profoundly affects levels of an enzyme. This specific site (Gln 243 ) in CCT␣ appears to serve as a structurally unique recognition signal for both calpains and CaM that may vie for occupancy within CCT␣. The studies are also the first to show that mutagenesis of a single residue within a CaM dock site blocks the ability of a proteinase to hydrolyze its substrate. This site, rather than a consensus PEST sequence, is integrally linked to CCT␣ proteolysis by calpains. Last, a new finding here is that CCT␣ is a CaM-binding protein and that CaM antagonizes CCT␣ degradation by calpains in vitro and in vivo. Evidence supporting CaM interactions with CCT␣ includes (i) co-immunoprecipitation of CCT␣ with CaM from cellular lysates, (ii) immunodetection of CaM in association with GST-CCT␣ fusion proteins in GST pull-downs, (iii) detection of CCT␣ in CaM-Sepharose binding assays, (iv) mammalian two-hybrid studies, and (v) FRET analysis. Manipulation of CaM expression by adenoviral overexpression or CaM siRNA in lung epithelia differentially altered stability of CCT␣ in the native state and in response to destabilizing effects of oxidized lipoproteins. Thus, CaM appears to be a physiologically relevant binding partner for CCT␣ that may serve to protect the enzyme from proteolytic cleavage.
Although the ratio of immunoreactive CaM versus CCT␣ levels in lung epithelia appears relatively low, this does not preclude CaM as a bona fide binding partner that protects CCT␣ under native conditions from calpains. The relative amounts of CaM, calpains, and CCT␣ in cells depend, in part, upon the served as a negative and positive control, respectively. B, top, map illustrating individual CCT␣ mutants constructed that harbor truncations within helix 1. All constructs tested harbor at least 243 upstream residues of CCT␣. Bottom, overexpressed GST-CCT fusion proteins were processed by GST pull-down analysis and CCT␣ (above) or CaM immunoblotting (below). C, top, CCTFL Q243A (a full-length CCT␣ with a Q243A mutation (FLQ243A)) and C267 Q243A (a truncated CCT␣ with 267 residues and a Q243A mutation (C267Q243)) mutants were constructed to test functionality of CaM binding within CCT␣. Bottom, overexpressed GST-CCT fusion proteins were processed by GST pull-down analysis as above and CCT␣ (above) or CaM immunoblotting (below). Data in each panel represent at least n ϭ 3 experiments.
avidity of antibodies for their detection. CCT␣ stability will also be governed by the binding affinities of calpains versus CaM for CCT␣. Thus, high affinity binding of CaM to CCT␣ might be sufficient to overcome lower stoichiometries of CaM relative to FIGURE 6. Calmodulin protects CCT␣ from proteolysis. A, left, varying amounts of purified CCT␣ were incubated with purified -calpain (0.6 pmol/ reaction for 30 min). Middle, varying amounts of CaM were incubated with purified CCT␣ (0.7 pmol) and recombinant -calpain (0.6 pmol/reaction). The ratios of CaM/CCT␣ are represented on the abscissa. Right, varying amounts of the calpain substrate, LLVY, were incubated with purified CCT␣ (0.7 pmol) and recombinant -calpain (0.6 pmol/reaction). In each panel, the digestion products were resolved by SDS-PAGE prior to CCT␣ immunoblotting. The percentage hydrolysis of the 42-kDa CCT␣ was then quantified densitometrically. B and C, MLE cells were infected with Adv-CaM or an empty virus (Adv-Con). MOI ϭ 40 for 6 h prior to exposure to Ox-LDL (100 g/ml) for an additional 24 h prior to harvest for analysis of CCT activity (B) or CCT␣, CaM, and ␤-actin immunoblotting (C). For CCT␣ immunoblotting, each lane of the gels was loaded with 20 g of total cellular protein. D-E, cells were transfected using nucleofection with CaM siRNA or a control siRNA. 72 h after transfection, cells were exposed to Ox-LDL (100 g/ml) for an additional 24 h prior to harvest for analysis of CCT activity (D) or CCT␣, CaM, and ␤-actin immunoblotting (E). F and G, cells were transfected with siRNA targeted against CaM or a control siRNA as above. 48 h after transfection, cells were infected with or without Adv-CaM for another 24 h. In some cells, Ox-LDL was added to the medium for an additional 24 h. Cells were then harvested for analysis of CCT activity (F) or for CCT␣, CaM, and ␤-actin immunoblotting (G). For CCT␣ immunoblotting in E and G, each lane of the gels was loaded with 5 g of total cellular protein. Results are mean Ϯ S.E. from n ϭ 3 experiments. other proteins. CCT␣ is proteolytically sensitive to calpains in vitro yet exhibits a relatively long half-life in cells, suggesting the presence of a stabilizing binding partner for the enzyme. The extended half-life of CCT␣ is somewhat unexpected, since calpains are constitutively active in cells that would shorten the enzyme's life span. Our data showing that CaM siRNA accelerates CCT␣ degradation in cells provide strong evidence that endogenous CaM is required to protect CCT␣ from calpain degradation (Fig. 7). Thus, the relatively slower turnover rate of CCT␣ compared with other metabolic enzymes could be explained by CaM-bound CCT␣ that masks recognition sites for calcium-activated proteinases both under native conditions and in settings of proteinase excess.
