NF-κB and IκBα Are Found in the Mitochondria

The transcription factor NF-κB has been shown to be predominantly cytoplasmically localized in the absence of an inductive signal. Stimulation of cells with inflammatory cytokines such as tumor necrosis factor α or interleukin-1 induces the degradation of IκB, the inhibitor of NF-κB, allowing nuclear accumulation of NF-κB and regulation of specific gene expression. The degradation of IκB is controlled initially by phosphorylation induced by the IκB kinase, which leads to ubiquitination and subsequent proteolysis of the inhibitor by the proteasome. We report here that NF-κB and IκBα (but not IκBβ) are also localized in the mitochondria. Stimulation of cells with tumor necrosis factor α leads to the phosphorylation of mitochondrial IκBα and its subsequent degradation by a nonproteasome-dependent pathway. Interestingly, expression of the mitochondrially encoded cytochromec oxidase III and cytochrome bmRNAs were reduced by cytokine treatment of cells. Inhibition of activation of mitochondrial NF-κB by expression of the superrepressor form of IκBα inhibited the loss of expression of both cytochromec oxidase III and cytochrome b mRNA. These data indicate that the NF-κB regulatory pathway exists in mitochondria and that NF-κB can negatively regulate mitochondrial mRNA expression.

The transcription factor NF-B has been studied extensively due to its interesting regulation and to the range of biological processes that it controls. Five members of the immediate NF-B/Rel family have been identified: p50/NF-B1, p65/RelA, p52/NF-B2, RelB, and c-Rel. The classic form of NF-B is the heterodimer of the p50 and p65 subunits (reviewed in Ref. 1). In most cells, NF-B is complexed with members of the IB family, IB␣, IB␤, and IB⑀, which typically function to inhibit the action of this group of transcription factors. Stimulation of cells with stimuli such as TNF␣ 1 leads to activation of the IB kinase (IKK), which phosphorylates IB␣ or IB␤ on N-terminal serines (1,2). Phosphorylated IB then is targeted for ubiquitination and subsequent degradation by the proteasome, allowing NF-B to accumulate in the nucleus (3). In the nucleus, NF-B is a positive regulator of gene expression through its ability to bind to target sequences in the regulatory regions of genes encoding cytokines, cytokine receptors, antiapoptotic proteins, and cell cycle regulators (1,4). Additionally, NF-B has been shown to negatively regulate MyoD mRNA through a post-transcriptional mechanism that appears to require the transcriptional activity of NF-B (5). Presumably through its ability to regulate gene expression, NF-B dysregulation contributes to a variety of diseases, including oncogenesis, arthritis, and cancer cachexia (4,6).
Recently, more complex aspects of NF-B regulation have been proposed. For example, studies indicate that NF-B shuttles into and out of the nucleus in unstimulated cells (7)(8)(9). Thus, it was found that leptomycin B, an inhibitor of nuclear export, leads to the accumulation of NF-B and IB in the nucleus without an external stimulus. Whether NF-B and IB shuttle in a complex has been questioned (7). Additionally, it has been proposed that IB can be degraded by a nonproteasomal mechanism following stimulation of cells with cytokines (10). In that study, the use of inhibitors with distinct specificities indicated that phosphorylated IB␣ can be degraded by calpain in addition to the more characterized proteasome pathway.
The vertebrate mitochondrial genome is circular with ϳ17 kb of DNA (11) (for a review, see Ref. 12). The mitochondrial genome encodes two ribosomal RNAs, 22 tRNAs, and at least 13 peptides, which contribute to complex I, complex III, complex IV, and complex V of the electron transport system (11,12). A regulatory region associated with the origin of replication also serves as a promoter region for two large mitochondrial RNA transcripts that are processed into individual RNAs for the structural RNAs and mRNAs (see Ref. 13). For example, the mRNAs encoding cytochrome c oxidase I, II, and III (CoxI, -II, and -III, respectively) are derived from a common precursor RNA (12).
