The animal protocol was approved by the Institutional Animal Use and Care Committee of Johns Hopkins University (protocol number SW23M119). Neonatal Yorkshire male piglets, broadly in the range of 2- to 5 days old (1–2 kg), were used for descriptive normative neuroanatomical and biochemical experiments and for drug testing experiments. Because of the different tissue needs and brain harvesting approaches for independent experiments, naïve piglets were used as separate groups for the different types of experiments. These animals were not subjected to any experimental or surgical procedures. For example, fresh brain samples were used for immunoblotting (n = 6 naïve piglets), optimally prepared paraformaldehyde perfusion fixed brains were used for immunohistochemistry and brightfield/confocal microscope imaging (n = 8 naïve piglets), and seasonal and littermate matched naïve piglets (n = 9) were used for the drug treatment experiments.

For immunoblotting, 2- to 4-day-old naïve piglets (n = 6) received a lethal dose of pentobarbital 50 mg/kg and phenytoin 6.4 mg/kg (SomnaSol) and, after thoracotomy and left myocardial puncture and aortic catheterization, ice-cold 100 mM phosphate-buffered saline (PBS, pH 7.4) was perfused (~2 L) for body exsanguination. After decapitation, the brain was removed quickly and placed on wet ice. On an ice-cold metal plate, the brain was slabbed and microdissected into individual anatomical regions (Figure 1) that were snap frozen in isopentane (−70°C) cooled by a dry ice/alcohol slurry. All samples were stored in individual Eppendorf tubes at −80°C until used.

Nasal mucosa biopsy samples were also studied. The samples were obtained with a Cellebrity^™ Cytology Brush (Boston Scientific) from sham piglets; these animals received surgical anesthesia and were instrumented for physiological monitoring. Nasal mucosa tissue was eluted from the bristles using PBS and collected by gentle centrifugation. The samples were resuspended in PBS, and contaminating red blood cells were removed using EasySep^™ RBC Depletion Reagent (StemCell Technologies). After centrifugation, the nasal mucosa samples were frozen on dry ice and stored at −80°C until used.

Immunoblotting was done to examine the protein levels of PSMB5 and PSMA3 in different brain regions (Figure 1), their subcellular fractions, and the nasal mucosa of piglets. Frozen tissue samples were homogenized with a Brinkmann polytron in ice-cold 20 mM Tris–HCl (pH 7.4) containing 10% (wt/vol) sucrose, 200 mM mannitol, complete protease inhibitor cocktail (Roche), 0.1 mM phenylmethylsulfonyl fluoride, 10 mM benzamidine, 1 mM EDTA, and 5 mM EGTA. Crude homogenates were sonicated for 15 s and then centrifuged at 1,000 g_av for 10 min (4°C). For subcellular fractionation of piglet brain tissue, we used a modification of the Percoll protocol developed by Wieckowski et al. (2009). With this protocol, we isolated pure mitochondria and mitochondrial-associated membranes (MAMs) along with pure nuclear and cytosolic fractions. Protein concentrations were measured by bicinchoninic acid assay (Smith et al., 1985) with a kit (Pierce, Thermo Scientific, Carlsbad, CA) using bovine serum albumin as a standard.

Tissue lysates and subcellular fractions were subjected to sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to the nitrocellulose membrane by electroelution (Martin et al., 2000; Santos et al., 2018; El Demerdash et al., 2021). Ponceau S staining of nitrocellulose membranes before immunoblotting verified the lane equivalency of sample loading and transfer in each experiment. Blots of crude tissue lysates were blocked with 2.5% non-fat dry milk with 0.1% Tween 20 in 50 mM Tris-buffered saline (pH 7.4), then incubated overnight at 4°C with rabbit polyclonal antibody to PSMB5 (GeneTex, GTX23330), a mouse monoclonal antibody to 20S proteasome (clone 863425, R&D Systems), or a rabbit monoclonal antibody to PSMA3 (clone D490, Cell Signaling Technology). To characterize the subcellular identities of the piglet brain fractions, a panel of antibodies was used to detect histone H3 (Cell Signaling Technology) for the nucleus, glyceraldehyde phosphate dehydrogenase (GAPDH) (Abcam) for the cytosol, complex V (Life Technologies) for the mitochondria, cytochrome P450 reductase (Upstate Biotechnology), and inositol triphosphate receptor (IP3R) (Fotuhi et al., 1993) for MAMs and endoplasmic reticulum. Proteasome presence in the blots of different subcellular fractions was detected using rabbit polyclonal antibody to PSMB5 (GeneTex, GTX23330) and mouse monoclonal antibody to 20S proteasome (clone 863425, R&D Systems). After the primary antibody incubation, blots were rinsed and then incubated with horseradish peroxidase-conjugated secondary antibody (0.2 μg/mL). For all blots, the primary and secondary antibodies were used at concentrations for visualizing protein immunoreactivity within the linear range. The blots were developed with enhanced chemiluminescence (Pierce) and imaged with a ChemiDoc Imaging System (Bio-Rad, Hercules, CA).

For immunohistochemistry, 2- to 5-day-old naïve piglets (n = 8) and sham piglets (n = 6) that were sedated, intubated, anesthetized, and instrumented for physiological monitoring were deeply anesthetized with pentobarbital 50 mg/kg and phenytoin 6.4 mg/kg (SomnaSol) and, after thoracotomy and left myocardial puncture and aortic catheterization, ice-cold 100 mM PBS (pH 7.4) was perfused (~2 L) for body exsanguination followed by freshly prepared 4% paraformaldehyde (PF) in 100 mM phosphate buffer (pH 7.4) for brain fixation (~4 L). Appropriate tissue fixation was judged by the stiffness of the body and the immovability of the jaw and limbs. After decapitation, the head was placed in 4% PF overnight. The following day, the calvarium was carefully removed by rongeur, and the brain was extracted from the skull base and placed in PF for overnight. Brains were cut in the coronal plane, and samples were paraffin processed in tissue cassettes. The paraffinized brain blocks were cut on a rotary microtome into 10-μm sections that were mounted on gelatin/chrome alum-coated glass microscope slides for proteasome subunit immunohistochemistry (IHC).

