|Home | About | Journals | Submit | Contact Us | Français|
We have shown previously that the ELR-negative CXC chemokines interferon-inducible protein 10, monokine induced by gamma interferon, and platelet factor 4 inhibit epidermal growth factor (EGF)-induced m-calpain activation and thereby EGF-induced fibroblast cell motility (H. Shiraha, A. Glading, K. Gupta, and A. Wells, J. Cell Biol. 146:243-253, 1999). However, how this cross attenuation could be accomplished remained unknown since the molecular basis of physiological m-calpain regulation is unknown. As the initial operative attenuation signal from the CXCR3 receptor was cyclic AMP (cAMP), we verified that this second messenger blocked EGF-induced motility of fibroblasts (55% ± 4.5% inhibition) by preventing rear release during active locomotion. EGF-induced calpain activation was inhibited by cAMP activation of protein kinase A (PKA), as the PKA inhibitors H-89 and Rp-8Br-cAMPS abrogated cAMP inhibition of both motility and calpain activation. We hypothesized that PKA might negatively modulate m-calpain in an unexpected manner by directly phosphorylating m-calpain. A mutant human large subunit of m-calpain was genetically engineered to negate a putative PKA consensus sequence in the regulatory domain III (ST369/370AA) and was expressed in NR6WT mouse fibroblasts to represent about 30% of total m-calpain in these cells. This construct was not phosphorylated by PKA in vitro while a wild-type construct was, providing proof of the principle that m-calpain can be directly phosphorylated by PKA at this site. cAMP suppressed EGF-induced calpain activity of cells overexpressing a control wild-type human m-calpain (83% ± 3.7% inhibition) but only marginally suppressed that of cells expressing the PKA-resistant mutant human m-calpain (25% ± 5.5% inhibition). The EGF-induced motility of the cells expressing the PKA-resistant mutant also was not inhibited by cAMP. Structural modeling revealed that new constraints resulting from phosphorylation at serine 369 would restrict domain movement and help “freeze” m-calpain in an inactive state. These data point to a novel mechanism of negative control of calpain activation, direct phosphorylation by PKA.
Signaling pathways from the epidermal growth factor (EGF) receptor (EGFR) play important roles in wound healing. Wound fluid EGFR ligands, including heparin-binding EGF-like growth factor and transforming growth factor α, are very strong stimulatory factors for fibroblast cell migration necessary during the repopulation phase of repair (4, 39, 48). During this migration, tail deadhesion is postulated to be rate limiting (34). In experimental models, we and others have demonstrated that failure to deadhere limits cell motility (24, 42). Calpain activity is critical to integrin-mediated tail deadhesion on moderately and highly adhesive substrata (40) and to growth factor-induced motility (17). This intracellular protease appears to be a key switch, as calpain inhibitors convert EGFR-mediated signals from cell motility to matrix contractility (2).
Calpains (EC 184.108.40.206) are a highly conserved family of intracellular proteases. The two ubiquitous forms are distinguishable by their in vitro requirements for calcium, while the substrate specificities of these two forms appear to be identical (44). Calpain I, or μ-calpain, is activated at near-micromolar calcium; calpain II, or m-calpain, requires millimolar calcium levels. While calcium fluxes have been postulated to regulate calpains, the physiologically relevant activators of the m-calpain isoform are unknown, since intracellular calcium levels fail to reach the near-millimolar concentrations required (21). m-calpain, which predominates in fibroblastoid cells (17, 24, 44), is required for growth factor receptor-mediated deadhesion and motility (17, 42). Interestingly, EGF triggers m-calpain downstream of extracellular signal-related kinase (ERK)/mitogen-activated protein kinase signaling and not phospholipase Cγ signaling, which mobilizes intracellular calcium (17, 49), suggesting a novel mechanism of activation. The physiological substrates of calpain are not known. However, a number of in vitro and in vivo substrates provide excellent candidates, as they are present at the inner face of the adhesion complex. These include the cytoplasmic domain of select β-integrins (12), focal adhesion kinase (11), and paxillin and talin (6). Despite the precise molecular basis for calpain-mediated deadhesion being unknown, it has been well established that tail deadhesion requires at least one of these forms to be acting (24, 40, 42).