We report here that the NF-B subunits p50 and p65 along with IB␣, but not IB␤, are found in the mitochondria as well as in the cytoplasm of proliferating cells. Quiescent liver cells exhibit largely p50 and IB␣ in the mitochondria with little detectable p65. Electron microscopy and biochemical approaches confirm the localization of these proteins to mitochondria. Recently, others have found IB␣ and p65 in the mitochondria of Jurkat T cells (14). Interestingly, our data indicate that mitochondrial IB␣ is phosphorylated on N-terminal serines in response to cellular TNF␣ stimulation, followed by nonproteasomal degradation. TNF␣ treatment leads to a significant reduction in CoxIII and cytochrome b mRNA. Inhibition of NF-B by mitochondrially localized superrepressor IB␣ blocked the loss of both CoxIII and Cyt B mRNA. These data show that NF-B is found in mitochondria, where it regulates mitochondrial specific gene expression.

Cell Culture and Treatment with Cytokines and Protease Inhibitors-
Cell lines were obtained from the Lineberger Comprehensive Cancer Center Tissue Culture Core Facility at the University of North Carolina (Chapel Hill, NC). U937 cells were grown in Dulbecco's modified Eagle's medium H supplemented with 10% fetal bovine serum and antibiotics. HT1080 V and HT1080 I lines were previously described (15). Cells were treated with 10 ng/ml human recombinant TNF␣ diluted in phosphate-buffered saline. Calpeptin, lactacystin, and MG132 (Calbiochem) were all resuspended in Me 2 SO and used at a final concentration of 40 ng/ml.
Electron Microscopy and Immunocytochemical Labeling-U937 cells were suspended in serum-free medium and were mixed with a fixative containing 0.5% glutaraldehyde in 0.1 M sodium cacodylate buffer, pH 7.4. The cells were cooled to 10°C on ice, followed by microwave irradiation using a 750-watt microwave oven until of a final temperature of 40°C was obtained (16). The fixed cells were pelleted, rinsed in 0.1 M sodium cacodylate buffer, and postfixed with 1% osmium tetroxide in 0.1 M sodium cacodylate buffer for 30 min. Following postfixation, the cell pellet was dehydrated through a graded series of ethanol washes and embedded in L. R. White resin (available from Ted Pella, Inc., Redding, CA). 70-nm ultrathin sections were taken of the embedded pellet, mounted on nickel grids, and stained using indirect immunocytochemical methods.
Immunocytochemical labeling, incubations, and washing steps were carried out at ambient temperature. The incubation and wash buffer for the primary antibody consisted of 0.05 M Tris-buffered saline with 0.1% fish skin gelatin and 0.01% Tween 20 (TBS/FGT) at pH 7.6. Gridmounted sections were blocked with 0.2 M glycine in Hanks' balanced salts for 10 min, followed by a secondary blocking step with 5% goat serum in TBS/FGT for 10 min. The grids were incubated for 1 h with primary antibody (p50/NF-B1; Upstate Biotechnology, Inc., Lake Placid, NY). After rinsing with TBS/FGT, the grids were incubated for 1 h in secondary antibody (goat anti-rabbit IgG 10 nm colloidal gold (BBI International, Ted Pella, Redding, CA), diluted 1:50 in 0.1 M TBS/FGT, pH 8.2. Negative controls were performed concurrently, consisting of incubation in normal rabbit IgG at the same repetitive dilutions and conditions. The immunogold-stained sections were poststained with uranyl acetate followed by lead citrate, and the grids were observed and photographed using a LEO EM-910 transmission electron microscope (LEO Electron Microscopy, Inc., Thornwood, NY) at 80 kV.