Proteasome subunit localization IHC was done using immunoperoxidase and immunofluorescence procedures. Negative controls for both methods were anatomically matched piglet brain sections incubated with rabbit or mouse non-immunized normal IgG isotypes (Sigma-Aldrich) at the same concentrations, and for the same time, as the primary antibodies. Immunoperoxidase IHC, with diaminobenzidine (DAB) as chromogen (immunoreactivity is seen as brown), was done on piglet brain paraffin sections as described (Santos et al., 2018; Lee et al., 2021) to localize PSMB5 with rabbit polyclonal antibody (GeneTex, GTX23330). Primary antibody incubation was done overnight at room temperature in a humidified chamber. This antibody was characterized for specificity in pig brain homogenates using Western blotting. It detects a major band at ~25 kDa (Figure 2A). Nissl counterstaining with cresyl violet (CV) was done for cellular and laminar identifications in the neocortex. Immunofluorescence IHC was done on piglet forebrain paraffin sections as described (Martin et al., 2013) to identify potential subcellular colocalizations of the proteasome. PSMB5 detection was with rabbit polyclonal antibody (GeneTex, GTX23330); mitochondria were detected with mouse monoclonal antibody to cyclophilin D (clone E11AE12BD4, Abcam) and mouse monoclonal antibody to voltage-dependent anion channel (VDAC) (N152B/23, BioLegend). Cyclophilin D (also known as peptidylprolyl isomerase D, PPIF) and VDAC are well-established mitochondrial proteins, and the antibodies to these proteins have been characterized (Martin et al., 2009, 2011; Chen et al., 2021). Secondary antibodies (Thermo Fisher Scientific, Molecular Probes), used at a dilution of 1:400, were AlexaFluor-594 goat-anti-rabbit IgG and AlexaFluor-488 goat-anti-mouse IgG.

Piglet brain sections were viewed and imaged using brightfield and confocal microscopy. For brightfield microscopy, an Olympus BH-2 microscope with a CaptaVision camera and software was used. The confocal imaging was done using a Nikon C2 confocal scanning microscope equipped with a Nikon LUN4 4-line solid-state laser system, a Ti2 motorized stage, a Perfect Focus System-3, and a Nikon NIS-Elements software package or a Leica Mica Microhub Widefield Imaging System.

Confocal microscope images were used to quantify the colocalization of VDAC and PSMB5 immunoreactivities in the neuropil and neuronal perikaryal cytoplasm using a modified method described in Chang and Martin (2009). Pixel particles with green, red, and yellow variant colors were determined per 1 μm² in randomly selected non-overlapping image microzones.

Immunofluorescence IHC was done on human cerebral cortex (postcentral gyrus and superior frontal gyrus) paraffin sections (10 μm), as described (Kageyama et al., 2018), to identify descriptively the cellular localizations of the proteasome using IHC conditions identical to those used for pig brains. Human postmortem autopsy brain samples were obtained from the Johns Hopkins Brain Resource Center as described (Martin, 1999; Kim et al., 2020). The human brain samples used here were from adult male control individuals (n = 3, 59–80 years old, postmortem delays to autopsy, 6–10 h) without any final diagnosis of neurologic disease and from a 1-year-old male. Staining of human and pig brain sections was done concomitantly with sodium citrate (boiling for 20 min) used for antigen retrieval and blocking/permeabilization with 10% NGS and 0.4% Triton X-100 for 2 h. PSMB5 was detected with rabbit polyclonal antibody (GeneTex, GTX23330) at 1:200. A mouse monoclonal antibody (clone 48, BD Transduction Laboratories) to the apoptosis inhibitor protein Aven was used as a general marker for the cytoplasm, nucleus, and mitochondria (Martin et al., 2007) at 1:250. Secondary antibodies were the same as those described for pigs. Coverslipping was done with VECTASHIELD^® containing DAPI. Piglet and human brain sections were imaged with a Mica widefield confocal scanning microscope (Leica Microsystems).

Neonatal piglets (2–3 days old, 1.5–2.0 kg, male) were sedated, intubated, anesthetized, and instrumented with a femoral artery catheter for thorough cardiovascular monitoring before, during, and after intravenous (iv) treatment with drugs delivered through an internal jugular vein catheter tunneled to the back for central venous access to deliver additional drug doses (Figure 3). We used a narrow age range in an effort to reduce variability. Normothermic body temperature was constantly maintained with a warming blanket during the procedure. The test drugs were pyrazolone (PYR) and chlorpromazine (CPZ). The PYR derivative used was 5-[(3,5-dichlorophenoxy)methyl]-1,2-dihydro-3H-pyrazol-3-one, C₁₀H₈Cl₂N₂O₂ (CMB-087229, Millipore-Sigma). The CPZ used was 2-chloro-10-(3-dimethylaminopropyl)phenothiazine hydrochloride, C₁₇H₁₉ClN₂S · HCl (C8138, Sigma-Aldrich, St. Louis, MO). PYR and CPZ were identified by others (Trippier et al., 2014; Jones et al., 2017) as activators of the 20S proteasome in cell culture-based small molecule screens. After establishing hemodynamic baselines with an arterial blood pressure catheter, piglets received an iv-cocktail bolus of PYR (10 mg/kg or 20 mg/kg) and CPZ (1 mg/kg) delivered over 2 min. In independent experiments, PYR and CPZ were given alone to piglets (data now shown). PYR was well tolerated at 10 and 20 mg/kg doses. However, higher concentrations of CPZ (>10 mg/kg) produced unacceptable lethargy, hypotonia, and acquired dystonia in piglets. This is why a lower dose of CPZ was used. Control piglets received vehicle (iv DMSO/ethanol/saline). About 3 h later, after blood pressure and heart rate monitoring, the piglets had their arterial catheter removed and the wound closed. They were awakened from anesthesia and extubated. A second dose of PYR/CPZ or vehicle was delivered iv 12 h after the first dose. A subset of piglets received a third iv dose of PYR/CPZ or vehicle 6 h later (because of insufficient power, these animals were combined into the total group of piglets receiving 20 mg/kg PYR and 1 mg/kg CPZ).

