Enhancement of proteasome function by PA28α overexpression protects against oxidative stress

Abstract

The principal function of the proteasome is targeted degradation of intracellular proteins. Proteasome dysfunction has been observed in experimental cardiomyopathies and implicated in human congestive heart failure. Measures to enhance proteasome proteolytic function are currently lacking but would be beneficial in testing the pathogenic role of proteasome dysfunction and could have significant therapeutic potential. The association of proteasome activator 28 (PA28) with the 20S proteasome may play a role in antigen processing. It is unclear, however, whether the PA28 plays any important role outside of antigen presentation, although up-regulation of PA28 has been observed in certain types of cardiomyopathy. Here, we show that PA28α overexpression (PA28αOE) stabilized PA28β, increased 11S proteasomes, and enhanced the degradation of a previously validated proteasome surrogate substrate (GFPu) in cultured neonatal rat cardiomyocytes. PA28αOE significantly attenuated H₂O₂-induced increases in the protein carbonyls and markedly suppressed apoptosis in cultured cardiomyocytes under basal conditions or when stressed by H₂O₂. We conclude that PA28αOE is sufficient to up-regulate 11S proteasomes, enhance proteasome-mediated removal of misfolded and oxidized proteins, and protect against oxidative stress in cardiomyocytes, providing a highly sought means to increase proteasomal degradation of abnormal cellular proteins.—Li, J., Powell, S. R., Wang, X. Enhancement of proteasome function by PA28α overexpression protects against oxidative stress.

The ubiquitin proteasome system (UPS) mediates the most important proteolytic pathway for the degradation of abnormal and most normal intracellular proteins. It includes two essential steps: polyubiquitination of a specific protein molecule and subsequent degradation of the ubiquitinated protein by the 26S proteasome (26S) (1–6). Structurally, the 26S consists of two subcomplexes: a 20S core particle (20S) and the 19S regulatory particle (19S) attached to one or both ends of the 20S (3). Besides 19S, the 11S proteasome activator (11S) can also regulate protein degradation by the 20S. The 20S can be capped by 19S or 11S at both ends (19S-20S-19S or 11S-20S-11S) or by 19S at one end and 11S at the other (19S-20S-11S, hybrid proteasomes) (7, 8). The 11S-associated 20S is found in many cell types (7–10), including cardiomyocytes (10, 11). The 11S is also known as proteasome activator 28 (PA28) or REG and can be formed by PA28α and PA28β in heteroheptamers (α3β4 or α4β3) or by PA28γ in homoheptamers (γ7) (12, 13). The 11S activates 20S differently from 19S and does not seem to need ATP for binding or activation (14, 15). Binding of an 11S to the 20S appears to increase 20S proteolytic capacity without affecting overall catalytic subunit activity. This is thought to occur as a result of insertion of the carboxyl-terminal tails of the PA28 subunits into pockets of the 20S, resulting in conformational changes in its α subunits. This, in turn, opens the access channel to a greater degree, enhancing access to the catalytic chamber (16, 17). PA28 subunits are induced by interferon-γ, and the 11S activator ring is thought to play a role in antigen presentation, although knockdowns have very little effect on antigen presentation (18–20). It has been suggested that the 11S may play a greater role in intracellular protein degradation than in antigen processing (14, 21). This is based partly on observations of attenuated ATP-dependent degradation of ornithine decarboxylase in PA28α/PA28β-knockout mice (15). Further, PA28γ knockout in mice has very limited effects on general antigen presentation, although the mice did show growth retardation and a shorter life span (20, 22). In recent relevant studies, it was observed that PA28γ degrades steroid receptor coactivator SRC-3 and p21 in a ubiquitin- and ATP-independent manner (14, 23, 24).

To facilitate the in situ assessment of UPS proteolytic function in the cell or organs, several fluorescence proteins have been engineered as specific substrates of the UPS. One of the commonly used proteins is created by fusion of degron CL1 to the carboxyl terminus of the enhanced green fluorescence protein (GFP) and is known as GFP^u, GFPu, or GFPdgn (25, 26). Like a misfolded protein, degron CL1 appears to signal for ubiquitination through surface exposure of hydrophobic residues (27). Proteins carrying CL1, such as GFPu/GFPdgn, can serve as surrogates of misfolded proteins, which are degraded by the UPS (5). Hence, changes in the degradation rate of GFPu/GFPdgn in the cell may arguably reflect the ability of the UPS to degrade misfolded proteins, a pivotal part of protein quality control in the cell.