Preparation of Mitochondria-Mitochondria were isolated from cell lines as described previously (17). Digitonin was added to the mitochondrial buffer A at 0.05% to aid in the disruption of the cells. Further purification was performed by ultracentrifugation at 30,000 rpm in gradient buffer (250 mM mannitol, 1 mM EGTA, 25 mM Hepes, 0.1% bovine serum albumin), pH 7.4, supplemented with 30% Percol (Sigma). The uppermost band was removed and washed two or three times with mitochondrial buffer A. Mitochondria that were analyzed by Western blot analysis were boiled in SDS sample buffer and loaded directly onto SDS-polyacrylamide gels. When electrophoretic mobility shift assays (EMSAs) were performed, purified mitochondria were incubated in nuclear extract buffer for 10 min in ice. The sample was then centrifuged for 10 min at 12,000 rpm at 4°C. The supernatants were used directly in the EMSA reaction (see below). Isolation of mitochondria from rat liver was as previously described (18).
EMSAs-EMSAs were performed as described previously (20). An oligonucleotide corresponding to an NF-B site in the H-2K b gene was radiolabeled using [␣-32 P]dCTP and the Klenow fragment of DNA polymerase I (Roche Molecular Biochemicals). For antibody supershift analysis, extracts were preincubated 15 min at room temperature with 1 g of antiserum before the addition of the radiolabeled gel shift probe. Antibodies used in supershift analysis were identical to those utilized for Western blotting analysis.
IB Kinase Assay-Cells were either treated or not treated with TNF␣ for specified times and harvested, and mitochondrial and cytoplasmic extracts were isolated in phosphate-buffered saline containing phosphatase inhibitors. Equal amounts of protein (500 g) were immunoprecipitated using antibodies against the ␤ subunit of IKK (a gift of Dr. F. Mercurio, Signal Pharmaceuticals, San Diego, CA). Kinase activity was determined by incubating the immunoprecipitates with 4 g of glutathione S-transferase-IB␣ (amino acids 1-54) wild type substrate or a mutated form of IB␣ (S32T,S36T) in the presence of [␥-32 P]ATP, as described (21). The immunoprecipitates were subjected to SDS-PAGE, dried, and visualized by autoradiography.
Northern Blot Analysis-RNAs were isolated using Trizol as recommended by the manufacturer (Invitrogen). Northern blot analysis utilized 10 -20 g of total cellular RNA separated on 1.5% formaldehydeagarose gels according to standard procedures. RNA samples were transferred overnight to nylon filter, UV cross-linked (Stratagene, La Jolla, CA), and probed with randomly labeled probe (Amersham Biosciences) corresponding to cytochrome c oxidase II or III or cytochrome B. Hybridization and wash conditions were obtained using Expresshyb (Stratagene) solution as described by the manufacturer. Blots were normalized for equal loading using a cDNA probe for glyceraldehyde-3-phosphate dehydrogenase.

Mitochondrial Localization of NF-B-Studies
were initiated in order to analyze the subcellular localization of NF-B and IB subunits. Electron microscopy of sections of U937 cells in association with immunocytochemistry indicated the presence of both the p50/NF-B1 and p65 subunits localized in the inner matrix of the mitochondria as well as in the cytoplasm (see Fig. 1). The control (see "Materials and Methods") showed no detectable labeling. Additionally, IB␣ was detectable in the inner matrix of mitochondria. 2 After observing the presence of the NF-B p50 and p65 subunits in the mitochondria using electron microscopy, we were interested in confirming the results biochemically. Mitochondria were first isolated from rat liver and subjected to increasing concentrations of digitonin, which functions to elute proteins in a manner dependent on the localization within the mitochondria. Supernatants and mitochondrial pellets at each concentration were analyzed by Western blotting using antibodies to IB␣, p50, or p65. The results show that both IB␣ and p50 are present in the rat mitochondria (Fig. 2). With the addition of low concentrations of digitonin, IB␣ protein was not detected in the supernatant. The addition of 0.4% digitonin (lane 5) leads to release of IB␣ from mitochondria so that it is detected in the supernatant. However, the p50 protein was not visible in the supernatant until 0.5-0.6% digitonin was added (lanes 6 and 7). When these results are plotted against known mitochondrial components, release of p50 from the mitochondria following digitonin treatment corresponds to the release of fumarase, 2 a marker for the inner matrix of the mitochondria, whereas IB␣ release from the mitochondria occurs earlier, indicating that IB␣ localization is closer to the mitochondrial surface and presumably also in the inner matrix. p65 was not detected in the mitochondria of quiescent rat liver (data not shown, but see below regarding cells in culture).
