Iron Regulatory Protein 2 Turnover through a Nonproteasomal Pathway*

Iron regulatory protein 2 (IRP2) controls the synthesis of many proteins involved in iron metabolism, and the level of IRP2 itself is regulated by varying the rate of its degradation. The proteasome is known to mediate degradation, with specificity conferred by an iron-sensing E3 ligase. Most studies on the degradation of IRP2 have employed cells overexpressing IRP2 and also rendered iron deficient to further increase IRP2 levels. We utilized a sensitive, quantitative assay for IRP2, which allowed study of endogenous IRP2 degradation in HEK293A cells under more physiologic conditions. We found that under these conditions, the proteasome plays only a minor role in the degradation of IRP2, with almost all the IRP2 being degraded by a nonproteasomal pathway. This new pathway is calcium-dependent but is not mediated by calpain. Elevating the cellular level of IRP2 by inducing iron deficiency or by transfection causes the proteasomal pathway to account for the major fraction of IRP2 degradation. We conclude that under physiological, iron-sufficient conditions, the steady-state level of IRP2 in HEK293A cells is regulated by the nonproteasomal pathway.

Iron metabolism is exquisitely regulated by all organisms, from bacteria to humans. In mammals, the iron-regulatory proteins (IRPs) 2 mediate the coordinate expression of proteins that participate in iron metabolism (1)(2)(3). When iron stores are low, the IRPs bind to an RNA stem-loop structure known as an ironresponsive element located in either the 5Ј-or 3Ј-untranslated region of mRNA. If the iron-responsive element is close to the cap site, binding of the IRP blocks initiation of translation, causing a decrease in the level of the protein encoded by that mRNA. Conversely, when the iron responsive element is located in the 3Ј-untranslated region, binding of the IRP stabilizes the mRNA by decreasing susceptibility to nuclease attack, causing an increase in the level of the protein encoded by the mRNA.
Mammals have two known IRPs, IRP1 and IRP2. The two IRPs are regulated by different mechanisms (3). When cellular iron stores are low, IRP1 lacks a functional iron-sulfur center and binds to its iron-responsive element targets. When iron stores are sufficient, IRP1 regains its full iron-sulfur center, loses the ability to bind to iron responsive element, and functions as a cytosolic aconitase (4,5). The cellular levels of IRP1 are unaffected by iron status in most cell types.
In contrast, IRP2 protein and iron responsive element binding activity are readily detected when iron stores are limited but are low or absent when iron stores are sufficient (6,7). The decrease in IRP2 protein occurs as a consequence of rapid degradation; synthesis of the protein is constitutive and generally does not vary substantially with iron status (6,8). The proteasome was implicated in this degradation soon after IRP2 was described in 1994 (6 -9). Identifying the E3 ligase specific for IRP2 took much longer and was reported in 2009 (10,11).
Quantitation of physiological levels of IRP2 is problematic because of its low abundance in the cytoplasm. Most studies of IRP2 have rendered cells iron deficient by treatment with chelators such as deferoxamine, thus greatly increasing the IRP2 level and allowing more confident quantitation. Addition of a high concentration of iron, usually as ferric ammonium citrate (FAC), triggers rapid degradation of IRP2 by the proteasome. As Dycke and co-workers (12) pointed out, the turnover of IRP2 in cells not perturbed by these manipulations has not been well characterized. Furthermore, most studies have been carried out with transfected cells overexpressing IRP2, often with an epitope tag to facilitate detection. These manipulations can also alter the kinetics and pathways of IRP2 turnover (13).
These considerations led us to investigate the turnover of IRP2 in nontransfected cells grown under standard culture conditions without manipulation of their iron status by chelators. Under these conditions, we found that the turnover of IRP2 is primarily mediated by a nonproteasomal pathway.

EXPERIMENTAL PROCEDURES
Cell Culture-HEK293A cells were a kind gift from Dr. Rui-Ping Xiao (National Institute on Aging) and are also commercially available from Invitrogen. Cells were maintained in high glucose Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (Invitrogen), in a 37°C humidified incubator (Thermo Fisher Scientific) with carbon dioxide controlled at 5%. Cells were grown on BioCoat TM collagen I-coated Petri dishes (BD Biosciences) and routinely passaged (ϳ1:5) using citrate saline (135 mM potassium chloride, 15 mM sodium citrate, pH 7.4). Cells were washed with citrate saline twice and then incubated with 1.5 ml of citrate saline (for a 100-mm dish) for 10 min at 37°C. Then, the dish was gently tapped to detach cells.
