Immunoproteasome Deficiency Modifies the Alternative Pathway of NFκB Signaling

Immunoproteasome is a protease abundant in immune cells and also present, albeit at lower concentrations, in cells outside the immune system. Recent evidence supports a novel role for the immunoproteasome in the cellular stress response potentially through regulation of NFκB signaling, which is the primary response to multiple stressors. The current study tests whether the Classical or Alternative Pathways are regulated by immunoproteasome following chronic TNFα exposure in cultured retinal pigment epithelial cells isolated from wild-type mice and mice deficient in one (LMP2, L2) or two (LMP7 and MECL-1, L7M1) immunoproteasome subunits. Assays were performed to assess the expression of NFκB responsive genes, the content and activity of NFκB transcription factors (p65, p50, p52, cRel, RelB), and expression and content of regulatory proteins (IκBα, A20, RPS3). Major findings include distinct differences in expression of NFκB responsive genes in both KO cells. The mechanism responsible for the altered gene expression could not be established for L7M1 since no major differences in NFκB transcription factor content or activation were observed. However, L2 cells exhibited substantially higher content and diminished activation of NFκB transcription factors associated with the Alternative Pathway and delayed termination of the Classical Pathway. These results provide strong experimental evidence supporting a role for immunoproteasome in modulating NFκB signaling.


Introduction
The proteasome is a proteolytic complex that regulates cellular processes essential for cell survival, such as cell cycle, signal transduction, gene expression, and degradation of damaged and misfolded proteins. Several proteasome subtypes, defined by their catalytic subunits, have been described [1,2]. In the standard proteasome, the catalytic subunits are b1, b2, and b5. These subunits cleave proteins after acidic, basic, and hydrophobic amino acids, respectively. In nascent proteasome cores, the standard catalytic subunits can be replaced by the inducible subunits, LMP2 (b1i), MECL-1 (b2i) and LMP7 (b5i) to form the catalytic core of the immunoproteasome (i-proteasome). While there are minor differences in the catalytic activities of the b2/ MECL and b5/LMP7 subunits for standard and i-proteasome, the activity of b1 and LMP2 differ. LMP2 preferentially cleaves after hydrophobic amino acids rather than after acidic amino acids, generating a population of peptides that are enriched in hydrophobic C-termini. This difference in cleavage specificity is important for i-proteasome's role in immune function. Intermediate cores containing a mixture of standard and inducible catalytic subunits have also been reported [3].
Proteasome subtypes differ substantially not only in their enzymatic characteristics but also in their pattern of expression, suggesting the potential for discrete contributions to cell processes. Standard proteasomes are constitutively expressed in nearly all mammalian cells. In contrast, i-proteasome is highly expressed in cells of the immune system, where it performs functions associated with generating peptide ligands for MHC class I antigen presentation [4]. I-proteasome is also present, albeit in low abundance under basal conditions, in cells outside the immune system, including neurons of the retina and brain, skeletal muscle and epithelial cells of the retina [5][6][7]. When these cells are exposed to various stressors, such as inflammatory cytokines, disease, or oxidative stress, i-proteasome is significantly upregulated [8][9][10][11]. In addition to its rapid induction, assembly of the nascent i-proteasome core particle is four times faster than the standard core and conversely, i-proteasome's half-life is substantially shorter [12]. This highly dynamic adjustment in i-proteasome content permits its rapid response to environmental challenges. Taken together, these results suggest a role for iproteasome in regulating processes associated with the cellular response to stress and injury.
The nuclear factor-kappa B (NFkB) pathway is the primary mechanism for responding to multiple stressors, such as toxic chemicals, UV light, and oxidative damage, as well as proinflammatory cytokines, viral and bacterial products. Activation of the NFkB pathway elicits rapid induction of early response genes that help protect the cell from damage. However, aberrant regulation or long-term activation of NFkB signaling can lead to pathologies, such as toxic shock, neurodegenerative and inflammatory diseases [13].
Prior to stimulation, NFkB dimers are sequestered in the cytoplasm by inhibitory proteins, including the IkBs (a, b, e) and the precursors of active transcription factors, p105 and p100. Upon stimulation of NFkB through either binding of a ligand to its cognate receptor or through a receptor independent event (i.e., oxidative stress), a cascade involving activation of multiple kinases and phosphorylation of select proteins in the NFkB pathway is initiated. Proteasome plays a direct role in regulating NFkB via complete degradation of the inhibitor IkB, releasing the active dimer and allowing its translocation into the nucleus. This pathway, typically involving p65/p50 dimer, is referred to as the Classical Pathway. A second mechanism involves endoproteolysis or partial degradation of the inhibitory portion of p105 and p100, generating the active transcription factors p50 and p52. In addition to p65, these transcription factors can associate with RelB or co-associate to form homo and heterodimers. This pathway is referred to as the Alternative Pathway. While initially believed to be functionally separate, more recent evidence suggests considerable crosstalk between the Classical and Alternative NFkB pathways [15]. Crosstalk between the two pathways has been well established in immune cells following TNFa-stimulation and is an integral part of the innate and adaptive immune response [16].
An important component of regulation built into the NFkB pathway is the induction of multiple inhibitors that serve to attenuate or terminate the signal. These inhibitors include the IkBs, p100, and p105, which sequester transcription factor dimers in the cytoplasm. Additionally, IkBa can terminate the signal by removing the p50/p65 dimer from the promoter. Specific to terminating TNFa-induced stimulation of NFkB is the inhibitor A20 (Fig. S1). This protein is an E3 ligase that ubiquitinates RIP1, a modular protein located at the TNF receptor, and targets it for proteasome degradation. The absence of RIP1 terminates NFkB signaling at the receptor [17].
Evidence in both KO mice and humans with i-proteasome mutations implicate a role for the i-proteasome in modulating NFkB signaling. In i-proteasome-deficient mice, defects in proteolytic processing of NFkB precursors (p100/p105) and decreased degradation of IkBa have been reported [18][19][20]. While these initial findings were later disputed [21][22], more recent reports in patients with mutations in the LMP7 protein [23] have re-energized this controversy. Notably, patients harboring mutations in LMP7 consequently have lower i-proteasome content and exhibit a spectrum of auto-inflammatory syndromes (i.e., higher IL-6 levels) that implicate aberrant NFkB signaling as part of the disease mechanism [23].
The current study tests the hypothesis that i-proteasome modulates NFkB signaling by comparing the early and late response to chronic TNFa exposure in cultured retinal pigment epithelial (RPE) cells isolated from wild-type mice and mice deficient in one (LMP2, referred to as L2) or two (LMP7 and MECL-1, referred to as L7M1) subunits of the i-proteasome. TNFa was used as the stimulant because (1) the mechanism of TNFa-induced NFkB signaling is well-described (Fig. S1), and (2) chronic stimulation with TNFa provides a means for generating an NFkB response similar to that observed with chronic inflammation. Cell lines deficient in i-proteasome subunits may provide important mechanistic details about NFkB signaling that are relevant to the chronic inflammation experienced by patients with mutations in i-proteasome genes. Cultured cells from the retinal pigment epithelium were used because they express iproteasome in measurable amounts under basal conditions and can strongly upregulate i-proteasome under conditions of stress or inflammation [8,10]. These cells are responsive to TNFa and have an immunomodulatory role in vivo, as demonstrated by their secretion of the pro-inflammatory molecule IL-6 and their ability to induce unresponsiveness in T-cells [24]. The current study tests whether RPE behave similar to immune cells in activating both the Classical and Alternative Pathways in response to TNFa [16]. The use of cells from KO mice allowed us to determine if altering the i-proteasome content affects the TNFa response. Results from the current study revealed distinct differences in expression of NFkB responsive genes in cells lacking either one (L2) or two (L7M1) i-proteasome subunits. Our data showed higher content and diminished activation of NFkB transcription factors associated with the Alternative Pathway and delayed termination of the Classical Pathway in cells deficient in the LMP2 subunit.