Proteasome functional insufficiency (PFI) has been observed in several animal models of human cardiovascular disorders and has been implicated in human cardiomyopathies (4, 28–33). More recently, PFI was shown to activate a major signaling pathway of cardiac pathological hypertrophy and facilitate maladaptive remodeling of stressed hearts (34). Interventions that normalize cardiac proteasome function could be extremely valuable for defining the pathophysiological significance of cardiac PFI and also in development of new therapeutic strategies. We have previously observed significant up-regulation of PA28α and PA28β in an experimental rat diabetic cardiomyopathy model (10). Aside from this past study, the role of 11S in cardiac pathophysiology has not been explored. Therefore, we investigated whether up-regulation of 11S can alter the overall proteolytic function of proteasomes in cardiomyocytes. Using a gain-of-function approach and taking advantage of a well-established UPS functional reporter, we tested the hypothesis that forced PA28α overexpression (PA28αΟE) can enhance proteasome-mediated removal of abnormal proteins in cardiomyocytes, via stabilizing PA28β and up-regulating 11S proteasome activators, thereby protecting against oxidative stress.

MATERIALS AND METHODS

Neonatal rat ventricular myocyte (NRVM) culture and recombinant adenoviral infection

These were carried out as described previously (35). Adenoviruses expressing PA28α (Ad-PA28α) and PA28β (Ad-PA28β) were newly constructed (35). Adenoviruses expressing β-galactosidase (Ad-β-gal), GFPu (Ad-GFPu), or PTEN (Ad-PTEN) were described previously (35, 36).

SDS-PAGE and Western blot analysis

Cells were lysed using 1× loading buffer (50 mM Tris-Cl at pH 6.8, 2% SDS, and 10% glycerol) for the preparation of total proteins (28). Immunoblots were performed using primary antibodies against PA28α (PW 8185; Affiniti Research Products, Devon, UK), PA28β (PW 8240; Affiniti Research Products), α-actinin (A7811; Sigma, St. Louis, MO, USA), RPN2 (PW 9256; Biomol, Plymouth Meeting, PA, USA), RPT-6 (PW 9265; Biomol), GATA4 (sc-9053: Santa Cruz Biotechnology, Santa Cruz, CA, USA), AKT (9272; Cell Signaling, Beverly, MA, USA), PTEN (9559; Cell Signaling), GFP (sc-9996; Santa Cruz Biotechnology), proteasome subunit β5 (anti-PSMB5; customized antibody; ref. 28), or 20S proteasome core subunits (this antibody recognizes 20S α5, α7, β1, β5, β5i, and β7 subunits; PW 8155; Biomol).

The commercially sourced antibodies against PA28α and PA28β were generated in rabbits against synthetic peptides specific for PA28α or PA28β. Specificity was verified by 2-D gel electrophoresis followed by sequential Western blot analyses of the same transfer blot, which revealed distinct isoelectric focusing migration spots for PA28α and PA28β (pI=5.37 and 5.08, respectively) which were not overlapping (data not shown).

Reciprocal immunoprecipitation (co-IP)

Cultured NRVMs or mouse ventricular tissues were homogenized in the radioimmunoprecipitation assay (RIPA) buffer. The supernatants were collected and incubated with anti-PA28α or anti-PA28β antibody and protein A/G conjugated agarose beads (protein A/G PLUS-agarose immunoprecipitation reagent; sc-2003; Santa Cruz Biotechnology) overnight at 4°C. Precipitated products were then fractionated by SDS-PAGE and detected by immunoblots for PA28α and PA28β.

Indirect immunofluorescence labeling and confocal microscopy

These techniques were performed as described previously (37). The primary antibodies used were anti-PA28α (customized), anti-PA28β (sc-23642; Santa Cruz Biotechnology), and anti-α-actinin (A7811; Sigma). Fluorescence-conjugated secondary antibodies were purchased from Molecular Probes (Carlsbad, CA, USA).

Transcript analysis

Total RNA was extracted from cultured cardiomyocytes or mouse ventricles using TRI Reagent (Molecular Research Center, Cincinnati, OH, USA). Total RNA transcript levels of PA28α or PA28β were examined with Northern blot analyses using P32-labeled PA28α or PA28β probes generated with the nick translation kit (Roche, Indianapolis, IN, USA). The mRNA levels of the fetal gene program and GFPu transcript levels were measured by RNA dot-blot analyses, as described previously (25, 37). Semiquantitative RT-PCR was carried out using specific primers toward GFPu, PA28α, and PA28β for their transcript levels. Relative transcript levels were obtained with normalization to 18S ribosome RNA or GAPDH transcript levels.

Cycloheximide chase

Cardiomyocytes were cultured in DMEM supplemented with protein synthesis inhibitor cycloheximide (50 μM). Cells were collected at indicated time points for the analysis of protein levels of PA28β by Western blots.