SDS-PAGE and Quantitative Western Blot-Cells were briefly washed twice with phosphate buffered saline from Sigma (1 mM KH 2 PO 4 , 3 mM Na 2 HPO 4 ⅐7H 2 O, 155 mM NaCl, pH 7.4, supplemented with 1 mM diethylenetriamine pentaacetic acid) followed by the addition of radioimmune precipitation assay buffer (50 mM HEPES, pH 7.4, 150 mM NaCl, 1% Nonidet P-40, 0.25% deoxycholic acid, freshly supplemented with 1 mM phenylmethylsulfonic fluoride, 1 mM diethylenetriamine pentaacetic acid, and a protease inhibitor mixture (Sigma)). Then, the dishes were frozen on dry ice for 10 min, followed by a 30-min incubation on ice. Cells were scraped and transferred into microcentrifuge tubes and centrifuged at 16,000 ϫ g at 4°C for 10 min. The supernatant was transferred to fresh microcentrifuge tubes, and protein concentration was determined by the BCA method (Thermo Fisher Scientific). Protein samples were mixed with 2ϫ Novex Tris-glycine SDS sample buffer (Invitrogen) supplemented with 1 mM 2-mercaptoethanol and heated for 10 min at 100°C. The same amount of protein (20 g) was loaded onto each lane, and SDS-PAGE was performed on Novex 4 -12% Tris-glycine gel in Novex Trisglycine SDS running buffer (Invitrogen) at maximal 125 volts or 35 mA for 2 h. Then, proteins were transferred to 0.45-m pore size nitrocellulose membranes (Invitrogen) at 70 volts for 3 h at 4°C using a Bio-Rad Tank Transfer System. Membranes were blocked for 30 min with blocking buffer (Li-Cor, Lincoln, NE), followed by incubation in blocking buffer supplemented with 0.05% Tween 20 and 1:10,000 diluted primary antibodies overnight on a rocking platform at 4°C. The next day, membranes were further incubated with the primary antibody for 1 h at room temperature and then washed with 1ϫ PBS, pH 7.4, twice for 10 min, followed by incubation in blocking buffer (0.05% Tween 20) supplemented with a 1:25,000 dilution of Alexa Fluor 680-conjugated secondary antibody (Invitrogen) for 1 h at room temperature on a rocking platform. Membranes were then washed with PBS for 40 min with a change of buffer every 10 min. Quantitative Western blot images were obtained by scanning the membrane on a Li-Cor Odyssey infrared imaging system according to the manufacturer's instructions. The integrated intensity (kilocounts-mm 2 ) associated with each protein band was independent of the size of the drawn area. The same Western blot result can be obtained with lysate from the cells lysed directly in SDS sample buffer and heated at 95°C.
Estimation of Endogenous IRP2 Protein Concentration-Recombinant IRP2 was kindly provided by Yi He (National Heart, Lung, and Blood Institute). Its mass calculated from sequence is 104,915.5 and the mass measured by electrospray mass spectrometry (14) was 104,915.9 Ϯ 0.2 (S.E. of four analyses). The recombinant IRP2 and bovine serum albumin (Sigma) were used as quantitative standards. They were subjected to SDS gel electrophoresis on the same gel, stained for 30 min with 0.25% Coomassie Blue in 45% methanol and 10% acetic acid, and destained overnight in the same solution without dye (15). The gel was scanned with an Odyssey infrared imaging system, and the amount of recombinant IRP2 was quantified from the standard curve for albumin (15). HEK293A cells were cultured with or without 300 M FAC for 6 h or 100 M deferoxamine for 16 h. Cells were detached from 60-mm Petri dishes in 1.5 ml of citrate saline, and the number of cells was determined by NucleoCounter (New Brunswick Scientific). One ml of cell suspension was centrifuged, and the cell pellet was dissolved by heating in 80 l of 1ϫ SDS-PAGE sample loading buffer (Invitrogen). Twenty l of cell lysate was analyzed on the gel with other lanes containing known amounts of recombinant IRP2 to provide a standard curve. Proteins were transferred to a nitrocellulose membrane followed by quantitative immunodetection. The concentration of endogenous IRP2 was calculated from the cell number and a cell volume of 2.0 picoliters (16,17).