This requirement of calpain activity for fibroblast migration during dermal repair provides a target for negative regulation. This would both prevent excess fibroplasia and convert the motile phenotype to one of matrix contraction (2). In our previous paper (42), we reported that ELR-negative CXC chemokines present during the resolution phase of wound repair (14) limited growth factor-induced cell motility. Interferon-inducible protein 10 (IP-10), monokine induced by gamma interferon (MIG), and platelet factor 4 (PF4) prevented EGFR- and platelet-derived growth factor receptor-mediated calpain activation and cell deadhesion. Although this was accomplished secondarily to cyclic AMP (cAMP) generation, the steps are unknown that bridge the presumption of protein kinase A (PKA) being activated and the prevention of m-calpain proteolytic actions. The only evidence-supported mechanism for negative regulation of calpains is a stoichiometric inhibition by the endogenous inhibitor calpastatin (5). However, it is unclear whether calpastatin suppresses m-calpain activity, as they do not fully colocalize in cells (33), and/or whether there are additional mechanisms for m-calpain attenuation. Our earlier work (42) demonstrated that the chemokine subclass that binds to the CXCR3 receptor, the ELR-negative CXC chemokines (IP-10, MIG, PF4, and IP-9), prevented EGFR-mediated calpain activation through a cAMP-dependent pathway. Thus, we hypothesized that PKA directly phosphorylates m-calpain and thereby prevents its activation by growth factors. This novel mechanism of attenuation was proposed because (i) one can identify a PKA consensus site in the m-calpain putative regulatory domain III (25) and (ii) it has been reported elsewhere that m-calpain can be phosphorylated in vitro by PKA (32). Herein, we report that PKA phosphorylation prevents EGF-induced m-calpain activation. This was demonstrated by genetically eliminating this site in human m-calpain and finding that this rendered the enzyme PKA resistant and that cells expressing this mutant calpain were resistant to cAMP inhibition of motility. These findings provide a novel regulatory mechanism for control of m-calpain activity.
The murine fibroblast line NR6WT was used in all experiments. These cells, lacking endogenous EGFR, express approximately 100,000 human wild-type EGFRs/cell (7) and respond to EGF with both motility and mitogenesis (8). These cells were passaged in minimal essential medium α (MEMα) supplemented with 7.5% fetal calf serum, nonessential amino acids, and pyruvate (all culture reagents from GIBCO/BRL, Rockville, Md.). Since the human EGFR construct was engineered with neomycin-phosphotransferase as a selectable marker, 350 μg of G418/ml was added into the culture medium to maintain human wild-type EGFR. Prior to all experimentation, the cells were made quiescent in MEMα supplemented with 0.5% dialyzed fetal calf serum (no G418).
Human m-calpain cDNA was obtained by reverse transcription-PCR-based cloning from human dermal fibroblasts (Hs68; American Type Culture Collection, Manassas, Va.). Briefly, total RNA was collected from Hs68 cells with Trizol (GIBCO/BRL). Reverse transcription was performed with purified Hs68 total RNA with m-calpain-specific oligonucleotide primer (5′-CCTCGTGTCCTTTGAGAGCG-3′) and Superscript II reverse transcriptase (GIBCO/BRL). To generate a poly-His (His)-tagged m-calpain large subunit, cDNA sense (5′-AGCTAGCGGACCGCAGCATGG) and antisense (5′-GCCTTGCCGGCCTCAATGATGATGATGATGATGGTCAAGTACTGAGAAACAGAGCC; including the six-His tag) primers were designed according to an m-calpain cDNA sequence (GenBank accession no. M23254.1). Purified cDNA was amplified by PCR using sense and antisense primer and Elongase (GIBCO/BRL) and cloned into PCR II TA cloning vector (Invitrogen, Carlsbad, Calif.). The size of the PCR product was ~2.2 kbp. After confirmation by sequencing, m-calpain cDNA was subcloned into pCEP4 (Stratagene, La Jolla, Calif.) downstream from a cytomegalovirus (CMV) promoter; the hygromycin resistance gene conferred selectability for stable expression. The CMV promoter was replaced with mouse mammary tumor virus promoter (MMTVp) for inducible expression (8). The PKA-resistant hCANP mutant clone was generated using a PCR-based mutagenesis kit (Stratagene) and primers that encoded the mutation (5′-CTGGAGGCGGGGCGCAGCTGCGGGAGGTTGCAG-3′ and 5′-CTGCAACCTCCCGCAGCTGCGCCCCGCCTCCAG-3′). This changed amino acids 369 and 370 from ST to AA and thus is referred to as ST369AA. The poly-His tag was replaced with cDNA for green fluorescent protein (GFP) from pEGFP-C1 (Clontech, Palo Alto, Calif.). Both His-tagged and GFP-tagged hCANP constructs were utilized for establishing stable transfected cell lines.
