Comprehensive Analysis of Proteasomal Complexes in Mouse Brain Regions Detects ENO2 as a Potential Partner of the Proteasome in the Striatum

Abstract

Defects in the activity of the proteasome or its regulators are linked to several pathologies, including neurodegenerative diseases. We hypothesize that proteasome heterogeneity and its selective partners vary across brain regions and have a significant impact on proteasomal catalytic activities. Using neuronal cell cultures and brain tissues obtained from mice, we compared proteasomal activities from two distinct brain regions affected in neurodegenerative diseases, the striatum and the hippocampus. The results indicated that proteasome activities and their responses to proteasome inhibitors are determined by their subcellular localizations and their brain regions. Using an iodixanol gradient fractionation method, proteasome complexes were isolated, followed by proteomic analysis for proteasomal interaction partners. Proteomic results revealed brain region-specific non-proteasomal partners, including gamma-enolase (ENO2). ENO2 showed more association to proteasome complexes purified from the striatum than to those from the hippocampus. These results highlight a potential key role for non-proteasomal partners of proteasomes regarding the diverse activities of the proteasome complex recorded in several brain regions.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10571-021-01106-2.

Introduction

The 26S proteasome is a multicatalytic complex that selectively degrades ubiquitinated and non-ubiquitinated proteins in the cell. The proteasome complex is not only involved in protein quality control but also performs essential functions in modulating homeostatic levels of proteins involved in key biological processes such as the cell cycle, apoptosis, and signal transduction in both physiological and pathological conditions. The proteasome is commonly formed by joining the 20S barrel and one 19S cap to form the 26S proteasome (Grumati and Dikic 2018). Nevertheless, various other proteasome subtypes with different 20S or regulatory subunit compositions are also found in cells (Coux et al. 1996). Examples include 20S with one or two regulatory 19S cap, the PA28-20S proteasome, PA200-20S proteasome complexes, and immunoproteasomes (Kors et al. 2019). In immune cells, transcriptional and post-translational modifications can induce changes in the 20S proteasome subunits to form an immunoproteasome with enhanced proteolytic activity (Murata et al. 2018). Posttranslational modifications such as phosphorylation, methylation, and ubiquitination combined with proteasomal regulation at the transcriptional levels by proteins such as NF-E2-related factor-2 (NRF2) and NRF1 modulate proteasomal activities in a tissue-dependent manner (Kors et al. 2019). In addition, cells in different brain regions can alter their proteasome compositions in response to inflammatory signals or the proteolytic needs of the cell (Früh et al. 1994; Stohwasser et al. 2000; Ding and Keller 2001; Zeng et al. 2005). While the addition of one or two 19S caps to the 20S core particle has commonly been documented in mammalian cells, hybrid proteasome complexes with 11S caps have also been found in brain tissues, highlighting the lack of mechanistic understanding of these subtypes (20S with a 19S cap or 20S with an 11S cap) on peptide substrate hydrolysis in the brain (Tanahashi et al. 2000; Dahlmann 2016). Overall, different brain regions are known to have unique proteasome profiles, a phenomenon common among other organs as well (Noda et al. 2000; Ding and Keller 2001). The proteasome, or more specifically, proteasomal subunit proteins have been associated with a network of over 450 interacting proteins. Through those interactions with these various protein complexes, the proteasome is indirectly implicated in regulating several biological processes in both a spatial and a temporal manner (Guerrero et al. 2008; Grabbe et al. 2011). Some proteasome accessory proteins have been shown to regulate proteasome activities in a tissue-dependent manner and are often found with other concomitant modulations such as post-translational modifications of proteasomal subunit proteins (Zong et al. 2006; Drews et al. 2007; Kors et al. 2019).

Therefore, we hypothesize that the molecular heterogeneity of proteasome subpopulations in different brain regions contributes to significant differences in proteasome activities and their sensitivities to proteasomal inhibitors, independently of proteasome concentration. Using a well-accepted gradient ultra-centrifugation technique for proteasomal isolation followed by a set of proteasomal assays, we provided direct evidence of distinct proteasome subpopulations in the striatum and hippocampus. Proteasome subpopulations are found to be distinct in their proteolytic activities, molecular composition, and responses to proteasome inhibitors. Furthermore, proteomic analysis followed by a set of immunoprecipitations with proteasome antibodies revealed gamma-enolase (ENO2) protein, a known proteasome partner, has closer association with proteasome complexes purified from the striatum than to those from the hippocampus in a mouse model.

Materials and Methods

Cell Culture and Protein Extraction

The ST HDH Q7/7 cell line, derived from the E14 wild-type Htt mouse (homozygous Huntingtin) Q7/Q7 knock-in embryo, was obtained (Coriell Institute for Medical Research, USA). Striatal cells were grown at 33 °C in 10% FCS in DMEM in three to five 75 cm cell culture flasks. Cells were differentiated to neuronal cells by incubating for 12 h in serum-free DMEM containing α-FGF (10 ng/ml), TPA (20 µM), forskolin (48.6 µM), and dopamine (5 µM), as previously described (Trettel et al. 2000). HC2S2 rat hippocampal cells were cultured in N2 medium using plates coated with Poly-O and Matrigel. Two days after the addition of tetracycline (1.0 mg/ml), the cells stopped dividing and started to extend processes that began to interconnect within 3 days after tetracycline addition. Nuclear and cytoplasmic fractions were extracted using NE-PER™ (ThermoFisher, USA) nuclear and cytoplasmic extraction reagents (Supplement Fig. S1) as previously described by our group (Abdullah et al. 2015).

