The structure of the deubiquitinase USP15 reveals a misaligned catalytic triad and an open ubiquitin-binding channel

Abstract

Ubiquitin-specific protease 15 (USP15) regulates important cellular processes, including transforming growth factor β (TGF-β) signaling, mitophagy, mRNA processing, and innate immune responses; however, structural information on USP15's catalytic domain is currently unavailable. Here, we determined crystal structures of the USP15 catalytic core domain, revealing a canonical USP fold, including a finger, palm, and thumb region. Unlike for the structure of paralog USP4, the catalytic triad is in an inactive configuration with the catalytic cysteine ∼10 Å apart from the catalytic histidine. This conformation is atypical, and a similar misaligned catalytic triad has so far been observed only for USP7, although USP15 and USP7 are differently regulated. Moreover, we found that the active-site loops are flexible, resulting in a largely open ubiquitin tail-binding channel. Comparison of the USP15 and USP4 structures points to a possible activation mechanism. Sequence differences between these two USPs mainly map to the S1' region likely to confer specificity, whereas the S1 ubiquitin-binding pocket is highly conserved. Isothermal titration calorimetry monoubiquitin- and linear diubiquitin-binding experiments showed significant differences in their thermodynamic profiles, with USP15 displaying a lower affinity for monoubiquitin than USP4. Moreover, we report that USP15 is weakly inhibited by the antineoplastic agent mitoxantrone in vitro A USP15-mitoxantrone complex structure disclosed that the anthracenedione interacts with the S1' binding site. Our results reveal first insights into USP15's catalytic domain structure, conformational changes, differences between paralogs, and small-molecule interactions and establish a framework for cellular probe and inhibitor development.

Keywords: USP15; catalytic triad; crystal structure; cysteine protease; deubiquitylation (deubiquitination); protease; protein conformation; protein degradation; ubiquitin; ubiquitin-specific protease.

Figure 1.

Kinetic parameters and crystal structure of the USP15 catalytic core. A, schematic representation of the human USP15 domain structure highlighting the location of the catalytic core region encompassing the subdomain halves D1 and D2 in blue and the catalytic triad residues (green) with additional domains labeled as DUSP (domain present in USPs) and UBL (ubiquitin-like). B, kinetic assays of FL-USP15 and USP15-D1D2 using ubiquitin-AMC as a substrate with FL-USP15 in green and USP15-D1D2 in blue. Each point represents the mean for data points measured in triplicate. Values for V_max and K_m were used to calculate the turnover number, k_cat, and catalytic efficiency, k_cat/K_m, and are listed in the table on the right. Error bars, S.E. C, cartoon representation of the crystal structure of the USP15 catalytic core with catalytic triad residues shown as a green stick representation and active-site loops and key secondary structure elements labeled. D, B-factor “putty representation” of USP15-D1D2 highlighting the variation of B-factors where the thickness is proportional to its local B-factor and thus its flexibility and is color-coded blue to red (for lowest to highest B-factors). The approximate location of the distal ubiquitin is modeled as a semitransparent surface representation (red) into the S1 binding site and shows that the finger region will need to flex to accommodate ubiquitin.

Figure 2.

Comparison of USP4 and USP15 catalytic core crystal structures. A, superposition of cartoon representations of the USP15 structure (in blue with catalytic triad residues in green) and the USP4 structure with β-mercaptoethanol bound to the catalytic cysteine (in gray with catalytic triad residues shown as sticks in wheat; PDB ID 2Y6E (24)). Note the differences in the catalytic triad residues and surrounding loop regions CCL, SL, BL1, and BL2. B, close-up view of the active-site region showing the different conformations of USP15 (blue) and USP4 (gray) residues in the CCL, SL, and BL2. The approximate location of the ubiquitin C-terminal tail is depicted in a red semitransparent cartoon representation. C, conformational differences of selected labeled residues shown in stick representation in USP15 (in blue on the left) and USP4 (in gray on the right). Equivalent residues in both structures located in the switching loop SL are highlighted in orange (note that in USP15, the SL is largely flexible, indicated by a dotted line), whereas other intervening residues are highlighted in magenta. The catalytic cysteines are depicted in wheat, and the histidine and aspartate of the catalytic triad are colored in green. D, surface representations of USP15 (left) and USP4 (right) catalytic cores. Residues that are dissimilar between the two paralogs are highlighted in yellow, residues with weakly similar properties are colored in orange, and residues with similar properties are colored in blue. The light blue or gray background, respectively, denotes fully conserved residues between USP15 and USP4. Catalytic triad residues are highlighted in green, and selected residues are labeled. Note the differences in conservation between the large distal ubiquitin-binding cavity (S1 pocket) and the S1′ regions in both structures.

