Tumor Necrosis Factor-α-inducible IκBα Proteolysis Mediated by Cytosolic m-Calpain

The cytokine tumor necrosis factor α (TNF-α) induces expression of inflammatory gene networks by activating cytoplasmic to nuclear translocation of the nuclear factor-κB (NF-κB) transcription factor. NF-κB activation results from sequential phosphorylation and hydrolysis of the cytoplasmic inhibitor, IκBα, through the 26 S proteasome. Here, we show a parallel proteasome-independent pathway for cytokine-inducible IκBα proteolysis in HepG2 liver cells mediated by cytosolic calcium-activated neutral protease (calpains). Pretreatment with either calpain- or proteasome-selective inhibitors partially blocks up to 50% of TNF-α-inducible IκBα proteolysis; pretreatment with both is required to completely block IκBα proteolysis. Similarly, in transient cotransfection assays, expression of the specific inhibitor, calpastatin, partially blocks TNF-α-inducible NF-κB-dependent promoter activity and IκBα proteolysis. In TNF-α-stimulated cells, a rapid (within 1 min), 2.2-fold increase in cytosolic calpain proteolytic activity is measured using a specific fluorescent assay. Inducible calpain proteolytic activity occurs coincidentally with the particulate-to-cytosol redistribution of the catalytic m-calpain subunit into the IκBα compartment. Addition of catalytically active m-calpain into broken cells was sufficient to produce ligand-independent IκBα proteolysis and NF-κB translocation. As additional evidence for calpain-dependent IκBα proteolysis and NF-κB activation, we demonstrate that this process occurs in a cell line (ts20b) deficient in the ubiquitin-proteasome pathway. Following inactivation of the temperature-sensitive ubiquitin-activating enzyme, IκBα proteolysis occurs in a manner sensitive only to calpain inhibitors. Our results demonstrate that TNF-α activates cytosolic calpains, a parallel pathway that degrades IκBα and activates NF-κB activation independently of the ubiquitin-proteasome pathway.

Nonlysosomal (cytoplasmic) protease systems have recently been identified as important regulators of intracellular activi-ties including programmed cell death, protein kinase abundance, and cell-cycle progression (1)(2)(3). In viable cells, two prominent cytoplasmic protease systems have been identified. These include the ubiquitin-proteasome pathway, mediating targeted turnover of misfolded and unstable proteins, and the calcium-activated neutral protease (calpain)-calpastatin system, initially thought to be important in regulating turnover of protein kinases and key structural proteins in the cell (1). More recently, however, inducible proteolysis has also been shown to be important in hormonal control of gene expression by modulating the nuclear abundance of certain transcription factors. These processes include cholesterol-induced cleavage of the sterol-regulated element binding protein (reviewed in Ref. 2) and, of special interest to the pathophysiology of inflammatory processes, mechanisms for intracellular signaling produced by the cytokine tumor necrosis factor-␣ (TNF-␣). 1 Following binding its receptor on the plasma membrane, TNF-␣ initiates de novo transcription of genetic networks, in part, through activating nuclear translocation of the cytoplasmic transcription factor nuclear factor-B (NF-B) (4,5). NF-B, a multiprotein complex inactivated in the cytoplasm by association with its IB inhibitor, translocates into the nucleus following dissociation of the NF-B⅐IB complex. TNF-␣ modifies NF-B⅐IB association through a process initiated by inducible IB␣ serine phosphorylation on its amino-terminal regulatory domain, a modification coupled to IB polyubiquitination (Ub n ) on adjacent lysine residues (6). NF-B⅐IB dissociation requires IB proteolysis because phosphorylated and ubiquitinated IB still inactivates NF-B (Ref. 6 and references therein).
Presently, the ubiquitin-proteasome system has been the only pathway identified in mediating cytokine-inducible IB proteolysis. Pretreatment with cell-permeant proteasome inhibitors blocks TNF-␣-inducible IB proteolysis concomitantly with the accumulation of Ub n -and phosphorylated IB intermediates (6,8). Independently, inducible IB␣ proteolysis in cells harboring thermolabile ubiquitin-activating enzymes is markedly slowed at non-permissive temperatures (9).