Iodixanol Gradient Ultra-Centrifugation

We used iodixanol for gradient fractionation because it maintains relatively constant osmolality and viscosity despite changes in the density of the gradient. Because of the mild iso-osmotic conditions, all organelles including proteasomes and endosomes can be fractioned intact, without loss of water, as the density of the gradient increases. A modified version of the protocol was developed according to the original publication (Graham 2002). In iodixanol gradient fractionation, each particle sediments to a position in which the gradient density is equal to the particle density. Total protein concentration was assessed using the BCA (bicinchoninic acid) protein assay technique (Pierce BCA Protein Assays, ThermoFisher).

Proteasomal Activity Assays and Analysis

Catalytic assays were performed in triplicate for chymotrypsin-like, trypsin-like, and caspase-like activity using Suc-LLVY-AMC, Bz-Val-Gly-Arg-AMC, and Z-Leu-Leu-Glu-AMC substrates (Enzo Life Sciences, USA), respectively, at final concentrations of 100 μM (see supplementary material for details).

Western Blot (WB)

Collected fractions (60 μl total volume mixed with 2× loading buffer; Bio-Rad, USA) after gradient fractionation were subjected to 4–12% SDS-PAGE and probed with pan alpha antibody (proteasome 20S α1, 2, 3, 5, 6 & 7 subunits, mAb [MCP231], Enzo Life Sciences) or β5 (PSMB5) antibodies to locate the 20S complex and with S8 antibody (proteasome 19S ATPase subunit Rpt6, mAb [p45-110], Enzo Life Sciences) to locate the 19S cap. The β5 (PSMB5), a generous gift from Dr. Xuejun Wang, successfully worked for the striatum and hippocampus tissues. IRDye infrared secondary antibodies (LI-COR, USA) provided high-sensitivity visualization. GM-130 (cis-Golgi marker, #ab52649) and calnexin (ER membrane marker, ab13504) antibodies were purchased from Abcam (USA). Anti-Eno2 polyclonal and monoclonal antibodies (sc-31859 and sc-376375) were purchased from Santa Cruz (USA).

Measurement of Proteasomal Catalytic Activities in Mouse Brain Regions

Wild‐type (WT) male C57BL/6 mice (2 months old) were purchased from Envigo (USA). Mice were euthanized at 8 ± 1 weeks old and brain regions collected. All procedures were approved by the Institutional Animal Care and Use Committee of the University of South Dakota. The results reported in this manuscript were generated from males. Tissues were subjected to immediate liquid nitrogen flash freezing and kept at − 80 °C prior to preparation of tissue lysates in the presence of 2 mM ATP. Tissues were disrupted with a disposable plastic pestle (Millipore Sigma) using two different lysates buffers. For proteasomal activity assays (Figs. 3 and 4), we used the NE-PER™ (ThermoFisher, USA) nuclear and cytoplasmic extraction reagent containing 2 mM ATP. All fractionation assays were done in the presence of 2 mM ATP. Figure S2 (panels A and B) indicates the linearity conditions for proteasome-substrate reaction recorded at 1,3, 5, and 10 h in the presence of 2 mM ATP. Therefore, all measurements were recorded at the 5-h time point to record maximum catalytic activities of present proteasome complexes in examined cell and tissue lysates. For the IP experiments in Fig. 6A and B, the tissues were lysed using a cytosolic lysis assay buffer containing 50 mM Tris–HCl, PH7.5, 250 mM sucrose, 5 mM MgCl₂ 2 mM ATP, 1 mM DTT, 0.5 mM EDTA, and 0.025% digitonin. DTT, ATP, and digitonin were made fresh for each experiment.

Fig. 3.

Proteasome catalytic activities in mouse striatum. A–C Cytoplasmic and D–F nuclear lysates from mouse striatum tissue were subjected to iodixanol gradient ultra-centrifugation. Collected fractions were treated with MG132, bortezomib, and TLCK, and activities of the proteasome complex were measured. Proteasome activities indicate differences between the cytoplasm and the nucleus. Arrowheads represent non-specific proteasome activities, and arrows show peaks for three proteasomal activities. G–I Based on the distribution pattern of proteasome activity in the collected fractions, proteasomal peaks were individually compared for the chymotrypsin-, trypsin-, and caspase-like activities in both cytoplasm and nucleus compartments with and without bortezomib treatments. Graphs represent the mean ± SD (P**** < 0.0001, P*** < 0.001, P** < 0.01 and P* < 0.05). This set of experiments was conducted on three male mice while corresponding assays were performed in triplicate. Panel J: Cytoplasmic fractions were subjected to WB using β5 (PSMB5) and RPT6 (S8) antibodies

Fig. 4.

Proteasome catalytic activity in mouse hippocampus. A–C Cytoplasmic and D–F nuclear lysates from mouse hippocampus tissue were subjected to iodixanol gradient ultra-centrifugation. Collected fractions were treated with MG132, bortezomib, and TLCK, and activities of the proteasome complex were measured. Proteasome activities indicate differences between the cytoplasm and the nucleus. Arrowheads represent non-specific proteasome activities, and arrows show peaks for three proteasomal activities. G–I Based on the distribution pattern of proteasome activity in the collected fractions, proteasomal peaks were compared for the chymotrypsin-, trypsin-, and caspase-like activities in both cytoplasm and nucleus compartments with and without bortezomib treatments. This set of experiments was conducted on three male mice followed by proteasomal assays in triplicate. Graphs represent the mean ± SD calculated (P**** < 0.0001, P*** < 0.001, P** < 0.01 and P* < 0.05). J Cytoplasmic fractions were subjected to WB using β5 (PSMB5) and RPT6 (S8) antibodies

Fig. 6.