Overexpression of Proteins-Cells were plated on 60-mm dishes 24 h before transfection. At the time of transfection, cells were at 70 -80% confluency. The culture medium was changed to 3 ml of prewarmed medium 1 h before transfection.
An IRP2 vector with an amino-terminal FLAG tag and a carboxyl-terminal HA tag, designed for expression in Escherichia coli, was kindly provided by Yi He (National Heart, Lung, and Blood Institute). A forward primer (5Ј-TCTAGATCTAGAG-ATGGACTATAAAGACGATGAT) and a reverse primer (5Ј-GGGCCCGGGCCCCTAAGCGTAATCTGGAACATC) were employed to clone FLAG-IRP2-HA from the vector. The PCR product was inserted between the XbaI and ApaI sites of the mammalian expression vector pRC-CMV. The fidelity of the resulting plasmid pRC-CMV-FLAG-IRP2-HA was verified by sequencing. Cells were transfected with Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions.
The calpastatin expression plasmid was a generous gift from Dr. David A. Potter (University of Minnesota, Minneapolis, MN) and Dr. Masatoshi Maki (Nagoya University, Nagoya, Japan). For each plate, 2-8 g of plasmid DNA was added to 400 l of 2.5 M CaCl 2 , and the tube was gently tapped to facilitate mixing. The solution was added drop-by-drop to 400 l of HEPES-buffered saline (20 mM HEPES, 140 mM NaCl, 5 mM dextrose, 50 mM KCl, 0.7 mM Na 2 HPO 4 ⅐7H 2 O, pH 7.4) with agitation to facilitate formation of a DNA-calcium phosphate complex. After 5 min without agitation, the transfection solution was added drop-by-drop, and the plates were rocked back and forth to facilitate mixing. Cells were incubated for 6 h in the incubator before changing to fresh growth medium. The cells were further incubated for 24 -48 h for experimental treatments, or the cells were split and allowed to grow until appropriate confluency was reached for experimental use.
Fluorescent Calcium Imaging and Measurement-For calcium imaging, cells were grown on a 60-mm dish until ϳ80% confluent before treatment with a proteasome inhibitor. After the treatment, cells were washed briefly with prewarmed PBS and then incubated in Dulbecco's modified Eagle's medium containing 10 M Fluo-4, 0.02% Pluronic F-127, and 1 mM probenecid (Invitrogen) for 30 min. Calcium indicator loading medium was replaced with fresh, prewarmed medium supplemented with 1 mM probenecid, and the cells were further incubated for 30 min. Fluorescent calcium images were captured on a Zeiss LSM 5 Pascal confocal microscope.
For quantitative calcium measurement, cells were grown on BioCoat TM 96-well collagen I-coated black wall/clear bottom cell culture plates (BD Biosciences) until fully confluent. Cells were then pretreated with 60 M PD150606 for 1 h and then with 1 M epoxomicin for 1, 2, 4, or 6 h. The calcium indicator was loaded using a Fluo-4 NW calcium assay kit (Invitrogen), and fluorescence was measured in well scan mode using a Spec-traMax Gemini EM fluorescent microplate reader (Molecular Devices) with excitation at 494 nm, emission at 516 nm, and a cutoff at 515 nm.
RNA Isolation-Cells were grown on 100-mm plates. After experimental treatment they were washed twice with 1ϫ PBS pH 7.4, total RNA was extracted using an RNeasy Mini Kit (Qiagen), and any DNA contamination was removed by an RNase-free DNase set (Qiagen). RNA concentration was measured at 260 nm on an Agilent 8453 spectrophotometer.