TNF␣ Treatment of U937 Cells Causes a Loss of IB␣ in the Mitochondria-To further analyze the localization of IB␣ and NF-B in the mitochondria and to determine potential responses to cytokine treatment, U937 cells were studied by Western blotting of mitochondrial and cytoplasmic extracts (Fig. 3). Cells were stimulated with TNF␣ over a 1-h time course. After separation of cytoplasmic and mitochondrial components, the mitochondrial pellet was resuspended in a volume of buffer equal to one-tenth volume of the cytosolic fraction, consistent with published estimates that mitochondria represent one-tenth the total volume of a human cultured cell (22). Protein assays confirmed that concentrations of the two fractions were approximately equivalent. Assuming that overall protein concentrations are similar in the mitochondria and cytoplasm of living cells, we loaded equal amounts of protein from each fraction in order to observe physiologically comparable levels of each subunit. First, the data support the EM data indicating that p50 and IB␣ were localized in the mito-chondria. Second, the data indicate that the p65 subunit of NF-B is located in the mitochondria of U937 cells. Additionally, the data indicate that the mitochondrial p65 exhibits slightly faster migration in SDS gels (e.g. compare the mobility of the mitochondrial p65 with cytoplasmic p65) (Fig. 3), possibly due to a protein processing event associated with mitochondrial import (see Ref. 12). The results also show that following TNF␣ treatment, there is degradation of IB␣ by 30 min and resynthesis at 60 min in both the cytoplasm and mitochondria of U937 cells (Fig. 3). The overall kinetics of loss of mitochondrial IB␣ is slightly different from what we observe in the cytoplasm. It is noted that a slower migrating form of mitochondrial IB␣ is detected following TNF␣ treatment, suggestive of induced phosphorylation (see below). There appears to be very little change in p65 in either the mitochondria or cytoplasm of these cells following TNF␣ treatment; however, p50 shows a reduction in the mitochondria following TNF␣ treatment of cells. To confirm the absence of mitochondrial contamination of the cytoplasmic fractions, the Western blot was probed with a cytochrome c antibody. Other experiments show that there is little or no cytoplasmic contamination of mitochondria and that IB␤ is not found in the mitochondria (see Fig. 5, B and C). Additionally, c-Rel was not detected at appreciable levels in the mitochondria of U937 cells. Identical results were found when using the HT1080 cell line, showing that the localization is not cell type-specific.
TNF␣ Stimulation Results in Increased p50 DNA Binding Activity in the Mitochondria-To determine whether the NF-B components in the mitochondria were able to bind DNA, mitochondrial extracts were prepared, and mitochondrial proteins were eluted in nuclear extract buffer. The results from an EMSA show two binding complexes (complexes I and II). The NF-B DNA binding activity is low in the mitochondria in untreated cells and increases with TNF␣ treatment, with the strongest binding occurring at the 1-h time point (Fig. 4). To identify which components of NF-B contribute to this binding activity, supershift analysis was performed on the mitochondrial extract that was treated with TNF␣ for 1 h. The use of a p50-specific antibody led to a supershifted band from complex II (Fig. 4). However, when a p65-specific antibody was added to the binding reaction, there was a reduction in the DNA binding in complex II without the appearance of a supershifted band. Complex I could not be supershifted with either antibody. These results indicate that p50 is the major NF-B subunit found in binding complex II, possibly existing as a homodimer. It remains possible that additional NF-B components may be part of the DNA binding activity. The data also suggest that the presence of p50 and p65 in the mitochondria is presumably not as a heterodimer, possibly suggesting distinct mitochondrial sublocalization for these two NF-B subunits.  3. Changes in p50 and IB␣ in the mitochondria of U937 cells upon TNF␣ stimulation. U937 cells were treated with 10 ng/ml TNF␣ over a 1-h time course, and mitochondria (M) and cytoplasmic (C) fractions were prepared and analyzed by Western blot analysis. Proteins were fractioned on SDS-polyacrylamide gels, transferred to nitrocellulose, and incubated with antibodies specific for p65, p50, IB␣, and cytochrome c.