The human m-calpain (hCANP) constructs were stably expressed in NR6WT cells by electroporation. Twenty micrograms of hCANP construct plasmid was added to 500 μl of cell suspension that contained approximately 2.0 × 107 cells. The cell suspension was transferred into an electroporation cuvette (0.2-cm gap) and electroporated (500 μF, 0.320 kV) using a Gene Pulser electroporator (Bio-Rad, Hercules, Calif.). Electroporated cells were plated into a six-well tissue culture plate. At 36 h after electroporation, cells were selected in complete medium containing 100 μg of hygromycin (Roche Diagnostics, Indianapolis, Ind.)/ml. Polyclonal lines consisting of more than 20 colonies were established. At least two independent electroporations and stably transfected lines were established for each construct.
EGF-induced cell migration was assessed by the ability of the cells to move into an acellular area (7). Cells were made quiescent for 24 h prior to being denuded by a rubber policeman. The cells were then treated or not with 10 nM EGF, CPT-cAMPS (1 μM) (Sigma, St. Louis, Mo.), Rp-8Br-cAMPS (5 μM) (Calbiochem, La Jolla, Calif.), Rp-8Br-cGMPS (5 μM) (Calbiochem), and/or H-89 (at a specified concentration) (Calbiochem). Cells were incubated at 37°C for 24 h. Photographs were taken at 0 and 24 h, and the relative distance traveled by the cells at the acellular front was determined.
EGF-induced calpain activation was determined by the BOC-LM-CMAC (t-butoxycarbonyl-Leu-Met-chloromethylaminocoumarin) assay (17, 41). Cleavage of this substrate yields fluorescence that is selective for calpain; specificity is ensured by blocking fluorescence by calpain-selective inhibitors and by molecular downregulation of calpains. Cells were plated at 50% confluence in a glass chamber (Labtek II; Nalge Nunc, Roskilde, Denmark). Cells were treated or not with CPT-cAMPS (1 μM) and/or Rp-8Br-cAMPS (5 μM) and incubated for 30 min in the presence of 30 μM BOC-LM-CMAC (Molecular Probes, Eugene, Oreg.). Cells were treated or not with EGF (10 nM) for 5 min prior to visualization using a cooled charge-coupled device camera (Spot II; Diagnostic Instruments, Sterling Heights, Mich.) (17). EGF-induced calpain cleavage of glutathione-conjugated BOC-LM-CMAC generates glutathione-conjugated CMAC and results in increased fluorescence (excitation, 330 nm; emission, 403 nm). The slides were observed by fluorescence microscopy with an Olympus (Tokyo, Japan) M-NUA filter; pictures were taken with the same exposure setting within each experiment. The signal intensities of cells for each experimental condition were measured in computer-captured pictures by using Photoshop (Adobe, San Jose, Calif.). The numerical data are the means ± standard errors of the means (SEM) of more than 100 cells.
Cells were grown to confluence in six-well tissue culture plastic plates. After 24 h of quiescence, cells were treated or not with dexamethasone (2 μM) (Sigma) for 18 h. Dexamethasone (2 μM for 18 h) was used to induce MMTV-driven hCANP expression. Cell lysates were separated on a sodium dodecyl sulfate (SDS)-10% polyacrylamide gel and transferred to a polyvinylidene difluoride (PVDF) membrane, Immobilon-P (Millipore, Bedford, Mass.). Blots were probed with anti-m-calpain (Santa Cruz Biotechnology, Santa Cruz, Calif.), anti-GFP (Clontech), or anti-poly-His (Santa Cruz) antibodies before visualization with alkaline phosphatase-conjugated secondary antibodies (Promega, Madison, Wis.) followed by development with a colorimetric method (Promega).
Cells were grown to confluence in a 10-cm-diameter tissue culture plastic plate. After 24 h of quiescence, cells were treated with dexamethasone (2 μM) for 18 h. His-tagged m-calpains were purified by Ni-nitrilotriacetic acid (NTA) agarose (Qiagen, Valencia, Calif.) affinity chromatography. Calpains were eluted with 50 mM Tris-HCl (pH 7.5) containing 250 mM imidazole and dialyzed with 50 mM Tris-HCl (pH 7.5). Purified His-tagged m-calpains were incubated for 15 min at 30°C with an assay mixture containing 50 mM Tris-HCl (pH 7.5), 10 mM MgCl2, 200 μM [γ-32P]ATP (final specific activity of 200 μCi/μmol), and 75 U of cAMP-dependent protein kinase catalytic subunit (New England Biolabs, Beverly, Mass.). The samples were separated by SDS-10% polyacrylamide gel electrophoresis, and then samples were transferred into an Immobilon-P (Millipore) PVDF membrane and exposed to X-ray film in a cassette for 72 h at −80°C. After the autoradiography the membrane was blotted with anti-m-calpain antibody (Santa Cruz) to confirm that the amounts of protein were consistent in each lane. The signals were visualized using alkaline phosphatase-conjugated secondary antibody (Promega) and a colorimetric development system (Promega).