Real-time Quantitative RT-PCR-A primer pair was designed to specifically amplify a 623-base pair section near the 5Ј end of the human IRP2 gene region. A BLAST search revealed less than a 50% match to any other gene. The forward primer was 5Ј-GCGATGGACGCCCCAAAAGCAGGATAC-GCC-3Ј, and the reverse primer was 5Ј-GGTTGCAAATGAA-AAGGACACAGGATGGGTG-3Ј. The human heat shock 70-kDa protein 6 (Hsp70B) forward primer was 5Ј-GTGA-AAGCCACTGCTGGAGATACCCACC-3Ј, and the reverse primer was 5Ј-ACGACGTCATGAATCTGGGCCTTGTCC-AGC-3Ј. These primers were synthesized by Bio-Synthesis (Lewisville, TX). QuantumRNA TM universal 18 S and ␤-actin internal standards were from Ambion. RT-PCR using this IRP2 primer pair and increasing amounts of total RNA isolated from 293A cells dose-dependently generated a single band of the predicted size on an ethidium bromide-stained agarose gel, demonstrating the specificity of the primer. Real-time quantitative RT-PCR was carried out on a 96-well PCR plate using Brilliant II SYBR Green QRT-PCR Master Mix Kit (Stratagene) according to the manufacturer's instructions. Triplicate measurements were performed for each experimental group. A mixture of 2ϫ SYBR Green QRT-PCR master mix, 200 nM each of upstream and downstream primers, diluted reference dye, RNase inhibitor, and reverse transcriptase was prepared, and 12.5 l of this mixture was pipetted into each well. 400 ng of RNA from each experimental group, and nuclease-free PCR grade water was added to give a final volume of 25 l. QRT-PCR was performed with a Stratagene Mx3000P QRT-PCR system (Stratagene) using comparative quantitation (calibrator mode). The PCR thermal cycling profile was as follows: step 1, 30 min at 50°C for reverse transcription; step 2, 10 min at 95°C to activate taq DNA polymerase and denature DNA; step 3, 40 cycles, 30 s at 95°C denaturation followed by 1 min 60°C annealing, and 1 min at 72°C elongation. Fluorescent data were collected at the end of each cycle.
Target Preparation and Hybridization to GeneChips-T7based RNA amplification was carried out on 1 g of the isolated total RNA as suggested by the manufacturer (Affymetrix). Total RNA was incubated with oligo(dT/T7) primers and reversetranscribed into double-stranded cDNA. In vitro transcription and biotin labeling of the purified cDNA was performed using T7 RNA polymerase at 37°C for 16 h using the Affymetrix in vitro transcription labeling kit. The yield of biotin labeled cRNA was determined spectrophotometrically with a Nanodrop ND-1000 spectrophotometer and integrity with the Agilent 2100 bioanalyzer. 20 g of biotin-labeled RNA was fragmented to an ϳ200-bp size by incubating in buffer containing 200 mM Tris acetate, pH 8.2, 500 mM potassium acetate, and 500 mM magnesium acetate for 35 min at 94°C prior to hybridization. Fragmented RNA was assessed for fragment size on the bioanalyzer and then hybridized to Affymetrix U133 plus 2.0 chips for 16 h, washed, and stained on an Affymetrix fluidics station. Microarray Data Processing and Analysis-Affymetrix GCOS (version 1.4) was used to calculate the signal intensity and the percent present calls on the hybridized Affymetrix chip. The signal intensity values obtained for probe sets in the microarrays were transformed using an adaptive variance-stabilizing, quantile-normalizing transformation. Transformed data from all of the chips were subjected to principal component analysis to detect outliers. To correct for multiple comparisons, fold-cutoff filters and false discovery rate analysis filters were applied. Two-way hierarchical clustering was used to bring together sets of samples and genes with similar expression patterns. The hierarchical clustering was performed with the JMP5.1 statistical software package (SAS Institute, Cary, NC) using the ward method. Pathway analysis was performed using Ingenuity Pathway Analysis (Ingenuity Systems, Inc., Redwood City, CA).

RESULTS
Quantitation of IRP2-To study the regulation of IRP2 protein levels in untransfected cells, we established a robust, sensitive, and quantitative assay. We screened available polyclonal and monoclonal antibodies and found that the mouse monoclonal 7H6 provided the best signal:noise ratio. Quantitation was accomplished by use of a fluorescent tagged secondary antibody, which was excited by an infrared laser, minimizing interference from autofluorescence of cellular components. Fluorescence was detected and quantitated with an infrared scanner. Fig. 1 shows the determination of endogenous IRP2 in HEK293A cells, confirming that we can measure IRP2 levels without transfection and without inducing iron deficiency. That the monoclonal antibody is specifically recognizing IRP2 is demonstrated by (1) the response of the detected band to iron status and (2) co-migration with purified, full-length recombinant IRP2 (Fig. 1A). The concentration of IRP2 is ϳ3.2 nM corresponding to ϳ3,800 molecules per cell (Fig. 1B).