Cell culture conditions
Retinal pigment epithelial (RPE) cells isolated from wild-type (WT) and knock-out (KO) mice deficient in one (LMP2, referred to as L2) or two (LMP7 and MECL-1, referred to as L7M1) subunits of the i-proteasome were immortalized as previously described [25]. This study includes results from two independently derived cell lines for WT and KO mice. An animal protocol for harvesting RPE was approved by the Institutional Animal Care and Use Committee of the University of Minnesota and followed the guidelines established by the National Institutes of Health. RPE cells were cultured in growth media containing Dulbecco's Modified Eagle Medium, 0.4 mM L-glutamine, 25 mM glucose (Gibco), MEM non-essential amino acids (Cellgro), 50 U/mL penicillin, 50 U/mL streptomycin (Gibco) and 4% heat inactivated fetal bovine serum (Atlanta Biologicals). Cells were cultured at 37uC in a humidified chamber containing 5% C0 2 . For TNFa treatments, cells were treated with recombinant human TNFa (10 ng/mL, R&D Systems) and harvested at times indicated in figures.

Preparation of tissue homogenates
In order to harvest murine RPE cells, the anterior segment was removed from enucleated globes. The retina was then removed and RPE cells were collected by gentle agitation in 1X PBS with a fine paint brush. The cells were pelleted by centrifugation at 6006g for 10 min. Whole cell lysates were prepared after two cycles of freeze thaw with liquid nitrogen, followed by homogenization with a glass dounce homogenizer using a buffer containing 50 mM Tris-HCl pH 7.8, 2% CHAPS.
Cerebellar tissue was dissected away from the brain stem after brains were removed from the skull. Tissue homogenates were prepared using a glass dounce homogenizer with a buffer containing 50 mM Tris-HCl pH 7.8, 2% CHAPS. Protein concentrations were determined using the bicinchoninic acid (BCA) assay (Pierce, Rockford IL) with bovine serum albumin as the standard.

IL-6 ELISA
Cells were grown in 96-well culture plates. Cell culture media was harvested from untreated cells or at 48 hrs after a single dose of TNFa (10 ng/mL)(16TNF). In a second experiment, cells were treated with TNFa (10 ng/mL) for 48 hrs, media was removed and replaced with either fresh media (wash-out) or media containing 10 ng/mL TNFa (26TNF). Cell media was harvested at 96 hrs after the initial treatment. The IL-6 content of culture media was determined by ELISA, using anti-mouse IL-6 (eBioscience, clone MP5-20F3) for the capture antibody, biotinconjugated anti-mouse IL-6 (eBioscience, clone MP5-32C11) for the detection antibody, and the enzyme-conjugated secondary antibody, AKP-Streptavidin (BD Pharmingen). A standard curve of recombinant mouse IL-6 (eBioscience) was used to calculate the amount of secreted IL-6. The content of secreted IL-6 was normalized to cell density for each well using the CyQuant NF cell proliferation Assay (Invitrogen). IL-6 production is reported as picograms IL-6 per cell (pg/cell).

