Neuroinflammation inhibition by small-molecule targeting USP7 noncatalytic domain for neurodegenerative disease therapy.

Neuroinflammation is a fundamental contributor to progressive neuronal damage, which arouses a heightened interest in neurodegenerative disease therapy. Ubiquitin-specific protease 7 (USP7) has a crucial role in regulating protein stability in multiple biological processes; however, the potential role of USP7 in neurodegenerative progression is poorly understood. Here, we discover the natural small molecule eupalinolide B (EB), which targets USP7 to inhibit microglia activation. Cocrystal structure reveals a previously undisclosed covalent allosteric site, Cys576, in a unique noncatalytic HUBL domain. By selectively modifying Cys576, EB allosterically inhibits USP7 to cause a ubiquitination-dependent degradation of Keap1. Keap1 function loss further results in an Nrf2-dependent transcription activation of anti-neuroinflammation genes in microglia. In vivo, pharmacological USP7 inhibition attenuates microglia activation and resultant neuron injury, thereby notably improving behavioral deficits in dementia and Parkinson's disease mouse models. Collectively, our findings provide an attractive future direction for neurodegenerative disease therapy by inhibiting microglia-mediated neuroinflammation by targeting USP7.

INTRODUCTION

The current drug development for human neurodegenerative diseases is a high-risk field; therefore, the identification of a promising therapeutic drug target based on complex pathological mechanisms is urgently needed (). Neuroinflammation occurs when the immune system reacts to various signals, such as infection, traumatic brain injury, ischemia, autoimmunity, or toxic metabolites, within the central nervous system (CNS) (). Accordingly, neuroinflammation has attracted considerable attention for their critical roles in neurodegenerative disorders as a common risk factor (, ). Typically, neuroinflammation is initiated by microglia, which are the resident immune cells of the CNS. Numerous activated microglia are widely found in the brains of preclinical models and patients with neuroinflammation pathology (). In this scenario, efficacious therapeutics by microglia inhibition may benefit the patients with inflammation-associated neuronal disorders, including dementia, Parkinson’s disease (PD), Huntington’s disease, and amyotrophic lateral sclerosis (). Therefore, it is of prime significance to discover novel therapeutic targets and potential drugs for microglia-mediated neuroinflammation.

Ubiquitination is a posttranslational modification that generally directs protein degradation in a wide variety of biological events (–). During this process, ubiquitin (Ub) molecules can be covalently attached to a lysine residue of the target protein through E1, E2, and E3 enzymes (). Meanwhile, deubiquitinating enzymes (DUBs) can reverse this process by removing Ub moieties (). Accumulating studies have shown that aberrant DUB function is implicated in multiple human diseases (–), providing DUBs as attractive drug targets. Thus, accumulating reports have demonstrated significant values in targeting ubiquitin-specific proteases (USPs)/DUBs for clinical translation (, ).

USP7 is an essential member of DUBs in all eukaryotes (). As a cysteine protease, USP7 cleaves the isopeptide bond between the C terminus of ubiquitin and ε-amine of lysine in substrate protein, causing a wide spectrum of biological effects, including DNA damage response, viral infections, tumorigenesis, and even physiology (, ). Structurally, USP7 is a multidomain protein that contains an N-terminal tumor necrosis factor (TNF) receptor–associated factor (TRAF) domain, a hand-like catalytic domain (CD), and a C-terminal HUBL domain that consists of five successive ubiquitin-like domains (Ubl 1 to Ubl 5). In particular, 19 C-terminal residues in HUBL domain play a fundamental role in USP7 function by binding to CD to stabilize USP7 conformation (, ). In the past decade, extensive studies focused on the important role of USP7 in regulating the p53-Mdm2 network for anticancer therapy (, ). However, whether USP7 regulates the microglia-mediated neuroinflammation process in neurodegenerative diseases is still poorly understood.

In this study, we identified a USP7-targeting natural small molecule, eupalinolide B (EB), with an obvious inhibition effect on microglia activation. Moreover, we presented a cocrystal structure of the noncatalytic HUBL domain of USP7 with EB at 2.35 Å, revealing Cys576 as an unprecedented covalent allosteric site for USP7 inhibition. Systematic screening revealed Kelch-like ECH-associated protein 1 (Keap1) as a key USP7 substrate protein. Inhibition of USP7 contributes to ubiquitination-dependent degradation of Keap1 and transcription factor Nrf2 activation, further inducing an array of antioxidant response element (ARE)–dependent gene expression for microglia inhibition. Furthermore, EB exerts an obvious therapeutic effect on microglia-mediated neuroinflammation by targeting USP7 in vivo and markedly alleviates neurodegenerative pathology on the senescence-accelerated dementia and PD mice.

In summary, our findings provide a fresh perspective for USP7-targeting therapeutics for neurodegenerative diseases via microglia inactivation strategy. Moreover, our studies underline the importance of noncatalytic HUBL domain in USP7 as a hotspot for structure-based drug design of USP7 inhibitors.

RESULTS

USP7 serves as a cellular target of anti-neuroinflammation small-molecule EB

During previous high-throughput screening assays in our laboratory, we identified the natural-derived small-molecule EB from Eupatorium lindleyanum to obviously inhibit BV-2 microglia activation by down-regulating inflammatory mediators including nitric oxide (NO), interleukin-6 (IL-6), and TNF-α (Fig. 1A and fig. S1A). The effect of EB was also verified in primary cultured mouse microglia (fig. S1B), while EB alone (10 μM) did not show an obvious effect on the proliferation of BV-2 cells (fig. S1K). Then, we used EB as a chemical probe to explore the potential cellular target for microglia inactivation. First, a biotin-tagged EB (bio-EB) was synthesized (Fig. 1B and fig. S1, C and D). Then, bio-EB was used to probe HuProt proteome microarray, followed by incubation with Cy3-conjugated streptavidin to present the positive EB-binding protein spots. Notably, a protein with the highest signal-to-noise ratio (SNR) was identified as USP7 (SNR of 19.06) (Fig. 1C). We next performed cellular thermal shift assay (CETSA) and drug affinity responsive target stability (DARTS) assays to confirm EB binding to USP7 (Fig. 1, D and E). Moreover, pull-down analysis was applied to directly capture USP7 from BV-2 lysates. Result showed that bio-EB markedly pulled down USP7, which was blocked by adding an excess amount of EB for competition (Fig. 1F). Next, surface plasmon resonance (SPR) analysis showed that EB specifically bound to USP7 with a dissociation constant (KD) of 6.85 μM (Fig. 1G). To explore the selectivity of EB binding to different USP family members, we checked the protein microarray data and found that EB specially bonds to USP7 with a substantially high SNR among 34 USPs (fig. S1E). Pull-down analysis using recombinant USP family members also confirmed this observation (fig. S1F). Meanwhile, USP7 activity assay using the Ub-Rh110 substrate showed that EB markedly inhibited USP7’s capacity for Ub-Rh110 hydrolysis [inhibition constant (KI) of 24.98 μM] (Fig. 1H and fig. S1H). Further pull-down assay and small interfering RNA (siRNA) knockdown of four other proteins with relatively high SNR (at SNR core above 13) in proteome microarray confirmed the specificity of USP7 in EB-mediated microglia inactivation (fig. S1, I and J).

Fig. 1.

USP7 serves as a cellular target of anti-neuroinflammation small-molecule EB.

(A) EB inhibited LPS-induced NO, IL-6, and TNF-α production in BV-2 cells. Cells were treated with LPS (1 μg/ml) and EB (1, 5, and 10 μM) for 4, 8, and 24 hours correspondingly for TNF-α, IL-6, and NO assay. One-way analysis of variance (ANOVA); multiple t tests with P values corrected for multiple comparisons using Dunnett’s post hoc method. (B) Chemical structure of EB and bio-EB. (C) Proteome microarray identified USP7 as a direct cellular target of EB. CETSA (D) and DARTS (E) assays confirmed the binding of EB to USP7 in BV-2 cells, and β-actin was used as the internal control. (F) Pull-down assay confirmed the engagement of EB with USP7 in BV-2 cells. Recombinant USP7 immunoprecipitated with EB beads was analyzed by Western blotting. (G) SPR assay showed the interaction of EB with USP7. (H) EB markedly inhibited USP7’s capacity for Ub-Rh110 hydrolysis. (I) USP7 knockdown effectively reversed EB-mediated NO, IL-6, and TNF-α inhibition in BV-2 cells (n = 3 independent experiments). Cells were treated with LPS (1 μg/ml) and EB (1, 5, and 10 μM) for 4, 8, and 24 hours correspondingly for TNF-α, IL-6, and NO assay. Two-way ANOVA; multiple t tests with P values corrected for multiple comparisons using Sidak’s post hoc method. (J) Immunofluorescence analysis suggested an interaction of EB (green) with USP7 (red) in BV-2 cells. Cells were treated with or without LPS and bio-EB (10 μM) for 2 hours. Scale bar, 20 μm. For (D) to (F), blots are representatives of at least three independent experiments with similar results. IB, immunoblot; FITC, fluorescein isothiocyanate.