Twenty-four hours after the initial dosing, the piglets received an overdose of SomnaSol and were quickly thoracotomized for cardiac puncture, aortic cannulation, and exsanguinated with ice-cold PBS perfusion (~2 L). Effective and complete exsanguination was necessary for removing proteasome-rich erythrocytes, with their ~20-fold excess of 20S proteasome over 26S (Neelam et al., 2011), from the brain. Brain gray and white matter regions (Figure 1) were harvested for proteasome chymotrypsin-like (CTL) activity (Brown and Monaco, 1993; Kisselev and Goldberg, 2005), Western blotting, an in-gel proteasome activity assay (Yazgili et al., 2021), or a novel in situ proteasome assay.

Native protein extracts were used for biochemical assays of functional proteasomes. Tissue samples were homogenized and sonicated in ice-cold tissue lysis buffer (40 mM Tris and 50 mM NaCl in distilled water at pH 7.2 with 10% glycerol), 2 mM β-mercaptoethanol (Sigma-Aldrich), 5 mM MgCl₂, and 2 mM adenosine 5′-triphosphate (ATP) (UBPBio, Dallas, TX). The samples were then centrifuged for 30 min at 4°C. The supernatant was collected and kept on ice for protein measurements using a bicinchoninic acid assay kit (Smith et al., 1985; Pierce). Proteasome CTL activity was measured using a fluorometric assay kit (UBPBio). Proteasome assay buffer was prepared by adding 50 μL of 20× proteasome assay buffer to 950 μL of sterile water at 37°C. The substrate was prepared by adding 2 μL of 1,000× Succinyl-Leu-Leu-Val-Tyr-7-amino-4-methylcoumarin (Suc-LLVY-AMC) stock to the assay buffer. All samples (50 μg total protein input) were assayed in duplicate with and without MG132 in 96-well, black-bottom plates. The proteasome inhibitor MG132 (Lee and Goldberg, 1998) was added to a parallel set of sample wells to distinguish peptidase activities. Suc-LLVY-AMC is also degraded by calpains, cathepsins, and other CTL proteases (Rock et al., 1994). Adding MG132 isolates proteasome substrate cleavage from non-proteasome proteolysis. Each plate included an AMC standard (blank, 1:500, 1:1,000, 1:5,000, and 1:10,000) to generate a concentration-dependent AMC fluorescence standard curve. Liberated AMC fluorescence from proteolysis of Suc-LLVY-AMC was immediately measured by a plate reader (TECAN Infinite M Plex) with 360 nm excitation/460 nm emission filters for 15 min of assay reaction time at 37°C. Each sample’s AMC fluorescence was determined from the AMC standard curve and reported as relative fluorescence units. The AMC fluorescence from tissue homogenate with MG132 was subtracted from that of homogenate without MG132 to obtain relative fluorescence levels specific to proteasome Suc-LLVY-AMC peptidolysis. Replicates were averaged, and their corresponding MG132-controlled wells were subtracted. Femtomoles of AMC released were calculated from relative fluorescence units using a standard curve of purified AMC included in each microplate assay. The AMC standard curve slope for each plate was standardized to the average activity slope of MG132-inhibited wells with cerebellar cortex input.

For the in-gel proteasome activity assay, we used the same tissue homogenates that were prepared under non-denaturing conditions and used for the proteasome fluorometric microplate assay. Native PAGE was done using 3–8% Tris-acetate gels. Purified human 20S from human erythrocytes (Enzo Life Sciences) was used as a positive control. The resulting gels were incubated for 30 min at 37°C in a dark box in a reaction solution containing Tris (100 mM, pH 7.5), ATP/MgCl₂, dithiothreitol (DTT), and Suc-LLVY-AMC (2 mM). Negative control gels were incubated in a reaction solution containing MG132 (10 μM). Proteasome CTL enzymatic activity in gels was imaged with a ChemiDoc imaging system (Bio-Rad).

We developed a novel in situ proteasome enzyme histochemical method to localize active proteasome in brain sections. Freshly frozen (unfixed) blocks of piglet somatosensory were cut (10 μm thick) on a cryostat and mounted on gelatin/chrome alum-coated glass microscope slides. The slides were stored at −80°C in closed boxes. Sections were incubated in the dark for 2 h at 37°C with a CTL proteasome reaction solution containing 100 mM Tris (pH 7.5), 250 mM ATP/MgCl₂, 1 mM DTT, 50 mM Suc-LLVY-2R110 (AAT Bioquest), and 20% polyvinyl alcohol. Negative control sections were pretreated with MG132 and incubated in a reaction solution containing MG132 (10 μM). Afterwards, the slides were rinsed in Tris buffer, dipped in 4% PF/20% sucrose for 5 min, counterstained with DAPI, and coverslipped. Sections were viewed with a Zeiss Axiophot microscope under epifluorescence and imaged with SPOT software. Suc-LLVY-2R110, upon CTL cleavage, generates a fluorescent product that is bright green (excitation 498 nm/emission 520 nm) when viewed with a fluorescein isothiocyanate filter set.