There are both methodologic and mechanistic explanations for our results showing that in pulse-chase studies CaM siRNA reduced CCT␣ half-life from 8 to ϳ2 h, yet we observed more modest (25-30%) reductions in CCT activity (Fig. 6). First, the kinetics of when CCT activity was assayed differs after exposure to CaM siRNA in Fig. 6 (96 h) versus our pulse-labeling studies (Fig. 7, ϳ80 h). It is likely that CaM siRNA reduces CCT activity to a nadir by ϳ80 h and that thereafter activity increases. This is consistent with studies using CaM pharmacologic inhibition, where CCT activity recovers after CaM actions are blocked (34). These temporal differences of CaM siRNA effects on CCT activity versus CCT half-life are also consistent with a loss of CaM siRNA efficacy by 96 h, since many siRNA duplexes exhibit time-dependent and labile effects (35). For example, recent studies indicate that maximal efficacy of nonvector synthetic RNA interference is highly dependent upon the half-life of the targeted protein; thus, shorter half-life proteins like CaM (t1 ⁄ 2 ϭ 10 -12 h) (36) will usually exhibit maximal knockdown between 12 and 48 h, whereas longer half-life proteins (Ͼ24 h) may exhibit maximal reductions after several days (35). Thus, loss of CaM siRNA effects after half-life measurements were concluded by ϳ78 h will be associated with partial restoration of CCT activity by 4 days, as observed in our studies. Second, in pulse-chase studies, the stability of only newly synthesized CCT␣ is determined and not preformed enzyme. The unmeasured preformed or preexisting pool of enzyme may be highly phosphorylated at specific sites or membrane-associated, providing greater stability (37). Because CCT␣ is regulated at multiple post-translational levels, our half-life measurements in the setting of CaM siRNA treatment do not take into account coregulation by such mechanisms. Third, even these more modest inhibitory effects of CaM siRNA on CCT activity were physiologically significant, because they were sufficient to reduce PtdCho synthesis (Fig. 8).
Calpains, in part, target PEST motifs within substrates to facilitate degradation (12,15). The program PESTFIND identified a strong PEST motif (residues 16 -32) within CCT␣, evidenced by a hydrophilic stretch of amino acids containing two prolines, several acidic residues, and one serine flanked by an NH 2 -terminal lysine. Mutagenesis of threonine 25 and serine 32 to alanine within CCT␣ removed this motif as a high value PEST target. Indeed, both the CCT␣ T25A/S32A double mutant and an internal deletion mutant devoid of the PEST motif retained sensitivity to calpain degradation in vitro and after cellular expression. As with c-Fos and Ca 2ϩ -ATPase, the CCT␣ PEST motif may not serve as a proteolytic signal (16,38) with the caveat that the PEST domain is sufficiently exposed in vivo. Unlike cyclic AMP-dependent kinase (39), the PEST motif is probably unmasked in our system because ϳ40% of CCT␣ was membrane-associated, a feature that activates CCT␣, exposing its NH 2 -terminal domain (40).