Enolase 2 (Eno2-gamma-enolase) protein has a higher association to mouse striatum proteasomes. A Hippocampus and striatum tissue lysates were subjected to immunoprecipitation using MCP21 (antibody against the 20S proteasome). Lane 1 demonstrates pulled-down 20S in association with the striatum Eno2 (bottom gel), while no association is observed with the hippocampus Eno2 (top gel). The last lanes (inputs) in the top and bottom gels in A confirmed the presence of Eno2 proteins in both mouse STR and mouse HC. B In another set of experiments, both mouse HC and mouse STR tissues were subjected to IP using RPT6 antibodies immobilized on agarose beads. Control groups were beads immobilized with mouse IgG. After IP, pulled-down proteins were subjected to WB. Interestingly, Eno2 showed association with both HC and STR 26S proteasomes pulled down by anti-RPT6 antibodies. However, measurement of pixels in lanes II and IV with UN-SCAN-IT gel (version 6.1) showed a pixel total of 9432 in lane II (hippocampus) versus a pixel total of 12,401 in lane IV (striatum). These findings indicate that Eno2 shows more association with mouse striatum 26S proteasome versus mouse hippocampus proteasome (30% higher association). C Fractions 6 to 9 (high peaks for proteasome complexes collected from hippocampus and striatum tissues) were subjected to WB using Eno2 and β5 antibodies. Eno2 substantially cofractionated with the striatum proteasome. D These results suggest higher association of Eno2 with striatum proteasome complexes

Pull-down Proteasome Complex

MCP21 antibody (targeting the α2 subunit of the 20S proteasome) and RPT6 antibody (targeting the RPT6 ATPase subunit in the 19S subunit) purchased from Enzo Life Sciences were used separately for immunoprecipitation experiments. These two antibodies, immobilized on agarose beads protein A/G (Millipore Sigma, USA), are able to pull down the 20S proteasome or 26S proteasome complex and their associated proteins, respectively (Bousquet-Dubouch et al. 2009; Rezvani et al. 2012; Sparks et al. 2014). Additionally, a control group was conducted in parallel using a normal mouse IgG serum (Santa Cruz Biotechnology) immobilized immunized on agarose beads. For proteomic experiments, we mixed fractions 8 and 9 corresponding to fractions with the highest proteasomal activities. The combined fractions and 10 μg MCP21 antibodies immobilized on protein A/G beads were incubated at 4 °C overnight with gentle rocking. The pulled-down 20S proteasome and its partners were washed with a moderate salt buffer (20 mM Tris–HCl, 150 mM NaCl, pH 8) for five times to separate the non-proteasomal partners associated with the proteasome complex (Wang et al. 2016). The supernatant containing proteasomal-associated proteins was collected and subjected to Vivaspin 2 concentrators with a molecular weight cutoff of 5 kDa (Sigmaaldrich, USA) to remove extra buffer. In the next step, the concentrated supernatants (equal volumes with 2× loading buffer) were subjected to 4–20% gradient SDS-PAGE to be visualized by SYPRO Ruby staining (ThermoFisher Scientific). Then, the entire isolated gels from each group were sent for mass spectrometry (details in supplemental materials, Appendix A; Proteomics Core Facility, https://www.usd.edu/medicine/basic-biomedical-sciences/proteomics-core).

The criteria for the MS/MS analysis were as follows: The eluted ions were analyzed through tandem mass spectrometry (MS/MS analysis), considering one full precursor MS scan (400–2000 m/z) followed by four MS/MS scans of the most abundant ions detected in the precursor MS scan, operating under dynamic exclusion or a direct data acquisition system (DDA). For the bioinformatics analysis (mass peaks or list converted into sequences), the following criteria were used: The MS/MS spectra were searched with MASCOT against the Swiss-Prot database with a taxonomy filter for mouse (Mus musculus) to analyze peptide matches (peptide masses and sequence tags) as well as protein searches. Also, the ion score or expected cutoff was set at 5, using a 95% confidence interval (CI%) threshold (P < 0.05), considering a minimum score of 36 as a criterion for peptide identification. Moreover, a minimum of two identified peptides that matched equally well to multiple protein IDs were considered, and only those proteins that appeared in at least two or more replicates were included in the list.

A similar IP procedure was used for the experiments illustrated in Fig. 6 using 10 μg MCP21 or RPT6 antibodies immobilized on beads. We used 600 μg total tissue lysates (striatum or hippocampus) followed by the IP experiments shown in Fig. 6A and B.

Statistical Analysis

Enzymatic assay experiments in cells were conducted in triplicate for iodixanol gradient, with average values shown. The fractionation assay for each sample (cells or tissues) was repeated three times. The results were consistent (Fig. S2C-E), and representative data are presented. Graphs were created using GraphPad Prism version 8.0.0 (GraphPad Software, San Diego, California, USA). The data obtained from groups of three mice were analyzed per brain region using Student’s t-test in the Prism software. The proteasome activities were expressed as averages of three independent fractionated tissues treated with and without bortezomib. P values of less than 0.05 were considered significant.