His-tagged hCANP proteins were purified using a Ni-NTA column as described above. The activity of the hCANP proteins expressed in cells was tested in vitro using a constitutively active ERK (Upstate Biotechnology, Inc., Lake Placid, N.Y.) to induce calpain (A. Glading et al., unpublished observations). Purified calpain from a confluent 10-cm-diameter dish was incubated with 0.5 μg of recombinant Tau protein in 25 mM HEPES and 1 mM dithiothreitol (DTT) (Panvera, Madison, Wis.) with or without calcium (final concentration of 40 to 41 μM free Ca2+ ion in assay sample) in the presence or absence of active ERK (300 U in 20 mM morpholinepropanesulfonic acid [MOPS; pH 7.2], 25 mM β-glycerol phosphate, 5 mM EGTA, 1 mM Na3VO4, 1 mM DTT, 75 mM MgCl2, and 500 μM ATP) for 10 min. The reaction mixtures were then separated on an SDS-10% polyacrylamide gel, transferred to a PVDF membrane, and probed for Tau (anti-Tau C-terminal sequence; Zymed, South San Francisco, Calif.). Reduction in size of the Tau band (65 kDa) was indicative of calpain activity. While calcium alone did not stimulate substantial Tau cleavage at this time point, longer incubations with calcium alone did cause Tau cleavage (data not shown).
To investigate the effect of PKA phosphorylation on in vitro activity of hCANP, a similar experiment was performed using microtubule-associated protein 2 (MAP2) as the substrate. The substrate was changed becuase PKA phosphorylates Tau and prevents its cleavage by calpain (29, 35). Although PKA similarly phosphorylates MAP2, longer incubations are able to overcome this effect (27, 30). Again purified hCANP from a confluent 10-cm-diameter dish was incubated with 1 μg of MAP2 protein (Cytoskeleton) dissolved in 25 mM HEPES-1 mM DTT-CaCl2 (final concentration of 23 to 28 μM free Ca2+ ion in assay sample) with or without ERK (300 U in same buffer as above) in the presence or absence of PKA catalytic subunit (25 U in PKA assay buffer [Cell Signaling Technology, Beverly, Mass.] with 75 mM MgCl2) (Cell Signaling Technology). The reaction mixture was then separated on an SDS-10% polyacrylamide gel, transferred to a PVDF membrane, and probed for MAP2 with a monoclonal anti-MAP2 antibody (Sigma). Loss of cleavage product bands (75 to 120 kDa) was indicative of calpain activity.
To avoid the influence of endogenous mouse calpain in NR6WT cells, cells were treated with antisense phosphothiorate-linked oligonucleotide specific for mouse m-calpain (5′-TGCCCGCCATGGTAGCGATC-3′) (17). Briefly, cells were treated with oligonucleotide and EGF (10 nM) for 6 h to deplete the preexisting calpain. Cells were then incubated for a further 18 h in the presence of oligonucleotide without EGF to recover from EGF stimulation but prevent de novo calpain synthesis. At the same time cells were treated with dexamethasone (2 μM). Then the cell motility assay was performed as described above.
The modeling template is the X-ray crystal structure of m-calpain (22). A phosphate group was manually placed on the side chain of S369 with the SGI graphics workstation (Silicon Graphics, Mountain View, Calif.). Appropriate side chain adjustments were made to eliminate any apparent steric conflict near the phosphorylation site. The S369-phosphorylated m-calpain was then energy minimized using the SYBYL software package (Tripos Associates, St. Louis, Mo.), employing a Tripos force field, Gasteger and Marsili charges, and dielectric constant. Energy minimization continued until the final system energy converged. This process took approximately 2,000 cycles. The resulting model was carefully examined to ensure that there was no geometry violation and that the interactions involving the phosphate group were reasonable (i.e., devoid of any unacceptable short contacts with other atoms in the vicinity). Diagrams were generated using Molscript (31).