A CaM binding motif within CCT␣ might serve as an alternative recognition signal for calpains. Data base analysis (available on the World Wide Web) of the CCT␣ sequence initially predicted a putative CaM binding domain within the distal catalytic domain-membrane binding domain interface (residues 205-240) on the basis of hydrophobicity, an average hydrophobic moment, and propensity for ␣-helix formation. Mapping studies using GST-CCT␣ carboxyl-terminal truncated mutants, however, localized a CaM binding motif to residues 242-250 exclusively within the membrane-binding domain (Fig. 9). This motif has some features resembling other CaM binding domains. This region resides within ␣-helix-1, consistent with the predilection of CaM binding domains to localize in amphipathic helices (20). Second, the presence of a calpain cut site juxtaposed upstream of this domain is in line with calpain cleavage of substrates at or near CaM binding domains (21) FIGURE 8. Mutagenesis of Gln 243 protects CCT␣. A, full-length CCT␣ (GST-CCT) and GST-CCT Q243A mutants were purified and used as substrates for calpain digestion using various amounts of the proteinase. Hydrolysis products were resolved using SDS-PAGE and immunoblotted for CCT␣. *, the upper ϳ40 kDa band represents a fragment (361 amino acids) excluding the GST tag extending to the NH 2 -terminal clip site at Ser 5 -Ser 6 , and the lower ϳ26 kDa band represents an internal CCT␣ fragment resulting from this cleavage and an additional cleavage as described (7). B and C, wild-type CCT␣ (FL) and a full-length CCT␣ variant harboring the Q243A mutation (FLQ243A) were transfected in cells. After 24 h, Ox-LDL (100 g/ml) was added for an additional 24 h. The cells were harvested and analyzed for CCT activity (B) and immunoreactive CCT␣ (C). Results are mean Ϯ S.E. from n ϭ 3 experiments. NT, nontransfected. (Fig. 4A). However, the sequence LQERVDKVKKKVK is not identical to but exhibits some similarity to IQ motifs (IQXXXRGXXXR) present in proteins that bind CaM in a calcium-independent manner (20). The CaM binding domain within CCT␣ harbors a highly conserved glutamine at position 2 and a basic residue at position 11, features characteristic of IQ motifs (20).
A remarkable observation from our studies is that a single amino acid substitution of glutamine within the CaM binding motif (Q243A) not only negates CaM binding but was sufficient to totally block degradation of CCT␣ by calpain (Fig. 8A). These effects were recapitulated in cells exposed to oxidized lipids, where the expressed Q243A CCT␣ mutant construct was resilient to calpains evidenced by the stability of the overexpressed protein and preservation of enzyme activity. We did not examine the functionality of other residues, since Glu is highly conserved within the CaM motif. Presumably, polarity and/or electrostatic interactions between Gln 243 or other residues in the motif, and calpain might enhance accessibility of the proteinase to its adjacent CCT␣ cleavage site or help sequester calcium (Fig. 9). Of note, Gln within the motif may also have several favorable electrostatic interactions with the backbone of CaM at Leu 111 , Gly 112 , and Glu 113 or via binding to domain IV of the large calpain subunit at basic loops (Protein Data Bank code 1AJI). Our results differ from studies of the calcium ATPase and inducible nitric-oxide synthase, where deletion of an entire canonical CaM binding region either attenuated or was insufficient to modify calpain activity (14,16). Conversely, removal of CaM binding domains within caldesmon and calponin do not alter substrate recognition by calpain (41). Unlike our results, Padanyi et al. (13) demonstrated that point mutagenesis at Trp 1093 in the calcium ATPase pump increased the accessibility of a calpain hydrolysis site within a CaM binding domain. Thus, our data suggest a somewhat unique molecular model whereby availability of CCT␣ will be influenced by stoichiometry and binding affinities of CaM versus calpain utilizing Gln 243 as a critical recognition site. Interestingly, a point mutation at a highly conserved Gln (Q3180P) was recently identified within an IQ motif of the gene encoding abnormal spindle-like microcephaly-associated protein; this mutation is linked to an inherited disorder characterized by neurodevelopmental arrest of brain growth, raising the possibility that interactions between abnormal spindle-like microcephaly-associated protein, calpains, and CaM might play a role in disease pathogenesis (42).
Our study demonstrates that CaM stabilizes a protein in cells, an issue not addressed in prior work (14,16,21,22). Of note, CaM inhibition reduces PtdCho synthesis and impairs lung growth, but these studies relied on the use of nonselective approaches (34,(43)(44)(45). Adenoviral gene transfer of CaM into lung epithelia was achieved with high efficiency (Ͼ95%) and specificity, allowing for gain-of-function analysis. Complementary loss-of-function experiments were facilitated by cellular electroporation of CaM siRNA constructs that also proved to be efficient (Figs. 6 and 7). Ad5-CaM overexpression effectively blocked Ox-LDL-induced CCT␣ breakdown and rescued acceleration of enzyme turnover after CaM knockdown, indicating that siRNA effects were selective for CaM. These stabilizing effects of CaM on CCT␣ protein were the dominant effect, since separate in vitro experiments showed that recombinant CaM only produced modest (ϳ2-fold) activation of purified CCT (data not shown). These studies coupled with our physical protein interaction data underscore a potentially important biochemical and physiologic role for CaM in regulating PtdCho biosynthesis. Future work using structural analysis of CaM-CCT␣ complexes will provide newer insight into the conformational environment for these interactions. Testing of these mechanistic associations in vivo also awaits the generation of suitable transgenic or knock-in animal model systems that conditionally express relevant molecular sites within CCT␣ or CaM in epithelia.