Results

Optimization of Iodixanol Gradient Fractionation for Effective Isolation and Identification of Protein Complexes, Including Proteasome Complexes in Neuronal Cells

Figure 1A shows a summarized protocol used to enrich proteasomal complexes in both cytoplasmic and nuclear cell lysates of mouse striatum (ST HDH Q7/7) and HC2S2 rat hippocampal cell lines. The iodixanol gradient fractionation allows separation of protein complexes including proteasomes based on their density (Fig. 1A). We initially used these relevant neuronal-like cell lines to troubleshoot our assays before moving on to striatum and hippocampus tissues, which offer a more complex system. Using neuronal-like cells allowed us to determine whether the iodixanol gradient fraction method can efficiently fractionate multibranched cells and whether the cell lysates of neuronal cells will not interfere with fractionation of protein complexes and cellular organelles. We used arylesterase and leucine aminopeptidase activities as ER (endoplasmic reticulum) and Golgi markers, respectively (see Supplementary Materials). These two enzymes are well known as processing enzymes in ER and Golgi compartments, respectively. Therefore, they represent ER and Golgi compartments with surrounding membranes. As previously shown (Ng et al. 2007), these membranes are associated with several cytoplasmic partners, including proteasome complexes. Panels B and C in Fig. 1 show fractions contain endoplasmic reticulum (arylesterase) and Golgi apparatus (leucine aminopeptidase). The ER marker (arylesterase) showed the presence of an ER compartment in fractions 1 to 6 (Fig. 1B). The presence of ER in fractions 4 through 6 was expected with our discontinuous 8, 16, 28, and 38% iodixanol gradient, as previously described (Pedrazzini et al. 2000). However, the presence of ER in fractions 1, 2, and 3 (low-density gradients; Fig. 1A) likely corresponds to a fragmented form of ER, which needs further characterization. Meanwhile, we probed collected fractions of a striatum-derived cell line with GM-130 (cis-Golgi marker) and calnexin (ER membrane marker) antibodies (Fig. 1D). GM-130 was present in those fractions with high leucine aminopeptidase activity (fractions 3–6). However, calnexin signals were observed in fractions 12 through 14, as previously reported for other cell lines (Rezvani et al. 2012). Similar to arylesterase ER marker, GM-130’s signals in fractions 1 and 2 may represent a fragmented Golgi apparatus or budded-off Golgi vesicles.

Fig. 1.

Iodixanol gradient fractionation of proteasome complexes and quality control tests. A Overview of the experimental study. Cytoplasmic and nuclear lysates were fractionated via iodixanol gradient ultra-centrifugation and separated into twenty fractions. B and -C Sedimentation of arylesterase (endoplasmic reticulum markers) and leucine aminopeptidases (Golgi apparatus markers) according to their densities verified that iodixanol gradients can provide satisfactory resolution of membrane particles and protein complexes in the cytoplasm segment of a mouse striatum cell line. D GM-130 (Golgi marker) and Calnexin (ER marker) revealed locations of Golgi and ER, respectively. E Density of fractions (g/ml-red line) as well as total protein concentration (mg/ml-columns) of collected fractions were determined in the cytoplasm segments of cells

In addition, an equal volume of each fraction was weighed to confirm the ideal discontinuous gradient formed by the iodixanol, which is necessary to separate protein complexes based on their density and composition (Fig. 1, panel E, red line). The total protein concentration in each fraction, as determined by the BCA assay, showed two peaks in fractions 2 and 3 as well as fractions 12 and 13. Based on previous publications using similar iodixanol gradient fractionation (Shintani et al. 2001; Onódi et al. 2018), these two peaks likely correspond to soluble proteins and ribosomal in lighter density fractions (2 and 3 fractions) and smooth and plasma membrane proteins in heavier density fractions (12 and 13 fractions). Ongoing proteomic studies in our team will determine the composition of these light and heavy protein peaks. We obtained similar results in an HC2S2 rat hippocampal cell line (Fig. S3).

Together, the above results indicate that our discontinuous iodixanol density gradient enables efficient isolation with improved resolution and recovery of intact subcellular compartments including ER, Golgi, and protein complexes in a neuronal cell line (Yee et al. 2008; Wang et al. 2014).

Mouse Striatum-Derived Cell Line Displays Different Proteasome Inhibition Profiles

Using the iodixanol gradient fractionation method, the three proteasome proteolytic activities of the cytoplasmic and nuclear lysates of a mouse striatum-derived cell line were measured (Fig. 2). To compare proteasome activities between the nucleus and cytoplasm compartments, we had two choices. We first wanted to use one of the subunits of the proteasome (Rpt6) to normalize the amount of loading. However, we found that this approach may not be accurate because proteasome complexes exist in different compositions in nucleic versus cytoplasmic compartments along with unassembled proteasome subunits. Therefore, we decided to use the same amount of total proteins of nucleic and cytoplasmic compartments for all gradient experiments (~ 6 mg total cell or tissue lysates).

Fig. 2.