Several lines of evidence indicate the presence of alternative * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  (nonproteasome-dependent) processing pathways for IB proteolysis. First, in pre-B lymphocytes, c-Rel:NF-B1 is constitutively nuclear as the consequence of a calcium-dependent proteolytic activity that preferentially affects IB␣ (rather than IB␤ (10)). Second, we have observed a non-proteasome-dependent pathway mediating inducible IB␣ proteolysis (and NF-B activation) following respiratory syncytial virus infection of human airway epithelial cells (11). However, whether additional nonproteasome-dependent pathways participate in cytokine-inducible NF-B activation have not been explored.
These studies prompted us to examine whether nonproteasome-dependent pathways participate in cytokine-inducible IB␣ degradation. Here we investigate the proteolytic mechanism involved in a well characterized model of NF-B activation in TNF-␣-stimulated HepG2 hepatocytes, where NF-B activation mediates the expression of acute phase reactants (12,13). By using calpain and proteasome-selective inhibitors, we demonstrate that inducible IB␣ proteolysis is partially blocked following inhibition of either pathway and completely blocked following inhibition of both. By using a specific fluorescent assay in intact cells, we describe for the first time that TNF-␣ rapidly activates cytosolic calpain proteolytic activity. In subcellular fractionations of TNF-␣-stimulated cells, the catalytic m-calpain subunit translocates from the particulate into the cytosolic fraction (the latter containing IB␣) coincidentally with IB␣ proteolysis. Moreover, TNF-␣-inducible IB␣ proteolysis occurs in cells conditionally deficient in the ubiquitin-proteasome pathway, and in cells expressing IB␣ mutations deficient in proteasome-dependent processing. Together, these data implicate calpains are a parallel pathway in mediating IB␣ proteolysis and NF-B activation.
Plasmid Construction-A reporter plasmid containing Ϫ162 to ϩ44 bp of the human IL-8 promoter driving expression of CAT was produced by subcloning the BamHI/HindIII restricted polymerase chain reaction product of the IL-8 gene into the same sites of a pGEMCAT plasmid. For this, an upstream primer 5Ј-ACTTGGATCCACTCCGTATTTGATAAG-G-3Ј (BamHI site underlined) and the downstream primer 5Ј-AGAAG-CTTGTGTGCTCTGCTGTCTCTGAA-3Ј were used to polymerase chain reaction amplify the IL-8 promoter (11). Plasmids were purified by ion exchange chromatography (Qiagen) and sequenced to verify authenticity prior to transfection.
In Vitro Protease Assays-For proteasome activity, 200 g of HepG2 lysates from control or TNF-␣-treated (30 ng/ml, 15 min) were incubated with 60 M in 1 ml of ATP-containing reaction buffer (15) in the presence or absence of indicated protease inhibitor (30 min, 30°C). AMC released was quantitated by measuring fluorescence emission intensity at 440 nm (I ex , 365 nm) normalizing to standards. Results are mean Ϯ S.D. of three experiments. To confirm Suc-LLVY-AMC hydrolyzing activity was dependent on proteasome activity, HepG2 lysate was proteasome-depleted by ultracentrifugation (150,000 ϫ g, 26 h, 4°C). Western blots before or after ultracentrifugation were done to detect the core proteasome subunits, RING 10 and . Both RING 10 and , present in the crude supernatant, were lost in the S 150 supernatant and recovered in the 150,000 ϫ g pellet. By contrast, the 80-kDa m-calpain catalytic subunit remained in the S 150 supernatant. Ninety four percent of the AMC generation was lost in the S 150 supernatant, indicating that proteasome activity was selectively being measured by this assay. For measurement of m-calpain caseinolytic activity, 0.025 units/ml purified human erythrocyte m-calpain was incubated with phosphorylated bovine casein (2 mg/ml) in the presence or absence of indicated inhibitors under standard conditions (4 mM CaCl 2 , 10 mM dithiothreitol, at 30°C, 1 h). Hydrolysis was quantitated by Coomassie Brilliant Blue colorimetric assay (16). Similar results were obtained with purified rabbit skeletal muscle m-calpain (not shown).