Then, we knocked down USP7 expression with siRNA and found that EB-dependent down-regulation of NO, IL-6, and TNF-α was markedly reversed in USP7 knockdown BV-2 cells (Fig. 1I and fig. S1G). In particular, we calculated and compared the inhibition rates of EB on different inflammation mediator production. As a result, we found that EB (10 μM) exhibited a high inhibition rate of NO production (92.7 ± 2.7%) in the N.C. siRNA group but a decreased inhibition rate (42.2 ± 5.1%) in the USP7 siRNA group. Similar results were also observed in IL-6 (99.7 ± 0.3% in the N.C. siRNA group to 45.1 ± 12.9% in the USP7 siRNA group) and TNF-α (99.6 ± 0.4% in the N.C. siRNA group to 57.5 ± 4.9% in the USP7 siRNA group). Therefore, these results suggest that USP7 is highly associated with microglia-mediated inflammation response and serves as a potential cellular target contributing to EB-mediated anti-inflammation effect. Furthermore, bio-EB showed an obvious fluorescence colocation with USP7 in cells (Fig. 1J). Collectively, these observations suggested that EB selectively bonds to USP7 in microglia.

Cocrystal structure of USP7 in complex with small-molecule EB

To explore the EB-binding domain on USP7, we performed SPR analysis and found that EB selectively interacted with noncatalytic HUBL domain (fig. S2, A to C). To investigate the mechanism for the EB-USP7 interaction, we determined the apo-HUBL structure (2.3 Å) and the cocrystal structure of HUBL in complex with EB (2.35 Å) (Fig. 2, A and C, and table S1). HUBL consisted of five Ubl domains and displayed an extended arrangement with three modules: Ubl 1–Ubl 2, Ubl 3, and Ubl 4–Ubl 5, which was similar to the reported HUBL structure [Protein Data Bank (PDB) ID: 2YLM] (). EB was anchored to a negatively charged cavity between Ubl 1 and Ubl 2 (fig. S3, B and C). The α, β-unsaturated moiety of EB served as a reactive Michael acceptor and formed a covalent bond with Cys576 (Fig. 2B and fig. S2H). Asp666, Arg723, and Glu572 were involved in stabilizing EB binding through hydrogen bonds and van der Waals interactions (Fig. 2B).

Fig. 2.

USP7 inhibition via small-molecule EB-dependent allosteric regulation mechanism.

(A) Structure of USP7 HUBL in complex with EB. Cartoon and surface representation of the overall structure. Ubl 1 to Ubl 5 are shown in slate, cyan, green, wheat, and pink, respectively. Bound EB is shown in sphere and colored by element. (B) Interactions between USP7 HUBL and EB. Residues involved in the USP7 HUBL-EB interaction are shown in lines, and hydrogen bond is shown as black dashed line. (C) Superposition of USP7 HUBL-EB complex (violet, residues 560 to 1084) with structures of USP7CD123 (white, PDB ID: 5FWI, residues 210 to 881), apo-HUBL (blue, residues 560 to 1083), and HUBL-DNMT1 complex (orange, PDB ID: 4YOC, residues 560 to 1083). (D) Schematic models of different USP7 conformations. (E) Fluorescence spectroscopy analysis of USP7 conformational change induced by EB. Fluorescence emission spectra (upon excitation at 280 nm) of recombinant USP7 in the presence of the indicated concentration of EB. (F) FlAsH-based FRET assay in BV-2 cells with or without EB (10 μM) treatment (2 hours) (n = 26 cells). Scale bar, 5 μm. One-way ANOVA; multiple t tests with P values corrected for multiple comparisons using Dunnett’s post hoc method.

The covalent bond between EB and Cys576 was also verified by liquid chromatography–tandem mass spectrometry (LC-MS/MS) analysis and EB washout experiment (fig. S2, E to G). SPR assay showed that mutation of Cys576 into alanine severely disrupted the interaction between EB and HUBL. Moreover, we established BV-2 cells with specific USP7C576A expression to test the effect of EB. First, we detected the effect of EB on the production of different inflammatory mediators from BV-2 cells with USP7WT and USP7C576A. As a result, we found that lipopolysaccharide (LPS) stimulation did not cause a discernible difference on NO, TNF-α, and IL-6 levels between USP7WT and USP7C576A BV-2 cells. However, EB deeply decreased the NO level in LPS-induced USP7WT BV-2 cells, but not in USP7C576A BV-2 cells. Furthermore, enzyme-linked immunosorbent assay (ELISA) results also confirmed that EB failed to decrease LPS-induced TNF-α and IL-6 levels in USP7C576A BV-2 cells compared with USP7WT (fig. S3A). Collectively, these observations indicated that Cys576 of USP7 was critical for the anti-inflammation effect of EB. Last, our data revealed a unique druggable site for specific USP7 inhibition in microglia.

USP7 inhibition via small-molecule EB-dependent allosteric regulation mechanism

The full-length structure of USP7 has remained undisclosed. To characterize a possible dynamic mechanism of EB-mediated USP7 inhibition, we incorporated structures of the USP7 HUBL-EB complex, USP7CD123 (residues 207 to 882, PDB ID: 5FWI) (), apo-HUBL, and the complex structure of HUBL with its substrate DNMT1 (HUBL-DNMT1; PDB ID: 4YOC) () through superposition of Ubl 1 and Ubl 2. Structure alignment revealed a rotation from Ubl 3, which generated three models of USP7 conformation (Fig. 2, C and D, and fig. S3, D and E). In the apo-HUBL structure aligned with USP7CD123, the distance from the C-terminal peptide of HUBL to CD was 60 Å, indicating an open conformation of USP7. Binding of DNMT1 reduced the distance to 24 Å and resulted in a closed USP7 conformation. In the USP7 HUBL-EB structure, EB dragged Ubl 3 to Ubl 5 to bend outward, separating the C-terminal peptide and CD by almost 80 Å, thereby forming an over-opened conformation of USP7 (Fig. 2, C and D). As the C-terminal peptide of HUBL was reported to be involved in USP7 activity (), we then proposed that USP7 may be in an autoinhibited state in open conformation. Ubiquitinated substrates induced the closed conformation of USP7 as an active state. Meanwhile, EB-mediated over-opened USP7 conformation disrupted the contact of the C-terminal peptide with CD, thus inactivating USP7 in an allosteric regulation mechanism.

Next, tryptophan fluorescence spectroscopy analysis was conducted to validate the conformational change of USP7 induced by EB. This analysis demonstrated that there was an obvious decrease in USP7 fluorescence intensity following the allosteric regulation of USP7 in cells (Fig. 2E). Next, we performed Fluorescein arsenical hairpin binder (FlAsH)–based fluorescence resonance energy transfer (FRET) assay (fig. S3F) (). We found that EB treatment markedly induced a low FRET efficiency between enhanced cyan fluorescent protein (ECFP) at the N terminus and a 12–amino acid peptide (FLNCCPGCCMEP) at the C-terminal disordered region of USP7 (7.8 ± 1.8% for control versus 0.9 ± 1.6% for EB-treated cells), which was consistent with our previous supposition that the over-opened conformation of USP7 was induced by EB (Fig. 2F). Overall, our findings suggested a previously undisclosed allosteric inhibition mechanism for USP7 by targeting the noncatalytic HUBL domain.

Keap1 is a direct substrate for USP7 deubiquitination in microglia

Next, we tried to identify the specific USP7 substrate proteins for degradation in microglia. Stable isotope labeling by amino acids in cell culture (SILAC)–based quantitative proteomics identified a reproducible set of 5821 putative proteins, and 5 of which were considerably down-regulated with EB treatment (Fig. 3A). Notably, Keap1 attracted our considerable attention for its critical role in the inflammation process. Thus, we speculated that Keap1 may be an underlying substrate protein of USP7 for microglia activation. To this end, we first used cycloheximide (CHX) to block protein synthesis and observed that EB-dependent Keap1 degradation was markedly accelerated in CHX-treated BV-2 cells (Fig. 3B and fig. S4A). Meanwhile, we found that CHX treatment induced a slight Keap1 degradation for 12 hours in N.C. siRNA–transfected cells; however, CHX comparatively promoted Keap1 degradation in USP7 knockdown cells during the same period. These results further supported the observation that Keap1 degradation upon EB treatment was mediated by USP7 (fig. S4H). Next, coimmunoprecipitation (co-IP) analysis further elucidated the interaction between USP7 and Keap1 (Fig. 3C and fig. S4B). Furthermore, immunofluorescence analysis revealed that EB suppressed USP7-Keap1 colocalization by promoting USP7 nuclear translocation in BV-2 cells (Fig. 3E). Genetic knockdown of USP7 abolished EB-mediated Keap1 degradation in BV-2 cells, suggesting that USP7 was a crucial determinant of Keap1 stability in microglia (Fig. 3D and fig. S4C). Next, the regions responsible for the USP7-Keap1 interaction were investigated with truncated versions of USP7 and Keap1. The interaction between domains was examined by co-IP in human embryonic kidney (HEK) 293T cells transfected with the indicated truncations of USP7 and Keap1. It turned out that the TRAF domain of USP7 contributed to the USP7-Keap1 interaction (Fig. 3F). Moreover, the intervening region (IVR) and broad complex/tramtrack/bric-a-brac (BTB) domains of Keap1 are mainly responsible for Keap1 binding to USP7, and Kelch plays a weak role to exert an auxiliary recognition function. Thus, we speculate that IVR and BTB domains play a central role in Keap1 binding to USP7, and Kelch functions as a supporting domain. SPR assay was further used to determine TRAF binding capacity to BTB with a KD value of 18 nM (fig. S4, D and E). Notably, no binding was detected between Keap1 and EB by SPR and pull-down assay, indicating that Keap1 is not a direct target of EB (fig. S4, F and G).