Proteasome peptidase activity assays

Cultured NRVMs were collected in HEPES buffer and processed for peptidase activities as described previously (37). Briefly, 20 μg of cell lysate was used in each peptidase activity reaction. Synthetic substrates Suc-LLVY-AMC (18 μM), Z-LLE-AMC (45 μM), and Ac-RLR-AMC (40 μM) (Biomol) were used to measure the chymotrypsin-like (β5), caspase-like (β1), and trypsin-like (β2) activities, respectively, without and with ATP (Sigma). The ATP concentrations used, which were determined by optimization (38, 39), were 28, 14, and 28 μM for β5, β1, and β2 activities, respectively. The reactions were performed in the presence or absence of specific proteasome inhibitors. Proteasome inhibitor MG132 (20 μM; Biomol) was used in β5 and β1 activities, while epoxomicin (5 μM; Calbiochem, San Diego, CA, USA) was used for β2 activities. The portion of activity that was inhibited by proteasome inhibition is attributed to the proteasome.

Radioactive protein pulse-chase analysis

Cultured NRVMs were coinfected with Ad-GFPu and Ad-PA28α or Ad-β-gal to determine the half-life of GFPu. NRVMs were first starved in methionine/cysteine-free DMEM for 60 min. Cells were then metabolically labeled with 150 μCi/ml 35[S]methionine/35[S]cysteine for 30 min. After being washed, cardiomyocytes were incubated in DMEM supplemented with FBS and 2 mM methionine and cysteine. At the indicated times, cells were collected in RIPA buffer for the immunoprecipitation of GFPu with the anti-GFP antibody and conjugated agarose A/G beads (sc-9996 AC; Santa Cruz Biotechnology). Collected samples were separated by SDS-PAGE, transferred to PVDF membranes, and exposed to a phosphor screen. The amounts of labeled proteins were detected by radioautography.

Creation of HEK293 GFPu/RFP stable cell lines

GFPu/RFP stable cell lines were generated as described previously for the GFPu-2 stable cell lines (40). Briefly, HEK293 cells from the American Type Culture Collection (ATTC; Manassas, VA, USA) were cotransfected with pGFPu and pDSRed2-C1 plasmids using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA). The pDSRed2-C1 encoding a red fluorescence protein (RFP) was purchased from Clontech (Mountain View, CA, USA). The pGFPu and pDSRed2-C1 plasmids share exactly the same regulatory elements in the vector (26).

Gel filtration

HEK293 cell lysates were fractioned by gel filtration, as described previously (37).

Protein carbonyl assay and vacuum-assisted protein dot blotting

Protein carbonyls were assessed using the OxyBlot protein oxidation detection kit (S7150; Millipore, Billerica, MA, USA). Briefly, cardiomyocytes were collected and lysed in the RIPA buffer. After centrifugation, the supernatant was collected and supplemented with DTT (50 mM, final concentration). Protein samples were then mixed with the same volume of 12% SDS and incubated with an equal volume of the 1× dinitrophenylhydrazine (DNPH) derivatization solution at room temperature for 15 min before reaction termination on addition of the neutralization solution. The DNPH-tagged proteins were then used for SDS-PAGE/Western blot or loaded directly onto a PVDF membrane via vacuum-assisted dot blotting using standard techniques. An anti-DNP antibody was used for detection of the DNPH-derivatized proteins.

TUNEL assay

Cultured cardiomyocytes on coverslips were fixed with 4% paraformaldehyde in PBS for 20 min and then permeabilized with 0.1% Triton X-100 in PBS for 15 min at room temperature. The label solution and enzyme solution from the in situ cell-death detection kit (11684795910; Roche) were mixed to obtain the TUNEL reaction mixture immediately before use, according to the manufacturer's instructions. TUNEL reaction mixture was spread across the cell layer, and then the coverslips were incubated at 37°C for 60 min in the dark. Finally, cardiomyocytes were stained by Alexa Fluor 565-conjugated phalloidin, and the nuclei were stained by DAPI. Samples were further analyzed under a fluorescence microscope (Zeiss Axiovert 200A; Carl Zeiss, Oberkochen, Germany) using an excitation wavelength of 488 nm. Fluorescence images were taken 5 times on each coverslip randomly. The number of TUNEL-positive nuclei (green) and the number of total nuclei (blue) within cardiomyocytes (red cytoplasm) were counted using Image-Pro Plus software (Media Cybernetics, Inc., Bethesda, MD). A total of >2000 cardiomyocytes/group was counted; data are presented as the number of TUNEL-positive nuclei per 1000 cardiomyocytes.

DNA laddering assay

Genomic DNA of cultured NRVMs was extracted with the phenol/chloroform method. The PCR kit for DNA laddering assay (Maxim Biotech, South San Francisco, CA, USA) was used to assess the degree of DNA fragmentation. The ligation step and subsequent PRC reaction were performed by following manufacturer's instructions.