Isolation of Nuclear/Cytoplamic Fractions and whole cell lysates
Nuclear and cytoplasmic fractions were prepared immediately from cell pellets using NE-PER Nuclear and Cytoplasmic Extraction Reagents kit (Pierce, Rockford IL). Fractions containing either cytoplasmic or nuclear proteins were stored at 280uC. Whole cell lysates were prepared by lysing cells in buffer containing 10 mM KCl, 25 mM Tris, pH 7.8, 2.5 mM EDTA, 0.5% NP40, 10% glycerol. Lysates were cleared by centrifugation at 6006g for 15 minutes. Cell pellets used to generate whole cell lysates were stored at 280uC until use. Protein concentrations were determined using bicinchoninic acid (BCA) assay (Pierce, Rockford IL) with bovine serum albumin as the standard.

Western blot analysis
Proteins were electrophoretically separated on SDS-polyacrylamide gels (10% or 13%) and transferred to either nitrocellulose (for NFkB antibodies) or PVDF membranes (all other antibodies), as previously described [25]. The linear range of detection was determined for each antibody from blots containing incremental increases in protein. Loading controls were GAPDH, Beta actin, or histone H2B (nuclear fraction). Information for primary antibodies, linear range, dilutions, and protein loads for each antibody are provided in Table 1. Images were captured using a ChemiDoc XRS (Bio-Rad Laboratories, Hercules, CA) and analyzed using Quantity One (Bio-Rad Laboratories). The density of sample immune reactions was normalized to a WT control (untreated cells) run on each blot. For quantification of proteasome subunits, a sample of 20S purified from liver was run on each blot to ensure that the immune reactions were from subunits incorporated into the core.

Phagocytosis measured by flow cytometry
Cells were cultured in 6-well plates. Growth medium was removed from confluent cells and replaced with fresh growth medium containing 1.0 mm polystyrene beads loaded with yellowgreen fluorescent dye (FluoSpheres, Molecular Probes; Eugene, OR) at a concentration of 1.4610 5 microspheres/mL. RPE cells were incubated with the beads for 24 hrs to allow for phagocytosis, then washed 26 and harvested by trypsinization. The cells were prepared for flow cytometry in growth medium to ensure the removal of beads outside the RPE. Cell pellets were resuspended in PBS containing 2 mM EDTA and 0.5% BSA. The resulting cell suspensions was used to determine microbead content (ex/ em = 505 nm/515 nm) per cell using flow cytometry (FACSCalibur with Cell Quest software, BD Biosciences) and analyzed using FlowJo software (Tree Star, Ashland, OR).

RNA isolation
Total mRNA was isolated from cells pellets using RNeasy Mini kit (Qiagen; Valencia, CA). cDNA was generated in a GeneAmp PCR System 9700 (Applied Biosystems, Forester City, CA) from RNA (70 ng) using SuperScript III reverse transcriptase (Invitrogen) and oligo-dTs (Integrated DNA technologies).

Quantitative RT-PCR
Quantitative real-time PCR was performed using an iQ5 Multicolor Real-time PCR I-cycler (Bio-Rad Laboratories, Hercules,CA). Reactions were performed with cDNA (2 ng/ reaction, performed in triplicate) using Immolase DNA polymerase (Bioline), 800 mM dNTP mix (Bioline), SYBR Green (Invitrogen), and 200 nM of forward and reverse primers. Normalized gene expression was determined using the iQ5 optical system software (Bio-Rad Laboratories, Hercules,CA) using acidic ribosomal phosphoprotein P0 (ARBP) expression as a reference gene for each sample. The sequence for primers used in qRT-PCR is provided in Table 2.

NFkB transcription factor activation assay
Activity of individual NFkB transcription factors was determined in nuclear extracts (5 mg) using the ELISA-based TransAM NFkB Transcription Factor assay kit (Active Motif; Carslbad, CA). Experiments were performed on WT, L7M1, and L2 cells cultured and processed in parallel. The assay was performed for three to five separate cell preparations, each measured in duplicate. Data were normalized to WT untreated cells in each experiment.

Statistical Analysis
Data were analyzed by two-way ANOVA with mouse strain and time as the main effects, using the statistical software NCSS (v. 2001, Kaysville, UT). One-way ANOVA was performed for cells from each mouse strain to determine if there was a significant response to TNFa. Tukey-Kramer post-hoc test was performed to determine differences between groups. Significance was set at p,0.05. Data are reported as mean 6 standard error of the mean.

