More 26S and 30S proteasomes are beneficial in proteinopathy

Significance

Until now, 26S and 30S proteasomes have been used together to describe proteasome holoenzymes without recognizing the regulation of their assembly and distinct activities. In this study, we identified a protein that regulates the ratio of 26S/30S and 20S proteasomes in mammalian cells and tissues. We also demonstrated that in mammals, the proteasome assembly process and the role of assembly chaperone may be distinct from that in yeast and that the increase in 26S and 30S proteasomes alleviates proteinopathy in the tau mouse model. This study expands our knowledge of 26S and 30S proteasome assembly and highlights the potential of S5b/PSMD5 as a therapeutic target in proteinopathies.

Abstract

Despite the many studies on the ubiquitin–proteasome system, our understanding of the proteasome itself is limited. The balance and regulation of 26S and 30S proteasomes are not yet known. Here, we show that among the proteasome base assembly chaperones, only S5b/PSMD5 determines the levels of proteasome holoenzymes, especially the 30S proteasome, in mammals. In a variety of cell lines and mouse tissues, we found that apart from its role in yeast 19S proteasome assembly, loss of S5b/PSMD5 increased assembly toward the 26S and 30S proteasomes. During the process, we identified proteasome complexes that may represent alternative assembly intermediates, including the 19S base complex and 20S complexes harboring 19S subunits, eventually leading to a shift in the overall steady-state level of the 26S and 30S proteasomes. Intriguingly, the addition of the S5b/PSMD5 protein in vitro and its increase in cells efficiently disassembled the 30S proteasome into the 20S and 19S complexes. Increase in the 26S and 30S proteasomes over the 20S proteasome enhances the degradation of the aggregation-prone proteins and ubiquitinated proteins in cells and ameliorates cognitive impairment through the reduction of tau pathology in PS19 mice. These results suggest that 26S and 30S proteasomes are manipulated by S5b/PSMD5 and are beneficial for mitigating proteinopathy in mammals.

Proteasomes exist in three main states: 20S, 26S, and 30S. The 20S proteasome, also known as the core particle (CP), contains proteolytic active sites responsible for substrate degradation (1). The 26S and 30S proteasomes comprise one 20S and one 19S regulatory particle (RP) (RPCP) and two RP (RP2CP), respectively. The 19S RP is composed of lid and base subunits, each with functions related to deubiquitination and substrate processing (2). Ubiquitylated proteins are targeted to the 19S RP on the proteasome and deubiquitylated, unfolded, and degraded through the 20S proteasome (3). Recently, the abundance of free 19S proteasome was shown to regulate neuronal synapses (4, 5). Thus, each proteasome may have distinct roles, and cells may adapt by adjusting the ratio of the proteasome complex depending on the cellular context (6–8). Thus, the discriminating roles and assembly of proteasomes are urgently required.

The intricate mechanisms governing proteasome assembly have been previously elucidated. In particular, the assembly of 19S RP was recently revealed by the presence of four base assembly chaperones, Nas2 (p27/PSMD9), Nas6 (p28/PSMD10), Hsm3 (S5b/PSMD5), and Rpn14 (PAAF1), in yeast (and mammals) (9–14). These chaperones independently form precursor assembly modules in the base assembly (15), and their deletion reduces the assembly of 19S, and thereby the activity of the 26S proteasome, exclusively in yeast (10) and partly in mammalian cells (16). Intriguingly, increasing evidence has described a distinct role for S5b in proteasome assembly in mammals. Transgenic overexpression of S5b in mice results in disassembly of the 26S proteasome (17), and its deletion in cancer cells increases the 26S proteasome (18). Moreover, the 26S proteasome currently encompasses both 26S and 30S proteasomes. However, the existence and distinct functions of the 30S proteasome remain unclear (19).

In the present study, we demonstrate that S5b is crucial for determining the level of 26S/30S proteasomes in mammals and that the abundance of the 26S/30S proteasome ameliorates tau proteinopathy by enhancing the degradation of aggregation-prone proteins.

Results

S5b Deficiency Enhances Assembly and Activity of 26S/30S Proteasomes in Mouse Tissues.

To examine the role of S5b in proteasome assembly in mammals and investigate the phenotype of S5b-deficient mice, we generated S5b knockout (KO) mice using the CRISPR-Cas9 system. DNA sequencing revealed a 126 bp deletion in exon 2 of Psmd5, which encodes the S5b protein, causing a frameshift and premature stop codon, resulting in S5b deletion at the protein level in mouse tissues (Fig. 1A). We have not yet observed any noticeable differences in phenotypes, including body weight, litter size, muscle strength, motor coordination and balance, abnormal behaviors such as hyperactivity, impaired learning, and memory, and external physical characteristics, between S5b KO mice and wild-type (WT) mice. We confirmed that S5b deletion did not affect other base assembly chaperones or proteasome subunits in different tissues (Fig. 1A). Interestingly, when we measured proteasomal chymotrypsin-like activity, we found that compared to WT mice, all tissues of S5b KO mice exhibited increased activity, with a less pronounced effect in the liver (Fig. 1B and SI Appendix, Fig. S1 A and B).