In our previous paper (42), we demonstrated that ELR-negative CXC chemokines IP-10, MIG, and PF4 inhibit EGFR-mediated cell deadhesion and thus motility secondary to cAMP generation in primary human fibroblasts. To validate that this occurs in our genetically amenable mouse fibroblast system, we demonstrated that the cell-permeable cAMP analog CPT-cAMPS had a similar negative effect on human wild-type EGFR expressing mouse fibroblast NR6WT cells (Fig. (Fig.1A).1A). CPT-cAMPS at 1 μM inhibited 10 nM EGF-induced cell migration by 55% ± 4.5%. To prove that cAMP inhibits cell motility through cAMP-dependent PKA, NR6WT cells were treated with PKA inhibitors. The cell-permeable antagonist of cAMP, Rp-8Br-cAMPS, prevented CPT-cAMPS inhibition of motility, whereas the protein kinase G (PKG)-preferential inhibitor Rp-8Br-cGMPs had no discernible effect (Fig. (Fig.1A).1A). A second PKA inhibitor, H-89, also eliminated cAMP's inhibition in a dose-dependent manner (Fig. (Fig.1B1B).
We postulated that calpain-mediated deadhesion is required for tail detachment during EGF-induced motility. Thus, cAMP should prevent this release. Direct visualization of cells by time-lapse microscopy demonstrated that the effect of attenuated rear release is to prevent productive motility (data not shown). The vast majority of cells extended protrusions that adhered to the substratum and pulled the nuclei forward. However, the tails were not efficiently detached in the presence of CPT-cAMPS, and the cells were observed to recoil and to retract their protrusions. With many cells, we could visualize repeated cycles of extension and retraction with the same point tail adhesion remaining constant throughout. Over 80% of the control, EGF-only-treated cells demonstrated productive motility over the same time period (data not shown), in agreement with previously published works (36, 47). These findings support the contention that the rate-limiting step for calpain in cell motility is at the point of rear release.
To demonstrate that this inhibition of motility occurred secondarily to inhibition of calpain activation, we found that cAMP prevents EGF-induced m-calpain activation as determined by BOC-LM-CMAC fluorescence (Fig. (Fig.2).2). Again, the PKA inhibitor Rp-8Br-cAMPS eliminated the effect of CPT-cAMPS (Fig. (Fig.2).2). Cell cytometry studies determined the numerical value of calpain activity from the BOC-LM-CMAC assay. CPT-cAMPS inhibited EGF-induced calpain activation by 87% ± 2.3%. This was significantly reduced by Rp-8Br-cAMPS (15% ± 2.2%) but not by the control PKG inhibitor Rp-8Br-cGMPS (86% ± 2.3%). These data demonstrate that the PKA attenuation cross talk of growth factor-induced m-calpain is functional in these cells, both extending our understanding of this counterregulatory pathway and validating the use of these cells in subsequent experiments.
The data above show that cAMP inhibits calpain activation through cAMP-dependent PKA. We hypothesized that this occurred directly through the novel mechanism of direct PKA phosphorylation of m-calpain. Human m-calpain large subunit contains a PKA consensus sequence, RRXS/T, in its putative regulatory domain III (46). Despite the previous report that calpain is not a phosphoprotein in vivo (1), we and others find m-calpain isolated from fibroblasts to be heavily phosphorylated as detected with antiserine and antithreonine antibodies (data not shown) and by standard biochemical assignment of phosphorylation sites (J. Cong, V. F. Thompson, and D. E. Goll, Abstr. Am. Soc. Cell Biol. 40th Annu. Meet., 2000, abstr. 2003). In fact, Cong and colleagues reported that the serine 369 in the PKA consensus site was phosphorylated. Independent of this report, we mutated this serine 369 to an alanine; the adjacent threonine 370 was simultaneously replaced with an alanine to rule out the possibility either that both amino acids are targets or that the elimination of serine 369 would alter the site to RRXXT with a shift to phosphorylation of threonine 370. Both the wild-type and ST639AA His-tagged constructs were expressed in cells, isolated using a nickel column, and subjected to in vitro phosphorylation by a cAMP-dependent PKA catalytic subunit. Wild-type hCANP was phosphorylated by PKA, while ST369AA hCANP was not phosphorylated (Fig. (Fig.3A).3A). These findings provide proof that m-calpain may be directly phosphorylated by PKA at this site in regulatory domain III.