Characterization of cells
To test how the absence of specific i-proteasome subunits affected NFkB signaling, we used two independently-derived cell lines that were generated from RPE cells harvested from WT mice and mice deficient in one (L2) or two (L7M1) i-proteasome subunits, as previously described [25]. All of the cell lines express PEDF, Best1, and RPE65 (Fig. 1A,B), and are able to phagocytose latex beads (Fig. 1E). These results confirm that the cultured cells have retained these in vivo characteristics specific to RPE cells.
Proteasome activity was measured in each cell line using fluorogenic peptide substrates to monitor the caspase-, trypsin-, and chymotrypsin-like activities (Fig. 1F). For L7M1 and L2 cells, the chymotrypsin-like activity was ,40% higher and lower, respectively. However, these results did not achieve statistical significance (p = 0.09). Trypsin-(p = 0.12) and caspase-like (p = 0.20) activities were also not different between WT and KO cells.
Proteasome subunit composition was evaluated from Western immunoblots using subunit-specific antibodies (Fig. 1C,D). The reaction of the a7 subunit was used to estimate proteasome total content since it is constitutively expressed in all proteasome cores [8]. For the b-subunits, alignment of the immune reaction of the 20S isolated from liver with sample immune reactions was used to ensure that measurements were performed on mature proteins that had incorporated into the 20S core. As expected, there was no immune reaction associated with L7M1 and L2 for the LMP7 and LMP2 subunits, respectively, as predicted from the gene disruption [27][28][29]. Densitometric analysis of the immune reactions for two independently derived RPE cultures showed no significant difference in both the a7 (data not shown) and b1 subunits when comparing WT and KO cells (Fig. 1D). While b5 content was ,80% higher in L7M1 cells compared to WT, this increase did not reach statistical significance (p = 0.12). For the immunoproteasome subunits, LMP2 was increased more than 2.5-fold in L7M1 cells (p = 0.01) and LMP7 was decreased ,50% in L2 cells (p = 0.004).
Subunit composition was also measured in RPE cells harvested from murine eyes to determine how closely the cultured cells replicate the in vivo state (Fig. 1C,D). The relative content of a7, b1, LMP2, and LMP7 were not different between WT and KO mice. Similar to cultured cells, there was a 60% increase in content of the b5 subunit in L7M1 RPE (p = 0.11). These results show the main difference between cultured RPE and cells in vivo is the significantly greater content of LMP2 in cultured cells from L7M1 mice.
The high LMP2 content in L7M1 cultures and equivalent incorporation of the LMP2 subunit in L7M1 RPE in vivo are in direct contrast to previous reports from other tissues, which showed that in the absence of LMP7 there was significantly less LMP2 incorporation [30][31][32]. While our previous results for the 20S purified from spleen are consistent with inhibited LMP2 incorporation in L7M1 mice, we have also shown LMP2 incorporation is not inhibited in L7M1 retinas or in RPE cell cultures derived from a different set of KO mice [8]. Taken together, these results suggest there are tissue-specific differences in proteasome subunit composition. To directly test this possibility, we monitored proteasome subunit composition in the cerebellum from WT and KO mice. We have previously shown that within this region of the brain, i-proteasome is expressed in measurable amounts in glia, neurons, oligodendrocytes, and bone marrowderived immune cells [5]. Under identical experimental conditions used to estimate proteasome subunit composition in RPE cells, we found that LMP2 content was three-fold lower in the cerebellum of L7M1 mice (Supplement Fig. S2). These results suggest that brain and retinal cells in vivo and in cultures have different proteasome populations in L7M1 mice. Furthermore, the LMP7 subunit is not absolutely required for LMP2 incorporation in certain cells. Thus, RPE cells provide a unique experimental setting for testing differences in immunoproteasome content.
To determine the optimal dose of TNFa to achieve NFkB activation, preliminary experiments tested a range of TNFa concentrations (0.5 to 15 ng/ml). Based on the dose-dependent expression of IL-6 and cell viability assays showing that cells remained viable over 48 hrs, a dose of 10 ng/ml TNFa was selected for experiments (data not shown). Western blotting was used to evaluate the proteasome population in each cell line and determine whether the subunit content changed with TNFa treatment over the time course of our experiments. Densitometry of the immune reaction showed no change in proteasome subunits throughout 16 hours of exposure to TNFa (Fig. 1G). These results indicate that the proteasome population remained stable over the time frame of our assays.

Altered Expression of NFkB responsive genes
An early outcome of NFkB activation is the upregulated expression of multiple genes containing the kB enhancer element in the promoter [14]. However, the magnitude and kinetics of activation depend upon the type of stimuli and cell. The expression of NFkB transcription factors (nfkb1, nfkb2, relb), negative regulators (ikba, a20), and three prototypic NFkB responsive genes (inos, cox2, il6), which are all regulated by NFkB, was measured by quantitative RT-PCR following treatment with TNFa.
In monitoring the time-dependent change in expression for nfkb1 (p105), all cells exhibited a 3-to 4-fold increase in expression at 2 hours ( Fig. 2A). For nfkb2 (p100), a robust upregulation was also observed by 2 hrs in all cells (Fig. 2B). However, nfkb2 expression in i-proteasome-deficient cells was 3-fold higher than in WT cells (p,0.01). For relb, expression was increased ,five-fold by 2-4 hrs post treatment in WT and L7M1 cells (Fig. 2C). The response was significantly greater in L2 cells; a 10-fold increase in relb expression was observed from 2 to 16 hrs post-TNFa (p,0.01).
Expression was also monitored for the inhibitory proteins ikba and a20. For the early phase inhibitor IkBa, TNFa stimulated an 18-fold increase in expression in WT cells by 30 min, followed by a slight decrease in expression throughout the 16 hrs (Fig. 2D). This sustained, elevated expression reflects the chronic stimulation by TNFa. I-proteasome-deficient cells exhibited similar kinetics of upregulated expression, but of significantly lower magnitude. The maximum expression was only 10-fold and 3-fold over unstimulated values for L7M1 and L2 cells, respectively.
Expression of the late phase inhibitor a20 was also robustly upregulated by 30 min post treatment, followed by a rapid decline to slightly above baseline levels in all cells (Fig. 2E). While the kinetics of the response was similar between cells, there were significant differences in the magnitude of response. Maximum levels of expression were 40-, 20-, and 12-fold over baseline values in WT, L7M1, and L2 cells, respectively (p,0.01).
Quantitative PCR of three prototypic NFkB responsive genes showed that in WT and L2 cells, inos was significantly upregulated by 2 to 4 hours, respectively (Fig. 2F). In contrast, L7M1 did not respond. For cox2, there was no response in WT and L2 cells, but a robust increase by 15 minutes post-TNFa treatment in L7M1 cells (Fig. 2G). Measures of il6 expression showed that all cells exhibited a similar response; a significant ,16-fold upregulation was observed by 2 hours (Fig. 2H). To explore potential differences downstream of expression, the content of IL-6 secreted into the culture medium was monitored. We found that the amount of IL-6 produced by L7M1 cells was significantly higher at baseline, at 48 hours after a single dose of TNFa, and at 96 hours after combinations of multiple treatments (Fig. 2I). These results show that while regulation of expression was approximately equal, there were differences in production and secretion of IL-6 that were influenced by the absence of LMP7 and MECL. In fact, the hyperactive response of L7M1 cells suggests potential problems in modulating the NFkB signal.
Results from quantitative PCR show that the absence of specific i-proteasome subunits altered the expression of multiple NFkB responsive genes. These results provide strong experimental evidence supporting a role for i-proteasome in regulating NFkB signaling.