Fig. 1.

The 26S/30S proteasomes increase in S5b KO mouse tissues. (A and B) Tissue extracts of WT and S5b KO mice (8-mo-old) were analyzed by western blotting (A) and assayed for chymotrypsin-like peptidase activity (B). Unpaired t test, two-tailed, n = 3. (C−J) Brain (C and D), kidney (E and F), liver (G and H), and muscle (gastrocnemius) (I and J) extracts of WT and S5b KO mice were separated by native-PAGE, analyzed by Suc-LLVY-AMC overlay assays, western blotting, and Coomassie staining (C, E, G, and I). The amounts (quantified by β5) and activities (quantified by LLVY-AMC) of 20S, 26S, and 30S proteasome complexes on the blots were quantified by ImageJ (D, F, H, and J). The asterisk indicates protein complexes highly detected in S5b KO tissues. The arrow indicates S5b module (S5b–PSMC2–PSMC1–PSMD2). The upper band observed in 26S proteasomes represents the Ecm29-bound form, while the lower band represents Ecm29-free form. Unpaired t test, two-tailed, n = 3. Bars represent mean ± SD. N.S, nonsignificant; *P < 0.05 **P < 0.01 ***P < 0.001.

Given that S5b is an assembly chaperone of 19S RP, we examined whether increased activity was linked to assembly. After native-PAGE of the WT and S5b KO mouse tissue extracts, the levels of 30S, 26S, and 20S proteasomes (β5, PSMD2) and their corresponding activities (LLVY-AMC) were measured. Surprisingly, the levels of 30S and 26S proteasomes were increased in S5b KO mouse tissues compared to those in WT tissues. Accordingly, the chymotrypsin-like activities of the 30S and 26S proteasomes increased concomitantly with S5b KO in the corresponding tissues (Fig. 1 C–J). In contrast, 19S RP decreased in all tissues. Notably, we detected two intermediate complexes (IC, asterisk) containing PSMD2 near the 20S proteasome in S5b-deficient tissues (Fig. 1 C, E, G, and I). These results indicated that the levels of 26S/30S proteasomes were increased by S5b deficiency in mouse tissues.

S5b Level Negatively Determines 26S/30S Proteasome Levels in Many Mammalian Cell Types.

To characterize this relationship at the cellular level, we isolated mouse embryonic fibroblasts (MEFs) from WT and S5b KO mice and immortalized them using the SV40 T antigen. Analysis of proteasome complexes revealed that 26S/30S proteasome levels and activities were highly increased in S5b KO MEFs compared to those in WT MEFs (Fig. 2 A and B and SI Appendix, Fig. S3A). We also analyzed proteasome complexes under different conditions and observed reduced levels of the 20S proteasome in S5b KO MEFs (SI Appendix, Fig. S2 A and B). We further analyzed other cell types, including C2C12, HT22, and HeLa, and deleted S5b using the CRISPR-Cas9 system. As observed in MEFs, 26S/30S proteasome levels increased more than 1.5-folds after S5b deletion (Fig. 2 C–E and SI Appendix, Fig. S3 B−G). As observed in mouse tissues, S5b negatively determines the amount and activity of 26S/30S proteasomes at the cellular level, implying that S5b may control the processes governing the levels of 26S/30S proteasomes in mammals.

Fig. 2.

S5b level inversely correlates with 26S/30S proteasome in mammalian cell lines. (A and B) WT and S5b KO MEF cells were separated by native-PAGE, analyzed by Suc-LLVY-AMC overlay assay, western blotting, and Coomassie staining (A) and the amounts (quantified by β5) and activities (quantified by LLVY-AMC) of proteasome complexes were quantified by ImageJ (B). The asterisk indicates protein complexes highly detected in S5b KO cells. The arrow indicates S5b module (S5b–PSMC2–PSMC1–PSMD2). Unpaired t test, two-tailed, n = 3. (C−E) Control and S5b KO C2C12 cells (#1, #6) (C), S5b KO HT22 cells (D), and S5b KO HeLa cells (E) were subjected to native-PAGE, analyzed by Suc-LLVY-AMC overlay assay and western blotting. The amounts and activities of proteasome complexes were quantified by ImageJ. Unpaired t test, two-tailed, n = 3. (F−I) Pearson’s correlation analysis between the expression of S5b (F), PSMD9 (G), PSMD10 (H), or PAAF1 (I) and the amounts of 30S proteasome in human cell lines. (J) HEK293T cells were transfected with 19S base assembly chaperone expression vectors for 24 h, subjected to native-PAGE and analyzed by western blotting. The amounts of proteasome complexes were quantified by ImageJ. Unpaired t test, two-tailed, n = 3. (K) Control and S5b, PSMD9, or PSMD10 KO C2C12 cells were subjected to native-PAGE and analyzed by western blotting, and the amounts of proteasome complexes were quantified by ImageJ. Unpaired t test, two-tailed, n = 3. Bars represent mean ± SD. N.S, nonsignificant; *P < 0.05 **P < 0.01 ***P < 0.001.