Effect of Proteasome Inhibition on IRP2 Levels-A common protocol employed in the study of the regulation of IRP2 is to induce iron deficiency by an overnight treatment with the iron chelator deferoxamine (DFO). In HEK293A cells, DFO treatment increases IRP2 levels by ϳ7-10ϫ, up to ϳ30 nM (Fig. 1B).
As expected, addition of FAC relieves the iron deficiency and causes a dramatic drop in IRP2 levels ( Fig. 2A). This drop is blunted by proteasome inhibitors such as epoxomicin, lactacystin, and MG-132. In DFO-pretreated cells that were not exposed to FAC, epoxomicin had no effect ( Fig. 2A).
In the absence of DFO, a steady-state level of IRP2 is maintained by constitutive synthesis of IRP2 balanced by continuous degradation. Modulation of the rate of degradation allows for rapid changes in IRP2 levels coincident with alterations in iron availability (18). Not surprisingly, addition of FAC decreased this steady state level by about half in 6 h (Figs. 1B and 2B). What was surprising was the finding that epoxomicin could not block this decrease (Fig. 2B, lane 3). Moreover, epoxomicin itself caused a decrease in IRP2 level without FAC treatment (Fig. 2B, lane 4). The effect was not a peculiarity of epoxomicin as two other proteasome inhibitors, lactacystin and MG-132, had the same effect (Fig. 2C). Transferrin is the physiological carrier of iron, with uptake via receptor-mediated endocytosis. Addition of iron-loaded transferrin (holotransferrin) caused a decrease in IRP2 levels, which also was not blocked by epoxomicin (Fig. 2D). Thus, the lack of effect of proteasome inhibitors was not due to the addition of ionic iron as FAC, nor was it due to failure to effectively inhibit the proteasome, as addition of epoxomicin caused marked accumulation of ubiquitinylated proteins (Fig. 3).
Without DFO pretreatment, exposure of HEK293A cells to epoxomicin reproducibly caused a decrease in IRP2 levels of ϳ50% at 6 h as shown in Fig. 2E. In addition, we noted that the levels of IRP2 increased by 10 -20% at 1-2 h, with the marked decrease occurring after 4 h, also shown in Fig. 2E. The initial small increase followed by a decrease is consistent with two pathways for degradation of IRP2, one via the proteasome and the other via a nonproteasomal pathway. The former appears to dominate in DFO-treated cells, whereas the latter predominates in cells not treated with DFO.
DFO is trapped in the lysosome after uptake by endocytosis (19,20). Lysosomal localization is to be expected because DFO has four amines with pK a values ranging from 8.3 to 10.8 (21). It is thus a lysosomotropic amine that becomes protonated and then trapped in the lysosome. Such lysosomotropic amines are well known to alkalinize endosomes and lysosomes, thereby interfering with their functions, including receptor-mediated endocytosis and protease activity. Ammonium chloride and chloroquine are well known lysosomotropic amines, which cause intracellular iron deficiency by inhibiting the release of iron from transferrin after receptor-mediated endocytosis of  holotransferrin (22)(23)(24). Chloroquine is not an iron chelator so that comparing its effects on IRP2 metabolism with those of DFO would allow assessment of whether the iron chelating property of DFO is required to observe proteasomal degradation of IRP2. Overnight treatment with chloroquine markedly increased IRP2, which was readily degraded by the proteasome upon addition of FAC; epoxomicin alone did not cause degradation of IRP2 in chloroquine-treated cells (Fig. 2F). Because the effects of chloroquine on IRP2 are the same as those observed with DFO, we conclude that chelation of iron is not required for degradation of IRP2 by the proteasomal pathway. Alkalinization of the endosome/lysosome suffices to greatly increase IRP2 levels and direct IRP2 degradation to the proteasome, as also observed by Dycke and colleagues (12). They suggested that the nonproteasomal degradation was mediated by a lysosomal pathway, but we do not think a lysosomal protease can be implicated by that observation because of the confounding effect caused by the simultaneous inhibition of receptormediated endocytosis of transferrin.
Mechanism of Change in IRP2 Levels-Previous investigators established that the primary mechanism of regulation of IRP2 levels is via modulation of the rate of its degradation and not of its mRNA level (6,8). However, given our unexpected finding that treatment with a proteasome inhibitor decreased the level of IRP2 in nontransfected cells (Fig. 2B), we measured the IRP2 mRNA levels by quantitative RT-PCR to determine whether epoxomicin treatment induced a change. Fig. 4A shows that it does not. Hsp70B mRNA levels are known to increase markedly upon epoxomicin treatment, so we also measured those levels as a positive control for epoxomicin action (Fig. 4B).