Phosphorylation of Mitochondrial IB␣ in Response to TNF␣ Occurs on N-terminal Serines, and Degradation of IB␣ in the Mitochondria following TNF␣ Stimulation Is Proteasome-independent-
The fact that TNF␣ induced the degradation of IB␣ in the mitochondria led us to explore whether IB␣ is phosphorylated and degraded in a manner similar to that of cytoplasmic IB␣. First, we asked whether IB␣ loss was dependent on a proteasome-mediated mechanism. Cells were pretreated with proteasome and calpain inhibitors and either left untreated or stimulated with TNF␣. Mitochondria were purified, and proteins were eluted and analyzed by Western blotting. The results (Fig. 5A) show that upon treatment of cells with TNF␣ for 15 min, there is a significant loss of IB␣ in both the mitochondrial and the cytoplasmic compartments. This loss in the cytoplasm can be blocked following pretreatment with either lactacystin, a proteasome-specific inhibitor, or with MG132, an inhibitor of both calpain and the proteasome (10) (Fig. 5A). When cells were pretreated with calpeptin, an inhibitor specific for calpain (10), there was a modest inhibition of the degradation of cytoplasmic IB␣ but considerably less inhibition than with the other inhibitors. However, in the mitochondrial fraction, lactacystin had very little effect in blocking the degradation of IB␣. Additionally, cells pretreated with calpeptin showed more inhibition of mitochondrial IB␣ degradation than lactacystin, unlike the cytoplasm, which showed the opposite effect. However, it is noted that calpeptin was able to inhibit the degradation of IB␣ in the mitochondria only partially.
When an antibody specific for the phosphorylated form of IB␣ (anti-phosphoserine 32) was used to probe the same blot, phosphorylated IB␣ was visualized when cells were pretreated with all of the inhibitors in the cytoplasm but only in cells pretreated with calpeptin or MG132 in the mitochondria (Fig. 5A). Again, these results point to a lack of proteasome effect in the mitochondria. It is not clear whether the degradation of IB␣ in the mitochondria is a calpain-specific pathway, since calpeptin does not completely block the degradation. This may be a function of the inhibitor, or there may be other factors involved in this process. Additionally, we cannot rule out that the lack of effect seen in the mitochondria when the cells are pretreated with lactacystin was due to the lack of permeability of the mitochondria to lactacystin.
To address whether the proteasome may not be relevant in mitochondrial IB␣ degradation, cytoplasmic and mitochondrial extracts were prepared from U937 cells, and Western blotting was performed using antibodies specific for Lmp-2, a functional subunit of the proteasome (23) and Skp-1, a component of the ubiquitin ligase associated with IB␣ ubiquitina-tion (3). The results show that both proteins can only be found in the cytoplasm of U937 cells (Fig. 5B). Blots were also probed for cytochrome c as a mitochondrial marker. These results further support the possibility that the mitochondria lacks proteasome function. They also confirm that there is no cytoplasmic contamination of the mitochondria preparations.