Statistical analysis

Unless indicated otherwise, all quantitative data are presented as means ± sd. Where applicable, Student's t test and 1-factor or multiple-factor analyses of variance (ANOVAs) were performed for statistical significance tests using SigmaStat 3.0 software (Systat Software, Inc., San Jose, CA, USA). The Holm-Sidak test was used for post hoc comparisons. Values of P < 0.05 were considered to be statistically significant.

RESULTS

PA28αOE up-regulates 11S proteasome activators in cultured NRVMs

To explore the pathophysiological significance of PA28 up-regulation, we used a gain-of-function approach using cultured cardiomyocytes and recombinant adenovirus-mediated gene delivery. To achieve PA28αOE, cultured NRVMs were infected with Ad-PA28α. As expected, PA28α protein levels were increased in a dose-dependent manner by Ad-PA28α infection (Fig. 1A). Interestingly, the increase of PA28α was accompanied by increased PA28β protein (Fig. 1A). Further analysis showed a linear correlation between increased PA28α and PA28β protein levels (r=0.93, Fig. 1B). Co-IP revealed that PA28α-bound PA28β and PA28β-bound PA28α were markedly increased in Ad-PA28α-infected cells, compared with the control cardiomyocytes (Fig. 1C), indicating that PA28αOE increases the amount of 11S in NRVMs. Immunofluorescence microscopy showed that the PA28α and PA28β colocalized in both the cytoplasm and nucleus of the PA28αOE cardiomyocytes and are distributed in the same pattern as the endogenous counterparts in the control cells (Fig. 1D). To dissect how PA28αOE caused the increase of PA28β, we first examined whether steady-state levels of PA28β RNA were increased. The transcript levels of PA28β in PA28αOE cells and in the control cells did not differ significantly (Fig. 2A, B), suggesting that up-regulation of PA28β by PA28αOE is post-transcriptional. Indeed, further testing using a cycloheximide chase assay revealed that the degradation rate of PA28β protein was decreased by PA28αOE in NRVMs (Fig. 2C, D).

Figure 1.

PA28αOE in cultured NRVMs up-regulates 11S proteasomes. A) Protein levels of PA28α and PA28β in cultured NRVMs after Ad-β-Gal or Ad-PA28α infection were analyzed by immunoblotting. Lane 1 is from Ad-β-Gal (β-galactosidase, 20 MOI)-infected control cells. α-Actinin was probed as a loading control. B) Correlation analysis of the protein levels of PA28α and PA28β detected in panel A. C) NRVMs were infected with Ad-β-Gal (PA28αΟE−) or Ad-PA28α (PA28αΟE+) at 20 MOI. Association of PA28α and PA28β was analyzed by immunoprecipitation (IP) followed by immunoblots. D) Confocal micrographs of NRVMs immunofluorescence-stained for PA28α (green) and PA28β (red). Superimposition (yellow) indicates colocalization of PA28α and PA28β in the cell. Sarcomeric α-actinin (blue) was labeled to identify cardiomyocytes. Imaging gain used for the control (CTL)-group images was 2-fold of that used for the Ad-PA28α group. AU, arbitrary units.

Figure 2.

PA28αOE stabilizes PA28β protein in NRVMs. A, B) Northern blot analyses of PA28α (A) and PA28β (B) in NRVMs. In-lane 18S rRNA images were included as loading controls. C, D) Degradation of PA28β was analyzed using cycloheximide (CHX) chase. C) PA28β protein levels for the control group (Ad-β-Gal) and PA28αOE group (Ad-PA28α) at different time points after CHX treatment were analyzed by immunoblotting. D) Densitometric data of PA28β were calibrated with the same-lane α-actinin. The PA28αOE elongated the half-life of PA28β from 14 h in control cells (triangles) to 79 h under PA28α OE (diamonds). AU, arbitrary units.

Up-regulation of 11S enhances UPS function in cultured NRVMs

To investigate the effect of PA28αOE on proteasome function, we performed conventional in vitro proteasome peptidase activity assays using the cell lysates from cultured NRVMs. Compared with the Ad-β-Gal-infected control cells, PA28αOE induced a modest but statistically significant increase in the ATP-independent proteasome chymotrypsin-like activities but showed no effect on caspase-like and trypsin-like activities. As expected, ATP significantly increased all three peptidase activities in both the PA28αOE and control groups. ATP-dependent chymotrypsin-like and caspase-like activities of the PA28αOE group were significantly higher than those of the control group, although no difference was detected in the trypsin-like activities (Fig. 3). Given the potential negative effect of ATP on proteasome peptidase activities (38, 39), the ATP concentrations used for these assays were determined by optimization. Results of these in vitro assays suggest that PA28αOE may have a positive effect on proteasome proteolytic function.

Figure 3.