To test this hypothesis, we examined the effect of S5b overexpression on the proteasome complex levels in HEK293T cells (SI Appendix, Fig. S3H). We noted that S5b overexpression reduced the levels and corresponding activities of 26S/30S proteasomes but markedly enhanced the 20S proteasome in a dose-dependent manner (SI Appendix, Fig. S3 I and J). We extended our analysis to examine the correlation between S5b and the 26S/30S proteasome in various cell lines. Quantification of proteasome subunits, complexes, and base assembly chaperones revealed that cells expressing low levels of S5b carried high levels of 26S/30S proteasome, and vice versa (SI Appendix, Fig. S4 A and B). Pearson correlation analysis revealed that S5b levels among the base assembly chaperones exhibited a significant inverse correlation with the 30S proteasome (r = 0.90), whereas PSMD9 and PSMD10 displayed some statistically insignificant correlations (Fig. 2 F–I). The S5b levels also exhibited a moderate (r = 0.62), albeit not statistically significant, correlation with 26S proteasome (SI Appendix, Fig. S4C). Thus, S5b is a negative regulator of the 26S/30S proteasome in mammalian cells.

Given that S5b is a base assembly chaperone, all of which are known to function in proteasome assembly in yeast, we probed the effects of other base assembly chaperones on proteasomes. The amount and activity of proteasome complexes were measured in HEK293T cells overexpressing S5b, PSMD9, PSMD10, or PAAF1 (SI Appendix, Fig. S4D). Among these chaperones, only S5b reduced 26S/30S proteasomes and increased 20S proteasomes (Fig. 2J and SI Appendix, Fig. S4 E and F). Moreover, when each chaperone was deleted in C2C12 cells (SI Appendix, Fig. S4G). Except for PAAF1, whose mouse version was not elucidated, the amounts and activities of 26S/30S proteasomes increased, largely due to S5b deficiency (Fig. 2K and SI Appendix, Fig. S4 K and L). As reported previously (10), PSMD10 deletion decreased 30S proteasome levels. Intriguingly, PSMD9 deletion also augmented the levels and activities of 26S/30S proteasomes, albeit to a lesser extent than the S5b deletion. This increase in 26S/30S proteasomes may be due to reduced S5b levels in PSMD9 KO cells (SI Appendix, Fig. S4H). In addition, we observed an increase in PSMD9 levels when PSMD10 was removed (SI Appendix, Fig. S4 I and J). In mammalian systems, distinct from yeast, there appears to be an intricate interplay with the potential for compensatory mechanisms between base assembly chaperones and S5b, potentially assuming a pivotal role in the assembly of 26S/30S proteasomes.

Low S5b Shift the Overall Balance Toward 26S/30S Proteasomes Potentially Via Two ICs.

Base assembly chaperones bind to the C-terminus of specific base subunits to facilitate early stage base (20–22). They also impede the assembly of immature 19S RPs with 20S proteasomes during later stages of 26S/30S proteasome assembly (23). To ascertain whether the increase in 26S/30S proteasomes due to the S5b deletion aligns with these established hypotheses, we conducted in vitro assays using purified proteasomes and assembly chaperones. Intriguingly, only the S5b protein induced the disassembly of holoenzymes, especially 30S, into 19S and 20S proteasomes (Fig. 3 A–E). This distinguishes our findings from earlier observations in yeast in which the S5b ortholog Hsm3 displayed only weak dissociative effects on proteasome holoenzymes (24). We further reaffirmed this role of S5b in proteasome disassembly by characterizing domains or residues, including the R184E mutant, that impede binding between S5b and PSMC2 (17). Deletion mapping analysis revealed that no other binding regions of S5b with PSMC1, PSMD2, or other sites were involved in the proteasome disassembly (SI Appendix, Figs. S5 A−D and S6 A and Bhttp://www.pnas.org/lookup/doi/10.1073/pnas.2422570122#supplementary-materials). As reported previously (16, 25), we confirmed that S5b binds to PSMC2 in mammals, allowing PSMC1 to act as a linker for the binding of PSMC2 to PSMD2 to form the S5b module. Unlike in yeast, S5b is likely to engage in direct binding exclusively to PSMC2 in mammals.

Fig. 3.