Altered NFkB transcription factor content and translocation
To investigate potential mechanisms responsible for the altered response to NFkB signaling in i-proteasome deficient cells, multiple assays were employed to monitor the phosphorylation state, protein content, and nuclear translocation of NFkB transcription factors. An early step in activation of p65 involves its phosphorylation following degradation of the inhibitor protein IkBa [33]. Western immunobloting was used to monitor both the phosphorylated p65 and the total p65 population (Fig. 3A). The data shows a rapid peak of phosphorylation within 5 minutes of TNFa treatment indicative of p65 activation in all cell lines. This rapid activation was followed by a decline in the ratio of phosphorylated to total p65 at 30 minutes that was maintained at just above baseline for at least 4 hours. The pattern of activation was replicated in the measures of p65 nuclear content; a significant 5-to 6-fold increase was observed for all cell lines at 30 min post-TNFa indicating robust translocation (Fig. 3B). By 3 hrs, p65 content was reduced approximately 50% in WT and L7M1 cells, but did not decrease in the L2 cells. Additionally, measures of the overall cytoplasmic content of p65 were not different between cells, although there was significantly less p65 in L2 cells at 30 min (Fig. 3C). These results suggest that p65 activation is not regulated by the i-proteasome, but there is evidence for delayed termination of p65 activation in cells deficient in LMP2.
RelB is part of the Alternative Pathway and a preferred partner of p52 [34]. In comparing the overall cytoplasmic content of RelB, L2 cells contained significantly higher levels than WT and L7M1 cells (p,0.01) (Fig. 3D). In all cells, treatment with TNFa resulted in a significant increase in the cytoplasmic content (p = 0.02) by 3 hrs post treatment. These results are consistent with the time dependent change in expression, which showed a significant increase in message by 2 hrs in all cells (Fig. 2C). Additionally, upregulation in L2 cells was greater than 2-fold more than both WT and L7M1, which is consistent with the higher RelB protein content. TNFa also induced nuclear translocation of RelB by 3 hrs, as suggested by the 40% (WT) to 80% (L7M1, L2) increase in RelB nuclear content (p,0.01) (Fig. 3E).
The nuclear content of cRel was also monitored before and after TNFa treatment (Fig. 3F). Overall, the nucleus of L2 cells contained significantly lower amounts of cRel compared with both WT and L7M1 (p,0.01). While there was a significant decrease in cRel content in WT cells after 30 min and 3 hr of TNFa stimulation (p = 0.014), nuclear cRel content did not change in L7M1 and L2 cells. Based on these results, showing minimal differences in cRel response, we did not pursue additional assays for this transcription factor.
Comparing the content of p105, the precursor for the active transcription factor p50, significantly higher levels were observed in L7M1 and L2 compared with WT cells (p,0.01) (Fig. 4A). For p50, the overall content was higher in L7M1 cells, but did not vary with TNFa treatment in any cell line. There was also no celldependent difference in the overall content of p50 in the nucleus at baseline and at 30 minutes post-TNFa treatment (Fig. 4B), which was increased ,50% at 30 min over basal levels in all cells. For WT and L7M1 cells, p50 levels returned to baseline by 3 hours. In contrast, cells deficient in LMP2 exhibited higher levels even at 3 hours, suggesting prolonged activation of p50. These data, along with the results showing elevated levels of nuclear p65 at 3 hrs (Fig. 3B) suggest that cells deficient in LMP2 have delayed termination of the Classical Pathway.
Protein content of the p100 precursor was significantly higher in L2 cells compared with WT and L7M1 cells (p,0.001). The p52 content was also 2 to 3-fold higher in L2 cells, although this increase was not statistically different (Fig. 4C). The higher overall content of p100 in L2 cells could be due to the 3-fold higher expression of the nfkb2 gene in i-proteasome-deficient cells (Fig. 2B). TNFa also stimulated the translocation of p52 into the nucleus (Fig. 4D). At 30 min and 3 hrs post treatment, nuclear content was ,2-fold higher than baseline levels (p,0.01). Of note, L2 had significantly less nuclear p52 compared with WT and L7M1 cells (p = 0.04). Considering the higher p100 present in the cytoplasm, the lower nuclear p52 suggests a defect in nuclear translocation of this transcription factor in cells deficient in LMP2.