Proteolytic activity and inhibitor response of the proteasome in a mouse striatum-derived cell line. A Schematic diagram of the 20S proteasome catalytic complex flanked by the 19S regulatory subunit made by 6 ATPase subunits (RPT-1–6) and regulatory particles of non-ATPase (RPN) subunits. B–D Cytoplasmic and E–G nuclear lysates of mouse striatum-derived cells were subjected to iodixanol gradient ultra-centrifugation. Subsequently, samples were analyzed for proteasomal catalytic activities with and without the proteasome inhibitors MG132, TLCK, and bortezomib. Arrowheads represent potential non-specific proteasomal activities, while arrows show peaks of three proteasomal activities recorded in collected fractions. H–I An equal volume of each fraction (cytoplasm and nucleus) was subjected to SDS-PAGE followed by WB using anti-pan alpha (20S proteasome) and anti-RPT6 (19S subunit-S8 ATPase) antibodies to illustrate the distribution of proteasome complexes in the collected fractions

Based on the order of the appearance of the peak in different fractions, the observed proteolytic activities of the fractions were typically associated with 20S, 26S, and 30S proteasomes (Fig. 2A) where the 20S is the lightest proteasome complex, followed by the 26S and 30S complexes. We used previous reports (Giannini et al. 2013) as well as data available on the Enzo website (https://www.enzolifesciences.com/BML-PW9310/proteasome-26s-human-purified/) to locate proteasome peaks (Fig. 2, arrows) in the presented results. Panels B-G in Fig. 2 demonstrate three proteasomal activities in both cytoplasmic and nuclear fractions. Bortezomib and MG132 proteasome inhibitors reduced the chymotrypsin- and caspase-like activities of both cytoplasmic and nuclear proteasomes (Fig. 2B, E, D and G). The amount of chymotrypsin- and caspase-like activities that were reduced by the proteasomal inhibitors MG132 and bortezomib were considered “real or true proteasome activities.” Bortezomib was unable to inhibit trypsin-like activities in either cytoplasmic or nuclear fractions, while TLCK, a selective trypsin inhibitor, induced a partial inhibition of trypsin-like activities (Figs. 2C and F). The absence of response to bortezomib for chymotrypsin- (Fig. 2B and E, arrowheads) and caspase-like activities (Fig. 2D and G, arrowheads) in the low-density fractions (fractions 1 to 4) suggests the proteolytic activities observed in the earlier low-density fractions may not correspond to proteasome complexes. However, WB results shown in Fig. 2H suggest a low number of pan alpha (20S) subunits are present in the cytoplasm compartments of those early fractions, which could be partly responsible for the peaks recorded in low-density fractions in Fig. 2B and D. On the other hand, MG132 inhibits the aforementioned proteolytic activities, as it is considered to be a non-specific inhibitor (Fig. 2B, D, E, and G). The absence of pan alpha signals in the early nuclear fractions in WB indicate that the chymotrypsin- and caspase-like activities recorded in the low-density fractions in nuclear fractions are not the 20S proteasome complex (Dahlmann 2016).

Western blot (WB) analysis using an anti-pan-α antibody confirmed the presence of the proteasome complex in the cytoplasm and nucleus (Fig. 2H-I). We also probed cytoplasmic and nuclear fractions with an anti-RPT6 (S8) ATPase antibody (Fig. 2H-I). The antibody detected RPT6 subunits in fractions 7–9 in the cytoplasm, which matched the catalytic activities recorded in panels B-D in Fig. 2. It is noteworthy that RPT6 levels corresponding to 19S regulatory lid components overlapped with 20S-positive fractions, which implies that fractions 6–9 would appear to contain 26S or 30S proteasomes in cytoplasmic fractions (Fig. 2H). However, anti-RPT6 detected RPT6 proteins only in fractions 4, 5, and 6 in nuclear fractions. The protein abundance of RPT6 in the nuclear compartments was low, and we used VIVASPIN 2 filter-5,000 MW PES to increase the protein concentrations of nucleus fractions followed by WB, as their lower catalytic activities indicate in Fig. 2E and G. As equal total cytoplasmic and nuclear proteins is used for gradient experiments (see “Materials and Methods” section), the results indicate the levels of proteasome complexes and their activities are markedly higher in the cytoplasm, particularly for chymotrypsin- and caspase-like activities (Fig. 2B versus E, and D versus G). Interestingly, the level of trypsin-like activity was similar in both cytoplasmic and nuclear compartments. We measured the three proteasome proteolytic activities of the cytoplasmic and nuclear lysates in another neuronal cell line using the iodixanol gradient fractionation method (Fig. S3). Interestingly, HC2S2 rat hippocampal cells showed distinct proteasome activities and responses to inhibitors, suggesting the existence of different proteasome subpopulations in the cytoplasm and the nucleus. Together, these findings suggest: (1) iodixanol gradient fractionation is an acceptable method to measure proteasomal catalytic activities in neuronal cells and (2) There are different proteasome subpopulations in the cytoplasm and the nucleus based on the origin of tissues.

Proteasomal Catalytic Activities in Mouse Striatum and Hippocampus

Well-established evidence indicates that alteration of proteasome activities in different neurodegenerative diseases is brain region-dependent (Lehtonen et al. 2019). In the next step, we used iodixanol gradient fractionation to determine proteasomal proteolytic activity in two mouse brain regions: striatum (STR) and hippocampus (HC) tissues extracted from three male C57bl/6 mice (Figs. 3 and 4). Analyzing chymotrypsin-, trypsin-, and caspase-like activities in the striatum tissue revealed a peak between fractions 7 to 11 (Fig. 3A-F, arrows). Additionally, several panels showed a peak of proteasome activities in low-density fractions (Fig. 3A-F, fractions 1–3, arrowheads) in the cytoplasm and nucleus. In comparison to the cytoplasm, proteolytic activities were lower in the nucleus fractions when the percentage of activities was determined for three catalytic activities (Fig. 3D-F, arrows), despite loading the same total protein tissue lysates. However, considering the effect of bortezomib inhibition on these proteasomal activities, different effects were observed for the cytoplasm and the nucleus using the catalytic activities recorded in those highest peaks (Fig. 3G-I). In the cytoplasm, bortezomib inhibition reduced chymotrypsin-, trypsin-, and caspase-like activities more than 50% (Fig. 3G-I). On the other hand, nucleus proteasomal activities decreased less than 40% (Fig. 3G-I). MG132 and TLCK had similar effects, as observed with cultured cells (Figs. 2 and 3S). Cytoplasmic fractions probed with β5 (PSMB5) and RPT6 antibodies (Fig. 3J) showed signals in fractions 6 to 9, which partially match with activities recorded in Fig. 3, panels A-C.