Figure 3.

Monoubiquitin and linear diubiquitin binding to USP4 and USP15 active-site mutants. ITC analyses of raw data measured at 25 °C and binding isotherms fitted to a one-site binding model of USP15-D1D2 C269S and USP4-D1D2 C311S with monoubiquitin and linear diubiquitin, respectively. Respective dissociation constants are listed in the table below, and associated ΔG, −TΔS, and ΔH values are graphically depicted on the right to highlight different contributions to the binding.

Figure 4.

Influence of blocking loop 2 mutations on the ubiquitin-binding behavior of USP15. ITC raw data and binding isotherms fitted to a one-site binding model of USP15-D1D2, USP15-D1D2 G860V, USP15-D1D2 bl2usp4, and USP4-D1D2 with monoubiquitin at 37 °C. The respective dissociation constants are given in the table below on the right, and the associated thermodynamic parameters ΔG, ΔH, and −TΔS are graphically represented on the left.

Figure 5.

Comparison with other USP catalytic core structures determined in the absence of ubiquitin. A, crystal structures of USP15 (blue) and USP4 (PDB code 2Y6E (24)), USP8 (PDB code 2GFO (27), USP7 (PDB code 4M5X (38)), USP14 (PDB code 2AYN (33)), and USP12 (PDB code 5K16 (41)) shown in a gray cartoon representation with catalytic triad residues highlighted in green and BL1, SL, and BL2 loop regions each depicted in red. B, superposition of the same structures as seen in A, highlighting the most variable regions in the structures. C, superposition of the active-site region of USP15 (blue) and USP7 (light blue), the only known USP catalytic core structures whose catalytic triad is misaligned in a similar way in the free enzyme. Note the difference in switching loop conformations and the significant distances (Å) between the catalytic cysteines and histidines.

Figure 6.

Inhibition of USP15 by mitoxantrone. A, IC₅₀ curve for mitoxantrone as an inhibitor of USP15 using diubiquitin gel shift cleavage assays. Error bars, S.E. B, mF_obs − DF_calc omit electron density map calculated with the mitoxantrone molecule removed contoured at 2.0σ with the density shown in light orange and corresponding final model in an orange stick representation. C, cartoon representation of the USP15–mitoxantrone complex crystal structure. The USP15 protease domain is depicted in blue with catalytic triad residues in green and mitoxantrone shown in a space-fill representation in orange. D, close-up view of the molecular basis of the interaction. Key residues involved in the interaction are labeled and shown as yellow sticks. Otherwise, the color code is the same as in C. E, same representation of the USP15–mitoxantrone complex as in C with a diubiquitin substrate shown as a semitransparent surface modeled into the active site based on the orientation seen in the crystal structure of the USP30–Lys⁶-diubiquitin complex (PDB code 5OHP (35)) to highlight clashes with mitoxantrone binding in the S1′ pocket. F, superposition of available crystal structures of USP7 (gray) in complex with small-molecule inhibitors in magenta stick representations (PDB codes 5N9R, 5N9T, 5NGE, 5NGF, 5UQV, 5UQX, and 5WHC (49–51)) and USP2 (gray; PDB code 5XU8 (52)), highlighting the interaction locations in predominantly the S1 pocket and active-site channel. USP15 is depicted in a blue cartoon representation with mitoxantrone in orange. Catalytic triad residues are depicted in green stick representations.

Figure 7.

Hypothetical model of an activation mechanism to align the USP15 catalytic triad upon substrate binding. A, superposition of USP15 (blue) and USP4 (gray; PDB code 2Y6E (24)) crystal structures with possible conformational changes upon substrate binding to align the remote Cys²⁶⁹ into an active conformation indicated based on the two structures. Hypothetical conformational changes of selected residues in the SL (orange in both structures), helices α5 and α6, CC loop, and surrounding regions are indicated by arrows between equivalent residues in the sequences. The active-site cysteines are both depicted in green stick representations (BME adduct in USP4). B, different view of A to highlight conformational differences between the two structures. The color code is as in A.