Calpain Assay in Intact Cells-For HepG2 and ts20b cells, calpain activity was measured by the rate of generation of the fluorescent product, AMC, from intracellular thiol-conjugated Boc-Leu-Met-CMAC (17). Cells were dispersed, grown on glass coverslips, continuously superfused with physiologic saline solution at 37°C, and sequentially imaged with a quantitative fluorescence imaging system (18). At t ϭ 0, Boc-Leu-Met-CMAC (10 M, Molecular Probes) was introduced into the superfusion solution, and mean fluorescence intensity (excitation 350 nm, emission 470 nm) of individual cells was measured at 60-s intervals. At 10 min, TNF-␣ (30 ng/ml) was added to the superfusion solution with 10 M Boc-Leu-Met-CMAC. The slope of the fluorescence change with respect to time represents the intracellular calpain activity (17). Hydrolysis of the thiol-conjugated substrate was rate-limiting for the generation of fluorescent product as shown by comparing the initial rate of cell fluorescence increase after exposure to Boc-Leu-Met-CMAC with that produced by CMAC. CMAC requires only the thiol conjugation step, not hydrolysis, for fluorescence. There was a 35.8-fold increase in the AMC fluorescence rate compared with Boc-Leu-Met-CMAC, demonstrating that hydrolysis and not conjugation was rate-limiting. For calpain assays in whole cell populations, suspension cultures of HepG2 cells were loaded with 10 M Boc-Leu-Met-CMAC, and changes in intracellular fluorescence was measured prior to and after TNF-␣ addition at 37°C using a FACS Vantage system. Cellular fluorescence of AMC was measured using a 360-nm excitation filter and a 405-nm long-pass emission filter.
Calpain-dependent IB␣ Proteolytic Assay in Cytosolic S 100 Extract-Two hundred g of protein 100,000 ϫ g supernatant (S 100 ) was incubated with 150 ng of recombinant human IB␣ for indicated times in Reaction Buffer (RB, 25 mM HEPES, pH 7.2, 65 mM KCl, 2 mM MgCl 2 , 1.5 mM CaCl 2 , 2 mM dithiothreitol) at 32°C in a final volume of 60 l. IB␣ degradation was quantitated by Western immunoblot.
IB Proteolysis and NF-B Activation Assay in Broken Cells-Indicated amounts of purified rabbit skeletal muscle m-calpain was added to a mixture of 200 g of HepG2 cytosol and 1 ϫ 10 6 sucrose cushionpurified nuclei (12) in 10% glycerol-containing RB (100 l, 32°C). IB proteolysis in cytoplasmic extract and nuclear Rel A was analyzed by Western immunoblot following sucrose-cushion purification of nuclei (12). Gel mobility shift assay was performed using the NF-B-binding site from the angiotensinogen promoter as described (12).
In contrast, IB␣ proteolysis was partially blocked by the proteasome inhibitors lactacystin and Z-LLF (Fig. 1a, lanes 3  and 4). Of the presumed proteasome-selective inhibitors, only Z-LLL was a completely effective inhibitor of IB␣ proteolysis. Surprisingly, moreover, the calpain inhibitors Z-LnL and E64 also partially blocked IB␣ proteolysis, even under conditions where IB␤ proteolysis was unaffected (compare lane 2 with 6 and 7). These data suggest a parallel contribution of calpainlike proteases in TNF-␣-inducible IB␣ hydrolysis.
To define the kinetics of proteasome-independent pathways mediating IB␣ proteolysis, IB␣ half-life in TNF-␣-treated cells was compared in cells containing with that in cells lacking proteasome activity (Fig. 1b). In cells not treated with protease inhibitors, IB␣ proteolysis is rapid (t1 ⁄2 of 1-3 min), occurring coincidentally with the generation of phosphorylated IB␣ intermediates (Fig. 1b, IB␣ P ). In cells pretreated with the potent irreversible proteasome inhibitor lactacystin, IB␣ proteolysis occurs with a detectably slower half-life (t1 ⁄2 of 7-15 min) and is incomplete, with the appearance of a Ͻ30-kDa intermediate (Fig. 1b, bottom). To determine whether any pathway other than the combined calpain/proteasome account for IB␣ proteolysis, the additive effects of the specific irreversible calpain inhibitor Z-LLY (21) and lactacystin were studied (Fig. 1c). At saturating doses, neither Z-LLY nor lactacystin alone could completely inhibit IB␣ proteolysis. In the presence of both inhibitor types, IB␣ proteolysis was completely blocked with accumulation of non-and phosphorylated IB␣ intermediates (Fig. 1c). We note consistently that IB␣ P intermediate was detectable at the 15-min time point in the presence of proteasome inhibitors but not in calpain inhibitors (see "Discussion").