S5b deletion tilts overall steady state balance toward proteasome assembly state. (A and B) Purified human proteasomes (26S/30S) (3 μg) were incubated with PSMD9 (10 μg) or S5b protein (5 μg, 10 μg, 20 μg) for 3 h and then subjected to native-PAGE followed by Suc-LLVY-AMC overlay assay or Coomassie staining (A). Each proteasome complex activity was quantified by ImageJ (B). n = 1. (C) Purified human S5b, PSMD9, and PSMD10 proteins (1 μg) were separated by SDS-PAGE and stained with Coomassie blue. (D and E) Purified 26S/30S proteasomes (3 μg) from WT mouse liver were incubated with purified S5b (10 μg), PSMD9 (6.3 μg), or PSMD10 (4.5 μg) proteins for 3 h and then subjected to native-PAGE followed by Suc-LLVY-AMC overlay assay or Coomassie staining (D). The activities of proteasome complexes were quantified by ImageJ (E). Unpaired t test, two-tailed, n = 3. Bars represent mean ± SD. (F) Control and S5b KO C2C12 cells were subjected to native-PAGE and analyzed by western blotting. The asterisk indicates a protein complex highly detected in S5b KO cells. The arrow indicates S5b module (S5b–PSMC2–PSMC1–PSMD2). (G and H) Control and S5b KO C2C12 cells were analyzed by western blotting or by native-PAGE followed by western blotting (G) and the amounts of proteasome complexes were quantified by ImageJ (H). IC indicates protein complexes highly detected in S5b KO cells. The arrow indicates S5b module (S5b–PSMC2–PSMC1–PSMD2). Unpaired t test, two-tailed, n = 3. Bars represent mean ± SEM. (I) Liver extracts of 9-mo-old WT and S5b KO mice were separated by native-PAGE and analyzed by western blotting (PSMD2). Proteasome complexes affinity-purified from 9-mo-old WT and S5b KO mouse livers were subjected to native-PAGE and analyzed by coomassie staining (cooma) or Suc-LLVY-AMC overlay assay. Purified proteasome complexes were separated by SDS-PAGE and stained with Coomassie blue (SDS-cooma). IC-1 and IC-2 indicate protein complexes highly detected in S5b KO tissues. The arrow indicates S5b module (S5b–PSMC2–PSMC1–PSMD2). n = 1. (J) IC-1 and IC-2 (asterisks) detected in the purified proteasome complexes (I) were cut-out and analyzed by LC−MS/MS. Relative amounts of base subunits consisting the protein complexes are indicated as fold change. n = 1. (K) WT and S5b KO MEFs were separated by native-PAGE and analyzed by western blotting using antibodies recognizing base and lid subunits. n = 1. *P < 0.05 **P < 0.01 ***P < 0.001 ****P < 0.0001.

Next, to understand how 26S/30S proteasomes assemble in the absence of S5b, we characterized the IC readily observed in S5b KO tissues (Fig. 1) and cells (SI Appendix, Fig. S3E). Compared to WT cells, 19S RP and S5b modules were barely detected, and IC containing PSMD2 was increased in C2C12 S5b KO cells (Fig. 3 F–H). Therefore, we hypothesized that these IC mediate the formation of 26S/30S proteasomes during S5b deficiency. Thus, we purified proteasome complexes from WT and S5b KO mouse livers using the UBL affinity method and confirmed the presence of two complexes, IC1 and IC2, with other proteasomes (Fig. 3I). We identified the components of IC1 and IC2 by LC−MS/MS analysis and immunoprecipitation assay (SI Appendix, Fig. S8 and S9), revealing that the upper IC1 is an intact 19S base and the lower IC2 is composed of partial components of the base and lid along with an intact 20S proteasome (SI Appendix, Fig. S8 D and E). IC2 was previously described; the 19S base assembly was templated using 20S proteasome (26, 27).

Consistent with the aforementioned results, S5b deletion resulted in an approximately 1.5-fold increase in all base components of both protein complexes (Fig. 3J). Western blotting confirmed the presence of IC1 and IC2 and their complete base composition in S5b KO MEF cells (Fig. 3K). We also tracked the position of exogenous FLAG–PSMC2 on proteasome complexes over time and found that IC complexes were detected prior to the formation of higher-order complexes such as the 19S RP. FLAG–PSMC2 was detected first in the S5b module, then in IC1 and IC2. Although we were unable to verify whether this assembly proceeds all the way to the 26S/30S proteasomes, we confirmed that the IC complexes are formed from the S5b module rather than resulting from impairments in the assembly process (SI Appendix, Fig. S7 A and B). Additionally, we verified that the subunit compositions of the 26S and 30S proteasomes in WT and S5b KO mouse liver were virtually the same (SI Appendix, Fig. S8 A and B). Among the base assembly chaperones, only S5b was detected in the 19S RP (SI Appendix, Fig. S8C), highlighting the unique role of S5b in the assembly of 19S and 20S proteasomes. Taken together, the results suggest that S5b deletion augments base assembly, possibly by utilizing the IC, leading to an overall steady-state balance toward the 26S/30S proteasomes.