GraphPad Prism 9 was used to analyze data. Brain regional proteasome activity data passed Shapiro–Wilk tests for normality and were compared using Kruskal–Wallis ANOVA on ranks. Correlations between nasal biopsy and brain regional proteasome activities were examined by Spearman correlation.

Western blotting was done to determine relative protein levels of the β5 proteasome subunit (PSMB5) in different regions of the neonatal pig brain (Figure 2A). Immunoreactivity for the β5 proteasome subunit was detected at ~25 kDa, consistent with expectations from the work of others (Sutovsky et al., 2004) and us (Santos et al., 2018; El Demerdash et al., 2021). In precisely microdissected white matter regions (Figure 1) such as parietal cortex subcortical white matter, corpus callosum, and internal capsule β5 levels were similar (Figure 2A). Microdissected gray matter regions of cerebral cortex (primary somatosensory cortex and frontal cortex) hippocampus, caudate nucleus, and cerebellar cortex β5 levels were similar to each other and to white matter (Figure 2A). Nasal mucosa biopsy samples from piglets also had enrichment of β5 immunoreactivity like the different brain regions (Figure 2A). From the Western blotting, we deemed the PSMB5 antibody suitable for immunohistochemical application on pig brain (Figures 2B–H). The α3 proteasome subunit (PSMA3) was also examined in pig brain by Western blotting (Figure 2I). The PSMA3 immunoreactive band was also detected at ~25 kDa, like the mobility of PSMB5 (Figure 2A). In selected brain regions, PSMA3 levels were similar in gray matter and white matter areas (Figure 2I).

IHC was used to localize PSMB5 in different regions of the piglet brain (Figures 2B–H and Supplementary Figures 1A–C). PSMB5 immunoreactivity was ubiquitous and not particularly defining of any brain regions. Essentially, every neuron in each layer of the cerebral cortex was positive (Figures 2B–E). The somatosensory cortex (Figure 2B) had a distinct layer 4 that was inconspicuous in the motor cortex (Figure 2C), consistent with other observations (Primiani et al., 2023). In addition to the horizontal laminar patterns, vertical column arrangements of cell bodies, organized orthogonally to the cortical surface, were visible in the primary somatosensory cortex (Figure 2B, yellow lines). These cortical columns had a width of 4–5 neuronal perikarya (40–50 μm in total) and a periodicity of about 50 μm. Virtually all pyramidal neurons in the hippocampus were positive for PSMB5 (Figure 2F). The principal neurons in the striatum were generally positive (Figure 2G). Glial cells in gray matter and white matter were positive or negative for PSMB5 (Figures 2E,G,H). PSMB5 was localized conspicuously in the nuclei of neurons (Figures 2D–G) and glia (Figures 2D,G,H) where it was strongly in the nucleoplasmic matrix but typically not within the nucleolus. In the cytoplasm and neuropil, PSMB5 immunoreactivity had an amorphous diffuse and a particulate appearance (Figures 2D,G).

We performed an experiment to compare the localization of PSMB5 in neonatal pig neocortex to young and adult human disease-free neocortex concomitantly using identical immunofluorescence methods (Supplementary Figure 1). In pigs (Supplementary Figures 1A–C), the PSMB5 staining pattern seen by immunofluorescence was like that seen by immunoperoxidase (Figures 2B–E). There was prominent nuclear localization in virtually all neurons in each cortical layer. PSMB5 was also present in the cytoplasm as a diffuse overlapping pattern with immunoreactivity for the antiapoptotic protein Aven (Supplementary Figures 1B,C).

The patterns of PSMB5 immunoreactivity in human neocortex in young (1-year-old) and aged (59–80 years old) individuals (Supplementary Figures 1D–H) were variations of the pattern seen in neonatal pigs. In the 1-year-old brain, most neocortical neurons in all layers were PSMB5-positive with prominent nuclear labeling but low colocalization with Aven (Supplementary Figures 1D–F). There was an inhomogeneity of immunoreactivity in the neuropil, with some areas slightly higher in PSMB5 immunoreactivity than nearby areas (Supplementary Figure 1D, white hatched outlines). In the older human brain, neurons were positive or negative for PSMB5 (Supplementary Figures 1G,H) compared to virtually all cerebrocortical neurons being positive in neonatal pig and young human brain (Figures 2B–E and Supplementary Figures 1A–C). The superficial layers (2 and 3) had more prominent neuropil PSMB5 immunoreactivity than the deeper layers. Positive neurons in aged human brains generally had more apparent PSMB5 immunoreactivity in the cytoplasm than in the nucleus (Supplementary Figure 1H), while in neonatal pigs and young human brain, PSMB5 nuclear immunoreactivity was prominent. The Aven immunoreactivity in the cytoplasm was very discreetly and strongly colocalized with PSMB5 in older human neocortical neurons in a pattern suggestive of mitochondria (Supplementary Figures 1E,F), which contrasted with the infant brain.

We performed a bioinformatic analysis on the human PSMB5 full-length amino acid sequence using MitoFates (Fukasawa et al., 2015) to search for mitochondrial targeting and mitochondrial processing peptidase (MPP) cleavage sites. This software has better discriminative and predictive performance than other predictive tools (Fukasawa et al., 2015). A well-known mitochondrial matrix protein, CyPD (PPIF), was used as a comparator protein. PSMB5 registered a low probability N-terminal mitochondrial targeting sequence (probability = 0.002) compared to CyPD (probability = 0.982). However, PSMB5 had an N-terminal MPP site in the vicinity of amino acid 17 and a C-terminal VKKVI TOM20 recognition motif (Saitoh et al., 2007).