Enzymatic activity of calpains are influenced by the effect of endogenous calpain inhibitor, calpastatin. As additional evidence for the role of calpains in NF-B activation, the effect of transiently expressed calpastatin was determined on NF-Bdependent reporter activity in transient cotransfection assay (18,22). We have previously shown that the human IL-8 promoter is TNF-␣-inducible in a manner solely dependent on a high affinity NF-B site (23). Cotransfection of calpastatin expression plasmid (pcDNA I-calpastatin) did not affect basal IL-8/CAT activity but significantly blocked TNF-␣-inducible CAT activity (Fig. 1d). As a control, the effect of calpastatin on IB␣ steady state levels was measured in transient transfectants. HepG2 cells cotransfected with IL-2 receptor expression plasmid in the absence or presence of various concentrations of pcDNA I-calpastatin were stimulated with TNF-␣. Following isolation of transfected cells, a Western immunoblot was done to determine changes in IB␣ in cytosolic lysates (Fig. 1e). In the presence of 2.5 g of pcDNA I-calpastatin, IB␣ proteolysis was inhibited by ϳ50%. Combined, these data strongly suggest a parallel contribution of the calpain system in TNF-␣-inducible proteolysis of IB␣ and NF-B activation.
Specificity of Protease Inhibitors-The specificity and effect of protease inhibitors for proteasome and calpain activities were directly measured in vitro (Fig. 2). Hydrolysis of the fluorogenic substrate Suc-LLVY-AMC was used to measure proteasome activity in whole cell lysates (15,19). As described under "Experimental Procedures," following proteasome depletion, 94% of the Suc-LLVY-AMC hydrolyzing activity was depleted, indicating that the assay is measuring bona fide proteasome activity. As shown in Fig. 2a, in both control and TNF-␣-treated cells, Suc LLVY-AMC hydrolytic activity was indistinguishable (12.5 and 13 nmol/min/mg, respectively). Also, greater than 90% inhibition of proteasome activity was seen following lactacystin, Z-LLF, and Z-LLL treatment, indicating these inhibitors potently inhibited (whereas Z-LnL, E64, Z-LLY and PMSF had no effect) cellular proteasome activity. This effect was consistent for either control or TNF-␣-stimulated cells.
Caseinolytic activity of purified m-calpain was measured in the presence of the same inhibitors (Fig. 2b). Surprisingly, the presumed "selective" proteasome inhibitors Z-LLF and Z-LLL, as well as the calpain inhibitors Z-LnL, E64, and Z-LLY, were potent inhibitors of m-calpain. Only lactacystin, therefore, was able to differentiate calpain from proteasome activity, and Z-LnL, E64, and Z-LLY, conversely, were specific for caseinolytic activity of calpain, without effects on the proteasome.
TNF-␣-inducible Changes in Intracellular Calpain Proteo-lytic Activity-Direct measurement of dynamic changes in intracellular calpain activity in broken cells has been hampered due to the presence of endogenous calpastatin inhibitor that rapidly associates with active calpains following cell disruption. However, the recent development of a specific fluorescent assay using a cell-permeant calpain substrate to measure changes in calpain proteolytic activity has obviated the need for broken cell assays (17). After diffusion of the substrate, Boc-Leu-Met-CMAC, into cells, it is conjugated with glutathione (GSH) to form a membrane-impermeant, nonfluorescent calpain substrate. Following its hydrolysis, the unquenched fluorescent product (AMC-GSH) accumulates, where its rate of accumulation is a measure of intracellular calpain activity (17). Specificity of this assay for calpain has been previously demonstrated by its inhibition by the specific calpain inhibitor, Z-LLY, and independence from lysosomal, serine, or cathepsin proteases (17). A basal rate of generation of AMC-GSH was observed in individual cells (Fig. 3a, 1-10 min). We further examined the assay specificity by measuring the effects of inhibitors on basal proteolytic activity. Basal generation of the fluorescent proteolysis product is calcium-dependent due to the inhibitory effects of the intracellular calcium chelator 1,2-bis(oaminophenoxy)ethane-N,N,NЈ,NЈ-tetraacetic acid tetra(acetoxymethyl) ester (76% inhibition, 100 M (n ϭ 30)) and is quantitatively inhibited by the calpain inhibitors, Z-LnL (59% inhibition at 100 mM (n ϭ 40) and 93% at 200 M (n ϭ 30)) and Z-LLY (62% inhibition at 100 M (n ϭ 70)), and not inhibited by the proteasome inhibitor, lactacystin (0% inhibition at 10 M (n ϭ 20)).