Fig. 3.

Keap1 is a direct substrate for USP7 deubiquitination in microglia.

(A) SILAC-based proteomic analysis for EB-induced protein degradation. Workflow for SILAC-based proteomic is shown on the left. Protein turnover analysis is shown in scatterplots (right). SILAC was applied as a medium supplement, then light (blue) and heavy (red) samples were mixed and analyzed by MS. Protein turnover analysis is shown in scatterplots on the right. The x axis represented the log2 fold change in the ratio of peak intensities of H (heavy) versus L (light) peptides. The y axis represents the nominal −log10 (P value). Plots with nominal −log10 (P value) less than 1.3 (P < 0.05) and absolute log2 fold change greater than 0.26 (fold change > 1.2) are colored in red. Other plots are indicated in gray. (B) Immunoblotting of Keap1 expression after EB treatment in the presence or absence of CHX (1 μM). (C) Co-IP assay with HA-tagged beads followed by Western blotting. His-Keap1 immunoprecipitated with overexpressed HA-USP7WT or HA-USP7C576A from the lysate of HEK293T cells treated with EB (10 μM) was analyzed. (D) Immunoblotting of EB-mediated Keap1 degradation with or without USP7 knockdown. (E) Immunofluorescence analysis for the distribution of USP7 (red) and Keap1 (green). Scale bar, 20 μm. (F) Schematic shows three different domains of USP7 (indicated in red) and three different domains of Keap1 (indicated in blue). Co-IP revealed USP7 (TRAF)–Keap1 (BTB-IVR) interaction domains. (G) In vitro ubiquitination assay of Keap1. Assay was performed with recombinant E1, E2-UbcH5a, ubiquitin, ATP, and the indicated concentration of EB (1, 10, and 50 μM). (H) Immunoblotting of EB-mediated WT and K48- and K63-linked ubiquitinations of Keap1. (I) Baf-A1, but not MG132, reversed EB-mediated Keap1 degradation. For (B) to (I), blots are representatives of at least three independent experiments with similar results.

To examine whether Keap1 is directly deubiquitinated by USP7, we performed in vitro ubiquitination assay. Results showed that Keap1-ubiquitin conjugates were readily observed in the presence of Cul3, Rbx1, ubiquitin, E1, and E2-UbcH5a in vitro, which were ablated by USP7. Encouragingly, EB treatment noticeably increased Keap1 ubiquitination (Fig. 3G and fig. S4I). To investigate which ubiquitination type (K48 or K63 dependent) is involved in USP7-mediated Keap1 deubiquitination, we performed co-IP experiment. Hemagglutinin (HA)–tagged USP7 and His-tagged Keap1 were cotransfected into HEK293T cells with or without EB treatment (10 μM). As shown in Fig. 3H and fig. S4K, Keap1 displayed remarkable increases in both K63- and K48-linked ubiquitinations upon USP7 knockdown or EB treatment, demonstrating that USP7 was site nonselective on deubiquitinating Keap1. Next, to explore which ubiquitination type enables EB-dependent Keap1 degradation, we used the proteasome inhibitor MG132 and autophagy inhibitor Baf-A1 to treat BV-2 cells. Keap1 expression after EB (10 μM) treatment in the presence or absence of Baf-A1/MG132 was detected by Western blotting. As shown in Fig. 3I and fig. S4J, Baf-A1 obviously reversed EB-mediated Keap1 degradation but not for MG132, indicating that Keap1 degradation was regulated in an autophagy-dependent manner. Collectively, Keap1 was identified as the ubiquitination degradation substrate of USP7 in microglia.

USP7 inhibition activates Keap1/Nrf2 signal in microglia

The Keap1/Nrf2 pathway plays a crucial role in the inflammation process. Luciferase-based reporter assay revealed that EB sharply increased luciferase activity upon LPS stimulation, which was blunted by USP7 knockdown (Fig. 4A). We also found that EB obviously up-regulated Nrf2 expression after LPS stimulation (Fig. 4B and fig. S5A). Moreover, we found that Nrf2 in nuclear extracts was effectively up-regulated by EB in BV-2 cells (Fig. 4C and fig. S5B). Meanwhile, EB noticeably drove the expression of Nrf2 target genes including HO-1, CAT, and PRDX-1, which was reversed by USP7 knockdown in BV-2 cells (Fig. 4, D and E, and fig. S5, C and D). These findings were also verified in primary microglia (fig. S5, E and F). Moreover, we observed that specific Nrf2 knockdown exerted a substantial antagonism on EB-dependent microglia inhibition (Fig. 4F). For example, only 0.3 ± 0.01 μM NO was detected in the EB (10 μM) group of negative control cells (N.C. siRNA) but still 2.0 ± 0.8 μM in USP7 down-regulated cells. Therefore, pharmacological inhibition of USP7 by EB boosted the Keap1/Nrf2 signal in microglia.

Fig. 4.

USP7 inhibition activates Keap1/Nrf2 signal in microglia.

(A) Luciferase reporter gene assay for EB-mediated Nrf2 transcriptional activity with or without USP7 knockdown. BV-2 cells cotransfected with Nrf2/ARE luciferase reporter plasmids and renilla (RL) plasmids were treated with LPS (1 μg/ml) in the absence or presence of EB (1, 5, and 10 μM) for 12 hours. (B) Immunoblotting of Nrf2 expression in BV-2 cells treated with LPS (1 μg/ml) and EB for 2 hours. (C) Immunoblotting of Nrf2 intracellular distribution in BV-2 cells treated with LPS (1 μg/ml) and EB (10 μM) for 2 hours. Histone and α-tubulin were used as the internal control in the nucleus and cytoplasm. (D) Immunoblotting of Nrf2 target gene expression in BV-2 cells treated with LPS (1 μg/ml) and EB (1, 5, and 10 μM) for 12 hours. (E) Immunoblotting of Nrf2 target gene expression in BV-2 cells treated with LPS (1 μg/ml) and EB (10 μM) for 12 hours, with or without USP7 knockdown. (F) EB-mediated NO, TNF-α, and IL-6 inhibition in BV-2 cells with or without Nrf2 knockdown. Cells were treated with LPS (1 μg/ml) and EB (1, 5, and 10 μM) for 4, 8, and 24 hours correspondingly for TNF-α, IL-6, and NO assay. For (A) and (F), statistical significance was determined by two-way ANOVA; multiple t tests with P values corrected for multiple comparisons using Sidak’s post hoc method. For (B) to (E), blots are representatives of at least three independent experiments with similar results.

USP7 inhibition alleviates microglia activation in vivo

EB was first evaluated for its pharmacokinetic property. In this experiment, the EB concentrations in the brain and plasma were measured at 0.25, 0.5, 1, 2, and 3 hours after a single intragastric administration of EB at the dose of 50 mg/kg. Pharmacokinetic analyses were estimated and correlated with previous studies. As a result, we found that the Cmax value of EB in plasma was above 1.2 μM and the AUC(0-∞) approached 7 μM*h. Meanwhile, the Cmax value of EB in the brain was 73.40 ± 27.79 ng/g at 0.25 hours (fig. S6A). Collectively, our current data demonstrated that EB had an acceptable oral bioavailability to achieve therapeutic levels as well as blood-brain barrier penetrance ability, which is similar to other small-molecule CNS drugs (–).

To further investigate the effect of USP7 in microglia activation in vivo, we generated a mouse model carrying a null allele of Usp7 (Usp7+/−), which can avoid the early embryonic lethality (fig. S6, B and C) (). Then, we examined the expression of Iba-1 and CD68 to determine whether EB affected microglia activation. Immunohistochemistry (IHC) analysis revealed that LPS-induced Iba-1 and CD68 expression was reduced after EB treatment in Usp7WT mice but not in Usp7+/− mice (Fig. 5, A and B). Meanwhile, EB protected microglia morphology by decreasing the number of branching points in Usp7WT mice rather than in Usp7+/− mice (Fig. 5, A and B, and fig. S6F). Moreover, EB alone (50 mg/kg) did not change the differentiation shape and number of Iba-1–positive microglia in the cortex (fig. S1, L and M). Furthermore, we quantificationally detected the inflammation mediators by ELISA and found that EB radically attenuated the proinflammatory cytokine production including TNF-α and IL-6 in the brains of Usp7WT mice, which was reversed in Usp7+/− mice (fig. S6D). Consistently, EB showed an obvious neuroprotective effect by Nissl staining and MAP2-immunoreactive dendritic length analysis on Usp7WT mice, rather than on Usp7+/− mice (Fig. 5, A and B, and fig. S6F). Moreover, the expression of major Nrf2 target genes, NQO1 and HO-1, was up-regulated in Usp7WT mice, rather than in Usp7+/− mice, after EB administration (Fig. 5C and fig. S6E). These findings indicated that EB targeted USP7 to exert an anti-neuroinflammation effect by inhibiting microglia activation in vivo.

Fig. 5.

EB inhibits microglia activation in Usp7WT but not Usp7+/− mice.