Measurement of proteasome catalytic activities in mouse hippocampus tissues extracted from the same three male mice in Fig. 3 showed the presence of proteasome activities (Fig. 4A-F, arrows) in both the cytoplasm and nucleus compartments. Similar to the striatum brain region, proteolytic activities were higher in the cytoplasm in comparison to the nucleus despite equal loading of total protein tissue lysates (Fig. 4A-C versus D-F). Additionally, several panels showed a peak in low-density fractions (Fig. 4, fractions 1–3, arrowheads) in the cytoplasm and nucleus. Bortezomib was significantly effective in inhibiting the proteasomal activity peaks of the cytoplasmic fractions (Fig. 4G-I). However, similar to the striatum, bortezomib significantly inhibited chymotrypsin-like activity in the nucleus fractions by 50%, while it failed to inhibit trypsin- and caspase-like activities in the nuclear hippocampus (Fig. 4G-I). Figure 4H indicates that the proteasome trypsin-like activities in nucleus do not respond to bortezomib, while the same level of trypsin-like activities significantly respond to bortezomib in mouse STR. Fractions probed with β5 (PSMB5) and RPT6 antibodies (Fig. 4J) showed signals in fractions 6 to 11 that partially match with activities recorded in Fig. 4, panels A-F. Proteasome subcomplexes and their non-proteasomal partners are the key factors for an intriguing diversity of proteasomal activities in different cellular compartments and particular brain regions.

Eno2, a Non-proteasomal Partner of the Proteasome Complex, Shows Different Association to Proteasomes in Mouse Striatum Versus Hippocampus

Non-proteasomal partners can play a key role in proteasome assembly and activity and regulate the post-translational modification of proteasomal subunits in fully assembled proteasome complexes (Bose et al. 2004; Zong et al. 2006; Drews et al. 2007; VerPlank and Goldberg 2017). Therefore, we hypothesized that proteasome complexes in different brain regions have unique non-proteasomal partners. We decided to investigate non-proteasomal partners of the proteasome complex in mouse striatum and hippocampus brain regions. To identify proteasome partners, pooled cytoplasmic fractions 8 and 9 with the highest proteasomal catalytic activities (Figs. 3A and 4A) obtained from the iodixanol ultra-centrifugation procedure were subjected to immuno-purification using MCP21 antibody immobilized on agarose protein A/G beads. Using a moderate salt buffer (see Materials and Methods section) allowed selective separation of associated proteins, including those with weak interactions with the proteasome complex, while the proteasome complexes and perhaps strongly attached partners dominantly stayed with agarose beads. Associated proteins visualized by SYPRO Ruby staining (Fig. 5A) were entirely cut out from the gel and sent for mass spectrometry. Figure 5B shows proteins uniquely pulled down with the 20S proteasome in mouse striatum and mouse hippocampus tissues. There were also several proteins shared by both regions. Proteins pulled down by the striatum proteasome were gamma-enolase (Eno2), 2',3'-cyclic-nucleotide 3'-phosphodiesterase, toll-like receptor 3, and enkurin (Fig. 5B). Hippocampus-exclusive proteins included clathrin heavy chain 1, ATP synthase subunit O, mitochondrial, acetyl-CoA acetyltransferase, Synaptogyrin-3, and Ig kappa chain V-V region (Fig. 5B). Among the shared proteins, we found several enzymes such as calcium/calmodulin-dependent protein kinase type II subunit alpha, which may be involved in post-translational modifications of proteasome subunits as previously described (Shimada et al. 2013), and 14–3-3 (ζ and δ), a proteasome partner that has a key neuroprotective role in neurodegeneration (Fig. 5B, Appendix A, Tables S1 and S2).

Fig. 5.

Proteasome-associated partners in mouse striatum and hippocampus found by an affinity purification method and a mass spectrometry approach. A Fractions 8 and 9 with the highest proteasomal catalytic activities recorded after iodixanol gradient fractionation of mouse striatum (STR) and hippocampus (HC) were subjected to IP using MCP21 antibodies immobilized on agarose beads. Control groups were protein A/G beads with mouse IgG serum. Associated proteins pulled down by a proteasome complex (dominantly 20S core) were separated by a moderate salt buffer and subjected to SDS-PAGE followed by mass spectrometry (details in Materials and Methods). B Mass spectrometry results identified known and unknown proteins associated with proteasome complexes enriched from mouse hippocampus and striatum tissues. A group of proteins were found to be associated with only one specific tissue, including Eno2, which was only detected in the STR brain region. The arrowheads in A point to a band below 56 kDa, which likely is Eno2 protein pulled down dominantly in striatum fractions

