Non-catalytic UBL2 domain directs deubiquitinase USP11 toward K48-linked polyubiquitin chains

Abstract

Ubiquitin-specific proteases (USPs), comprising the largest deubiquitinase family, are generally thought to have poor discrimination of ubiquitin (Ub) linkage types, but a number of USPs show preference toward certain linkages. USP11, a USP-family member implicated in cancer and neurodegeneration, carries an atypical catalytic domain which is split into two segments through the insertion of a UBL2 domain and an intrinsically disordered region (IDR). In addition, the chain-type selectivity of USP11 remains unclear based on the conflicting data from in vitro and in vivo studies. Here, we identify an important role of the UBL2-IDR in altering the ability of USP11 to cleave K29, K33, and K48 chains, with K48 chain showing the most significant effect. Using in vitro studies with Ub-tetramer and ubiquitinated proteins as well as cell-based analyses, we demonstrate that UBL2 domain endows USP11 with a selectivity towards the K48-linked Ub chains. Importantly, this function of UBL2 is not observed in its paralogs USP4 and USP15, which display broad activities towards most chain types. By leveraging AI-based virtual screening, we have identified selective USP11 inhibitors, including the FDA-approved drugs Fenoldopam and Olanzapine and their analogs, which act through a unique chemical scaffold and display significant efficacy both in vitro and in cells. Our findings not only uncover a previously unrecognized mechanism of linkage selectivity within the USP family but also provide a robust platform for the rational design of USP11-targeted therapeutics, underscoring the critical role of non-catalytic domains in deubiquitinase regulation and offering promising avenues for therapeutic intervention.

Keywords: DiffDock; K48-linkage; UBL2; USP1; deubiquitinase; polyubiquitin chain.

Figure 1

Domain architecture of USP11 and design of USP11 variants.A, full-length USP11 consists of multiple structured domains and one intrinsically disordered region. The catalytic domain is split into D1 and D2, with a UBL2-IDR inserted in between. Two USP11 variants were designed: USP11ΔDUSP, representing the catalytically active form, and USP11-D1D2, representing the catalytic core-only conformation. Notably, D1 and D2 are connected by five residues (ASTSK) (PDB ID: 8OYP (32)). B, AlphaFold3-predicted 3D structure of USP11, with structural domains color-coded as in panel A. The catalytic core, formed by D1 and D2 segments, resembles the typical USP-type catalytic domain. The UBL2-IDR insert contains a UBL2 domain, a previously uncharacterized UBL-like domain, and a disordered region. C, the AlphaFold pLDDT confidence scores for USP11 suggest high confidence in the structured domains; however, the domain–domain interactions may require further experimental validation and refinement.

Figure 2

Deubiquitination and Ub chain selectivity of USP11-D1D2 and USP11ΔDUSP.A, FL-USP11 expressed and purified from HEK293T cells was employed to cleave K48 or K63-linkaged Ub₄ chains, resulting in reduced amounts of Ub₄ and increased short Ub chains. B, C, seven types of Ub₄ chains were incubated with USP11-D1D2 (B) or USP11ΔDUSP (C), revealing distinct substrate selectivity and cleavage efficiency. D, the amounts of residual Ub₄ chains after 4-h digestion were evaluated from immunoblots in (B) and (C) and plotted. (E, F) USP11-D1D2 (panel E) and USP11ΔDUSP (panel F) failed to cleave the K29 or K48-linked Ub₂ chain.

Figure 3

Selectivity of USP11-D1D2 and USP11ΔDUSP toward K48 and K63 chain-modified substrates and their catalytic efficiency.A, polyubiquitinated substrates were generated using UBA1, UBCH7, wild-type or mutant (Q808M-E809L) Rsp5 E3 ligase, together with fluorescein-labeled Ub. The wild-type Rsp5 produced K63-linked polyubiquitin chains, while the mutant form generated K48-linked chains. B, USP11-D1D2 and USP11ΔDUSP displayed distinct cleavage patterns toward these substrates. The D1D2 variant was weakly active toward K48-linked chains, whereas USP11ΔDUSP efficiently cleaved both K48- and K63-linked chains, consistent with their activity against Ub₄ substrates. Chain-specific antibodies were used to confirm the composition and deubiquitination of the K48 and K63-linked substrates. C, Ub-Rho was used to evaluate catalytic activity. Initial cleavage rates were measured and plotted against USP11 concentration to determine apparent k_cat/K_M values.