(A) Iba-1, CD68, TNF-α, IL-6, MAP2 IHC staining, and Nissl staining of the cortex in Usp7WT and Usp7+/− mice, which were orally treated with vehicle or EB (10 and 50 mg/kg). Arrowheads indicate the activated microglia and positive expression of inflammatory factors or synapses. Black arrowheads in Nissl staining indicate the normal neurons, and red arrowheads indicate the damaged neurons; arrows point to synapses. Scale bars, 10 μm. (B) Quantitative analysis of Iba-1, CD68, TNF-α, IL-6, and MAP2 IHC staining of Usp7WT and Usp7+/− mice. Percent of positive expression cells was calculated by InForm software. Ten individual mice per group were applied in Usp7WT and Usp7+/− mice. Three of them per group were randomly selected and subjected to immunohistochemical analyses. For an unbiased analysis, six individual 20× photomicrographs from three mice of each group were measured in the positive cell analysis (fold of control %). Statistical significance was determined by one-way ANOVA; multiple t with P values corrected for multiple comparisons using Dunnett’s post hoc method. (C) Western blotting of Nrf2 target genes (NQO1 and HO-1) in brain tissues derived from Usp7WT and Usp7+/− mice. Representative immunohistochemical images from three biological replicates of each group are shown. N.S., not significant.

To confirm whether USP7 downstream of Nrf2 is involved in the anti-neuroinflammation effect of EB, Nrf2−/− mice were also used to monitor microglia activation in vivo (fig. S6, B and C). First, EB administration induced a more notable decline of Iba-1 and CD68 expression in Nrf2WT mice compared with Nrf2−/− mice (Fig. 6, A and B). Then, EB-mediated decrease in proinflammatory cytokines and microglia branching points in Nrf2WT mice was hugely antagonized in Nrf2−/− mice (Fig. 6, A and B, and fig. S6F). EB showed an obvious neuroprotective effect by increasing the number of MAP2-positive cells, neuronal synaptic length, and Nissl substance in Nrf2WT mice, but not in Nrf2−/− mice (Fig. 6, A and B, and fig. S6F). In particular, under EB treatment (50 mg/kg), 69.7 ± 9.3% were MAP2+ cells (fold of control %) in Nrf2WT mice, compared to 55.4 ± 7.3% in Nrf2−/− mice. As for Nrf2 target genes, the levels of HO-1 and NQO1 in Nrf2WT mice were appreciably increased in response to EB administration, rather than in Nrf2−/− mice (Fig. 6C and fig. S6E). Collectively, these results indicated that the USP7-dependent Keap1/Nrf2 axis played a fundamental role in modulating microglia-dependent neuroinflammation in vivo.

Fig. 6.

EB inhibits microglia activation in Nrf2WT but not Nrf2−/− mice.

(A) Iba-1, CD68, TNF-α, IL-6, MAP2 IHC staining, and Nissl staining of the cortex in Nrf2WT and Nrf2−/− mice, which were orally treated with vehicle or EB (10 and 50 mg/kg). Arrowheads indicate the activated microglia and positive expression of inflammatory factors or synapses. Black arrowheads in Nissl staining indicate the normal neurons, and red arrowheads indicate the damaged neurons; arrows point to synapses. Scale bars, 10 μm. (B) Quantitative analysis of Iba-1, CD68, TNF-α, IL-6, and MAP2 IHC staining of Nrf2WT and Nrf2−/− mice. Percent of positive expression cells was calculated by InForm software. Ten individual mice per group were applied in Nrf2WT and Nrf2−/− mice. Three of them per group were randomly selected and subjected to immunohistochemical analyses. For an unbiased analysis, six individual 20× photomicrographs from three mice of each group were measured in the positive cell analysis (fold of control %). One-way ANOVA; multiple t tests with P values corrected for multiple comparisons using Dunnett’s post hoc method. (C) Western blotting of Nrf2 target genes (NQO1 and HO-1) in brain tissues derived from Nrf2WT and Nrf2−/− mice. Representative immunohistochemical images from three biological replicates of each group are shown.

Translational study of USP7 inhibitor to improve neurodegenerative pathogenesis

Microglia-involved neuroinflammation is highly associated with neurodegenerative disorders. Thus, we tried to explore the therapeutic potential of EB on senile dementia. Here, spatial learning and memory performance of senescence-accelerated mouse-prone 8 (SAMP8) mice were evaluated by Morris water maze test. As shown in Fig. 7A, EB-treated SAMP8 mice showed a substantial improvement in memory behavior at the fifth day. Moreover, Nissl staining revealed that the decreased Nissl substance in the cortex and hippocampus of SAMP8 mice was dramatically restored by EB (Fig. 7B). These observations were also confirmed by MAP2 staining (fig. S7, A, E, and F). Furthermore, we found that SAMP8 brain showed obvious activation of microglia with increasing Iba-1 and CD68 immunoreactivity compared with SAM-resistant 1 (SAMR1), which was markedly restrained by EB treatment (Fig. 7B and fig. S7, A and E). The decreased number of branching points in the EB-treated group indicated the protective effect of EB on microglial morphology (fig. S7F). Moreover, EB treatment reduced TNF-α and IL-6 release as well as amyloid plaques in SAMP8 mice (fig. S7, A and E). Similar results were also observed in quantificational ELISA assay (fig. S7B). Furthermore, EB also improved the expression of NQO1 and HO-1 in the brains of SAMP8 mice (fig. S7C). Collectively, EB effectively alleviated microglia-mediated neuroinflammation to improve cognitive function.

Fig. 7.

Translational study of USP7 inhibitor to improve neurodegenerative pathogenesis.

(A) Morris water maze test of SAM mice (n = 7 mice). The mean escape latency of SAMR1 and SAMP8 mice, which were orally treated with vehicle or EB (10 and 50 mg/kg) for 30 days. (B) Iba-1 IHC staining and Nissl staining performed in the cortex of SAM mice. Arrowheads in IHC staining indicate positive expression of microglia. Black arrowheads in Nissl staining indicate the normal neurons, and red arrowheads indicate the damaged neurons; arrows point to synapses. Scale bars, 10 μm. (C) Behavioral assessment tests of MPTP-induced PD mice (rotarod test, pole test, and grip strength test), which were orally treated with vehicle or EB (10 and 50 mg/kg) for 15 days. Symbols represent individual data (n = 7 mice). One-way ANOVA; multiple t tests with P values corrected for multiple comparisons using Dunnett’s post hoc method. (D) IHC staining of TH and Iba-1 performed in the SNpc region of MPTP-induced PD mice. Arrowheads in IHC staining indicate positive expression of TH or the activated microglia. Iba-1 staining scale bars, 40 μm (10 μm in the enlarged image). TH staining photomicrographs were captured with a stereomicroscope. Scale bars, 200 μm (100 μm in the enlarged image). Representative immunohistochemical images from three biological replicates of each group are shown.

We also asked whether EB exerted beneficial effects on alleviation of PD pathology. Rotarod test showed that 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) injection steeply increased the frequency of falling compared with the saline group. However, treatment with EB for 15 days showed notable improvement in rotarod activity (20.3 ± 8.3 s for the 50 mg/kg EB treatment group) (Fig. 7C). Moreover, pole test showed that the times for mice to climb down the pole were significantly shortened after EB treatment (45.3 ± 27.0 s for the 50 mg/kg EB treatment group). Grip strength analysis also revealed that grip strength was obviously increased in the EB-treated group (187.6 ± 36.5 s for the 50 mg/kg EB treatment group) (Fig. 7C). IHC analysis of tyrosine hydroxylase (TH) showed that MPTP resulted in a sharp decrease in the number of TH-positive cells in substantia nigra pars compacta (SNpc), which was markedly rescued by EB (Fig. 7D and fig. S7G). Next, we detected the effect of EB on microglia-associated inflammation marker in SNpc. As shown in Fig. 7D and fig. S7G, Iba-1–positive microglia with larger somas and more branches in SNpc increased after MPTP injections, whereas EB remarkably suppressed microglia activation as evidenced by the decrease in Iba-1–positive cells. USP7 has been linked to various neurodegenerative diseases, and loss of USP7 leads to delay/intellectual disability, hypotonia, eye anomalies, feeding difficulties, spinal and bulbar muscular atrophy, spinocerebellar ataxia, and various other ataxic neurological phenotypes (–). In our study, analyses of the protein-protein interaction by using InBio Map have shown that there are high-confidence sets of the interaction between USP7 and UVSSA, ERCC8, and XAB2 (fig. S7D), the associated proteins in impaired development of the nervous system and PD pathology (, ). Together, these data showed that EB exerted a potential therapeutic effect in PD pathogenesis by alleviating microglia-mediated neuroinflammation.

DISCUSSION

The USPs are the largest family of DUBs, regulating a wide variety of cellular processes. Therefore, USPs are emerging as a promising class of drug targets. In this study, USP7 was first identified as a therapeutic target for neurodegeneration by suppressing microglia activation. Moreover, we reported previously undisclosed structural and mechanistic insights into USP7 allosteric inhibition by small-molecule EB by targeting the noncatalytic HUBL domain. In addition, we confirmed USP7 as a key regulator of the Keap1/Nrf2 signaling pathway in microglia for anti-neuroinflammation effect. Since there are currently no sufficient therapeutic targets for neurodegeneration, our findings thus provide access to wider perspectives for neurodegenerative disease therapy by inhibiting microglia activation by targeting USP7.

Nowadays, small molecules specially targeting the HUBL domain of USP7 have not been reported. Therefore, our study found the first druggable site in HUBL, which is far from classic CD on USP7, bearing important implications for the future structure-guided USP7 inhibitor design. Moreover, EB inhibits USP7 via a distinctive covalent and irreversible mode. To our knowledge, EB is the first covalent USP7 inhibitor targeting the noncatalytic domain, which quite differs from all other current USP7 inhibitors binding to CD. Thus, our findings provided a new insight and direction for designing a USP7 small-molecule inhibitor with previously unidentified binding pocket and characteristic regulation mechanism.