Altered content of NFkB Regulators
NFkB signaling is regulated by multiple accessory proteins that either enhance or inhibit expression of target genes. Ribosomal protein S3 (RPS3) is a multifunctional protein that enhances NFkB transactivation by stabilizing p65 binding to specific kB sites [35]. Measures of the cytoplasmic content showed significantly elevated levels of RPS3 in L2 compared with WT cells (Fig. 5A, left panel) (p,0.01). While TNFa had no effect on RPS3 cytoplasmic content in L2 cells, there was a significant increase in both WT and L7M1 cells at 3 hours. There was also no measurable change in nuclear RPS3 content in response to TNFa, although RPS3 content was higher in L2 compared with both WT and L7M1 cells (Fig. 5A, right panel). The elevated nuclear content of RPS3 in L2 cells may enhance binding of the p65/p50 dimer on the promoter. This mechanism could help explain the extended presence of p65 and p50 in the nucleus of L2 cells following TNFa stimulation.
The content of two inhibitors of NFkB signaling, IkBa and A20 was also examined. In monitoring IkBa protein content, basal levels were equivalent between cells (Fig. 5B). In response to TNFa stimulation, all cells displayed the expected initial decrease in IkBa content at 30 min, followed by a recovery to baseline levels by 3 hrs. This rapid rebound in IkBa content is due to the upregulated expression of the ikBa gene that occurred within 15 minutes in all cells during the early phase of signaling (Fig. 2D). Cells had similar basal content of A20 and exhibited a significant 3-to 5-fold increase by 3 hrs and remained elevated through 24 hrs post-TNFa treatment (Fig. 5C). Figure 6 provides a graphical summary of study results comparing the L7M1 and L2 cells with WT cells. Cells deficient in i-proteasome subunits exhibited a number of striking differences in basal content of NFkB proteins, as well as in their response to TNFa treatment. Additionally, there were substantial differences between cells lacking one (lmp2 2/2 ) or two (lmp7 2/2 /mecl-1 2/2 ) iproteasome catalytic subunits.