In addition, we isolated several proteins that formed complexes with the 30S proteasome, including Ecm29/ECPAS and USP14 (SI Appendix, Fig. S10A). Among them, Ecm29, an adapter protein that binds to the 26S proteasome (28), was detected with an approximately sixfold increase in the 30S proteasome in S5b KO mice. Given that Ecm29 stabilizes the association between 19S RP and 20S proteasome and is a candidate that promotes the formation of the 30S proteasome (29), we tested this possibility. By analyzing C2C12 Ecm29, S5b, and Ecm29/S5b double KO cells, we confirmed the marked presence of Ecm29 in the 30S proteasome of S5b KO cells (SI Appendix, Fig. S10 B and C). The amount and activity of the 30S proteasome were slightly reduced in Ecm29 KO cells. However, the increase in 30S proteasome activity resulting from S5b deficiency was maintained by Ecm29 deletion, as seen in the double KO cells (SI Appendix, Fig. S10D). Thus, formation of the 30S proteasome by S5b deficiency is independent of Ecm29.

26S/30S Proteasomes Promote Degradation of Aggregation-Prone and Ubiquitinated Proteins.

Considering the integral role of proteasomes in protein degradation, we assessed the substrate selectivity of the 26S/30S proteasomes using the proteasome reporter substrates GFP-CL1 (ubiquitin-dependent and aggregation-prone), ODC-GFP (ubiquitin-independent), and Ub^G76V-GFP (ubiquitin-dependent soluble) (30). Cellular degradation assays revealed that, compared to control cells, the degradation of GFP-CL1 and Ub^G76V-GFP increased in S5b-deficient cells, whereas ODC-GFP exhibited no substantial difference (Fig. 4 A–C and SI Appendix, Fig. S11A). ODC degradation is uniquely mediated by its interaction with antizyme (31), whereas GFP-CL1 and UbG76V-GFP are directly targeted by the 26S/30S proteasomes. Consequently, these differences may result in little variation in the degradation of ODC. Thus, ubiquitin-dependent protein substrates appeared to be more efficiently degraded in S5b KO cells than in control cells. Because 26S/30S proteasomes are primarily involved in the degradation of ubiquitinated proteins compared to the 20S proteasome, these results could reflect the enhanced activity of 26S/30S proteasomes in S5b-deficient cells. We further assessed the degradation of the well-known aggregation-prone proteins tau, α-synuclein, FUS, TDP-43, and SOD1, which are associated with neurodegenerative diseases. We found that all these proteins, except SOD1, were degraded more efficiently in S5b KO cells (Fig. 4D and SI Appendix, Fig. S11 B−F). SOD1 is predominantly degraded by the 20S proteasome (32). These findings highlight the advantageous role of the 26S/30S proteasome over the 20S proteasome in degrading aggregation-prone proteins.

Fig. 4.

Abundance of 26S/30S proteasomes promotes degradation of aggregation-prone proteins. (A−C) Control and S5b KO C2C12 cells were transfected with GFP-CL1 (A), ODC-GFP (B), or Ub^G76V-GFP (C). After 24 h, cells were treated with MG132 (20 μM) for 1 h. After washing, cells were treated with 50 µg/mL cycloheximide (CHX) for the indicated times and analyzed by western blotting. The signals of GFP relative to β-actin were quantified by ImageJ. Unpaired t test, two-tailed, n = 3. Bars represent mean ± SD. (D) Control and S5b KO C2C12 cells were transfected with the indicated constructs with RFP (transfection control) for 24 h and analyzed by western blotting. The signals on the blots were quantified by ImageJ. Unpaired t test, two-tailed, n = 3 Bars represent mean ± SD. (E) Control and S5b KO C2C12 cells were left untreated or treated with MG132 (5 μM), A23187 (2 μM), thapsigargin (1 μM), tunicamycin (1 μg/mL), H₂O₂ (500 μM), CoCl2 (1 mM), or etoposide (50 μM) for 24 h and cytotoxicity was examined by FACS analysis following PI staining. Unpaired t test, two-tailed, n = 3. Bars represent mean ± SD. (F) WT and S5b KO primary cortical neurons (DIV 7) were left untreated or treated with H₂O₂ (100 μM), thapsigargin (2 μM), tunicamycin (2 μg/mL), MG132 (10 μM), or Aβ_1-42 (10 μM) for 24 h and analyzed by FACS analysis following PI staining. Unpaired t test, two-tailed, n = 3. Bars represent mean ± SD. N.S, nonsignificant; *P < 0.05 **P < 0.01.

In addition, we analyzed stress tolerance in S5b-deficient C2C12 cells and primary cortical neurons, particularly under proteinopathic conditions. S5b KO C2C12 cells were less vulnerable than control cells to cell death when exposed to the proteasome inhibitor MG132 or the SERCA inhibitor thapsigargin (Fig. 4E). In contrast, the same cells were more vulnerable to the massive oxidative stress inducer H₂O₂. The endoplasmic reticulum stressor tunicamycin, calcium ionophore A23187, and the DNA-damaging agent etoposide did not show any noticeable differences between these cells. Unlike C2C12 cells, S5b deletion enhanced cell viability only after MG132 treatment of primary cortical neurons (Fig. 4F). Indeed, these observations show that cellular responses to proteopathic stress may vary depending on the cell type, but align well with the known cellular responses of 26/30S and 20S proteasomes to stress conditions, again confirming the notable role of 26S/30S proteasomes in proteostasis.