The subcellular localization of PSMB5 was explored directly in optimally prepared pig brains using immunofluorescence and confocal microscopy. PSMB5 immunoreactivity was enriched in the nucleus and cytoplasm (Figures 4A,B and Supplementary Figures 1A–C). PSMB5 colocalized partially with two different well-known mitochondrial markers (VDAC and cyclophilin D) in the cytoplasm of neurons and in the neuropil of the cerebral cortex and striatum (Figures 4A,B). Quantification of the colocalization showed that about half of the total mitochondria in the neuropil and perikaryal cytoplasm detected with VDAC colocalized with PSMB5 (Figures 4B,C). There was a significantly greater presence of PSMB5 in perikaryal cytoplasmic mitochondria compared to neuropil mitochondria (Figures 4B,D).

To confirm that the proteasome is possibly associated with mitochondria, piglets were completely exsanguinated by vascular perfusion of ice-cold PBS (to remove proteasome-enriched contaminant erythrocytes), and the fresh (not frozen) brain tissue was used immediately for subcellular fractionation into nuclear, cytosolic, crude mitochondrial, pure mitochondria, and mitochondrial-associated membranes (Figure 4E). The effectiveness of the subcellular fractionation method was confirmed using a panel of antibodies to different cellular constituents (Figure 4F). Histone H3, seen as a 17 kDa band, was detected only in the crude homogenate and in the nuclear fraction (Figure 4F). Complex V (ATP synthase), seen as a 57 kDa band, was concentrated in the pure mitochondrial fraction (Figure 4F). Critically, GAPDH, a cytosolic marker, was not present in the pure mitochondria fraction (Figure 4F), and thus, the pure mitochondrial fraction was deemed free of cytosolic proteasome contamination. The MAM fraction was enriched in IP3R, seen at 260–300 kDa, and cytochrome P450 reductase, detected at ~78 kDa (Figure 4F).

Fresh unfrozen naïve piglet brain (N = 2) was microdissected and subjected to tissue subcellular fractionation into cytoplasmic (soluble), nuclear, and pure mitochondrial compartments that were compared to the total activity in crude sonicated homogenate (Figure 5). Western blotting confirmed the brain tissue subcellular fractionation method (Figure 4F). The results were remarkably consistent among different pigs (Figure 5A), though this is only a descriptive pilot experiment with a small number of naïve piglet brains. The majority of proteasome CTL activity was in the cytosolic fraction (Figure 5B). About 40% of the total activity was in the nucleus (Figure 5B), and about 15–20% was in the pure mitochondrial fraction (Figure 5B).

The proteasome CTL activity assay was validated. Proteasome activity was detected as an increase in fluorescence units over time that was blocked effectively by MG132 (Figure 6A). Total AMC fluorescence at the reaction termination (15 min) was reduced by ~85% by MG132 (Figure 6A). This level of proteasome inhibition was reproduced in every independent assay (more than 20 different experiments) and was shown elsewhere (El Demerdash et al., 2021).

In naïve neonatal pigs, no significant difference in activity was observed among different brain regions, including gray and white matter areas (Figure 6B). Proteasome activity in different gray and white matter regions was examined based on initial t₀ activity, peak fluorescence (the end of the reaction progress curve), and the slope of the reaction progress curve. These parameters did not differ among brain regions, except for the corpus callosum (Figure 7). The t₀ proteasome activity in the corpus callosum was significantly higher (p < 0.05) compared to the internal capsule and the parietal subcortical white matter (Figure 7D). Proteasome activities in the nasal mucosa and in different brain regions were examined for correlations (Figure 8). In white matter, no significant correlations were observed (Figures 8A,B). For example, there was no relationship between proteasome peak activity in the internal capsule and nasal mucosa proteasome activity (Spearman’s r = 0.80, p = 0.14). However, the proteasome activities in the nasal mucosa and some gray matter regions were correlated (Figures 8D–F). Proteasome t₀ activity in the hippocampus is correlated with nasal mucosa proteasome activity (Spearman’s r = 0.9746, p = 0.0001). Proteasome peak activity in the hippocampus is correlated with nasal mucosa proteasome activity (Spearman’s r = 0.900, p = 0.02). In contrast, the hippocampal proteasome activity slope did not correlate with nasal mucosa proteasome activity (Spearman’s r = 0.632, p = 0.23). Caudate nucleus proteasome t₀ activity did not correlate with nasal mucosa proteasome activity (Spearman’s r = 0.80, p = 0.065).

Piglet hemodynamics were monitored for 3 h after iv drug/vehicle delivery, after which piglets were recovered from anesthesia and survived. The average mean arterial pressure was 65 ± 4 mmHg in drug-treated piglets and 72 ± 6 mmHg in vehicle-treated piglets. The average heart rate was 166 ± 18 beats/min in drug-treated piglets and 198 ± 25 beats/min in vehicle-treated piglets.