Following the mass spectrometry results, the previously demonstrated association between Eno2 and the proteasome was further explored in two mouse brain regions using total tissue lysates. We confirmed the presence of the Eno2 protein in both striatum and hippocampus brain tissue (Fig. 6A, lane III). Next, we used MCP21 immobilized on agarose beads to pull down the 20S proteasome in both tissues. Western blot (WB) of pulled-down proteins using anti-Eno2 antibodies confirmed that the 20S proteasome complex from the striatum is able to be associated with Eno2, while no Eno2 band was detected with the hippocampus 20S proteasome (Fig. 6A). Similarly, the arrowheads in Fig. 5A point to a dominant band under 56 kDa pulled down in striatum tissue and is notably weaker in hippocampus tissue pull-down. Subsequently, we used anti-Rpt6 immobilized on agarose beads to pull down the 26S proteasome from striatum and hippocampus tissue lysates. Similar to the MCP21 pull-down, associated proteins were enriched and subjected to WB. Anti-Eno2 antibodies showed association of Eno2 with both the hippocampal and striatal 26S proteasome (Fig. 6B). However, measurement of the total signal in those two bands with UN-Sac-It software showed a 30% higher signal density of Eno2 in the striatum lane in comparison to the HC tissue. WB experiments with Eno2 and β5 proteasome antibodies further confirmed that Eno2 substantially co-sediments with the striatum proteasome (Fig. 6C). The results illustrated in Fig. 6 indicate a higher binding association of Eno2 for the proteasome in the striatum versus the hippocampus (Fig. 6D). These findings indicate that iodixanol gradient fractionation can be a valuable tool to identify selective non-proteasomal partners involved in proteasomal regulation in different proteasome subcomplexes.

Discussion

The function of the proteasome machinery is to catalyze the degradation of ubiquitinated proteins in the cell. This machinery is composed of the core 20S proteasome in complex with several regulatory particles, most notably the 19S and 11S regulatory particles. Different combinations of particles are seen in cells, such as free 20S, 26S (20S single-capped with 19S), 30S (20S double-capped with 19S), and a hybrid proteasome (20S capped with 19S and 11S subunits) (Peters et al. 1993; Yoshimura et al. 1993; Jung et al. 2009; Dahlmann 2016, Tanaka. 2009). Different groups have determined the relative percentage of these complexes in various cell types. For example, in HeLa cervical cancer cells, 40% of the proteasome is found unassociated with regulatory particles, while the remaining 60% is capped with 19S and 11S subunits (Tanahashi et al. 2000). In a U937 monocytic cell line stimulated with IFN-γ, a higher proportion of free 20S (~ 53%) was seen in the cell (Fabre et al. 2013). In addition to cancer and immune cell lines, a census of the proteasome content of hippocampal neurons showed 27% of the proteasomes are 30S, and the rest function as 26S proteasomes (Asano et al. 2015). Proteasome activities are positively or negatively regulate by the variability in proteasome compositions; non-proteasomal partners of proteasomes (Stanhill et al. 2006; Sbardella et al. 2018; Lee et al. 2018); post-translational modifications that, in turn, modulate activation and cellular localization of proteasomes (Kim and Goldberg 2018; Zhang et al. 2019; Kors et al. 2019; Liu et al. 2020); and, finally, cell conditions such as stress (Asano et al. 2015). In addition to the heterogeneity of the proteasome content of the cell, proteasomal activity is affected by age and neurodegenerative disorders (Zeng et al. 2005; McKinnon and Tabrizi 2014; VerPlank et al. 2019), indicating that a diverse set of cellular signaling pathways modulate proteasomal activities in non-pathological and pathological conditions. During aging, oxidative stress caused by high oxygen consumption and low antioxidant enzyme content damages the proteasome (Keller et al. 2000c, d; Grimm et al. 2012). These changes are reflected by a decrease in proteasomal activity during aging; in fact, a significant decrease in proteasome activity could be a risk factor for neuronal disorders (Keller et al. 2000a; Zeng et al. 2005; Kelmer Sacramento et al. 2020). Moreover, a similar reduction in proteasome catalytic activity has been reported in Alzheimer’s disease (Keller et al. 2000b; Salon et al. 2000), amyotrophic lateral sclerosis (Kabashi et al. 2012), and Parkinson’s disease (McNaught and Jenner 2001; McNaught et al. 2002). In addition to neurodegenerative diseases, the heterogeneity of proteasome function is relevant to cancer progression, as the progression of these cells is crucially dependent on appropriate proteasome activity (Morozov and Karpov 2019).

In this study, iodixanol ultra-centrifugation was first optimized to fractionate and obtain the proteasome from two neuronal-like cell lines. Then, optimized iodixanol gradient fractionation was used to measure proteasomal catalytic activities in two different mouse brain regions. As previously described (Persaud-Sawin et al. 2009), the sucrose gradient and iodixanol gradient fractionations result in different yields based on the density of the proteins. We found our modified iodixanol gradient fractions provided substantially more isolated peaks for the several forms of proteasome complexes using brain tissues.

In vivo studies were carried out in two regions of mouse, striatum and hippocampus, since these two regions particularly are affected in neurodegenerative diseases (Moodley and Chan 2014; Nopoulos 2016). Mouse tissues showed all three proteasomal catalytic activities in the cytoplasm and the nucleus in both brain regions. Recorded proteasomal activities in the presence and the absence of proteasome inhibitors revealed the presence of diverse forms of proteasome complexes in these two brain regions. Further, experiments will certainly be necessary to verify the 20S proteasome complexes with one or two caps based on current results. It is worth noting that the specific functionality of 30S versus 26S has not yet been adequately addressed experimentally. Additionally, all tissues demonstrated bortezomib was significantly less effective in inhibiting the proteasomal activities of the nucleus than the cytoplasm. Together, these results indicate cytoplasmic and nuclear compartments have their own proteasome subcomplexes. These diverse subcomplexes may be necessary for specific functions in these two cellular compartments, as previously suggested in other organs (Muratani and Tansey 2003; Dang et al. 2016). Gender is another factor that can modify the activity of proteasome in normal and pathological conditions (Bellavista et al. 2014; Jenkins et al. 2020). Our future studies will investigate whether gender is also a factor for these proteasome subcomplexes and their different sensitivities to proteasome inhibitors in studied brain regions.