Discussion
In this study, we tested the hypothesis that the i-proteasome is involved in regulating NFkB signaling using RPE cells derived from WT mice and those deficient in one (L2) or two (L7M1) iproteasome catalytic subunits. A single, sustained dose of TNFa was used to stimulate NFkB signaling. Downstream early-and late-phase events, such as activation and nuclear translocation of NFkB transcription factors, expression of NFkB responsive genes and content of regulatory proteins, were monitored to begin identifying specific sites in the NFkB pathway that are altered due to i-proteasome deficiency.
Results showed equivalent TNFa-induced activation of the Classical Pathway, including rapid p65 phosphorylation, degradation of the inhibitory protein IkBa, and translocation of p65 and its prototypic binding partner p50 into the nucleus (Fig. 6). These results suggest activation of the Classical Pathway is independent of i-proteasome regulation. Termination of the signal, as evidenced by decreased content of nuclear p65/p50 by 3 hr, was similar between WT and L7M1. In L2 cells, there was no reduction in nuclear content of these transcription factors, suggesting delayed termination of the Classical Pathway. This delay may be due to increased nuclear content of RPS3, a protein that stabilizes the p65/p50 dimer on the promoter. Cells lacking the LMP2 subunit demonstrated other substantial changes in transcription factor content and activation. There was decreased nuclear translocation of p52 despite having higher cytosolic content of the precursor p100. The lower nuclear p52 content could a result of p100 sequestering p52 in the cytosol and preventing its translocation. Notably, RelB expression and content were significantly higher in L2 cells. Since RelB and p52 are the prototypic binding partners in the Alternative Pathway, the reciprocal change in content for these two transcription factors suggest the composition of dimers that signal via the Alternative Pathway is altered. These alterations in transcription factor content and activation in L2 cells could help explain the significant difference in expression of NFkB responsive genes. For the L7M1 cells, cytoplasmic content of p105/p50 and the inhibitory protein A20 were elevated. Changes in these NFkB regulators and in regulation at points in the NFkB pathway outside of those monitored in this study, as well as the 2.5-fold increase in LMP2 content compared with WT cells, could contribute to the mechanism responsible for the altered expression of NFkB , p65,and bactin in cell homogenates from WT and i-proteasome-deficient (L7M1 and L2) cultured RPE cells at baseline (no TNFa) and from 5 min (50) to 4 hrs with TNFa (10 ng/mL). Graph shows the ratio of phosphorylated to total p65 content before and after TNFa. *5 min is significantly higher than no TNFa for each cell type. Data are the mean (6 SEM) of five independent experiments from n = 2 cell lines/group. (B) Transcription factor p65 activation was monitored in nuclear fractions using the Trans-Am NFkB transcription factor assay kit. responsive genes and increased levels of secreted IL-6 at baseline and after TNFa treatment.
Our results showing that activation of the Classical Pathway is not affected by the absence of i-proteasome subunits is consistent with a recent study that used a chemical genetic approach [36]. The authors observed no defects in TNFa-induced activation of the Classical Pathway (i.e., degradation of IkBa, nuclear translocation of p65/p50) using small molecule inhibitors that specifically target either the LMP2 or LMP7 subunits. The Classical Pathway is the signaling pathway that is most consistently activated following many different stressors. The observation that this pathway was unaffected by i-proteasome-deficiency could explain why i-proteasome KO mice are viable and exhibit limited changes in processes associated with NFkB signaling.
Data from the current study show the most substantial differences in NFkB signaling occur in cells lacking the LMP2 subunit. Previous studies performed in T2 human lymphocyte cell lines, embryonic fibroblasts and spleen cells from LMP2 KO mice, and spleen tissue from non-obese diabetic mice also showed altered NFkB signaling. These model systems, which were deficient in the LMP2 subunit, showed defects in proteolytic processing of NFkB precursors (p100/p105) to the active transcription factors (p52/p50), and decreased degradation of IkBa after TNFa stimulation [18,19]. In studies of B cells from LMP2-deficient mice, delayed and less complete degradation of IkBa following lipopolysaccharide stimulation was observed [20]. In contrast to these published reports, the present study showed L2 cells had higher overall levels of p100 and p105, but there was no evidence for delayed proteolytic processing of the precursors since the levels of the active transcription factors p50 and p52 were equivalent to WT cells. Runnels et al also reported that the presence of LMP2 was not obligatory in the processing of p105 based on the generation of mature p50 protein in T2 cells with the LMP2 gene deleted [22]. Potential explanations for the discrepancies between studies include cell-and tissue-specific differences in i-proteasome populations and their response to TNFa. Substantial cell-specific differences in subunit content was clearly demonstrated in the current study when comparing cultured RPE cells, RPE in vivo, and cerebellum, and in our previous studies comparing retina and spleen proteasomes in KO mice [8]. In cultured RPE, we observed a dramatic increase in LMP2 content in L7M1 cells, which could be a compensatory response to the loss of LMP7 and MECL that is compounded by the absence of an in vivo environment and regulation from neighboring tissue. This overcompensation in LMP2 content by L7M1 could help explain the differences in response to TNFa exhibited by the two iproteasome-deficient cell lines.
The p100 precursor is constitutively processed to p52. In all cells, a steady level of precursor and processed protein is maintained, suggesting this is a tightly regulated and ongoing process. In L2 cells, the significantly higher levels of p100 could have a significant impact on gene transcription due to its multiple regulatory roles. For example, p100 sequesters both p65 and RelB in the cytoplasm, thus inhibiting both the Classical and Alternative pathways. Induced expression of p100 following TNFa stimulation facilitates the exchange of p65-containing dimers for p52/RelB, which is insensitive to negative feedback from IkBa and promotes late-phase transcription of genes [14]. It is also possible that the higher content of p100 in L2 cells is a compensatory response to higher levels of the RPS3 enhancer. Higher p100 may be required to keep NFkB activation in check.
In L7M1 cells, increased cox2 expression, higher basal levels of secreted IL-6 protein, and sustained IL-6 production following TNFa treatment suggests cells lacking the LMP7 and MECL-1 subunits have a heightened inflammatory response. Our results are consistent with earlier reports of enhanced TNFa-induced production of IL-6 in cultured fibroblasts, tissues, and serum from patients with Nakajo-Nashimura Syndrome [37]. This syndrome is one of a spectrum of recently described auto-inflammatory diseases that are associated with mutations in the LMP7 protein and lower i-proteasome content [23]. Analysis of tissues and cells from patients harboring mutations in the lmp7 gene has shown an accumulation of ubiquitinated and oxidized proteins that are likely due to the loss in i-proteasome function. The proposed mechanism responsible for the heightened inflammatory response included the loss of i-proteasome-dependent degradation of damaged proteins,  which triggers hyperactivation of the p38 MAP kinase pathway and overproduction of the inflammatory cytokine IL-6 [23].
One of the most intriguing results from our study is that the absence of specific i-proteasome catalytic subunits produced a unique response in the NFkB pathway. For example, multiple genes were differentially affected by the absence of specific iproteasome subunits. Potential mechanisms include differences in content of NFkB transcription factors, cross-talk between signaling pathways, and differences in NFkB regulation outside of the current study. There is a strong possibility that differences in composition of the active transcription factor dimers in L2 cells, where p100, p105, and RelB are found in greater abundance, plays a critical role in the differential regulation of NFkB genes. Studies in knock-out mice have suggested that each dimer carries out specific regulatory roles in transcription via selective proteinprotein interactions with other transcription factors, co-regulators (i.e., RPS3), and chromatin proteins [38]. Also contributing to the dimer-specific response is their unique DNA-binding specificities and affinities for target genes. Recent evidence has also suggested that dimers can adopt specific conformations when bound to different DNA sequences, which may also influence the binding affinity and kinetics of transcription [39]. Post-translational modifications can also influence the location, stability and interactions with DNA and other transcription factors. These modifications are regulated by not only the NFkB family of proteins, but also other signaling pathways, demonstrating the importance of cross talk between pathways in regulating signaling.