NF-B is regulated by multiple forms of IB, including IB␣ and IB␤ (1). Mitochondria blots from U937 cells were probed with IB␣ and IB␤ antibodies (Fig. 5C). The results indicate that whereas IB␣ is found in both the mitochondria and the cytoplasm, IB␤ is only found in the cytoplasm. Again, cytochrome c was used as a mitochondrial marker.
Evidence for Mitochondrial IKK Activity-In order to understand the mechanism of the induced phosphorylation of IB␣ in the mitochondria in response to TNF␣ treatment, we explored whether IKK family members were present in or potentially associated with the mitochondria. Western blots of fractions from U937 cells enriched for mitochondria were probed with IKK␣-, IKK␤-, and IKK␥-specific antibodies. The results show that all three IKK family members are present in the mitochondrial fraction (Fig. 6A). Blots were also probed with a cytochrome c antibody as a marker for mitochondria. Based on the phosphorylation on IB␣ in the mitochondria and the presence of the IKK family members, we wanted to determine whether functional kinase activity could be detected in the mitochondria fraction. Large scale mitochondria and cytoplasmic extracts were prepared from U937 cells either left untreated or treated with TNF␣. The results show that both in the cytoplasm and mitochondrial extracts, low levels of IKK kinase activity were seen in untreated cells (Fig. 6B). Following stimulation with TNF␣, the IB kinase activity in both the cytoplasm and mitochondria was greatly increased. These results indicate that the IKK family members are associated with the mitochondrial fraction and are potentially responsible for the phosphorylation of IB␣ seen there.
NF-B Negatively Regulates Expression of CoxIII and Cytochrome b mRNAs-In order to address a possible function of NF-B components in the regulation of mitochondrial gene expression, we utilized HT1080 cells either containing an empty vector or expressing the modified form of IB␣ known as IB␣SR (15). Interestingly, IB␣SR was shown to be in the mitochondria as well as the cytoplasm through its ability to be recognized by a FLAG tag antibody. 2 Importantly, the superrepressor form of IB␣ was shown to be resistant to degradation in mitochondria of HT1080 cells in response to TNF␣ stimulation. HT1080 cells stably transfected with either a vector control or the IB␣SR were treated with TNF␣, and RNA was isolated at several time points. The results show that HT1080 vector control cells exhibit a loss of CoxIII mRNA beginning at 1 h following TNF␣ treatment and continuing through 2 h (Fig. 7). The mRNA returns to near normal level following 4 h of TNF␣ stimulation. However, when HT1080 cells containing the IB␣SR were treated with TNF␣, only a modest change in the CoxIII mRNA was observed. Since cytochrome c oxidase III plays an integral role in complex IV of the electron transport machinery (12), it is expected that the loss of CoxIII would impact changes in ATP and reactive oxygen species. Efforts to address such changes in response to TNF signaling have been hampered by significant differences in ATP and ROI levels between IB␣-expressing cells and vector control cells in the absence of TNF treatment (see "Discussion"). We have explored whether another mRNA, encoded in a region downstream of CoxIII, is also regulated differentially in the HT1080 V and HT1080 I cells. As with CoxIII, we find that CytB mRNA is down-regulated by TNF␣ in the HT1080 vector cells but not affected in the IB␣-expressing cells. Overall, FIG. 4. TNF␣ stimulation increases DNA binding activity of p50 in the mitochondria. Mitochondria isolated from U937 cells that were treated with TNF␣ were incubated in nuclear extract buffer, and the resulting protein was analyzed by EMSA. Supershift analysis was performed on extract from the TNF␣ 1-h time point. Extracts were preincubated for 15 min with either a p65 or p50 antibody prior to the addition of the probe.
these results indicate a role for NF-B in the regulation of specific mitochondrial gene expression.