Because the mRNA levels of IRP2 were unchanged, we then determined whether the rate of IRP2 protein degradation was  altered. Treatment of cells with cycloheximide blocks protein synthesis, allowing the rate of protein degradation to be measured. To validate the experimental approach, we also quantitated levels of ERK3, a protein known to be degraded by the proteasome (25). Treatment of HEK293A cells with cycloheximide allowed the observation of ERK3 degradation, which was blocked by epoxomicin (Fig. 5A). Thus, both cycloheximide and epoxomicin were pharmacologically effective. Treatment of the cells with cycloheximide also caused a decrease in IRP2 levels, but addition of epoxomicin increased the rate of loss of IRP2. A first order plot of IRP2 levels (Fig. 5B) allows determination of the half-life of IRP2. In the absence of epoxomicin, the half-life was 3.0 h. With a 3-h pretreatment of epoxomicin, the half-life decreased to 1.7 h. We conclude that epoxomicin lowers cellular levels of IRP2 by increasing its rate of degradation.
Characterization of Nonproteasomal Pathway of IRP2 Degradation-We noted above that DFO and chloroquine both alkalinize the lysosome and thus inhibit lysosomal proteases and induce iron deficiency by blocking unloading of iron from holotransferrin. Either effect could cause the observed increase in IRP2 level. We considered the possibility that lysosomal dysfunction was not required to observe proteasomal degradation and that an increased IRP2 level alone would suffice. We therefore transfected HEK293A cells with a full-length construct of IRP2 carrying an epitope tag so that we could compare the behavior of endogenous and transfected IRP2. We used increasing amounts of vector DNA to assure that the increase in IRP2 would encompass levels achieved by DFO, that is, 5-10fold. Fig. 6 shows that increasing the level of cellular IRP2 by transfection also eliminates the epoxomicin stimulated degradation of IRP2. Epitope-tagged IRP2 exhibited the same degradation behavior, indicating that the presence of an epitope tag did not perturb the characteristics of degradation. This observation is consistent with the non-proteasomal pathway having a high affinity for IRP2 but relatively low capacity. Only the activity of the proteasomal pathway is observed upon epoxomicin exposure at higher levels of IRP2 when the nonproteasomal system becomes saturated.
We performed DNA microarray analysis in an attempt to detect mRNA of protease(s) that were up-regulated via proteasome inhibition by epoxomicin. We used the Affymetrix U133 human genome array, which interrogates 33,000 human genes, but no up-regulated protease genes were detected. The microarray data, which may also be of interest to those studying the effects of proteasome inhibition, have been deposited in the GEO database (National Center for Biotechnology Information) under accession no. GSE14429. We then tested chemical inhibitors of various classes of proteases for their ability to  block the epoxomicin-triggered degradation of IRP2. Table 1 lists those that had no effect on IRP2 degradation. Calcium Is Required for Epoxomicin-induced Nonproteasomal Degradation of IRP2-The one compound that did blunt the epoxomicin-triggered degradation was PD150606 (3-(4-iodophenyl)-2-mercapto-(Z)-2-propenoic acid), a calpain inhibitor (Fig. 7A). Without exposure to epoxomicin, treatment with PD150606 increased IRP2 levels above those in untreated cells, without altering the mRNA level of IRP2 (Fig. 7B). This observation suggests that the non-proteasomal pathway of IRP2 degradation operates normally even in cells not perturbed by epoxomicin. However, ϳ60 M PD150606 was required to see a clear effect (Fig. 7A), whereas effective inhibition of calpain within cells has been observed at 10 M PD150606 (27). Moreover, other calpain inhibitors did not have the ability to block the degradation of IRP2 triggered by epoxomicin (Table 1).
To reach a more confident assessment of whether calpain was the intracellular protease responsible for epoxomicin-triggered degradation of IPR2, we expressed calpastatin, the specific intracellular inhibitor of calpain. Fodrin ␣ is known to be degraded by calpain, so the appearance of its 150-kDa cleavage product was also followed as a positive control for calpastatin activity. Fig. 7C demonstrates that epoxomicin causes the degradation of both fodrin ␣ and IRP2, and calpastatin protects fodrin ␣ but not IRP2.