Effects of PA28α OE in cultured NRVMs on proteasome peptidase activities. At 48 h after Ad-PA28α or Ad-β-Gal infection, cultured NRVMs were collected, and cell lysates were extracted for in vitro proteasome peptidase activity assays. Chymotrypsin-like (β5), caspase-like (β1), and trypsin-like (β2) activities were assayed using specific fluorogenic substrates (see Materials and Methods) with or without ATP; concentrations of ATP were 28, 14, and 28 μM, respectively. For each peptidase activity reaction, 20 μg of cell lysate was used. Portion of activity that was inhibited by proteasome inhibition is attributed to the proteasome. Proteasome inhibitor MG132 (20 μM) was used in β5 and β1 activities; epoxomicin (5 μM) was used for β2 activity assessments. Inclusion of ATP (+ATP) in the reaction system significantly increased the activities of all three peptidases in both groups, compared to the system without ATP (−ATP) (P<0.05 or 0.01). AU, arbitrary units; CTL, control. *P < 0.05, Student's t test.

A surrogate substrate for the UPS was previously created by carboxyl fusion of degron CL1 to GFP and referred to as GFPu (35, 41). Conventional GFP is not an efficient substrate for the UPS, whereas GFPu has a much shorter half-life, and its degradation in the cell is proteasome dependent (26). Therefore, in the absence of a change in its synthesis, GFPu protein level changes reflect inversely the dynamic change in UPS proteolytic function in the cell (5). To further determine whether PA28αOE enhances protein degradation in intact cells, we employed the GFPu reporter system. NRVMs were coinfected with Ad-GFPu and Ad-β-Gal or Ad-PA28α at a dose ranging from 1 to 20 MOI. At 48 h after the viral infection, the cells were harvested for extraction of total proteins or total RNA. Western blot analysis revealed that GFPu protein levels decreased with PA28αOE in a dose-dependent manner (Fig. 4A), and GFPu protein levels negatively correlated with the protein levels of PA28α and PA28β (Fig. 4B, P<0.05). Steady-state GFPu transcript levels were not significantly changed by the PA28αOE (Fig. 4C, D), but rather, the half-life of GFPu protein was significantly shortened from 12 min in the control cells to 4 min in PA28αOE cardiomyocytes, as determined by the radioactive protein pulse-chase assay (Fig. 5). The radioactive pulse-chase assay was done twice. In both runs, PA28αOE consistently shortened GFPu half-life by a factor of 3. These results provide compelling evidence that PA28αOE enhances proteasome proteolytic function in cultured cardiomyocytes.

Figure 4.

Effects of PA28α OE on GFPu protein and transcript levels in cultured NRVMs. NRVMs were coinfected with Ad-GFPu at 20 MOI and either Ad-β-Gal (CTL) or Ad-PA28α. Total proteins or RNA were extracted from the cells 48 h after the viral infection for subsequent analyses. A) Compared to CTL infection (lane 1), PA28αOE markedly decreased GFPu protein levels (lanes 2–5) in a dose-dependent manner. B) GFPu protein levels negatively correlate with the protein levels of PA28α or PA28β. C, D) RNA dot-blot images (C) and a summary of the densitometric data (D) of GFPu in the PA28αOE and the CTL groups. GAPDH was used as a loading control. AU, arbitrary units; CTL, control; N.S., no significant difference (n=3/group; Student's t test).

Figure 5.

Pulse-chase assays of the GFPu protein in cultured NRVMs. A) Representative radioautograph of the GFPu protein radioactive pulse-chase assay. Radioactively pulse-labeled GFPu was immunoprecipitated at the indicated chase time points in the control (CTL) group and the PA28αOE group. B) Decay in relative GFPu radioactivity, plotted against chasing time. Representative GFPu half-life (t_1/2) values derived from the pulse-chase are also presented.

We then tested whether PA28αOE affected the homeostasis of endogenous proteins by examining the steady levels of bona fide UPS substrate proteins (e.g., GATA4, AKT, and PTEN). Interestingly, PA28αOE did not alter the abundance of any of these bona fide endogenous UPS substrates (Fig. 6A, B). We further tested whether PA28αOE selectively enhanced the degradation of transgenic proteins, because the reporter protein GFPu is transgenic. The same MOI of Ad-PTEN gave rise to the same PTEN protein levels in the cells without PA28αOE and in those with various degrees of PA28αOE (Fig. 6C, D), indicating that PA28αOE does not alter the stability of a transgenic wild-type protein.

Figure 6.

Endogenous proteasome substrate content is not affected by PA28αOE in NRVMs. A, B) Western blot images (A) and densitometry data (B) of the abundance of some of the known proteasome substrates. Protein levels of GATA4, AKT, and PTEN show no significant difference between the control and PA28α OE cardiomyocytes; n = 4. *P < 0.001; Student's t test. C, D) NRVMs were coinfected with Ad-PTEN (20 MOI; lanes 2–4) and Ad-β-Gal (20 MOI; lane 2) or Ad-PA28α (2, 5, and 20 MOI; lanes 3–5) for 2 d. Compared with coexpression of β-Gal, coexpression of incremental doses of PA28α did not decrease protein levels of overexpressed PTEN. P > 0.05, ANOVA analysis from 3 repeats. AU, arbitrary units; CTL, control.