S5b Inhibition Rescues Cognitive Impairment Via Reducing Tau Pathology in PS19 Mice.

Impaired proteostasis often leads to neurodegeneration through proteotoxic effects caused by the accumulation of aggregation-prone proteins. Thus, we investigated whether enhanced levels of 26S/30S proteasomes resulting from S5b KO were beneficial in a mouse model of proteinopathy. We generated PS19/S5b double KO mice by crossing S5b KO mice with PS19 mice that express human P301S mutant tau under the control of the prion protein promoter and exhibit neurofibrillary tangles and memory deficits at 6 mo of age (33–35). Behavioral analysis at 6 mo of age revealed that learning and memory deficits in PS19 mice were significantly rescued by S5b KO (Fig. 5 A–C), indicating that the abundance of the 26S/30S proteasome is crucial for the rescue of tau-mediated memory deficits.

Fig. 5.

S5b deletion in PS19 mice ameliorates cognitive impairment and tau pathology. (A−C) Cognitive performance of 6-mo-old WT, S5b KO, PS19, and PS19/S5b KO (Double) mice were assessed by Y-maze (A), novel object recognition (B), and passive avoidance (C) tests. (A) WT (n = 12), S5b KO (n = 8), PS19 (n = 10), double (n = 11). Two-way ANOVA with Tukey’s post hoc test. (B) WT (n = 12), S5b KO (n = 8), PS19 (n = 10), double (n = 15). Two-way ANOVA with Tukey’s post hoc test. (C) WT (n = 16), S5b KO (n = 10), PS19 (n = 9), double (n = 15). Paired t test, two-tailed. (D) Hippocampal tissues of each group were subjected to native-PAGE followed by western blotting and the amounts of proteasome complexes on the blots were quantified by ImageJ. Unpaired t test, two-tailed, n = 3. Bars represent mean ± SD. (E−K) Hippocampal tissues of each group were analyzed by western blotting (E) and the signals of α7 (F), 12E8 (G), AT8 (H), DA9 (I), PHF-1 (J), and MC1 (K) on the blots were quantified by ImageJ. (F, G, I, and J) Two-way ANOVA with Tukey’s post hoc test, n = 3. (H and K) Unpaired t test, two-tailed, n = 3. (L and M) Hippocampal dentate gyrus regions were analyzed by immunohistochemistry (AT8) and Hoechst staining (L), and the signals were quantified by ImageJ (M). Two-way ANOVA with Tukey’s post hoc test, n = 4. (Scale bar, 50 μm.) Bars represent mean ± SEM except for (D). N.S, nonsignificant; *P < 0.05 **P < 0.01 ***P < 0.001. AU, arbitrary unit.

Because increasing evidence shows that proteasomal activity is diminished in patients with Alzheimer’s disease (AD) (36), we assessed whether the observed memory recovery was associated with impaired proteasomal activity. Compared to WT mice, we found that the assembly and activity of the 30S proteasome were significantly reduced in the hippocampus of PS19 mice (Fig. 5D and SI Appendix, Fig. S12 A and B). Accordingly, 26S/30S proteasome levels were enhanced by S5b KO in PS19 mice, without changes in proteasome subunit levels (Fig. 5 E and F). To further explore whether this augmented proteasome assembly in Double KO mice affected tau pathology in PS19 mice, we examined the phosphorylated tau levels in the hippocampal tissues of each group. Western blot analysis revealed that the phosphorylated and pathologic forms (PHF-1, AT8, and MC1) of tau proteins, which increased in PS19 mice, were highly reduced in double KO mice (Fig. 5 E–K). These findings were corroborated by immunohistochemical analysis of the cortex and hippocampus of PS19 and double KO mice (Fig. 5 L and M and SI Appendix, Fig. S12 C and D). Taken together, these results suggest that the increase in proteasome holoenzymes resulting from S5b inhibition contributes to the reduction of pathological forms of tau protein and restoration of impaired memory function in a model of AD.

Discussion

Unlike in yeast, in which deletion of Hsm3 impairs the assembly of 19S and 26S proteasomes, we showed here that deletion of S5b in mammals stimulates the formation of 26S/30S proteasomes. Both Hsm3 in yeast and S5b/PSMD5 in mammals form a precursor assembly module, Hsm3/PSMD5-Rpt1/PSMC2-Rpt2/PSMC1-Rpn1/PSMD2, for 19S assembly. There was a minute difference in the binding of Hsm3 and S5b to their binding partners. In yeast, the central region and C-terminus of Hsm3 bind to Rpt1 and Rpt2, respectively, and in vivo formation of the Hsm3 module, Hsm3-Rpt1-Rpt2-Rpn1, is dependent on the interaction between Hsm3 and Rpt2 (21). Unlike yeast, PSMC1 does not bind to the C-terminus of S5b in mammals.