Piglets were treated (2–3 doses) with iv 10 mg/kg PYR or 20 mg/kg PYR with both dosages in a cocktail with 1 mg/kg CPZ (Figure 3). Twenty-four hours after the first dose, brain tissues (Figure 1) were harvested freshly for proteasome assays (Figure 9). In an analysis of eight different brain regions, though trends were apparent for dose-related changes in proteasome activity with PYR/CPZ, proteasome activity was not increased significantly (p < 0.05) above vehicle or naïve for any brain region (Figure 9A). When brain regions were stratified by functional network, trends were again apparent for increased proteasome activity with PYR/CPZ, but they did not achieve significance (Figure 9B). When all brain regions were grouped (Figure 9C), piglets treated with 20 mg/kg PYR/1 mg/kg CPZ had significantly higher proteasome activity compared to naïve piglets (p = 0.0006), vehicle-treated piglets (p < 0.0001), and 10 mg/kg PYR/1 mg/kg CPZ-treated piglets (p = 0.0004). Piglets treated with 10 mg/kg PYR/1 mg/kg CPZ did not differ significantly from naïve and vehicle-treated piglets.

Western blotting was done to determine if PYR/CPZ treatment altered the levels of PSMB5 in the somatosensory cortex (Supplementary Figure 2A) at 24 h after treatment (see Figure 3 for treatment protocol). PSMB5 levels were not significantly different (p > 0.05) in drug-treated piglets compared to vehicle-treated piglets (Supplementary Figure 2B).

We did in-gel assays to determine the activity and composition of functional proteasome complexes in the piglet cerebral cortex (Figure 10). In-gel proteasome assays revealed 26S-singly capped and 30S (26S-doubly capped) active complexes in piglet neocortex (Figure 10A). These activity bands were inhibited by 10 μM MG132 (Figure 10B), demonstrating that they are proteasomes (Lee and Goldberg, 1998). We did not identify free 20S proteasome in the piglet motor cortex (Figure 10A).

In a side-by-side descriptive comparison of in-gel activities with short incubation times, piglets treated with PYR/CPZ showed enhanced neocortical proteasome activity compared to vehicle-treated piglets. Activity enhancement in somatosensory cortex with PYR/CPZ treatment occurred in the 26S-singly capped proteasome and as a smear of activity between 26S-singly capped and 30S (Figure 10C). No enhancement of activity was seen for free 20S.

We adapted the proteasome microplate fluorometric biochemical assay to cryostat sections of unfixed piglet brain, naming it the in situ proteasome activity assay (Figure 11). With Suc-LLVY-2R110 as substrate and the addition of 20% polyvinyl alcohol to the reaction mixture as an enzyme stabilizing and diffusion limiting agent and tissue protectant (Chayen et al., 1973; Martin et al., 1986; van Noorden and Vogels, 1989), in situ proteasome activity was histochemically localized in brain tissue (Figure 11). Proteasome CTL activity was present throughout the neuropil and in isolated cell bodies, putatively neurons and glia (Figures 11A,B,E). The activity was completely inhibited by MG132 (Figure 11C). In contrast to the antibody-based IHC (Figure 2), surprisingly, the proteasome CTL activity was not uniformly or homogeneously distributed throughout the neuropil; rather, it had a patchy localization in the primary somatosensory cortex, with some microregions showing higher activity than nearby regions (Figures 11A,B). Some cell bodies appeared full of proteasome activity (Figure 11E), but other cells had only very discrete inclusions of activity within the nucleus (Figure 11D) or at a perinuclear location reminiscent of the centrosome (Figures 11B,D). The number of cells with perinuclear inclusions of CTL proteasome activity was higher in the cerebral cortical gray matter compared to the subcortical white matter (Figure 11F).

After carefully characterizing antibodies to PSMB5 and PSMA3 for specificity by Western blotting and determining their acceptability for use in immunohistochemistry, we mapped the localization of PSMB5 in 2- to 5-day-old piglet forebrain. PSMB5 was present in virtually all neurons but only in subsets of glial cells. It had low regional specificity. In the neocortex, the PSMB5 distribution was discerning of laminar and columnar arrangements of neurons. In the hippocampus, PSMB5 was present in essentially all pyramidal neurons within each subfield. In the striatum, PSMB5 delineated virtually all neurons but not all glia. The striking neuronal positivity for PSMB5 in the normal pig brain contrasted with the moderate to low PSMB5 immunoreactivity described for the human control cerebral cortex (van Scheppingen et al., 2016). We did our own descriptive analysis of PSMB5 localization in human neocortex and found that age is likely to be an important consideration for PSMB5 immunoreactivity. A caveat of antibody-based localization and protein levels is that the information is devoid of functionality.

Proteasomes are known already for their cellular localizations in yeast and many mammalian cell types in culture (Peters et al., 1994; Rivett, 1998; Rockel et al., 2005). In human hepatoblastoma cell subnuclear fractions, proteasome α subunits were detected in the nucleoplasm but not in the nucleolus or nuclear envelop (Rockel et al., 2005). Less information is available for neurons in situ (Adori et al., 2005, 2006). Proteasome localization in optimally prepared brains from different animals is worthy because species differences might exist in the neuronal and glial localizations of proteasomes (Adori et al., 2006). The nuclear and cytoplasmic comparisons are also of importance because there is evidence that the nuclear proteasome lacks activity (Dang et al., 2016). The cellular localization of PSMB5 in neonatal pig neurons was conspicuous for its cytoplasmic and nuclear immunoreactivities. Much of the nuclear PSMB5 appeared to be in the nucleoplasmic matrix, but it was not seen within the nucleolus as predicted from other studies (Rockel et al., 2005). Cytoplasmic and nuclear localizations of PSMB5 were described for human neurons (Adori et al., 2006; Nakamura et al., 2006) and rat neurons (Adori et al., 2006).