Having established that PD150606 inhibited epoxomicintriggered degradation of IRP2 by a mechanism other than inhibition of calpain, we considered the possibility that epoxomicin induced a change in intracellular calcium, which was required for IRP2 degradation, and, consistent with published studies (28), PD150606 was able to inhibit the change in calcium. Imaging of HEK293A cells with Fluo-4 AM demonstrated that a 6-h exposure to 1 M epoxomicin did increase the intracellular calcium concentration (Fig. 8A). The time course of calcium increase and the inhibition by PD150606 are shown in Fig. 8B.
Thapsigargin, an inhibitor of the sarcoplasmic/endoplasmic reticulum Ca 2ϩ -ATPase, raises intracellular calcium concentration. Exposure of HEK239A cells to 1 M thapsigargin caused an even more rapid degradation of IRP2 than epoxomicin (compare Figs. 2E and 8C). This thapsigargin-induced degradation was not mediated by the proteasome (Fig. 8D), and PD150606 inhibited the effect (Fig. 8E). As with epoxomicin, the inhibition required 40 -60 M PD150606. A double-tagged IRP2 (FLAG-IRP2-HA) whose expression was carefully controlled to match the endogenous level was also subject to thapsigargin-induced degradation (Fig. 8F). The rate of degradation of endogenous and transfected IRP2 were comparable. This result also provides additional confirmation that the anti-IRP2 monoclonal antibody recognized endogenous IRP2. Thapsigargin did not alter the IRP2 mRNA level (Fig. 8G).
Calmodulin is required to mediate many calcium-dependent processes (29). However, treatment with the calmodulin inhibitor W-7 did not blunt the epoxomicin-triggered degradation of IRP2 (Fig. 9). We conclude that the epoxomicin-induced degradation of IRP2 requires an increase in intracellular calcium, but calmodulin may not mediate the process.

DISCUSSION
Utilizing a monoclonal primary antibody and a secondary antibody labeled with a chromophore that fluoresces in the near-infrared, we can readily quantitate physiological levels of IRP2. This methodology allowed characterization of the degradation of IRP2 in HEK293A cells without elevation of IRP2 levels by transfection or induction of iron deficiency. This quantitative, sensitive method demonstrated that the steadystate turnover of IRP2 in iron-replete cells is maintained by a nonproteasomal degradation system. Cells and tissues may vary in utilization of the two pathways (6), as we did not detect the nonproteasomal pathway in two other cell lines that we tested, HeLa and COS-7. We also observed that the activity of the nonproteasomal system varied among lineages of the HEK293 cell line. Although HEK293A cells utilized the nonproteasomal system under physiological conditions, Guo et al. demonstrated that a proteasome inhibitor blocked iron-induced IRP2 degradation in the rat FTO2B hepatoma cell line even without DFO pretreatment.
The proteasomal degradation pathway is elegantly regulated by an iron-sensing E3 ubiquitin ligase (10,11). The pathway does not require the 73-amino acid "intervening domain" that is present in IRP2 but not IRP1 (30 -32). We have not yet characterized the regulation of the nonproteasomal pathway, other than demonstrating that calcium is required to observe its action. Characterization will be facilitated when the protease is identified. Neither a battery of protease inhibitors nor microarray analysis identified candidate proteases, suggesting that it may have novel properties. The behavior of the nonproteasomal pathway toward IRP2 is that of a high affinity, low capacity system whose action can no longer be observed when IRP2 increases 5-10-fold above basal level. The proteasomal system comes into action when the cell requires degradation of elevated levels of IRP2, such as upon relief of chelator-induced iron deficiency.
We found that proteasome inhibitors increase cellular calcium levels, an effect that has not been studied in detail previously. In undifferentiated PC12 cells, MG-132 caused an increase of cytosolic calcium and cell death (33). In myeloma cells, another proteasome inhibitor, bortezomib, caused an increase of cytosolic calcium, mitochondrial calcium loading, capacitative calcium influx from the extracellular space, activation of caspases, and cell death (26). Although calcium is required for the non-proteasomal pathway, our experiments with calpain inhibitors, especially the specific calpastatin, establish that calpain is not the nonproteasomal protease.
Fifteen years passed between when the proteasome was shown capable of degrading IRP2 and the enabling E3 ligase was identified. Hopefully, the interval from recognizing the nonproteasomal pathway to identification of its protease will be much shorter.