DISCUSSION
The results presented here demonstrate the localization of certain NF-B subunits and IB␣ to the mitochondria and are consistent with a recent publication showing that IB␣ and p65 are found in the mitochondria (14). Interestingly, the major components of the NF-B regulatory pathway are also apparently localized with the mitochondrial fraction, including the IB kinase subunits IKK␣ and IKK␤ as well as IKK␥. Consistent with this, TNF␣ treatment of U937 cells leads to the phosphorylation of degradation of the mitochondrially associated IB␣. In contrast with the degradation that occurs in the cyto-plasm, the mitochondrial degradation of IB␣ appears to be proteasome-independent, since a specific inhibitor of the proteasome (lactacystin) does not block degradation and because key components of the proteasome and ubiquitin ligase associated with IB␣ ubiquitination are not found in the mitochondria. The use of inhibitors suggests that calpain may be partly involved with IB␣ degradation in the mitochondria. The DNA binding assay indicates that an NF-B complex containing p50, possibly as a homodimer, is released from IB␣ and is capable of binding to DNA following TNF␣ stimulation. This response is correlated with the loss of CoxIII and CytB mRNAs following TNF␣ treatment. The involvement of NF-B in this process is suggested by the use of the superrepressor form of IB␣, which inhibits the loss of CoxIII mRNA. The data raise many questions about the functions of NF-B in the mitochondria.
Although great efforts were made to ensure the purity of the mitochondrial preparations, we cannot rule out the possibility of small amounts of contamination from other cellular organelles. We performed electron microscopy on our mitochondrial preparations and found that they were ϳ95% pure mitochondria. The other 5% of the preparation consisted of membranous debris, which may include Golgi, plasma membrane, and endoplasmic reticulum. Western blots were also performed on these preparations using various organelle markers, confirming that the mitochondrial preparations were highly purified. We did see some staining for endoplasmic reticulum in our mitochondrial fractions; however, the percentage of nonmitochondrial membrane in our preparations was low enough that we feel it did not contribute significantly to the protein content of the preparations. In addition, lack of endo-FIG. 7. CoxIII mRNA is maintained following TNF␣ stimulation in the absence of NFB. HT1080 cells transfected with either a vector control or IB␣-SR were treated with 10 ng/ml TNF␣ for the specified times. RNA was isolated, and 10 g was electrophoresed on 1.0% formaldehyde gels. RNAs were transferred to nylon membrane and probed with 32 P-labeled cDNAs corresponding to CoxIII and cytochrome b. The blots were stripped and reprobed with glyceraldehyde-3-phosphate dehydrogenase as a loading control.  2, 4, 6, and 8). Mitochondrial (M) and cytoplasmic (C) proteins were isolated and analyzed on 12% SDS-polyacrylamide gels. Blots were first probed with a phospho-IB␣ (pIB␣). Blots were stripped and reprobed with an IB␣ antibody. B, proteins isolated from untreated U937 cells were analyzed on 12% SDS-polyacrylamide gels. Blots were probed with Lmp-1, Skp-1, and cytochrome c antibodies. C, mitochondrial (M) and cytoplasmic (C) extracts isolated from U937 cells were analyzed on 12% SDS-polyacrylamide gels. Blots were probed with either IB␣, IB␤, or cytochrome c antibodies.
FIG. 6. IKK␣, IKK␤, and IKK␥ are all present and functional in the mitochondria. A, mitochondria (M) and cytoplasmic (C) extracts isolated from U937 cells were run on 12% SDS-polyacrylamide gels and transferred to nitrocellulose. Blots were probed with either IKK␣, IKK␤, IKK␥, or cytochrome c. B, U937 cells were either treated or not treated with TNF␣, and mitochondrial and cytoplasmic extracts were prepared. 500 g of protein was immunoprecipated with an anti-IKK␤ antibody. Immunoprecipates were incubated with wild-type glutathione S-transferase-IB␣ fusion protein in the presence of [␥-32 P]ATP. The immunoprecipitates were run on a 10% SDS-PAGE gel, dried, and exposed to film. The data represent three independent experiments. plasmic reticulum staining of p65 or IB␣ in the electron micrographs further supports the conclusion that NF-B staining in our fractions is mitochondrial.