There could be differential enrichments of PSMB5 in cellular compartments in pig neurons (Figure 2) compared to those shown for human neurons (Adori et al., 2005). We therefore did a pig and human brain side-by-side comparison with the same proteasome antibody, staining protocol, and imaging (Supplementary Figure 1). We found that aged adult (59–80 years old) human brain neocortical neurons show greater variation in their proteasome enrichments compared to neonatal pig (2–5 days old) neurons. For example, virtually every neuron in the neonatal pig neocortex was strongly positive for PSMB5; in contrast, many cortical neurons in the older human postcentral gyrus were negative for PSMB5 or had very low immunoreactivity, but other nearby neurons were strongly positive. In some human cortical neurons, PSMB5 immunoreactivity was enriched in the cytoplasm but low in the nucleus, while neocortical neurons in neonatal pigs universally had enrichment in the nucleus (Supplementary Figure 1). Age could be a factor in these differences, so we examined the brain of a 1-year-old infant. We found that the PSMB5 localization pattern in the human infant brain more closely resembles the neonatal piglet localization than does the brain of older humans (Supplementary Figure 1). This pilot experiment suggests brain maturation and aging have important roles in the neuronal localization of at least one proteasome subunit in the cerebral cortex.

The proteasome has variations in forms attributed to protein regulators. The specific forms of proteasome have different functions. The free 20S proteasome, without regulators, can selectively degrade oxidized and misfolded proteins through a facilitated unfolding mechanism in a ubiquitin- and ATP-independent manner (Pickering and Davies, 2012). Two 19S regulators can be added at either end of the 20S barrel to generate 26S-singly and -doubly capped forms, serving to limit protein targets to those that are ubiquitinated in a degradation process that is ATP-dependent (Pickering and Davies, 2012). In our in-gel activity assay, we found that the 26S-singly capped proteasome type dominated in the naïve piglet motor cortex with scant presence of the free 20S form. This proteasome composition differs from that of HeLa cells and rat tissues, where free 20S proteasome represented 40–60% of the total 20S (Tai et al., 2010). This might mean that in a naïve piglet brain with replete ATP, misfolded proteins are degraded by the ATP-dependent 26S proteasome and oxidized proteins are not formed excessively because endogenous oxidative stress is managed effectively by antioxidant systems. However, we have detected protein carbonyl accumulation in naïve piglet brain (Santos et al., 2018). More work needs to be done in piglet brain injury models to explore the presence of free 20S proteasome.

An important controversy is whether cytoplasmic and nuclear proteasomes both possess protease activity. Some studies have identified proteasome proteolysis activity in the nucleus (Kim et al., 1999; Ullrich et al., 1999; Rockel et al., 2005), but another study has not (Dang et al., 2016). Proteasome subunits in the nucleus might have non-peptidase functions involving DNA repair and transcription factor-like mechanisms (Russell et al., 1999; Xie and Varshavsky, 2001). Using a validated subcellular fractionation protocol combined with proteasome CTL activity assay, we showed that both nuclear and cytoplasmic compartments possess major biochemical activities. The cytosolic proteasome is the main contributor to the total catalytic activity in crude brain tissue homogenates.

More novel is our discovery that the proteasome can localize to mitochondria, where it possesses CTL activity. This finding was demonstrated microscopically and by subcellular fractionation, Western blotting for PSMB5, and an activity assay. By immunofluorescence and confocal microscopy, we found that PSMB5 partly colocalized with VDAC, cyclophilin D, and Aven in pig forebrain neurons and human neocortex. VDAC and cyclophilin D are well-known mitochondrial proteins (Martin et al., 2009, 2011). Though cytoplasmic proteasomes have recently been associated with mitochondrial protein quality control in a process in yeast called mitochondria-associated degradation (Rödl and Herrmann, 2023), a mitochondrial localization of proteasome has not been shown before. The proteasome could be tethered to the outer mitochondrial membrane, like DNA methyltransferase-1 (Wong et al., 2013), potentially associated with the translocator of the outer mitochondrial membrane (Bertolin et al., 2013), or within the intermembrane space, inner membrane, or mitochondrial matrix. Sub-mitochondrial fractionation and intact mitochondria-protease digestion experiments need to be done to decipher how the proteasome is associated with mitochondria. Intriguingly, the mitochondrial associated proteasome has CTL activity. It contributes about 15–20% of the total activity of crude homogenate in the cerebral cortex.

A mitochondrial-associated proteasome would be a complement to known mitochondrial ATPases associated with diverse cellular activities (AAA+) proteases found within the intermembrane space (e.g., iAAA), inner membrane (e.g., SPG7/paraplegin), and matrix (e.g., LONP1 and ClpXP) (Feng et al., 2021). The AAA+ mitochondrial proteins have many non-peptolytic functions, in addition to protease activities that degrade preferential mitochondrial protein targets that are misfolded or oxidatively damaged. For example, LONP1 functions in mitochondrial DNA replication and the degradation of oxidatively damaged aconitase (Bota and Davies, 2002). SPG7/paraplegin may be a critical component of the mitochondrial permeability transition pore (Shanmughapriya et al., 2015) and assemble mitochondrial ribosomes, in addition to proteolyzing damaged cytochrome c oxidase subunit 1 (Almontashiri et al., 2014). Interestingly, there are several non-mitochondrial cytoplasmic proteins in aberrantly misfolded, oxidized, cleaved, and otherwise pathological states (e.g., SOD1, αSyn, β-amyloid) that gain access to mitochondria to induce mitochondriopathy (Martin et al., 2006, 2007; Ruan et al., 2017). These abnormal proteins might not be substrates for highly specialized and dedicated mitochondrial AAA+ proteases. A mitochondrial-associated proteasome with broad peptidolytic capacity could provide an adaptive advantage to cells, and possibly individuals, for degrading xenoproteins binding to the surface of mitochondria or anomalously gaining access to mitochondria with potential toxicities, such as SOD1, αSyn, and β-amyloid. However, caution is warranted regarding evidence for an intrinsic mitochondrial proteasome until a comparative protease substrate profile is done with mitochondrial AAA+ proteases. While Suc-LLVY-AMC is a preferred substrate of PSMB5, and this proteasome subunit was detected in lysates of pure mitochondria, there could be some, albeit non-preferred, substrate overlap with ClpXP (Arribas and Castaño, 1993).