In the past few years, several small-molecule inhibitors of USP7 have been reported (). Although these inhibitors mainly target CD on USP7, the potential regulation mechanisms are quite different. P22077 and P50429 bind to the active site of USP7 to directly block its activity. However, FT827 and FT671 bind to a grove next to the active site, further inducing changes in switching loop and catalytic site misalignment (–). Different from the FT827-dependent conformation change within CD, EB causes a long-range allosteric regulation to affect the multidomain interaction between HUBL and CD, further inhibiting USP7 activity. Furthermore, we found that EB inhibits the interaction between USP7 and Keap1. Notably, Keap1 is specially recognized by the TRAF domain of USP7, indicating that pharmacologically targeting HUBL may modulate substrate recognition of TRAF. This finding sheds light on a possible mechanism of synergistic effect or cross-talk among different USP7 domains. In addition, previous studies have shown that Ubl domains of USP7 can also function as a substrate-binding platform (, ). Thus, developing small molecules targeting the HUBL domain to disrupt the USP7 interaction with potential substrates may further open an avenue to design USP7 inhibitors with specific functions. Overall, HUBL-targeted long-range allosteric regulation on USP7 not only helps us deeply understand the function and regulation mechanisms of the USP family but also offers a clue to develop novel therapeutic agents allosterically regulating multidomain enzymes. Here, we showed compelling evidence that USP7 down-regulated cells became less sensitive to EB inhibition. On the basis of previous reports (–), we speculate a possible mechanism that USP7 knockdown causes an obviously reduced expression in the cellular target of EB (target-ligand interaction metrology), thereby leading to a weakening effect of EB on the USP7-associated downstream inflammation signaling pathway. This may explain the reason for the impaired anti-inflammation effect of EB upon USP7 knockdown in our study. Meanwhile, USP7 knockdown may also drive the cells to establish compensatory biological pathways by bypassing USP7 function, which may reduce drug sensitivity (, ).

The Keap1/Nrf2 pathway is a major regulator of cytoprotective responses to oxidative stress (, ). The activation of Nrf2 via dissociation of the Keap1-Nrf2 inhibitory complex has been well illuminated, leading to Nrf2 nuclear translocation, as well as subsequent antioxidant and anti-inflammation gene expression (). However, in this study, we revealed a previously undiscovered strategy for Keap1/Nrf2 function modulation by targeting USP7. Thus, USP7/Keap1/Nrf2 may represent a crucial signaling pathway involved in microglia-mediated neuroinflammation. Furthermore, our work provided a innovative insight for senile dementia and PD therapy by alleviating the neuroinflammation process by targeting the USP7/Keap1/Nrf2 signaling axis, thereby suggesting a clinical translation significance.

Collectively, our findings revealed USP7 as a promising therapeutic target for microglia-mediated neuroinflammation. Moreover, we identified the noncatalytic HUBL domain in USP7 as a valuable druggable hotspot to design lead compounds for human neurodegenerative disease therapy. Meanwhile, in this study, we only focused on the potential therapeutic effects of USP7 inhibition in neuroprotection via microglia, which are major immune cells in the CNS; however, whether other cell types like astrocytes, oligodendrocytes, or cerebrovascular endothelial cells are also involved in this biological progress may need further investigation. Moreover, our study identified EB as an original molecular template for allosteric USP7 inhibition. Future studies on chemical synthesis–guided structural modification may provide a more promising candidate molecule with higher biological activity and less off-target effects for clinical translation. Furthermore, whether the potential clinical availability of USP7 inhibitors can be expanded to other neurodegenerative diseases including Huntington’s disease and amyotrophic lateral sclerosis still warrants examination in the future.

MATERIALS AND METHODS

Mice and cell lines

Male C57BL/6 (6 to 8 weeks old, 20 to 25 g) and SAM (7 to 8 months old, 35 to 40 g) were obtained from the Department of Laboratory Animal Science of Peking University Health Science Center. Male Usp7+/− (6 to 8 weeks old, 20 to 25 g) mice were obtained commercially from Cyagen Biosciences. Male Nrf2−/− (6 to 8 weeks old, 20 to 25 g) mice were given by S. Yu (Peking University). Mice were maintained under specific pathogen–free conditions with 12-hour light/dark cycles. All care and treatment of experimental animals were in strict accordance with the guidelines of the Association for Assessment and Accreditation of Laboratory Animal Care approved by the Institutional Animal Care and Use Committee of Peking University (license no. LA2020069).

Murine microglia (BV-2) and HEK293T cells were obtained from Peking Union Medical College Cell Bank and were cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS), streptomycin (100 μg/ml), and penicillin (100 U/ml). Cells were routinely maintained at 37°C in humidified air containing 5% CO2.

Preparation of biotin-EB

Synthesis of [2-[2-(2-hydroxyethoxy)-ethoxy]-ethoxy]-acetic acid benzyl ester

Benzyl 2-diazoacetate (5.7 g, 32.4 mmol) in dichloromethane (DCM; 5 ml) was added dropwise to a mixture of triethylene glycol (30.2 g, 201.3 mmol), BF3, Et2O (46 mg, 324 μmol), and DCM (50 ml). The mixture was stirred at 20°C for 5 hours. Thin-layer chromatography (TLC) (PE/EtOAc = 2:1, Rf = 0.02) showed that the reaction was completed. The reaction mixture was washed with water (30 ml) for five times and dried over anhydrous sodium sulfate. Then, the mixture was filtered and concentrated to afford the [2-[2-(2-hydroxyethoxy)-ethoxy]-ethoxy]-acetic acid benzyl ester as yellow oil (6.5 g, 21.8 mmol, 67.3% yield). The crude product was used in the next step directly. 1H nuclear magnetic resonance (NMR; 400 MHz, CDCl3) δ 4.20 (m, 2H), 3.59 to 3.74 (overlapped, 2H), 4.20 (s, 2H), 4.36 (m, 1H).

Synthesis of compound 2

Biotin (4 g, 16.4 mmol) and 1-ethyl-[3-(3-dimethylamino)propyl]-carbodiimide hydrochloride (4.3 g, 22.6 mmol) were added to a solution of [2-[2-(2-hydroxyethoxy)-ethoxy]-ethoxy]-acetic acid benzyl ester (4.5 g, 15.1 mmol) and 4-dimethylaminopyridine (DMAP) (3.7 g, 30.2 mmol) in DCM (80 ml). The mixture was heated at 70°C for 4 hours. TLC [DCM/methanol (MeOH) = 10:1, Rf = 0.3] showed that the reaction was completed. The mixture was cooled to 20°C and then washed with 1 N HCl (80 ml) followed by water (80 ml) for two times. After that, the mixture was dried over anhydrous sodium sulfate, filtered, and concentrated to afford crude compound 2 (8.0 g, 15.3 mmol) as yellow gum. The crude product was used in the next step directly. 1H NMR (400 MHz, CDCl3) δ 3.16 (m, 1H), 4.23 (m, 1H), 4.50 (m, 1H), 2.73 (d, J = 13.2 Hz, 2H), 2.91 (dd, J = 13.2, 4.8 Hz, 2H), 1.44 to 1.72 (overlapped, 2H), 2.37 (t, J = 7.6 Hz, 2H), 4.22 (m, 2H), 3.65 to 3.75 (overlapped, 2H).

Synthesis of compound 3

Pd/C (4 g, 10% purity) was added to a solution of compound 2 (8.0 g, 15.3 mmol) in MeOH (80 ml). The suspension was degassed under vacuum and purged with H2 (2.5 MPa) for three times. Then, the reaction mixture was stirred at 30°C under H2 atmosphere (2.5 MPa) for 20 hours. LC-MS showed that a little compound 2 remained. The reaction mixture was cooled to 20°C, filtered through a pad of celite, and washed with MeOH (20 ml) for three times. The combined filtrate was concentrated to afford crude compound 3 (6.5 g, crude) as a gray solid. The crude product was used in the next step directly. 1H NMR (400 MHz, CDCl3) δ 3.15 (m, 1H), 4.22 (m, 1H), 4.50 (m, 1H), 2.74 (d, J = 13.2 Hz, 2H), 2.89 (dd, J = 13.2, 4.8 Hz, 2H), 1.42 to 1.72 (overlapped, 2H), 2.35 (t, J = 8.0 Hz, 2H), 4.21 (m, 1H), 3.65 to 3.73 (overlapped, 2H).

Synthesis of bio-EB

EB (200 mg, 433 μmol) and EDCI (166 mg, 865 μmol) were added to a solution of compound 3 (6.0 g, 1.4 mmol) and DMAP (106 mg, 865 μmol) in DCM (5 ml). The mixture was heated at 70°C for 7 hours. LC-MS showed that the reaction was completed. The reaction was cooled to 20°C and washed with water (3 ml) for two times. Then, the organic layer was concentrated. The residue was diluted with MeOH (4 ml) and purified by prep-HPLC (high-performance liquid chromatography) (column: YMC-Actus Triart C18 5 μm, 150 mm × 30 mm); the content was eluted by a two-solvent system: (A) trifluoroacetic acid/H2O 0.056% (v/v) and (B) acetonitrile (ACN) in a 40 to 70% gradient of solvent B over 10 min. Bio-EB (300 mg, 324 μmol, 75% yield, 95% purity) was obtained as an off-white solid after lyophilization. 1H NMR (400 MHz, CDCl3) δ 6.66 (t, 1H), 6.39 (s, 1H), 5.80 (s, 1H), 5.48 (dd, 1H), 5.35 (m, 1H), 4.99 (d, 1H), 4.82 (d, 1H), 4.61 (d, 1H), 4.42 (m, 1H), 4.24 (m, 4H), 3.72 (m, 10H), 3.22 (m, 1H), 3.05 (d, 1H), 2.99 (m, 1H), 2.94 (m, 1H), 2.79 (m, 1H), 2.37 (t, 2H), 2.30 (m, 1H), 2.12 (s, 3H), 2.02 (s, 3H), 1.87 (s, 3H), 1.81 (s, 3H), 1.70 (m, 4H), 1.49 (m, 2H), (+)-ESIMS m/z 879.3 [M + H]+.