However, the question remains whether the ST369AA m-calpain is functional. We expressed both wild-type and ST369AA hCANP as poly-His-tagged constructs in cells and used nickel columns to purify them. To calibrate ERK activation and any inhibition by PKA, we used the physiologically attainable micromolar concentrations of free calcium rather than the pharmacological millimolar levels employed for maximal activation in vitro. This lower level of calcium leads to calpain-dependent Tau degradation, but that is noticeable only after extended incubation periods (data not shown). Both wild-type and ST639AA hCANP were similarly inactive as isolated and subsequently activable by ERK in vitro as determined by the ability to clip the reporter protein Tau (29) (Fig. (Fig.3B).3B). Thus, the mutation of the serine and threonine in domain III did not adversely affect either basal or activated enzymatic activity. A second substrate, MAP2, was used to determine what effect PKA has on calpain functioning in vitro, as PKA phosphorylates Tau to render it calpain resistant (29, 35). Here, too, ERK activated both wild-type and ST369AA m-calpains to cleave MAP2 (Fig. (Fig.3C).3C). However, addition of the catalytic subunit of PKA inhibited MAP2 cleavage only by wild-type, not ST369AA, hCANP (Fig. (Fig.3C).3C). This strongly suggests that the elimination of this PKA phosphorylation site eliminates a negative attenuation mechanism.
To determine the biological role of PKA phosphorylation of m-calpain, the constructs needed to be expressed in live cells. As there have been previous reports of instability of expressed calpain in cells (13), we chose a dexamethasone-inducible system (8). The expression of both the GFP- and His-tagged clones was tightly regulated by dexamethasone (Fig. (Fig.4).4). The total level of m-calpain in the cell increased by 32% ± 5%; the GFP-tagged m-calpain constructs migrated at ~27 kDa higher. By and large, the clones are expressed at similar levels in all of the cells as determined by GFP fluorescence (data not shown). This expression of the m-calpain constructs in the entire cell population allows for easy investigation of cell responses.
To test our hypothesis that direct phosphorylation of m-calpain by PKA eliminates EGFR-mediated effects, we induced expression of the calpain constructs in quiescent cells by using dexamethasone. EGF-induced calpain activity of cells expressing the control wild-type hCANP was inhibited by 83% ± 3.7% by CPT-cAMPS, similar to the level of inhibition in mock-transfected cells, while that of cells expressing the PKA-resistant ST369AA construct was inhibited by only 25% ± 5.5% (Fig. (Fig.5).5). This only partial elimination of inhibition is likely due to the fact that the mutant construct constitutes only about one-third of the cellular m-calpain with the endogenous calpain contributing to the activation level while still being inhibitable.
EGF-induced cell migration activity of cells expressing ST639AA was unaffected by CPT-cAMPS (−7% ± 12% and 13% ± 11% for His- and GFP-tagged cells, respectively), while the motility of cells expressing the control wild-type hCANP constructs was fully eliminated by CPT-cAMPS (63% ± 6.2% and 75% ± 5.4% for His- and GFP-tagged cells, respectively) (Fig. (Fig.6A;6A; the data for GFP-tagged cells are not shown). In both His- and GFP-tagged cells, the absolute basal and EGF-induced cell migration activities are quite consistent in the presence or absence of dexamethasone. The motility parameters of these cells in the absence of dexamethasone induction were identical to those for parental cells (Fig. (Fig.6B;6B; the data for GFP-tagged cells are not shown). That this was due to expression of the PKA-resistant construct was shown both by the control wild-type construct being inhibited and by the cells that expressed ST369AA being inhibited by CPT-cAMPS in the absence of dexamethasone induction.
To eliminate confounding effects of endogenous mouse m-calpain, antisense oligonucleotides downregulated the murine m-calpain (17). Antisense m-calpain successfully downregulated endogenous m-calpain expression (Fig. (Fig.6E)6E) and inhibited ~60% of EGF-induced cell migration (Fig. (Fig.6D).6D). This result is consistent quantitatively with our previous report of partial motility elimination by downregulation of m-calpain (17). In the cells treated with antisense oligonucleotide and dexamethasone, hCANP constructs replaced endogenous m-calpain and restored the full EGF-induced motility response. Only the cell migration of hCANP wild-type cells was inhibited by CPT-cAMPS (56% ± 8.2%; P < 0.05), while that of hCANP ST369AA was not inhibited by CPT-cAMPS (−7.7% ± 8.7%; not significant) (Fig. (Fig.6C).6C). Thus, the ST369AA construct replaced the functionality of endogenous m-calpain while remaining PKA insensitive.
Structural studies would provide a molecular basis for this mechanism of inhibition and why it appears dominant over EGF-induced activation. A structural model of S369-phosphorylated m-calpain was generated by molecular modeling. The initial amino acid localizations are based on high-resolution X-ray procedures. Since S369 of domain III is located in the interface region between domains III and IV, the phosphorylated S369 is able to make intimate contacts with two residues in domain IV. These two residues are R628 and H634, which can form two and one interaction with S369-P, respectively (Fig. (Fig.7).7). These new interactions are predicted to rigidify the mobility of both domains III and IV and prevent the formation of an active cleft.