How does the assembly of mammalian holoenzyme proteasomes increase in the absence of S5b? Our finding of both an intact base (IC1) and a base or lid subunit-harboring 20S (IC2) in S5b-deficient tissues and of only S5b on the 19S RP among the base assembly chaperones suggests the following mechanism. A previous study suggests that the S5b module may function as a reservoir for PSMC2 (37). In S5b deficiency, the components of the S5b module can be integrated into IC1 and IC2 via module formation or independently, leading to an assembly of the base. In silico modeling of the mammalian S5b module supports the idea that PSMC2–PSMC1–PSMD2 complexes can form even in the absence of S5b/PSMD5. It is possible that some assembly occurs using the 20S proteasome as a platform. In addition, base assembly chaperones in yeast prevent immature 19S RP from binding to 20S to avoid the formation of immature 26S proteasomes (23). Similarly, S5b in mammals may bind to free 19S RP to regulate its assembly with 20S into 26S/30S proteasomes. Overall, low levels of S5b increased base assembly through IC2 and binding between the 19S and 20S proteasomes, leading to augmented levels of 26S and 30S proteasomes.

Additionally, IC2 likely corresponds to complexes found in the 20S proteasome-templated base assembly model (15, 26, 27). However, unlike the complexes, IC2 contains each of the subunits of 19S and 20S, suggesting that 20S serves as a platform for the formation of not only the base but also the lid. Despite containing 19S subunits, the size of IC2 is comparable to that of 20S, indicating that it may not represent a singular entity, but rather a heterogeneous mixture in which certain base and lid subunits bind to 20S. Presumably, the initial assembly of the base and lid utilized 20S as the platform, followed by additional maturation processes. Moreover, the notion that the assembly process progresses to the base and does not reach 19S in S5b deficiency suggests the existence of a checkpoint along the pathway where the base and lid form 19S. However, because we have not yet confirmed the fidelity of the increased base in S5b KO cells, we cannot rule out the possibility that the increase in the IC1 and IC2 complexes in S5b KO cells represent an aberrant off-pathway. Still, S5b is a factor that enhances base assembly fidelity but decreases 26S/30S assembly efficiency, perhaps by interfering with the 19S-20S association.

Thus, another question arises regarding S5b as a possible assembly chaperone in mammals. However, we cannot say that this is not the case, as we did not examine the KO effects of multiple chaperones on 26S/30S proteasome assembly or monitor other possible small proteasome subcomplexes in mammals. Furthermore, our findings are inconsistent with an earlier report suggesting that S5b depletion reduces 26S/30S proteasome activities in mammalian cells (16), highlighting the need for further investigation. The UBL domain used to purify proteasomes in our assay is known to bind to Rpn1/PSMD2 (38) in yeast and Rpn13/ADRM1 and Rpn10/PSMD4 in mammals (39). Our LC−MS/MS results showed that the S5b module was purified along with the protein complexes in which PSMD2 was largely detected, suggesting that the UBL domain is also likely to bind to PSMD2 in mammals. Therefore, not all proteasome complexes can be purified using the UBL affinity method. In addition, other chaperones or proteins could compensate for the absence of S5b, as the proteasome assembly in mammals is more complicated than that in yeast.

Using LC−MS/MS analysis of the purified proteasomes, we identified Ecm29, a proteasome adaptor and scaffold (28) in holoenzymes. We observed both Ecm29-free and Ecm29-bound 26S proteasomes on the native gel but could not distinguish the Ecm29-free 30S proteasome from the Ecm29-bound 30S proteasome (40). Therefore, we could not measure the ratio of the Ecm29-bound 30S proteasome to the Ecm29-free 30S proteasome resulting from the S5b deletion. Although recent reports have suggested a role for Ecm29 in proteasome disassembly under stress conditions (7, 41), Ecm29 depletion does not affect 26S/30S proteasomes in S5b-deficient cells. As seen in the subcellular localization of the 26S proteasome (42–47), Ecm29 in the 30S proteasome may also play a similar role. Since Ecm29 is also known to associate with aberrant complexes and assists in their repair during proteasome assembly, however, it is possible that the 30S proteasomes we observed include aberrant forms (48).

It is interesting to note that the proteasome activities are variable among mouse tissues and that the loss of S5b results in a greater increase of proteasome activity in the brain. The proteasome activities following S5b KO increased approximately 1.5-fold in mouse tissues; the increase was 2.1-fold in the brain and 1.3-fold in the lung. This variation may be partly due to different levels of the S5b-bound 19S complexes in the tissues and the combined activities of the hybrid proteasomes containing the β5 subunit (e.g., 20S-PA200, 20S-PA700/PA28) and other proteases as well as 20S, 26S, and 30S proteasomes. Unlike that in cell lines, however, we know that this difference is not associated with S5b levels in the mouse tissues, which is of great interest and remains to be further characterized. In addition, we observed that the increase in the 26S/30S proteasome activities of S5b KO mouse tissues was more than that in the 26S/30S proteasome levels. Thus, the 26S/30S proteasomes formed in the absence of S5b may differ slightly from those formed in WT cells. However, we were unable to detect these differences using the UBL affinity method and need further characterization.