PA28αOE did not alter the abundance of 19S and 20S proteasomes but increased 11S-associated 20S proteasomes in the cell

To determine whether the enhanced proteasome function is caused by an up-regulation of 19S and/or 20S subcomplexes, we determined the effect of PA28αOE on the abundance of 19S and 20S representative subunits in the NRVMs using Western blot analyses (Fig. 7). Compared to the respective controls, PA28αOE did not cause significant changes in RPN2 and RPT6 subunits of the 19S or in the abundance of the core subunits of the 20S, suggesting that 19S and 20S abundance is not altered by PA28αOE. To examine the effect of PA28αOE on proteasome function in another type of cell, we created clonal HEK293 cell lines that stably express both GFPu and RFP under the same expression control elements, known as GFPu/RFP double-stable cell lines. Transient transfection of a GFPu/RFP double stable cell line (clone 19) with a mammalian expression vector (pShuttle-CMV) harboring PA28α or the empty pShuttle-CMV as the transfection control revealed that PA28αOE enhanced the degradation of GFPu protein but not RFP in the cell (Fig. 8A). To determine whether PA28αOE alters the proteasome subcomplex distribution, the crude protein extracts from PA28α- or control (β-Gal) plasmid-transfected HEK293 cells were fractionated by gel filtration, which separates native protein complexes by size, with the larger sizes eluted earlier. The size distribution of 19S, 20S, and 11S was assessed by measuring the abundance of their representative subunits, RPN2, PsmB5, and PA28β, respectively, in the sequential fractions of the eluted filtrate. Note that PA28αOE significantly increased the amount of PA28 subcomplexes, as shown by marked increases in PA28β in the complex form (Fig. 8B). Notably, it was also observed that the presence of the 20S proteasome (as marked by PsmB5) in larger protein complexes (fractions 13–17) decreased, but its presence in a relatively smaller protein complex (fractions 18–21) increased in PA28α-transfected cells, compared to control cells. This is consistent with 11S-associated 20S being smaller than 19S-associated 20S due to the size differential between 11S and 19S. These results suggest that PA28αOE increases 11S-associated 20S proteasomes.

Figure 7.

PA28αOE does not alter 19S and 20S proteasome abundance in cardiomyocytes. Western blot images (A) and densitometry data (B) of the protein levels of RPN2 and RPT6 of the 19S and the core subunits of 20S along with PA28α in cultured NRVMs. 19S- and 20S-subunit protein levels were not changed by PA28αOE. AU, arbitrary units. *P < 0.001 vs. control (CTL); Student's t test (n=4).

Figure 8.

PA28α overexpression enhances proteasome proteolytic function in noncardiomyocytes. A) Representative images of Western blot analyses for the indicated proteins in HEK 293 cells stably expressing GFPu and RFP at 48 h after transient transfection of a PA28α expression vector (PA28α) or the empty vector control (pShuttle-CMV). Overexpression of PA28α dose-dependently decreased GFPu but not coexpressed RFP. B) Western blot analyses following gel filtration of crude native proteins extracted from PA28α- and empty plasmid vector (CTL)-transfected cultured HEK293 cells. Antibodies against Rpn2, PsmB5, and PA28β were employed to probe the 19S, 20S, and 11S subcomplexes, respectively, in different elution fractions. Fraction numbers are indicated at top. Size of protein complexes decreases as fraction number increases. CTL, control.

PA28αOE protects against oxidative stress

To examine the effect of PA28αOE on removal of endogenous abnormal proteins generated under pathological conditions, cultured NRVMs were treated with H₂O₂ (50 μM) followed by comparison of protein carbonyl levels in PA28αOE and control cells. As expected, exposure to H₂O₂ significantly increased protein carbonyl levels in the cell. PA28αOE markedly attenuated H₂O₂-induced increases of the oxidized protein levels, especially in the molecular mass range between 50 and 220 kDa (Fig. 9).

Figure 9.

PA28αOE attenuates H₂O₂-induced increases in oxidized proteins in cultured NRVMs. Cultured NRVMs were infected with Ad-PA28α or Ad-β-Gal for 2 d and then treated with 50 μM H₂O₂ or vehicle (water) for 8 h. Protein carbonyls in cell lysates were marked by DNPH derivatization and probed by an anti-DNP antibody using quantitative protein dot blot (A, B) and Western blot (C) analyses. PA28αOE remarkably attenuated H₂O₂ induced increases of the oxidized protein levels (A, B), especially in the molecular mass range between 50 and 220 kDa (C).