Primary microglia culture

Primary microglia were obtained from mouse pups of postnatal days 1 and 2 (P1 and P2). Brains were isolated in cold Hanks’ balanced salt solution (HBSS) containing gentamicin (10 μg/ml) (Sigma-Aldrich). Cortices were mechanically dissociated, and blood vessels were removed. Tissue was chopped thoroughly into small pieces and transferred to HBSS containing 0.15% trypsin. After incubation at 37°C for 20 min, the supernatant was removed and DMEM/Ham’s F12 medium supplemented with 10% FBS, gentamicin (10 μg/ml), penicillin (100 U/ml), and streptomycin (100 μg/ml) was added. Cells were seeded on a poly-d-lysine–coated flask, and the culture medium was changed the next day to remove debris. After 14 days of culturing, primary microglia were collected by shaking the flask at 100 rpm for 1 hour. The purity (>95%) of these cultures was confirmed by Iba-1 immunocytochemistry.

Cell survival assay

Cell viability was assessed using the 3-(4,5-dimethyl thiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) solution (Sigma-Aldrich). BV-2 cells were treated with LPS (1 μg/ml; Escherichia coli, O111:B4, Sigma-Aldrich) in the absence or presence of EB (1, 5, and 10 μM). After 24 hours of incubation, culture supernatant was removed and staining solution containing MTT (5 mg/ml) was added to each well. Cells were then incubated for 4 hours in the dark at 37°C. Thereafter, formazan crystals were dissolved by dimethyl sulfoxide (DMSO), and absorbance at 570 nm was determined spectrophotometrically on an ELX800 UV universal microplate reader (Bio-Tek). Data were graphically displayed using GraphPad Prism (version 6.0).

Proinflammatory mediator analysis

BV-2 cells were treated with LPS (1 μg/ml) in the absence or presence of EB (1, 5, and 10 μM). The cell supernatant was collected at different time points for proinflammatory mediator assay (4 hours for TNF-α, 8 hours for IL-6, and 24 hours for NO). NO production was quantified using a NO assay kit, and concentrations of TNF-α and IL-6 were detected using commercial ELISA kits (ExCell) according to the manufacturer’s instructions.

Quantitative real-time PCR analysis

Total RNA was isolated using the MolPure Cell/Tissue Total RNA Kit YEASEN. RNA was reverse-transcribed to complementary DNA (cDNA) with Hifair Ш First Strand cDNA Synthesis SuperMix YEASEN. Quantitative real-time reverse transcription polymerase chain reaction (RT-PCR) was performed using Hieff qPCR SYBR Green Master Mix YEASEN. The 20-μl reaction mixture contained 200 nM primers, 10 μl of SYBR Green Master Mix, 0.5 to 2 μl of template cDNA, and nuclease-free water. The cDNA amplification was conducted on an Agilent Technologies Stratagene Mx3005P following the manufacturer’s instructions. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as an internal control.

Target identification of EB

Target protein of EB was identified by human protein microarray at room temperature. Commercial HuProt proteome microarray (Bc Bio) was blocked with blocking buffer [3% bovine serum albumin (BSA) in phosphate-buffered saline (PBS)] for 1 hour at room temperature. Bio-EB (10 μM) was used to incubate with the proteome microarray at room temperature for 1 hour in reaction buffer (1% BSA in PBS). The microarray was washed with PBS three times, followed by incubation with Cy3-conjugated streptavidin at a 1:1000 dilution for 1 hour. The microarray was then spun dry, and bio-EB–protein interactions were detected with a GenePix 4200A microarray scanner (Axon Instruments) at 635 nm. Data were analyzed using GenePix Prospector software (version 6.0).

SPR analysis

The interaction between EB and USP7 was detected by SPR using the Biacore T200 system at 25°C (GE Healthcare). Briefly, recombinant human USP7 protein (Sino Biological) was immobilized on an activated carboxymethylated 5 (CM5) sensor chip using amine coupling method. Gradient concentrations of EB were injected at a flow rate of 30 μl/min in running buffer [0.05% (v/v) Tween 20 and 5% (v/v) DMSO in PBS]. The results were analyzed with the Biacore evaluation software (T200 version 2.0). Data were fitted to the 1:1 Langmuir binding model, and kinetic parameters were derived.

Single-cycle affinity analysis was performed on Biacore T200 at 25°C for the protein-protein interaction. Recombinant TRAF-USP7 (residues 55 to 208), BTB-Keap1 (residues 48 to 180), and IVR-Keap1 (residues 180 to 318) were expressed and purified according to the previously reported methods (, ). Recombinant BTB-Keap1 or IVR-Keap1 was immobilized on a CM5 chip to almost 7000 response units (RU). Recombinant TRAT-USP7 at concentrations of 2.6, 7.7, 23, 70, and 210 nM flowed over the surface sequentially without regeneration. BSA in the same concentrations with TRAF-USP7 was used as a negative control. Data were fitted to a single-cycle kinetic model.

Preparation of EB-tagged beads

Bio-EB or 0.2% DMSO as a control was incubated with avidin-agarose for 2 hours at 4°C. Then, immobilized beads were washed 10 times with washing buffer [50 mM Hepes, 30 mM NaCl, 1 mM EDTA, 2.5 mM EGTA, and 0.1% Tween 20 (pH 7.5)] and kept at 4°C before use.

Cellular thermal shift assay

BV-2 cells were treated with EB (10 μM) or DMSO for 2 hours, then aliquoted, and heated at different temperatures (56° and 74°C) for 3 min. After cooling to room temperature, cells were lysed by freeze-thawing in liquid nitrogen for five cycles. Soluble proteins were collected by centrifugation at 12,000g for 5 min at 4°C and then detected by Western blotting.

DARTS assay

BV-2 cells were collected, and total protein was isolated. The lysate was diluted with TNC buffer [50 mM tris-HCl (pH 8.0), 50 mM NaCl, and 10 mM CaCl2] followed by treatment with EB (1, 5, and 10 μM) or DMSO. After incubation for 1 hour at room temperature, pronase (25 μg/ml) was added for a further 30 min at 37°C. Reactions were ceased by adding SDS–polyacrylamide gel electrophoresis (PAGE) loading buffer and detected by Western blotting.

Fluorescence spectroscopy

Intrinsic tryptophan fluorescence spectroscopy was used to determine conformational changes in USP7. The indicated concentration of EB was added in 20-μl aliquots of recombinant USP7, and the mixture was incubated at 25°C for 150 min. All samples were collected in a black 96-well plate. Fluorescence emission spectra were taken between 300 and 500 nm upon excitation at 280 nm (Cary Eclipse, Varian Instruments).

Transient transfection

Transient transfection was performed using Lipofectamine reagent according to the manufacturer’s instructions. Briefly, HEK293T cells were seeded to achieve a confluency of 70% the following day. Plasmids or siRNA was transfected using Lipofectamine 2000 or Lipofectamine RNAiMAX reagent (Thermo Fisher Scientific) separately in Opti-MEM. Cells were incubated with transfection complexes for 6 hours. Then, the medium was replenished and cells were harvested 48 hours after transfection.

Immunoprecipitation

Cells were collected and lysed in NP-40 lysis buffer supplemented with protease inhibitor cocktail (Roche). Ten microliters of antibody-conjugated magnetic beads or 50 μl of avidin-conjugated Sepharose was added to the lysate and incubated with rotation overnight at 4°C, followed by five washes with lysis buffer. Immunoprecipitation complexes were eluted with loading buffer and detected by Western blotting.

In vitro assay of USP7 activity

USP7 activity was monitored by fluorescence intensity measurements using a monoubiquitinated ubiquitin-rhodamine substrate (ubiquitin-Rho110, Boston Biochem). Activity assay was conducted for 1 hour at room temperature with a final USP7 concentration of 5 nM in black polystyrene 96-well plates (PerkinElmer, Waltham, MA, USA). The substrate concentration of ubiquitin-Rho110 was 25 nM, and EB was diluted in DMSO in double ratio dilution series from 64 μM. Samples were incubated in assay buffer [20 mM tris-HCl (pH 8.0), 2 mM CaCl2, 1 mM reduced glutathione, 0.01% (v/v) Triton X-100, and 0.01% (v/v) Prionex]. Upon addition of the substrate, fluorescence intensity was immediately monitored at 485-nm excitation/520-nm emission every 60 s over a period of 50 min with an EnSpire fluorescence spectrophotometer (PerkinElmer). Unless otherwise stated, all other reagents were purchased from Sigma-Aldrich (St. Louis, MO). Kinetic curves were plotted in GraphPad Prism (version 6.0), and kinact and KI values were derived.