The cellular mechanisms in which proteasome assembly is regulated in mammals are not well documented. In cancers, especially those resistant to the proteasome inhibitor bortezomib, low levels of S5b are frequently observed because of methylation of the promoter region of S5b (18, 49–51). Cancers may adapt to this condition by increasing proteasome assembly and activity, possibly in response to protein overload or by counteracting bortezomib, as evidenced by our results showing the resistance of S5b KO cells to MG132. In addition, a notable increase in 30S proteasome activity has been observed under chronic irradiation in Caenorhabditis elegans (52). We were excited to find that S5b determines 26S/30S proteasomes in many mammalian cells, but not much in tissues, probably because of variations in the baseline quantity of proteasomes.

In conclusion, we now have a key to modulate the overall ratio of the 26S, 30S, and 20S proteasomes in mammals. In particular, 26S/30S proteasomes efficiently govern the proteinopathy of neurodegenerative diseases such as AD, providing new therapeutic opportunities.

Materials and Methods

Antibodies.

MCP21 (Enzo BML-PW8105-0025), PSMD2 (Abcam, catalog no. ab140675), PSMC2 (Invitrogen, catalog no. PA5-96024), PSMC1 (ABclonal, catalog no. A15712), PSMD11 (ABclonal, catalog no. A15306), PSMD5 (S5b) (Invitrogen, catalog no. PA5-30137), PSMD9 (p27) (ABclonal, catalog no. A5357), PSMD10 (p28) (Santa Cruz, catalog no. sc-101498), PSMB5 (β5) (ABclonal, catalog no. A1975), β-actin (Santa Cruz, catalog no. sc-47778), PSMA7 (α7) (ABclonal, catalog no. A4052), PAAF1 (Sigma-Aldrich, catalog no. HPA039952), PSMC4 (ABclonal, catalog no. A2505), PSMD4 (Cell Signaling, catalog no. 12441), PSMD6 (ABclonal, catalog no. A18263), HA (Santa Cruz, catalog no. sc-7392), Ecm29 (Invitrogen, catalog no. PA3-035), GFP (Santa Cruz, catalog no. sc-9996), RFP (MBL, catalog no. PM005), and AT8 (p-Tau Ser202, Thr205) (Invitrogen, catalog no. MN1020) antibodies were purchased. MC1 (pathologic Tau), PHF-1 (p-Tau Ser396, Ser404), 12E8 (p-Tau Ser262, S356), and DA9 (Tau) antibodies were kindly provided by Dr. Peter Davies.

Mice.

B6;C3-Tg(Prnp-MAPT*P301S)PS19Vle/J (PS19) mice (Strain #:008169) were purchased from The Jackson Laboratory. Psmd5 (S5b) KO mice were generated by embryo injection and transfer to C57Bl/6 N mice using CRISPR-Cas9 system with target sequences 5′ TTC ACT TGG CCA GGA ACC TCA GG 3′ and 5′ CCT CAG GCT TGA CCT GCA GAG GG 3′ (ToolGen Inc and Macrogen Inc, Korea). The genotyping was confirmed using polymerase chain reaction and enzyme cutting (Bsu36I). Primer sequences 5′ CCG AGT TAC TGC TCC TGG AA 3′ and 5′ TCA GGA TGA AAG CCG CTA CT 3′. Both male and female mice were included in the experiments. All animal experiments were approved by and conducted in accordance with the guidelines of the Seoul National University Institutional Animal Care and Use Committee.

Statistics.

The number or size of samples and number of replicates are indicated in the figure legends. Statistical analysis was performed using GraphPad Prism 9 Software. Statistical significance was determined using an unpaired two-tailed t test for two independent groups, a two-way ANOVA followed by Tukey’s post hoc test for multiple groups, and a paired two-tailed t test for passive avoidance test data. Pearson’s correlation test was performed, analyzing the expression levels of base assembly chaperones and the amount of 26S or 30S proteasome complex in human cell lines.

Additional details on materials and methods are provided in the SI Appendix, including mouse tissue preparation, protein purification, plasmid constructs, proteasome purification, identification of proteasome complexes, and binding proteins by LC−MS/MS, in vitro binding assay, sucrose density gradient fractionation, immunoprecipitation, measuring cell viability, immunohistochemistry, Native-PAGE (different condition), Native-PAGE, and overlay assay, measuring proteasome activity, cell culture and transfection, western blotting, behavior tests, and generation of KO cell lines.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Mass spectrometry proteomics data have been deposited in ProteomeXchange (10.6019/PXD051472) (53).