In individual cells, TNF-␣ increased calpain activity ϳ2.2fold over the basal rate within 1 min of exposure (Fig. 3a). The TNF-␣-induced stimulation of calpain activity is blocked by the calpain inhibitor Z-LnL but not lactacystin (Fig. 3b), indicating an exact parallel of inhibitor specificity for intracellular calpain activity as for purified m-calpain in vitro (cf. Figs. 3b and 2c). This assay was also applied by FACS analysis to determine the portion of TNF-␣-responsive HepG2 cells. As shown in Fig. 3c, in HepG2 populations, mean cellular fluorescence intensity, as an indicator of calpain activation, increased an average of 2.5 Ϯ 0.3-fold in greater than 95% of cells following TNF-␣ administration for 60 -800 s. Specificity of changes in mean fluorescence intensity also follows the same inhibitor profile as shown in the single cell assay (not shown).

TNF-␣ Induces Changes in m-Calpain
Abundance-For mcalpain proteolytic activity to be relevant for IB␣ proteolysis, we sought to determine the subcellular distribution of m-calpain in control and TNF-␣-stimulated cells. For this, particulate (S 100 pellet) and cytosolic (S 100 supernatant) fractions were prepared at various times following TNF-␣ treatment by ultracentrifugation at 100,000 ϫ g and analyzed for both 80-kDa m-calpain catalytic subunit and IB␣ by Western immunoblot (Fig. 4a). Although m-calpain could be detected in both cytosolic and particulate fractions, normalizing each fraction per microgram of protein, the highest specific activity of m-calpain was found in the particulate fraction. In the cytosolic fraction, m-calpain abundance increased 2-fold within 2 min following TNF-␣ stimulation in the cytosolic fraction. In both fractions, however, m-calpain abundance fell (compare 15-min time points with control, Fig. 4a). Importantly, we note that the cytosolic fraction contained IB␣ and that changes in m-calpain subunit occurred concomitantly with IB␣ proteolysis (Fig. 4a) and synchronously with calpain proteolytic activity (Fig. 3). Internal control immunostaining of inert Rel B in the same membrane was used to document equivalent protein loading (Fig. 4a, bottom).