Striatum and hippocampus samples were also used to analyze protein partners of the proteasome complexes using mass spectrometry. This led to the identification of an array of protein partners that were shared, in addition to some exclusive to one tissue. The presence of 14–3-3ζ, a known proteasomal partner, confirmed the accuracy of pull-down experiments for mass spectrometry studies (Fig. 5) using iodixanol gradient fractionation. Several protein-modifying enzymes that may be involved in the post-translational modifications of proteasome subunits were pulled down (Fig. 5). In addition, proteomic results revealed the presence of a known proteasome partner, gamma-enolase (Eno2) (Weinkauf et al. 2009). The gamma-enolase (ENO2) is dominantly present in mature neurons and in cells of neuronal origin. A previous study using the proteasome from Saccharomyces cerevisiae had shown a direct association between the 19S proteasome and Eno2 (Verma et al. 2000).

We initially chose fractions 8 and 9 for proteomic assay since we recorded the highest peaks of proteasomal activities in fractions 8 and 9 (Figs. 3A, B and C; and 4A, B and C). After receiving the proteomic results, we decided to probe all collected fractions with β5 or Pan α as well as RPT6 antibodies. Unexpectedly, we found a higher signal for β5 or Pan α as well as RPT6 in fractions 6 and 7 (Figs. 3 and 4, Panel J). Subsequently, we decided to probe fractions 6 to 9 with β5 and Eno2. Interestingly, Eno2 shows the highest signals in 6 and 7 where there was also the highest level of β5 and RPT6 signals (Fig. 6C). We could obtain even higher hits for Eno2 in fractions 6 and 7. Future proteomic assays should be based on both the highest catalytic proteasomal activities as well as highest signals for proteasome complexes to accurately generate a map of non-proteasomal partners of proteasome in a fraction-dependent manner. The different regulatory partners associated with endoplasmic reticulum proteasome complexes versus those cytoplasmic proteasome complexes has been previously reported.

Using western blot analysis and immunoprecipitation, Ji et al. showed that apoptosis signal-regulating kinase 1 (ASK1), a MAP kinase, negatively regulated the function of the 26S proteasome by associating with the 19S regulatory particle (Ji et al. 2010). Similarly, due to the function of Eno2 as a phosphopyruvate hydratase, the association with the proteasome was suggestive of the regulatory function of Eno2. Western blot and immunoprecipitation experiments were performed in our study, demonstrating the stronger association of Eno2 with the mouse striatal proteasome. It is well established that differences in proteasome partners can fine-tune the function of the proteasome in a tissue-dependent manner. For instance, PI31 (a proteasomal inhibitor of 31kD) regulates the assembly of the proteasome and acts as an adaptor for proteasome motility in axons (Liu et al. 2019; Minis et al. 2019). Localized changes in the activity of PI31 are associated with neurodegenerative disorders, indicating the importance of tissue-dependent proteasome regulation. Further studies will determine the biological impact of the Eno2 protein on proteasome complexes (20S, 26S, and 30S) in terms of catalytic activities, cellular localization, sensitivity to proteasome inhibitors, and proteasome stability.

Conclusion

The proteasome activities and their responses to proteasome inhibitors including bortezomib in the striatum and hippocampus examined in this study reveal a fuller picture of proteasome sub-classes in brain regions. More importantly, these differences indicate the presence of different regulatory systems specific to different regions of the brain. This study revealed that the association of Eno2 with the 19S subunit could regulate the proteasome in the striatum differentially than in the hippocampus. Further studies need to be conducted to understand the regulatory function of Eno2 using gain- and loss-of-function experiments in striatum and hippocampus cells and animal models. In this study, we compared the non-proteasomal partners of proteasomes in healthy brain regions. Comparison of non-proteasomal partners in diseased tissue versus normal can provide a better picture of proteasome alteration during the progression of neurodegenerative diseases in a brain region-dependent manner. Based on current findings in normal ST and HC, the future direction is to repeat similar experiments in a 2XFAD transgenic mouse model of Alzheimer’s disease.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

MN and RA carried out the experiment. NE wrote the manuscript with support from EC and with inputs from all authors. EC performed the proteomic experiments. KR supervised the project. JSP participated in experimental design and contributed to writing and editing the final manuscript. Current Address, NE: Department of Biochemistry and Biomedical Sciences, McMaster University, 1280 Main Street West, Hamilton ON L8S 4L8, Canada. Current Address, MN: Department of Biostatistics, 170 Rosenau Hall, University of North Carolina—Chapel Hill, Chapel Hill, NC 27599, USA.

Funding

Support for this work was provided by the University of South Dakota Division of Basic Biomedical Sciences (READ award) and CBBRe (Center for Brain and Behavior Research, University of South Dakota). The Proteomics Core facility at the University of South Dakota was supported by NIH Grant Number 2P20 GM103443-19 from the INBRE Program of the IDeA National General Medical Sciences, NIH.

Data Availability

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Conflict of interest

The authors declare no conflict of interest.