Immunofluorescence assay

For immunofluorescent colocalization analysis, BV-2 cells were cultured on glass coverslips. Cells transiently cotransfected with appropriate plasmids were treated with or without LPS and EB (10 μM). For the USP7-EB interaction assay, the cells were transfected with two plasmids (HA-USP7 and His-Keap1) for 48 hours and then further treated with LPS and EB for 2 hours. For distribution of USP7 and Keap1 detection, the cells were transfected with only HA-USP7 plasmid for 24 hours, followed by LPS and EB treatment for 24 hours. Cells were incubated with His-tagged mouse primary antibody and HA-tagged rabbit primary antibody, followed by adding Alexa Fluor 594–labeled (red) anti-rabbit and Alexa Fluor 488–labeled (green) anti-mouse secondary antibodies. After staining with 4′,6-diamidino-2-phenylindole (DAPI), the cells were imaged at 594/618 nm for Rabbit Dylight-594 and at 488/519 nm for mouse Alexa Fluor 488 by confocal microscopy (Leica TCS SP8 X, Leica Microsystems).

Determination of EB-binding site on USP7

Recombinant human USP7 protein was incubated with EB (10 μM) or DMSO with rotation overnight at 4°C. Reactions were ceased by adding loading buffer, and proteins were separated by 10% SDS-PAGE. After visualization by silver staining, the bands corresponding to USP7 were excised and digested with trypsin.

Binding sites were identified by LC-MS/MS using a nano-HPLC-tandem LTQ-Orbitrap Velos pro mass spectrometer (Thermo Fisher Scientific). Peptides were loaded onto an analytical column (75 μm, 10 cm) packed with 3-μm RP C18 using an EASY-nLC II system in solvent A [0.1% (v/v) formic acid in water]. Separation was achieved with a linear gradient of 2% solvent B to 95% solvent B [0.1% (v/v) formic acid in acetonitrile] over 95 min. Eluted peptides were injected onto a nano-HPLC-tandem LTQ-Orbitrap Velos pro mass spectrometer using a nano-electrospray ion source. Full MS scans were collected in the Orbitrap with a resolution of 60,000 (full width at half maximum). MS/MS spectra were searched using Proteome Discoverer (1.4) software against a human proteome database (UniProt).

Cell fractionation and Western blotting

Nuclear and cytoplasmic proteins were extracted from BV-2 cells with the cytoplasmic and nuclear protein extraction kit (Beyotime). Briefly, BV-2 cells were lysed with buffer A [50 mM tris-HCl (pH 8.0), 10 mM NaCl, 5 mM MgCl2, and 0.5% (v/v) NP-40]. After 20 min of incubation, cytosolic extracts were collected after centrifugation at 12,000g for 5 min. Nuclear pellets were washed with lysis buffer A and then resuspended in lysis buffer B [20 mM Hepes (pH 7.9), 0.5 mM NaCl, 1 mM EDTA, and 1 mM dithiothreitol (DTT)]. After 20 min of incubation, nuclear extracts were collected after centrifugation at 12,000g for 10 min.

Proteins were resolved by 4 to 15% gradient SDS-PAGE and transferred to a polyvinylidene difluoride membrane (Merck) using a wet transfer system (Tanon). The membrane was blocked with blocking buffer [20 mM tris-HCl (pH 7.8), NaCl (3 mg/ml), 0.2% (v/v) Tween 20, and 5% (w/v) skimmed milk] at room temperature for 1 hour. The membrane was probed with primary antibodies at a 1:1000 dilution for 2 hours at room temperature. Then, the membrane was incubated with secondary antibodies at a 1:1000 dilution for 1 hour at room temperature. After washes, blots were developed with enhanced chemiluminescence (Thermo Fisher Scientific) and visualized on a Tanon 5200 imaging analysis system (Tanon).

SILAC-based proteomic analysis of EB-induced protein degradation

Stable isotope–labeled BV-2 cells were cultured in RPMI 1640 medium (deficient in l-lysine and l-arginine) (Invitrogen) supplemented with 12C614N2 l-lysine and 12C614N4 l-arginine (light) or 13C615N2-lysine and 13C615N4 l-arginine (heavy) (Thermo Fisher Scientific) over six passages to reach 95% incorporation. BV-2 cells containing light protein were treated with LPS (1 μg/ml) and EB (50 μM), while cells containing heavy protein were treated with LPS and DMSO as control. After 4 hours of treatment, light and heavy lysates were mixed at a 1:1 ratio. Proteins were resolved by SDS-PAGE and identified by nanoLC-MS/MS as described above.

Protein expression and purification

The HUBL domain of USP7 (residues 560 to 1102) was cloned into the Bam HI/Xho I sites of pGEX-6P-1 vector. Recombinant plasmid was transformed into E. coli BL21 (DE3) cells. Cells were cultured in LB medium at 37°C to an OD600 (optical density at 600 nm) of 0.8, and expression was induced with 0.5 mM isopropyl-β-d-thiogalactopyranoside (IPTG) at 16°C overnight. Harvested cells were resuspended in lysis buffer [20 mM Hepes (pH 7.5), 200 mM NaCl, 1 mM DTT, and 1 mM EDTA] and then lysed by sonication. Cell debris were removed by centrifugation, and the supernatant was loaded onto preequilibrated Glutathione Sepharose 4B beads (GE Healthcare) with lysis buffer. Protein was eluted using 20 mM reduced glutathione. Glutathione S-transferase (GST) tag was removed by PreScission Protease at 4°C overnight. Cleaved HUBL was further purified using a HiTrap-Q column (GE Healthcare) with buffer A [20 mM Hepes (pH 7.5), 50 mM NaCl, and 1 mM DTT] and buffer B [20 mM Hepes (pH 7.5), 1 M NaCl, and 1 mM DTT]. Protein fractions were concentrated and loaded on a Superdex 200 column (GE Healthcare) preequilibrated with 20 mM Hepes (pH 7.5), 150 mM NaCl, and 1 mM DTT. The final protein was concentrated to 15 mg/ml and stored at −80°C.

Crystallization of the HUBL-EB complex

Crystallization of HUBL was screened using Mosquito in sitting-drop vapor diffusion experiments at 4°C. Crystals were obtained in 2% (v/v) Tacsimate (pH 7.0), 0.1 M imidazole (pH 7.0), 8% (w/v) polyethylene glycol 3350 (PEG3350), and 5% (v/v) 2-propanol. To obtain the structure of the HUBL-EB complex, crystals of HUBL were soaked in a solution containing 10 mM EB, 2% (v/v) Tacsimate (pH 7.0), 0.05 M imidazole (pH 7.0), 14% (w/v) PEG3350, 5% (v/v) 2-propanol, 5% PEG400, and 5% DMSO for 40 hours. Socked crystals were flash-frozen in liquid nitrogen after cryoprotection in drops composed of 20% ethylene glycol.

Data collection and structure determination

Diffraction data for the apo-HUBL and HUBL-EB complex structure were collected on the beamlines BL17U1 and BL19U1 at the National Center for Protein Science Shanghai (NCPSS). The data were processed with XDS package (). Structures were solved by molecular replacement with Phaser () using the previously reported HUBL structure (PDB ID: 2YLM) as a search model. The final model was obtained after iterative cycles of manual rebuilding with COOT () and refined with Phenix.refine (). For the apo-HUBL structure, Ramachandran plot indicated that 97.91 and 2.09% of residues fell into the favored and allowed regions, and for the HUBL-EB structure, 94.27 and 5.73% of residues fell into the favored and allowed regions, respectively. Data collection and structure refinement statistics are shown in table S1. All structure figures were generated using PyMOL ().

Tryptophan fluorescence quenching study

Recombinant USP7 protein was incubated with EB (1.56 to 100 μM) or vehicle in a quartz plate. Tryptophan fluorescence spectra of reaction solutions were detected by setting the excitation wavelength at 280 nm and emission wavelengths from 300 to 550 nm on a fluorescence spectrophotometer (PerkinElmer). The fluorescence spectra of USP7 alone and the complex of USP7 and successive concentrations of EB were recorded. Spectra were corrected by the buffer and ligand contribution.

FlAsH-based FRET

HEK293T cells were cultured in six-well plates for transient transfection of FRET plasmids using Lipofectamine 2000. The medium was replaced 6 hours later, and cells were seeded onto confocal special 24-well plates 48 hours after transfection. Sixteen hours after seeding, cells were treated with 10 μM EB for 2 hours. Plasmids that contained only an ECFP tag or a 12–amino acid peptide were used as control to exclude intermolecular FRET.

Labeling of FlAsH was conducted after treatment of EB using commercial TC-FlAsH kits (Invitrogen) according to the manufacturer’s instructions. Briefly, cells were washed twice with HBSS to remove the medium thoroughly and then incubated with 250 nM FlAsH-EDT2 in HBSS at 37°C for 20 min. After incubation, cells were washed with 250 μM 2,3-dimercapto-1-propanol (BAL) in HBSS at 37°C for three times. At last, cells were maintained in HBSS for later imaging.

Fluorescence imaging was performed on a Leica TCS SP8 system. A 24-well plate was mounted on a Leica DMi8 inverted microscope using a typical holder, and images were taken with 40× oil objective lens. ECFP was excited at 405 nm, and FlAsH was excited at 514 nm directly. Emission of ECFP and FlAsH was measured at 480 ± 20 nm and 530 ± 20 nm, respectively. FRET efficiency was calculated as the emission ratio of FlAsH to ECFP (F530/F480). Fluorescence signals were digitalized using LAS X software.