To determine whether cytosolic fractions from TNF-␣treated cells containing translocated m-calpain catalytic subunit (determined by Western immunoblot, Fig. 4) also contain IB␣ proteolytic activity, an in vitro protease assay was established (Fig. 4b). In this assay, purified recombinant human IB␣ (rhIB␣) was added to TNF-␣-stimulated cytosolic lysates (S 100 supernatant), and the effect on rhIB␣ proteolysis was determined by Western immunoblot. We observed a time-dependent proteolysis of rhIB␣, a proteolysis that was blocked either by calcium chelation or the addition of calpain inhibitors, E64 or Z-LnL but not lactacystin or PMSF (Fig. 4b, bottom). m-Calpain Is Sufficient for Ligand-independent IB␣ Proteolysis and NF-B Translocation-To determine whether mloaded with calpain substrate at time ϭ 0. Accumulation of calpain hydrolytic product measured before and after TNF-␣ addition (arrow, at 10 min). b, relative stimulation of single cell calpain activity in the presence of protease inhibitors. Relative calpain activity was measured after no addition (control), TNF-␣ (30 ng/ml), TNF-␣ ϩ Z-LnL (100 M), or TNF-␣ ϩ lactacystin (10 M, 30 min). The basal activity in each cell was measured for use as its own control (open bars) prior to TNF-␣ stimulation (solid bars). Increased proteolytic rate occurred only for TNF-␣ (p Ͻ 0.001) and TNF-␣ with lactacystin (p Ͻ 0.05). c, kinetics of change in intracellular fluorescence intensity in TNF-␣-exposed cell populations. FACS analysis. A representative experiment is shown (n ϭ 6). Fluorescence intensity increased 2.5 Ϯ 0.3 fold in Ͼ95% cell population. FIG. 3. Effect of TNF-␣ on calpain activity in intact HepG2 cells. a, fluorescence intensity changes in a single cell. Cells were calpain proteolyzed native IB within the NF-B⅐IB complex and could produce ligand-independent NF-B activation, activated m-calpain was added to broken cell lysates (containing nuclei, Fig. 5). m-Calpain produced a time-dependent degradation of endogenously expressed IB␣ (Fig. 5, top). IB␣ was proteolyzed into transiently stabilized carboxyl-terminal intermediates of ϳ30 kDa (arrow, Fig. 5), an intermediate also seen in TNF-␣-stimulated cells lacking proteasome activity (see Fig.  1b). The effect of m-calpain was dose-dependent and required m-calpain proteolytic activity (Fig. 5, bottom). Inducible phosphorylation is apparently not required for calpain-induced IB␣ proteolysis because the phosphorylation-defective IB␣ mutant, S32/36A, is inducibly degraded, and nonphosphorylated recombinant IB␣ (IB␣ expressed in Escherichia coli) is efficiently degraded in vitro (not shown).
To determine whether calpains could result in NF-B activation, nuclei were purified on sucrose cushions and nuclear proteins extracted. Gel mobility shift assays showed that mcalpain induced a time-and dose-dependent increase in Rel A:NF-B1 DNA binding activity (Fig. 6a, indicated as complex 2, a species previously characterized by supershift assay (12)). To additionally demonstrate NF-B translocation, changes in 65-kDa Rel A nuclear abundance was measured by Western immunoblot (Fig. 6b), where a 2.3-fold increase in Rel A in the m-calpain treated nuclei was seen. We therefore conclude that ligand-independent IB␣ proteolysis and NF-B activation can be effected by m-calpain.
Degradation of IB␣ in Ubiquitin-Proteasome-defective Cell Lines-As additional evidence for calpain-mediated, proteasome-independent pathway for IB proteolysis and NF-B activation, we analyzed the effect of TNF-␣ in Balb/c 3T3 cells conditionally defective in the ubiquitin-proteasome pathway. ts20b cells express a temperature-sensitive E1 responsible for initial ATP-dependent step in the Ub n reaction (14,18), whereas control H38-5 cells are corrected ts20b stably transfected with the wild-type E1 (14). Relative stimulation of calpain activity was observed in individual cells measuring hy-  Fig. 1c). Bottom panel, dose response and sensitivity to calpain inhibitors. Calpain proteolytic activity is required for IB␣ proteolysis.