To further assess the biological relevance, the incidence of apoptosis using both TUNEL assay and DNA laddering was determined. PA28αOE significantly reduced the number of TUNEL-positive NRVMs both at baseline conditions and during H₂O₂ stimulation (Fig. 10A, B). DNA fragmentation is a hallmark of apoptosis and can be detected through analysis of DNA laddering. Exposure to H₂O₂ markedly increased DNA laddering in cultured NRVMs, which was attenuated by PA28αOE (Fig. 10C). Taken together, these data strongly indicate that PA28αOE enhances the removal of oxidized proteins and protects against apoptosis induced by oxidative stress.

Figure 10.

PA28αOE protects cultured NRVMs from undergoing apoptosis. NRVMs were cultured and treated as described in Fig. 9. Apoptosis in cultured NRVMs was assessed via TUNEL assays and DNA laddering assays under both baseline conditions and during H₂O₂ stimulation. A) Representative photographs of TUNEL labeling. B) Summary of quantitative TUNEL assay data. TUNEL-positive cardiomyocytes (green) were counted and expressed as per 1000 nuclei (DAPI staining, blue). Cardiomyocytes were identified by Alexa Fluor 565-phalloidin staining (red). Results of 4 experimental repeats are summarized. In the absence of H₂O₂ treatment, the PA28αOE group showed fewer apoptotic cells, compared with the control Ad-β-Gal-infected cells. H₂O₂ treatment significantly increased TUNEL-positive cells, but the increase was markedly attenuated by PA28αOE. *P < 0.05 vs. control (CTL) Ad-β-Gal group; ^#P < 0.01 vs. H₂O₂-treated Ad-β-Gal group; 1-way ANOVA. C) Representative image of PCR-based DNA-laddering assays. Genomic DNA was extracted from cultured cardiomyocytes. Equal amounts of total genomic DNA from different samples were used for the ligation (500 ng/reaction) and subsequent PCR amplification (20 ng/reaction). MM, DNA molecular size markers; bp, base pairs.

DISCUSSION

It is generally believed that the 26S proteasome or 19S-associated 20S proteasomes are uniformly responsible for the targeted degradation of soluble polyubiquitinated proteins (42). However, heterogeneities in the subcomplex distribution and functionality of the proteasome complex (7, 8, 43, 44), in the subunit composition of a subcomplex (11, 45, 46), and in the biochemical property of a subunit have begun to be revealed (45, 47, 48). It becomes increasingly clear that there is more than one population of proteasomes in the cell and that proteasomes at different subcellular locations or at same location but different times may differ significantly in terms of biochemical and biophysical properties. Using cultured cells, the present study has demonstrated for the first time that forced PA28αOE is sufficient to up-regulate the 11S proteasome activator in cardiomyocytes where the overexpressed PA28α binds and stabilizes its partner, PA28β. By monitoring the degradation of a surrogate misfolded protein substrate, we have also discovered that PA28αOE enhances proteasome proteolytic function in the cell and protects cardio-myocytes from being damaged by oxidative stress. Given the prevalence of PFI in heart disease (1, 5), this new discovery provides a highly sought means to enhance cardiac proteasomal activity.

The 11S can be either a homoheptamer of PA28γ or a heteroheptamer of PA28α and β (49). Induction of the synthesis of PA28α and PA28β by interferon-γ has been observed in both immune and nonimmune cells (7, 12). In the present study, we found that up-regulation of 11S can be achieved by forced overexpression of PA28α in both NRVMs and HEK293 cells. We hypothesize that the concurrent up-regulation between PA28α and PA28β is mediated by mutual stabilization at the protein level since transcript levels of PA28β were not significantly changed during forced PA28αOE, and as shown previously and in the present study, these proteins interact with each other. The proposed hypothesis is further supported by the significant decrease in the degradation rate of PA28β during PA28αOE in NRVMs. PA28α and PA28β interaction may prevent them from being recognized by the degradation machinery and thus result in mutual stabilization. Consistent with this notion, it was previously reported that no PA28α proteins were detected in PA28β-knockout mice (50).