We present data demonstrating a novel mechanism for negative regulation of EGF-induced m-calpain activation--direct phosphorylation by PKA. We previously reported that EGF-induced m-calpain activation is negatively regulated by ELR-negative CXC chemokines (42). However, the molecular mechanism for this regulation was undefined. Herein, we report that PKA phosphorylation of m-calpain at amino acids S369 and T370 blocks EGFR-mediated activation in vivo. The addition of a negatively charged side chain is modeled to “freeze” m-calpain in an inactive state. These findings support an unexpected model of m-calpain regulation that involves protein phosphorylation of the regulatory domain.
In brief, mutation of a putative consensus PKA site at amino acids S369 and T370 to alanines generated a calpain molecule that was resistant to CPT-cAMPS negative attenuation of EGF-induced calpain activation and cell motility. One concern about mutagenesis is that it might render the target molecule inactive. We do not feel that our failure to attenuate EGFR signaling is due to such a false-positive result since the ST369AA mutant calpain remains activable by EGF. The in vitro activity of the ST369AA mutant was similar to that of wild-type m-calpain as determined by two different calpain activity assays, cleavage of Tau and cleavage of MAP2 (Fig. (Fig.3).3). Our other biologic reporter assay, the BOC-LM-CMAC assay, relies on the mutant calpain being activable in the face of normally negative regulation. This finding is buttressed by the fact that two different tagged constructs acted indistinguishably, expressed either transiently (single-cell calpain activation [data not shown]) or stably.
A second caveat is that we have not mapped the PKA phosphorylation site directly. This was not attempted either in vitro or in vivo. In vitro, purified m-calpain is multiply phosphorylated on both serines and threonines even in its nonactivated state (data not shown; Cong et al., Abstr. Am. Soc. Cell Biol. 40th Annu. Meet., 2000). In addition, nonphysiological conservative replacement of serine 369 could easily shift the PKA phosphorylation to the adjacent threonine 370. In vivo, the multiple, seemingly cotranslational phosphorylation would confound attempts to metabolically label m-calpain. Furthermore, as EGF activates on the small fraction of m-calpain that is in the plasma membrane (18), we are likely dealing with a substantially substoichiometric and potentially short-lived modification. However, in our preliminary study, we could detect low-level phosphorylation of wild-type hCANP caused by PKA activator CPT-cAMPS stimulation, but not in ST369AA hCANP (data not shown). As J. Cong et al. (Abstr. Am. Soc. Cell Biol. 40th Annu. Meet., 2000) reported, calpain can be phosphorylated even in a nonactivated state; the elevation in signal above the background phosphorylation of hCANP appeared to be less than twofold. We speculate that only a low level of phosphorylation is noted because PKA phosphorylation is a regulatory event that in vivo may both only involve a small fraction of total cell m-calpain and be transient or lead to rapid m-calpain degradation. Nevertheless, PKA will phosphorylate wild-type but not ST369AA m-calpain both in vitro and in vivo, strongly suggesting that this is the target site. Additionally, the MAP2 cleavage assay data (Fig. (Fig.3)3) demonstrate that removal of the PKA target site renders m-calpain resistant to PKA. Rather, we hypothesized that PKA directly phosphorylates the putative consensus site in the proposed regulatory domain III (25). Alteration of this site by replacement with alanines yields a construct resistant to both PKA phosphorylation and enzymatic repression in vitro and CPT-cAMPS attenuation in vivo. These findings demonstrate that PKA phosphorylation of domain III prevents activation of m-calpain and strongly support the structure-based prediction that domain III serves as a regulatory domain (46).
A third caveat is that calpain modulation may also occur via actions on proteins other than m-calpain. cAMP-dependent PKA also has been reported elsewhere to phosphorylate calpastatin (32), the endogenous calpain inhibitor, with this phosphorylation affecting the distribution of calpastatin in neuroblastoma cells (3). We do not exclude the possibility, implied in these reports, that phosphorylation of calpastatin by PKA might affect the regulation of calpain. However, our hypothesis that PKA direct phosphorylation of calpain inhibits calpain activation was sufficiently verified by both in vitro and in vivo experiments. These findings herein strongly suggest that direct phosphorylation of m-calpain is the major regulatory mechanism preventing EGF-induced m-calpain activation in fibroblasts. Another study reports that in vitro serine phosphorylation of bovine m-calpain, at the equivalent of S369, by calmodulin-dependent protein kinase II increases general calpain activity (37). The relevance of this finding to our report is uncertain for two reasons. First, CaM kinase II phosphorylated only autoproteolyzed m-calpain and had no effect on full-length calpain; this is interesting in light of full-length calpain now being considered to be as active as the autolyzed form, which might simply be an intermediary of the degradative attenuation process (28). Second, this report examined calpain activity only in vitro in the presence of supraphysiological concentrations of the activator calcium, whereas our studies were performed under cytosolic calcium concentrations in vivo; it is conceivable that opposite effects of phosphorylation at identical sites could be seen under such diverse circumstances.