One obvious question is the potential mechanism whereby NF-B regulates CoxIII and CytB mRNA levels. One can envision at least four mechanisms. One mechanism may be that NF-B, possibly as a p50 homodimer, inhibits the processing of the RNA precursor at a level inhibiting CoxIII and CytB mRNA release. In this model, the release of mitochondrial p50 from IB␣ may lead to an interaction with the precursor RNA and an inhibition of processing. Another mechanism could be that binding of NF-B to mitochondrial DNA inhibits a transcriptional mechanism, possibly elongation, again somehow affecting CoxIII and CytB mRNA production. Another possibility is that p50 in association with IB␣ serves as a positive mechanism in regulating CoxIII and CytB mRNA but that dissociation of the p50 and IB␣ complex following TNF treatment leads to a loss of this control. Finally, we cannot rule out that a nuclear gene or genes regulated by NF-B encode proteins that regulate specific mitochondrial mRNA accumulation. Obviously, more complex experimentation will be required to determine how NF-B may regulate mitochondrial gene expression and whether other transcripts may be positively or negatively controlled by NF-B. Along these lines, it is presently unclear whether p50 and IB␣ are complexed in the mitochondria and whether the detection of p65 in the mitochondria is due to an interaction with p50 or IB␣. The recent paper from Bottero et al. (14) indicates that p65 and IB␣ are associated in mitochondrial extracts. Western blotting did not reveal significant levels of other NF-B subunits in the mitochondria of U937 cells (data not shown).
Another interesting question is how mitochondrial IB␣ becomes phosphorylated and how mitochondrial or mitochondriaassociated IKK becomes activated in response to TNF signaling (Fig. 6B). For cytoplasmic IKK, one current model is that IKK is recruited to factors associated with cytokine receptors, such as the TNF receptor (24), and becomes activated to phosphorylate IB␣ or IB␤. Presumably, rapid activation of mitochondrially associated IKK must occur by a distinct mechanism that is somehow dependent on TNF binding to its receptor on the membrane. The nature of the signal linking cell surface TNF binding to its receptor and the activation of the mitochondrial or mitochondria-associated IKK is unclear. We cannot rule out the possibility that the mitochondrial IB␣ becomes phosphorylated by a kinase distinct from IKK or becomes phosphorylated in the cytoplasm and is transported into the mitochondria.
The localization of NF-B to the mitochondria raises interesting questions concerning a potential role in regulating apoptosis. NF-B activation in response to TNF signaling has been shown to inhibit apoptosis. Thus, it has been shown that NF-B activates several gene products (namely Bcl-xL, A1/Bfl-2, IAP proteins, TRAF proteins, etc.) to inhibit the caspase cascade and to block cytochrome c release from mitochondria (4,6). It is interesting to speculate that mitochondrial NF-B may play a role in the suppression of apoptosis. As described above, it is predicted that modulation of CoxIII mRNA may ultimately impact ATP production, which is required for apoptosis induced by TNF. Along these lines, Bottero et al. (14) provide evidence that mitochondrial IB␣ interacts with ANT, the adenine nucleotide transporter, which has been speculated to be involved with apoptosis through its ability to regulate the mitochondrial permeability transition (see Ref. 14). Additionally, we have observed that expression of superrepressor IB␣ in cells alters basal levels of ATP (data not shown). Future studies will address mechanisms whereby mitochondrial NF-B may control cell death mechanisms.
NF-B is not the first nuclear encoded transcription factor to be localized to the mitochondria. Interestingly, the glucocorticoid receptor has been found in the mitochondria (25). Potentially important are observations that glucocorticoid receptor can physically interact with NF-B subunits (26); thus, NF-B/ glucocorticoid receptor interactions in the mitochondria may function to regulate key processes involved in cell growth and apoptsosis. Experiments addressing these issues may reveal novel regulatory mechanisms associated with mitochondrial function.