FIG. 6. Ligand-independent activation of Rel A:NF-B1 by purified m-calpain catalytic subunit in broken cell assay. a, gel mobility shift assay for nuclear DNA binding activity. Purified mcalpain was added to broken cell assay at the indicated concentrations and times. Following incubation, nuclei were prepared by sucrose cushion centrifugation. Assays were performed using 15 g of nuclear extracts binding to radiolabeled angiotensinogen NF-B-binding site (12). Migration of RelA⅐NF-B1 heterodimer (complex C2) demonstrated previously by supershift assay is shown (12). drolysis of the cell-permeant calpain substrate Boc-Leu-Met-CMAC incubated with TNF-␣ (30 ng/ml). In ts20b cells, calpain activity increased from 1.70 Ϯ 0.15 arbitrary units/min (n ϭ 29) to 2.49 Ϯ 0.15 arbitrary units/min (n ϭ 30, 60 min, 32°C) after TNF-␣ incubation (p ϭ 0.0005, two-tailed t test). After E1 inactivation by culture in ts20b cells at the restrictive temperature (39°C), TNF-␣-induced IB␣ proteolysis was still detectable at 15 min and continued until 60 min (Fig. 7a, top). By contrast, identically treated H38-5 cells showed a more rapid IB␣ proteolysis with a nadir at 15 min, followed by its resynthesis over 60 -120 min. This observation excludes nonspecific temperature effects on the TNF-␣ signaling pathway. In both cell types, DNA binding activity of the Rel A:NF-B1 heterodimer was induced in parallel to IB␣ proteolysis (Fig. 7a,  bottom). Finally, IB␣ proteolysis in restricted ts20b cells is blocked by calpain (Z-LnL, E64, and Z-LLY), and not by proteasome inhibitors (Fig. 7b). We note the slower kinetics of IB␣ proteolysis in the temperature-restricted ts20b cells are remarkably similar to those of lactacystin-treated HepG2 cells (Fig. 1b). Together, these data indicate that calpain-induced IB␣ turnover is slower than when both proteolytic systems are intact.

DISCUSSION
Calpains are intracellular calcium-dependent cysteine proteases whose ubiquitously expressed subunits include milli (m)-calpain and micro ()-calpain. Although these heterodimeric isoforms have indistinguishable substrate affinities, m-and -calpain are found in distinct subcellular local-izations and therefore may subserve distinct physiological roles (24,25). Here we show for the first time that the calpaincalpastatin system is a parallel pathway partly responsible for TNF-␣-inducible IB proteolysis and NF-B activation. TNF-␣, therefore, activates NF-B through the participation of two distinctly regulated cytoplasmic (nonlysosomal) protease systems as follows: (i) the constitutive proteasome pathway, where IB␣ proteolysis is governed by its rate-limiting post-translational modification (coupled phosphorylation/ubiquitination), and (ii) the inducible calpain-calpastatin system, where protease activity is directly modified by TNF-␣.
In the past, distinguishing between the effects of calpains and the proteasome in intracellular regulatory processes has been difficult because few selective inhibitors of the two cytoplasmic protease systems were identified. Data presented herein indicate that pathway-selective probes exist that can be used to dissect the parallel function of these protease systems in cytokine signaling. A role for the calpain-calpastatin pathway mediating NF-B activation is based on the convergence of the following observations. 1) Inducible IB␣ proteolysis is only be partially blocked by either calpain-selective or proteasomeselective inhibitors and completely blocked by both. 2) In TNF-␣-stimulated cells, a rapid (within 1 min), 2.2-fold increase in cytosolic calpain proteolytic activity in intact cells is measured.
3) Calpain proteolytic activity occurs indistinguishably with the particulate to cytosol redistribution of the catalytic mcalpain subunit. 4) IB␣ proteolysis occurs coincidentally with increases in m-calpain abundance in the cytosol. 5) Introduction of catalytically active m-calpain is sufficient to produce ligand-independent NF-B activation. 6) Calpain-dependent IB␣ proteolysis is demonstrated in cells lacking proteasome activity (ts20b cells).
The mechanism for activation of calpains in intact cells is unknown. In vitro, calpains exposed to nonphysiological concentration of calcium acquire enzymatic activity through autoproteolysis of its constituent subunits (1). In intact cells, evidence for autolytic activation or activation following changes in intracellular calcium concentration is weak. In other studies, calpains are known to be long-lived proteins with half-lives of Ͼ5 days; this observation would not be consistent with an autolytic protease (27). Our data indicates that TNF-␣-stimulated calpain activity occurs in the absence of detectable autolysis because autolytic products are not detected at times when changes in protease activity can be measured. Moreover, calpain activation in the absence of detectable changes in intracellular calcium concentrations has been described in hepatocytes (26). In data not shown, we have not observed any changes in total intracytoplasmic calcium concentrations in HepG2 cells. Nevertheless, intracellular calcium is apparently required for calpain activity in intact cells, because intracellular calcium chelators block calpain activity and IB␣ proteolysis (Table I, data not shown).