Computer modeling of the phosphorylation at S369 supports our in vivo findings. We chose to focus on S369 since this is the best consensus PKA site and has been reported elsewhere to be phosphorylated as determined by phosphopeptide mapping (J. Cong et al., Abstr. Am. Soc. Cell Biol. 40th Annu. Meet., 2000); T370 was also mutated to prevent possibly nonphysiological usage of an alternate phosphorylation acceptor. Without doubt, the multiple interactions enabled by S369-P add new constraints in the interface between domain III and domain IV. In the proposed activation mode of calpain upon addition of calcium, various domains of calpain would undergo domain movement in the process of assembling the active site (22). Experimental evidence supporting this hypothesis has been recently reported (23), where disruption of critical interdomain constraints resulted in an increase in calcium sensitivity. By the same reasoning, if the extra constraints were added then opposite effects would occur. In the case of calpain 3 (or p94), a thorough structural analysis has again revealed that the effect on interdomain movement is crucial for the activity (26). In the case of S369-P, S369 of domain III is strategically located at the interface between domains III and IV. Phosphorylation of S369 not only provides a highly charged group but, more importantly, “extends” the length of the side chain. Thus, it enables interactions with a couple of residues of domain IV. As a consequence, these interactions give rise to extra constraints in the interface, thereby severely restricting the freedom of both domains. The PKA consensus sequence RRxS369 is present only in the closely related μ-calpain (calpain I) and the testis-specific calpain 11 (45). However, it remains to be demonstrated experimentally whether this other ubiquitous calpain is negatively attenuated by PKA since only R628 but not H643 is present for cross-bridging in μ-calpain. Parenthetically, the supraphysiological concentrations of calcium used to demonstrate CaM kinase II-induced calpain activity (37) might either overcome this movement restriction or disrupt salt bridging. The rigidification imposed by phosphorylation of S369 would essentially hamper the movement of these domains in the assembly of the active site and result in the loss of activity. This modeling, by being theoretical like all modeling, provides a potential molecular basis for the inhibitory action of PKA and forwards predictions for both m-calpain structure and dominance of inhibitory signals that might guide future experimental studies that lie beyond the scope of the present work.
Stable transfections yielded only a fractional increase in total calpain levels. This was not surprising. Initially, we attempted overexpression of both wild-type and ST369AA calpains using the strong CMV promoter. In transient transfections, robust GFP fluorescence was noted shortly after electroporation, but most of the cells rounded and detached within 24 h (data not shown). We did not pursue whether this was due to calpain-mediated deadhesion (6, 11, 12, 17) or actual apoptosis as it lay beyond the scope of the present study. Calpain has been implicated elsewhere in some mechanisms of apoptosis (43), and excess calpain activity might trigger caspase-mediated apoptosis (38). We established NR6 cell sublines containing the MMTV-driven calpain constructs. Even in the presence of dexamethasone, we attained exogenous expression at only ~30% of m-calpain. That this was sufficient to transmit EGFR-mediated calpain activity and motility in the presence of CPT-cAMPS suggests that endogenous calpain levels are in excess of those needed for robust deadhesion during motility. This is consistent with our ancillary studies that find that EGFR-mediated deadhesion requires only the submembrane subset of m-calpain (18).
In our previous paper, we presented evidence that the counterregulatory ELR-negative CXC chemokines inhibit EGF-induced cell migration but not proliferation (42). Herein, we demonstrate that this occurs via direct PKA phosphorylation of m-calpain. This provides for testable hypotheses concerning fibroblast functioning during wound repair. In the inflammatory and reparative stages the high levels of EGFR ligands in the wound bed would promote repopulation through both motility and mitogenesis. Later in the resolution phase, the presence of IP-10 from ingrowing endothelial cells (19) and a related CXCR3-binding chemokine from basal keratinocytes (IP-9 or I-TAC) (9) would channel the motile phenotype to matrix contraction (2). As cAMP has been shown elsewhere to be antiproliferative in fibroblasts (10, 15, 16, 20), a second PKA-mediated pathway would limit fibroplasia.
We thank Latha Satish, Philip Chang, and Douglas A. Lauffenburger for comments and discussions.
This work was supported by a grant from the National Institute of General Medical Sciences (NIGMS/NIH).