Damaged or misfolded proteins usually undergo conformational changes and expose their hydrophobic sequences, which are believed to serve as a signal to activate ubiquitination. GFPu was engineered by carboxyl fusion of degron CL1 to GFP. Degron CL1 signals for ubiquitination via the surface-exposed hydrophobic structure of its predicted amphipathic helix (27, 51). GFPu is, hence, considered a surrogate of misfolded proteins (42, 52), serving as a full-length protein negative reporter for UPS proteolytic function. In the present study, when GFPu was expressed in the NRVMs or HEK293 cells, PA28αOE decreased GFPu protein levels in a dose-dependent manner via a post-transcriptional mechanism since GFPu transcript levels were not altered, and the half-life of GFPu protein was significantly shortened. These findings suggest that up-regulation of 11S, at least those formed by PA28α and PA28β, can enhance the degradation of misfolded proteins in cardiomyocytes. Consistent with an enhancement of proteasome proteolytic function by PA28αOE, both ATP-independent and dependent proteasome chymotrypsin-like activities and ATP-dependent proteasome caspase-like activities were higher in the cell extracts from PA28αOE NRVMs than that from the CTL cells. Notably, the direct effect of 11S on 20S activities is not, but the direct effect of 19S on 20S is, ATP-dependent; therefore, the enhancement of ATP-dependent proteasome peptidase activities by PA28αOE suggests that PA28αOE might have increased the abundance hybrid proteasomes (11S-20S-19S), where the association of an 11S and a 19S with the same 20S could enhance the effect of the 19S over the 20S. Indeed, our gel filtration data showed an increase in the hybrid proteasomes in PA28αOE cells.

To be degraded by the proteasome, a mature normal protein molecule, in general, must be first polyubiquitinated (2), which is generally considered to be the rate-limiting step. This is because post-translational modifications to normal protein substrates and/or the accessibility of their specific E3 ligases are often the prerequisite for triggering the ubiquitination of the normal proteins. Hence, enhancement of proteasome proteolytic function should not significantly affect the stability of normal cellular proteins. This postulate was tested in the present study and supported by our findings that the abundance of representative known normal protein substrates of the UPS, and even the steady-state level of a transgenic wild-type protein, was not affected by PA28αOE in cardiomyocytes. Thus, the increase of proteasome proteolytic function by the 11S up-regulation only enhances the removal of abnormal proteins but has little effect on the turnover of normal proteins. Consistent with this overall concept discussed, we observed that PA28αOE can significantly decrease the rate of apoptosis in cultured NRVMs at both baseline condition and during H₂O₂ stimulation, suggesting that PA28αOE can protect against oxidative stress. Increased production of oxidized proteins is one of the mechanisms underlying the detrimental effects of oxidative stress. The protection exerted by PA28αOE is likely through facilitating removal of oxidized proteins as the increase in protein carbonyls induced by H₂O₂ exposure was nearly completely abolished.

The mechanism by which PA28αOE enhances UPS-mediated degradation of abnormal proteins is not clear at this time, but it is likely that the 11S particle formed by PA28α and PA28β works as an alternative activator for 20S and increases proteasome proteolytic activities directed at abnormal or denatured/misfolded proteins. It is known that the 20S core can be capped by 11S at one end and 19S at the other to form a hybrid proteasome or by 11S at both ends (7, 8). Association of the 11S with the 20S stimulates the 20S peptidase activity (7, 8) and this study shows that PA28αOE increased the coexistence of the 11S with the 20S in HEK293 cells without resulting in any detectable changes in the protein levels of representative 19S or 20S subunits. An alternative explanation for the enhanced proteasome-mediated degradation of GFPu is that PA28s work as chaperones helping channel misfolded/unfolded proteins into 20S for degradation. Heat shock protein (Hsp) 90 plays a role in protein quality control by adjusting protein conformation and modulating the UPS (53–55). Hsp90 appears to be necessary for the assembly of 26S but not for the assembly of 20S (54) and is thought to potentially modulate the dynamic exchange of 11S and 19S to affect proteasome function (54). Interestingly, there have been studies showing crosstalk between 11S and Hsp90, as 11S is essential for Hsp90 to refold inactivated luciferase (56). In addition, Hsp90 association with 20S inhibited 20S proteolytic activity, which is effectively reversed by the competitive binding of 11S (54, 57). Collectively, these data suggest that PA28α and PA28β may behave as chaperones that can supervise the protein pool and assist in targeting terminally misfolded proteins for degradation.

The UPS plays an indispensible role in intracellular protein quality control through degradation of misfolded/unfolded and damaged proteins. PFI or protein quality control inadequacy has been observed in animal models of common forms of heart disease and also implicated in failing human hearts (58–60). Protein quality control inadequacy has been proposed as an important pathogenic process in the progression of heart failure and cardiomyocyte senescence (5). Even though PA28 was first identified as a proteasome activator during antigen processing, the 11S formed by PA28γ can mediate degradation of important cell cycle regulators (14). Here, we demonstrated that up-regulation of the 11S via forced PA28αOE may potentially be used as a strategy to intervene with PFI in cardiac disease. It will be interesting and important to test whether forced PA28αOE enhances proteasome proteolytic function in intact animals. Notably, 11S can be up-regulated by interferon-γ (7, 15), a cytokine that has been clinically used to treat human disease (61). It will also be useful to determine whether up-regulation of 11S by genetic or pharmacological means slows down the progression of heart disease in a model that displays PFI.