Figure 4

Ub chain selectivity of USP4 and USP15 is not affected by the UBL2-IDR insert.A, B, the domain architectures of USP4 and USP15 follow the same pattern as USP11 (shown in Fig. 1A). ΔDUSP and D1D2 variants were also constructed to assess whether the UBL2 insert affects chain selectivity. The D1D2-assembled structures of USP4, USP11, and USP15 are highly similar, with RMSD values less than 0.4 Å. C, D, both USP4-D1D2 and USP4ΔDUSP cleaved polyubiquitinated Rsp5 substrates with nearly identical patterns, indicating that the UBL2-IDR does not influence chain selectivity in USP4. Similar findings were obtained with USP15-D1D2 and USP15ΔDUSP.

Figure 5

USP11 UBL2 domain is critical for the cleavage of K48-polyubiquitinated substrates in vivo.A–C, HEK293T cells were transfected with indicated constructs and cell lysates were used for immunoprecipitation followed by Western blot analysis with indicated antibodies.

Figure 6

Repurposing FDA-approved drugs to identify USP11-specific inhibitors.A, the structure of USP11-D1D2 was used for virtual screening of 3113 FDA-approved compounds via DiffDock. The top 100 ranked candidates were manually inspected to ensure docking at the catalytic core or S1 Ub-binding site. A total of 32 compounds were selected for in vitro assays. B, 96 compounds were tested for USP11 inhibition using Ub-FAM as a substrate. Among them, 64 compounds were originally selected from USP45 and PLpro-CoV2 screens followed by the same workflow as in (A) and were included as controls. Fenoldopam and Olanzapine were identified as effective inhibitors of USP11-D1D2. C, the indicated compounds were tested for their abilities to block the deubiquitination activity of USP11ΔDUSP2 toward K48- or K63-poyubiquitin modified substrates. D, chemical structures and predicted USP11–inhibitor complex models are shown. Mitoxantrone, previously reported to inhibit USP11, and PR-619, a broad-spectrum DUB inhibitor, are included for comparison. E, structural models of USP11 in complex with Ub, Fenoldopam, or Olanzapine suggest that the two compounds bind to a cleft overlapping with the C-terminal tail of Ub (L73–G76). Key residues involved in the interactions are labeled.

Figure 7

Inhibitory validation of Fenoldopam and Olanzapine analogs.A, Five analogs of Fenoldopam and Olanzapine were selected for USP11 inhibition assays. Original compound names or CAS number are provided. The top-ranked DiffDock-generated structures suggested the analogs target to the same site Fenoldopam and Olanzapine target for. B, the indicated compounds were assayed for their activity to inhibit the deubiquitination activity of USP11-D1D2 toward K63-polyubiquitinated substrates. C, NMR WaterLOGSY spectra of 100 μM SRO-01 in the absence (blue) or presence (red) of 5 μM USP11-D1D2 showed opposite signal phases, while DMSO peaks remained in the same phase. These results confirm that SRO-01 binds to USP11 in solution. D, SRO-02 was also analyzed by WaterLOGSY, but no significant signals with opposite phase were observed, suggesting no binding or only very weak interaction below detection limits.

Figure 8

Identified inhibitors effectively suppress USP11-mediated deubiquitination of SOX11.A, Ubiquitinated Flag-SOX11 was immunoprecipitated from lysates of HEK293T cells transfected with His-Ub and Flag-SOX11 and then subjected to in vitro deubiquitination assays with USP11ΔDUSP, together with indicated inhibitors. B, C, in vivo analysis of Flag-SOX11 deubiquitination in HEK293T cells transfected with His-Ub, Myc-USP11 and Flag-SOX11 (B) or HA-tagged c-Myc (C) and treated with indicated inhibitors. Among the three panels, SRO-02 was used as a negative control for showing the contrast of inhibitory efficiency by the selected inhibitors.

Figure 9

Model of USP11-mediated polyubiquitin chain selectivity.A, domain organization of USP11 and USP11-D1D2 is shown with color coding consistent with Figure 1B. B, Di-Ub substrates contain a single isopeptide bond, allowing both USP11 and USP11-D1D2 to cleave K6-, K11-, K33-, and K63-linked di-Ub chains with similar specificity. C, in the context of tetra-Ub chains, USP11 distinguishes between linkage types through different recruitment mechanisms. K29 or K48-linked Ub₄ is recognized via the UBL2 domain. In contrast, recognition of K6, K11, or K63-linked Ub₄ does not require UBL2. This model suggests that UBL2 contributes to selective recognition and processing of K29 or K48-linked chains. M1 or K33-linkage chains may fail to bind to USP11ΔDUSP due to structural colliding to UBL2 or UBL2-IDR.