One other mechanism for calpain activation could include changes in subcellular localization. Calpains are not randomly distributed throughout the cell. In cultured cell lines, m-calpain is distributed in a fine reticular network in the cytosol, implicating an association with cytoskeletal elements (28), and in the central nervous system, m-calpain content is membraneassociated (29). In cultured HepG2 cells, we observe consistently that m-calpain redistributes into the soluble cytoplasmic fraction, a fraction containing IB␣, following TNF-␣ treatment. Whether redistribution is the mechanism for m-calpain activation will require additional investigation. Although mcalpain activity in the membrane fraction was previously thought to be important for proteolysis of protein kinase C (30), cytosolic calpain activity appears to be important in turnover of FIG. 7. TNF-␣-induced IB␣ degradation in ubiquitin pathway-defective cells is calpain-dependent. a, top, Western immunoblots for IB␣ in E1-deficient ts20b and wild-type H38 -5 cells. Upon TNF-␣ treatment, relative to control values, IB␣ was 83% (15 min), 11% (1 h), and 43% (2 h) in ts20b cells, and 43% (15 min), 63% (1 h), and 68% (2 h) of controls in H38-5 cells. Bottom, gel shift assay. Rel A:NF-B1 DNA binding activity (C2) increased in parallel with cytoplasmic IB␣ proteolysis. b, effect of protease inhibitors on inducible IB␣ degradation in ts20b cells. Western immunoblots are shown using IB␣ and Rel B (internal control) antibodies. Inducible IB␣ degradation is blocked by Z-LnL, E64, and Z-LLY, but not by lactacystin (Lact) and PMSF, indicating IB␣ proteolysis occurs via a calpain-sensitive pathway. the p53 oncoprotein (18). Based on our subcellular fractionation experiments, the intracellular site of proteolysis of IB␣ probably also occurs in the cytosol.
Calpain activity is inducible following activation of other hormone receptors, including the hepatic purinergic receptor (17), and the pituitary thyrotropin-releasing hormone receptor (31), perhaps indicating a role for second messenger involvement. Others have shown that phospholipid mediators, including second messengers implicated in TNF-␣ signaling, can activate calpain catalytic activity through a mechanism that may involve their direct binding to the 30-kDa regulatory subunit (29,32,33). These lipids apparently lower calcium requirements to a range normally found intracellularly (29). Lipid mediators may be important intermediates for TNF-␣-induced calpain activity for several reasons. First, TNF-␣ is known to increase ceramide production through its effects on acid sphingomyelinase activity in endosomal compartments (34); this second messenger has been linked to NF-B activation (35). Second, ceramide directly stimulates intracellular calpain activity in permeabilized cells (32).
Calpains are increasingly recognized to be important regulators of intracellular signaling processes. Initially described in turnover of activated protein kinase C, erythrocyte ankyrin, and calmodulin-binding proteins (Ref. 1 and references therein), calpains have recently been implicated in mediating turnover of the c-Fos transcription factor (36,37) and the tumor suppressor gene product p53 (18). Interestingly, both c-Fos and p53 were initially described to be proteasome substrates. Our data adds IB␣ to the emerging list of key regulatory proteins acted upon by a parallel calpain-proteasome pathway. Of relevance to IB␣, erythrocyte ankyrin itself is a substrate for calpain proteolysis (18,38).
The TNF-␣-inducible calpain pathway mediating IB␣ proteolysis described here is probably distinct from the two previously reported nonproteasome-dependent IB␣ proteolytic pathways (10,11). In the first report, constitutive IB␣ turnover in an undifferentiated pre-B lymphocytic cell line was not inhibited by the potent calpain inhibitors, calpain inhibitors I and II, or Z-LLF, agents that interfere with IB␣ proteolysis in our system ( Fig. 1 and see Ref. 12). Second, potent calpain inhibitors MG132 and Z-LLF do not have significant effects on IB␣ proteolysis in respiratory syncytial virus-infected epithelial cells (11). The relationship of these pathways to calpaincalpastatin pathway, therefore, seems unlikely.
In summary, we implicate the calpain-calpastatin and proteasome pathways are parallel mechanisms mediating inducible IB␣ proteolysis by the cytokine TNF-␣. These data indicate that calpains contribute to rapid IB␣ proteolysis through a mechanism involving changes in total cytosolic calpain proteolytic activity.