In vitro ubiquitination assay

An efficient in vitro ubiquitination assay system was performed for ubiquitinated proteins (). For substrates, plasmids expressing His-Keap1 or His-Cul3, and His-Rbx1 were transfected into HEK293T cells individually. Cell lysates were prepared in lysis buffer [15 mM tris-HCl (pH 7.4), 500 mM NaCl, 1 mM DTT, 1 mM NaF, and 0.25% NP-40] supplemented with protease inhibitors. Expressed proteins were purified through pull-down assay by incubating with His-tagged beads at 4°C for 4 hours. For Keap1-ubiquitin conjugation, 300 pmol of ubiquitin, 2 pmol of E1, 10 pmol of E2-UbcH5a, 2 mM adenosine triphosphate (ATP), and EB were added and the mixture was incubated in 30 μl of reaction buffer [50 mM tris-HCl (pH 7.5), 5 mM MgCl2, 2 mM NaF, and 0.6 mM DTT] at 37°C for 1 hour. To stop the reaction, the loading buffer was added and target proteins were detected by Western blotting.

Reporter gene activation assay

Nrf2 transcriptional activity was analyzed with a SuperLight dual luciferase reporter gene assay kit (BioAssay Systems) (). Briefly, BV-2 cells were transiently cotransfected with Nrf2/ARE luciferase reporter plasmids and renilla (RL) plasmids using Lipofectamine 2000 reagent as described above. Then, cells were treated with LPS (1 μg/ml) in the absence or presence of EB (1, 5, and 10 μM) for 12 hours. Thereafter, cells were lysed with lysis buffer. The firefly luciferase (FFL) luminescence was first measured by addition of the FFL reagent using a luminometer (PerkinElmer). Next, the RL luciferase luminescence serving as a baseline response was measured subsequently by addition of the RL reagent. Relative luminescent unit (RLU) of ARE reporter was calculated as the ratio of firefly to RL luciferase activity.

Anti-neuroinflammation and neuroprotection analysis in transgenic mice

To investigate the physiological functions of USP7, Usp7+/− mice were generated by CRISPR-Cas9–mediated genome engineering. Exons 3 to 12 and exons 2 to 31 of Usp7 were selected as target sites. Six- to 8-week-old wild-type (WT), Usp7+/−, and Nrf2−/− mice were orally treated with 0.5% CMC-Na solution (vehicle) or EB (10 or 50 mg/kg body weight). One hour later, mice were injected intraperitoneally with LPS (1 mg/kg body weight). Mice were sacrificed 24 hours later. Nrf2 target genes, NQO1 and HO-1, from brains were analyzed by Western blotting. Nissl staining and IHC staining of MAP2, CD68, Iba-1, TNF-α, and IL-6 were conducted to show neuronal damage and neuroinflammation in the cortex of mice.

Senescence-accelerated mice

The SAM was established by phenotypic selection of the AKR/J strain. In addition, SAMP and SAM-resistant (SAMR) littermates were separately inbred (). When compared to the SAMR strain, the SAMP mice show age-related learning and memory deficits. The SAMP strain can be used as an accelerated aging model in studying aging-related disorders. The SAMP8 mouse strain is one of SAMP strains, and SAMR1 was used as a control strain. Twenty-week-old male SAMP8 and SAMR1 mice were orally treated with 0.5% CMC-Na solution (vehicle) or EB (10 or 50 mg/kg body weight) for 30 days. Spatial learning and memory performance were evaluated by Morris water maze test. Nissl staining and IHC stainings of β-amyloid, MAP2, CD68, Iba-1, TNF-α, and IL-6 were then conducted to show neuronal damage in the cortex of mice.

Parkinson’s model mice

Parkinson’s model was established by administration of 30 mg/kg body weight of MPTP solution (50 mg/ml in saline) via intraperitoneal injection per day for five consecutive days. The control group received an equivalent volume of saline as vehicle. CMC-Na solution (0.5%; vehicle) or EB (10 or 50 mg/kg body weight) was given for 15 days during the whole experiment. According to a typical treatment method of subacute PD model, administration was started 10 days before the first MPTP intoxication and continued to be administered daily, 6 hours before the MPTP injections. Immunohistochemical analyses of TH and Iba-1 were performed in SNpc of mice.

Behavioral evaluation

Spatial learning and memory were assessed with Morris water maze test. Mice were given five cycles of trials (60 s per trial) each day to find a platform in a circular pool of water. At the end of the fifth day, the latency to find the escape platform (escape latency) was recorded and analyzed with Ethovision XT software (Noldus, Attleboro, MA, USA). Seven mice were included in each experimental group (n = 7).

Coordination and fatigue resistance were evaluated using a rotary rod apparatus (Yiyan Scientific Research). Briefly, mice were placed on an accelerating rod at a final speed of 30 rpm. They were safely dropped from the rotating lane when exhausted. Then, the number of dropping during a 180-s period was recorded. Mice were given three cycles of trials (180 s per trial) for pretraining before the experiment, and the formal experiment was conducted at the end of the fifth day. Seven mice were included in each experimental group (n = 7).

Pole test was performed by a wooden inclined pole system (YLS-Q15, Yiyan Scientific Research). Mice were placed on top of the ball of the pole. The time of mice climbing down was recorded. Tests were repeated three times to evaluate the average. Seven mice were included in each experimental group (n = 7).

Muscle strength was analyzed with a computerized grip strength meter (YLS-13A, Yiyan Scientific Research). To measure total grip strength (forelimbs and hindlimbs), mice were led to grip the apparatus using their forelimbs and hindlimbs. Mice were then gently pulled away horizontally until the grasp was broken. The peak tension was recorded, and the test was repeated three times. Seven mice were included in each experimental group (n = 7).

Evaluation of EB pharmacokinetic profile and brain penetration

Six- to 8-week-old male Balb/c mice were treated with a single oral administration of EB (50 mg/kg body weight). The blood and brain were collected at sacrifice, immediately after myocardial perfusion, at 0.25, 0.5, 1, 2, and 3 hours. Tissue homogenates were prepared in saline using a Minilys tissue homogenizer (Bertin Technologies) after adding 400 μl of acetonitrile/H2O (4:1, v/v) to homogenize 0.16 g of brain tissue. After that, a 60-μl aliquot of tissue homogenates or plasma was subjected to protein precipitation using 20 μl of MeOH containing internal standard (IS; 40 ng/ml). After vortex and centrifugation for 15 min at 13,000 rpm, 60 μl of supernatant was mixed with 150 μl of MeOH. Then, the analysis was conducted by ultra-high performance liquid chromatography (UPLC) (ACQUITY H-CLASS, Waters) coupled with mass spectrometric detection (AB SCIEX TRIPLE QUAD 4500, Applied Biosystems). Standard working solutions were made by addition of EB (a concentration range of 0.5 to 500 ng/ml). Micheliolide was used as the IS, and the response of IS was used to adjust the quantities of analytes. Fragmentor voltage and collision energy were 75 V and 34 eV for EB and 30 V and 14 eV for IS; the transitions (precursor to product) monitored were mass/charge ratio (m/z) of 480.1→181.1 for EB and 266.4→231.1 for IS. A 5-μl aliquot of the supernatant was analyzed, and the analysis was conducted by UPLC (ACQUITY H-CLASS, Waters) coupled with mass spectrometric detection (AB SCIEX TRIPLE QUAD 4500, Applied Biosystems). The mobile phase consisted of a linear gradient of deionized water with 10 mM ammonium acetate (A) and acetonitrile (B). A linear gradient was used at a flow rate of 0.3 ml/min, starting from 20 to 80% (B) for 5 min and holding for 6 min at 90% (B). Four mice were included in each experimental group (n = 4).

Tissue preparation and protein extraction

Mouse brains were removed immediately after euthanasia and stored at −80°C for subsequent biochemical analysis. Mouse brains were homogenized in lysis buffer (radioimmunoprecipitation assay) containing protease inhibitors, followed by centrifugation at 12,000g for 15 min. The supernatants were used as total protein lysates for Western blotting analysis of the indicated proteins.

Histopathologic analysis

Mouse brains were removed immediately after euthanasia and fixed in PBS containing 4% paraformaldehyde at 4°C for 48 hours for histopathologic test. Fixed brains were sectioned by a sliding microtome at a thickness of 4 μm.

For Nissl staining, coronal slices were washed twice in PBS, air-dried, and stained with 0.2% cresyl violet (Sigma-Aldrich) solution containing acetic acid for 15 min. Then, slices were washed and dehydrated in graded ethanol (70, 95, and 100%). After dehydration, sections were mounted under a coverslip using resin medium.

For immunocytochemistry, coronal slices were socked in blocking buffer (PBS containing 2% goat serum, 0.2% Triton X-100, and 3% H2O2) for 25 min to remove endogenous peroxidase activity and increase the antibody permeability. Then, slices were incubated with primary antibodies at 4°C overnight, followed by incubation with secondary antibody at room temperature for 2 hours. After dehydration and mounting, sections were imaged with a digital slide scanner (Nanozoomer-SQ, Hamamatsu Photonic).

Quantification and statistical analysis

Unless otherwise mentioned, data are presented on the basis of at least three independent experiments, and representative IHC images from a mouse of each group are shown. Each immunoblot was performed at least three times. All statistical data are presented as means ± SEM. Statistical significance was determined by GraphPad Prism 6.0 software.