We adapted the standard homogenate-based microplate fluorometric assay for proteasome CTL activity to an unfixed brain cryostat section assay. This assay is novel. The Y-2R110 reaction product formed by CTL cleavage of Suc-LLVY-2R110, stabilized by polyvinyl alcohol, was completely inhibitable by MG132. With this assay, proteasome CTL activity was visualized in cells and in the neuropil. This histological-based assay yielded different data compared to the antibody-based IHC that gave the impression of brain regional uniformity and generalized enrichment in neurons. However, the histochemical proteasome activity assay suggests a cellular heterogeneity and proteasome complexity hitherto undisclosed by homogenate- and antibody-based assays.

The histochemical activity assay was used on the piglet somatosensory cortex. It revealed that neurons and glial cells do not appear to be enriched similarly in proteasome CTL activity. It also directly showed nuclear proteasome catalytic activity. Within the neuropil, there was patchiness in catalytic activity, giving the impression of a functional mosaic of the proteasome in the cerebral cortex. We also found a distinguished proteasome compartment residing in a perinuclear position. We observed it with difficulty by standard immunolocalization, but our new in situ histochemical activity assay particularly highlighted this organelle because of the striking CTL activity. This perinuclear proteasome satellite might be the centrosome-associated proteasome seen structurally and biochemically in cultured HEK293 and HeLa cells (Wigley et al., 1999; Fabunmi et al., 2000). The structure was more common in cortical gray matter than in subcortical white matter, though it was not seen in every cell. To the best of our knowledge, this structure has not been described before in the brain. It might correspond to the aggresome described in yeast (Enenkel et al., 2022), but its robust visualization in naïve piglets suggests that this structure is not pathological and was formed only in response to an overload of misfolded proteins. If this perinuclear CTL activity-enriched proteasome satellite seen in the pig brain is centriole-related, then the centriole pair would be embedded in a matrix enriched in pericentriolar material 1 protein (PCM1) (Dammermann and Merdes, 2002). We have discovered that PCM1 is particularly sensitive to carbonyl oxidation in neonatal piglets with hypoxia–ischemia and treatment with hypothermia (El Demerdash et al., 2021) and, thus, the centriole might destabilize in neonatal brain injury. Because the centriole functions as a site for microtubule assembly for neurites and dendrites (Yu et al., 1993), PCM1 oxidation could be a pathological mechanism for dendritic or oligodendrocyte neurite beading and obliteration that we have seen in piglet encephalopathy (Lee et al., 2021).

We studied the not-often-discussed presence of the proteasome in the neuropil throughout the piglet forebrain. We observed this by immunohistochemistry and with our novel in situ enzyme histochemical assay. The catalytically active proteasome in the neuropil was completely inhibited by MG132. This brain parenchymal neuropil proteasome distribution appeared elegantly macro-compartmentalized as a mosaic in pig and human neocortex. A neuropil proteasome is interesting because it could be mitochondrial associated, as supported by the data shown here, and synaptic. A synaptic proteasome has been suggested to participate in endosomal uptake of glutamate receptors (Bingol and Schuman, 2004), activity-dependent structural plasticity (Bingol and Schuman, 2006), and long-term potentiation (Fonseca et al., 2006; Dong et al., 2008). Neuropil proteasomes could also be glial-associated (Tydlacka et al., 2008). It is unknown whether pharmacologically forced activation of proteasome activity would have deleterious effects on these proteasomal functions.

We have identified an apparently safe drug combination to increase CTL proteasome activity generally throughout the neonatal gyrencephalic brain with systemic treatment. In piglets treated intravenously with boluses of pyrazolone derivative and chlorpromazine cocktail over 24 h, brain proteasome CTL activity was modestly increased. Western blotting for PSMB5 indicated that upregulation of PSMB5 levels did not account for this catalytic increase (Supplementary Figure 2). Pyrazolone analogs are known to bind and activate the proteasome in cell-free and cell systems (Tripper et al., 2014) and showed promise in a mouse model of ALS involving mutant SOD1 proteinopathy. Chlorpromazine, a clinical drug used safely for over 5 decades, also activates the proteasome in cell systems (Jones et al., 2017). In piglets treated with CPZ, the sedation and motor effects were dose-limiting. Proteasome activation with PYR/CPZ seems possible for the cerebral cortex and deep forebrain, gray matter, and white matter. Importantly, these regions are disease- and injury-vulnerable in many human neurological conditions in which proteasome insufficiency is pathogenically relevant. Many future experiments are needed, including additional PYR dose ramping, drug effects on the other two catalytic activities of the proteasome, pharmacokinetics and brain half-life determinations, brain effect durations, and long-term dosing with toxicological and behavioral profiling for safety.

We discovered that proteasome functional activity in some brain regions correlates with proteasome activity in the nasal mucosa. Hippocampal and nasal mucosal activities are notably correlated. Even the protein levels of PSMB5 in the nasal mucosa matched those of several different brain regions. This finding could be relevant to CNS drug therapy development. A substantial impasse in CNS drug discovery for human brain and spinal cord injury and disease is demonstrating bioavailability and target engagement in the CNS of patients (Kingwell, 2011). If the nasal mucosa epithelial lining can serve as a surrogate for CNS proteasome activity, then this could advance the preclinical search for proteasome-related drug development of therapeutics for neurological disorders related to proteinopathy.