While our study provided a limited examination of specific proteins, it is possible that differences in components upstream and downstream of our focus could also influence the outcomes measured. For example, macrophages isolated from mice lacking the LMP7 and MECL subunits had significantly reduced iNOS expression following lipopolysaccharide stimulation [40]. The difference in expression was due to defects in signaling from the toll-like receptor 4 via the TRIF/TRAM pathway, which are upstream of NFkB signaling. It is also possible that i-proteasome's effect is broader in scope and affects other pathways, including those that cross talk with NFkB. One pathway with suggested regulation by the i-proteasome is Protein Kinase B (Akt) signaling, which is controlled by the molecule phosphatase and tension homologue deleted on chromosome 10 (PTEN). This protein is an important regulator of cardiac muscle size and cellular changes associated with cardioprotection during ischemia preconditioning. Of note, PTEN content is regulated by the presence LMP2. As demonstrated in KO mice, the absence of LMP2 resulted in higher PTEN levels, significantly smaller hearts, cardiomyopathy, and insensitivity to ischemia preconditioning [41,42]. The defects in cardiac function could be rescued in LMP2 KO heart muscle by treating with a PTEN inhibitor.
Additional reports provide compelling evidence for i-proteasome's role in regulating signaling in diverse biological processes. A recent study showed i-proteasome is involved in maintenance of pluripotency in human embryonic stem cells via degradation of proteins that regulate the cell cycle and control differentiation [43]. L7M1 KO mice and humans with LMP7 mutations both exhibit lipodystrophies that stem from defects in adipocyte maturation [23]. I-proteasome deficiency also had an adverse effect on visual function, suggesting defects in retinal signaling in mice lacking one (LMP7) or two (LMP7, MECL) i-proteasome subunits [44]. In L7M1 mice, we have also documented defects in corneal wound healing, a process involving both p38 MAP kinase and NFkB signaling [45].
Several mechanisms have been proposed to explain how the presence of specific proteasome subtypes (i-proteasome, standard and intermediate proteasomes) could differentially affect biological processes. Since the catalytic activity (i.e., degradation rate and substrate specificity) of each proteasome subtype differs, the repertoire of peptides produced will depend on the cell's proteasome population. Some of these peptides could be biologically active and regulate cell processes. For example, in a study examining the production of a specific peptide that was recognized by cytotoxic T lymphocytes, the authors reported that the peptide was generated by the i-proteasome and destroyed by the standard proteasome due to the presence of an acidic amino acid within the peptide's sequence [46]. Preservation of the peptide by the i-proteasome was explained by the exchange of b1 for LMP2, which leads to reduced cleavage of proteins after acidic amino acids. It is also possible that some intermediate proteasomes containing a mixture of different subunits could either enhance or compromise cell function, as shown previously for immune cells of LMP2 KO mice [47]. Decreased lymphocyte survival and reduced cytokine production, defects associated with impaired NFkB signaling, were attributed to the intermediate proteasomes contained in KO cells. The presence of different catalytic subunits could also change the conformation of the catalytic chamber, as suggested by the crystal structure of proteasome subtypes [48], and alter the specificity for substrates. Different catalytic subunits could also provide binding sites for protein and non-protein regulatory molecules, i.e., fatty acids, which could affect the kinetics of degradation. These ideas are supported by studies that reported different results when inhibiting subunit activity versus genetically eliminating the subunit [49,50].
Another mechanism that could help explain i-proteasome's effect on diverse biological processes includes its enhanced ability to degrade oxidized, misfolded, and polyubiquitinated proteins. This property is particularly relevant in responding to conditions of stress and injury, which often involves oxidative stress and the subsequent production of damaged proteins. Considerable evidence has accrued in support of this function for i-proteasome. For example, oxidized and ubiquitin-marked proteins accumulate in cells and tissues deficient in i-proteasome with stress, aging, or disease [37,[50][51][52][53]. Conversely, cells overexpressing i-proteasome regulatory complex PA28 exhibit enhanced degradation of oxidized and misfolded proteins [54,55]. Thus, the presence of iproteasome provides protection from the proteotoxic accumulation of damaged proteins.
In summary, results from this study show substantial differences in regulation of the NFkB pathway that were unique to cells deficient in either the LMP2 or the LMP7 and MECL catalytic subunits of the i-proteasome. As a caveat, our results may be specific to cultured RPE cells and therefore, studies in different cells from KO mice are essential to determine the applicability of these results. Never the less, these data provide unequivocal evidence that i-proteasome performs functions that go beyond its role in generating peptides for antigen presentation and includes a central role in modulating NFkB signaling. Regulation may occur through degradation of proteins directly involved in the signaling cascade, or by an indirect mechanism via degradation of regulatory molecules or modulators of other pathways that crosstalk with NFkB. The presence of multiple signaling pathways that coordinate to convey compensatory effects help ensure cell viability is maintained. This strategy is likely to apply to the functions performed by the i-proteasome since mice with genetic ablation of all three i-proteasome subunits are viable [50]. It is possible that i-proteasome plays a role in fine-tuning the rate of signaling or downstream events of not only NFkB, but other signaling pathways. Considering that NFkB is the main pathway for responding to stress, our results firmly establish the iproteasome at the center of the response to stress and injury. Figure S1 Overview of NFkB signaling following TNFa stimulation. (A) In the Classical Pathway of NFkB activation, TNFa binding to TNF receptor (TNFR) triggers receptor trimerization and recruitment of adaptor proteins such as RIP1 (receptor interacting protein) to the intercellular domain of TNFR. RIP1 is modified by K63-linked ubiquitin and contributes to the recruitment, phosphorylation, and activation of the IkK complex (c, a, b). Ikk phosphorylates the inhibitory protein IkBa, leading to ubiquitination and subsequent degradation by the proteasome.-Proteolysis of IkBa releases the p50/p65 dimer and allows it to bind to the facilitator protein Ribosomal Protein S3 (RPS3). Following nuclear translocation, RPS3 enhances p65 binding on the NFkB promoter of target genes, including inos, cox2, il-6, ikba, a20, nfkb1, nfkb2, and relb. The Alternative Pathway involves activation of IkKa, which phosphorylates p100 and p105 and initiates degradation of the inhibitory portion of p100 and p105 to generate p52 and p50, respectively. The prototypic binding partner of p52 is RelB.p50 can form heterodimers with p65 and take part in the Classical Pathway. Alternatively, p50/p50 homodimers can form, translocate into the nucleus and inhibit transcription. (B) Negative feedback includes transcription of several regulatory proteins that terminate NFkB signaling. Newly synthesized IkBa facilitates removal of p65/p50 from the nucleus and sequesters this dimer in the cytoplasm. Production of p100 also inhibits p65/p50 nuclear translocation. Expression of a20 produces a dual functional enzyme that replaces K63-linked ubiquitin with K48-linked ubiquitin on RIP1, which target it for proteasomal degradation. A loss of RIP1 ends NFkB activation. (TIF) Figure S2 Proteasome content in Cerebellum. Western blots showing reactions for proteasome subunits in cerebellar homogenates harvested from WT and i-proteasome deficient mice (L7M1 and L2). Proteins loads were 5 mg per lane for a7, b1, and b5. Protein loads were 35 mg per lane for LMP2 and LMP7. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as a loading control. The 20 s reaction was used as a positive control. Graph summarizes proteasome content in the cerebellum of WT and i-proteasome deficient mice (n = 3/group). One-way ANOVA results, *p = 0.01. (TIF)