br STAR Methods br Author Contributions br
Acknowledgments The authors thank Monika Kuhn for excellent technical assistance and Dr. Antje Schäfer for valuable scientific input. H.S. was supported by the DFG (FZ 82; Graduate School of Life Sciences and SCHI 425-6/1), and A.V.S. by Northwestern University and the Pew Charitable Trusts as Pew Scholar in the Biomedical Sciences. We thank Dr. Titia Sixma and Dr. Sonja Lorenz for kindly providing the plasmid constructs for human UBA1 and ubiquitin, respectively. We also thank BESSY and ESRF for synchrotron beamtime and support.
Introduction Aberrations in post-translational modifications by ubiquitin or ubiquitin-like proteins (Ubl), such as the small ubiquitin-like modifiers (SUMO), are associated with the pathogenesis of life-threatening diseases, such as cancer (Sarge and Park-Sarge, 2011, Zhu et al., 2010), neurodegenerative disorders (Steffan et al., 2004, Subramaniam et al., 2009), and viral infection (Jaber et al., 2009, Kim et al., 2010). For example, multiple studies indicate that SUMOylation is dysregulated in many types of cancers and that the SUMO-activating enzyme (SAE, SUMO E1) could be a potential target to inhibit c-Myc- and KRas-dependent oncogenesis (He et al., 2017, Kessler et al., 2012, Luo et al., 2009, Yu et al., 2015) and reduce cancer cell stemness and resistance (Bogachek et al., 2016, Du et al., 2016). The activating enzyme catalyzing ubiquitin-like Atg8 and Atg12 modifications in autophagy, known as Atg7, has been shown as an indirect target for KRas-dependent oncogenesis (Guo et al., 2013, Rosenfeldt et al., 2013). Despite the importance of Ubl modifications in dysregulated signaling pathways in diseased cells, only a handful of U.S. Food and Drug Administration-approved drugs targeting this type of post-translational modifications have been developed. This deficiency illustrates knowledge gaps in targeting these d-glucose by small molecules and underscores the need to discover novel chemotypes and mechanisms to inhibit Ubl modifications. An Ubl modification requires several steps that are catalyzed by three enzymes, referred to as E1 (activating enzyme), E2 (conjugation enzyme), and E3 (ligase). The SUMO E1 is a heterodimer of SAE1 and Uba2 (also known as SAE2). In brief, an Ubl is first activated by E1 through ATP hydrolysis and forms a thioester conjugate with E1. The Ubl is then transferred to E2, forming a thioester conjugate with E2. Finally, the Ubl is transferred to target proteins, a step usually catalyzed by an E3. Usually, Ubl modifications add new docking sites to target proteins. For example, SUMO modifications enables new protein-protein interactions through the SUMO-interacting motif in receptor proteins (Song et al., 2004, Song et al., 2005). At least three members of the SUMO family (SUMO1, 2, and 3) are ubiquitin-like proteins that can conjugate to other cellular proteins by a biochemical mechanism similar to ubiquitylation (Hay, 2005, Sarge and Park-Sarge, 2009, Yeh, 2009). Currently, the only known mechanism to inhibit the E1 enzymes targets their ATP-binding sites (Brownell et al., 2010, Soucy et al., 2009). Historically, covalent drugs have had great success (e.g., aspirin and penicillin), and covalent drugs have become a focus in anticancer and antiviral drug discovery (Kalgutkar and Dalvie, 2012, Singh et al., 2011). These compounds contain low reactivity warheads that allow covalent adducts to form only when a non-covalent complex forms first. Their duration of action depends on the turnover rate of their protein targets, independent of the drug's stability in the blood.
Discussion In this study, we have discovered an allosteric inhibitor chemotype that inhibits SUMO E1 through a covalent mechanism. Our further studies demonstrated covalent adduct formation of COH000 to Cys30 of SUMO E1, a cysteine that is not involved in binding ATP or SUMO. This is consistent with the results that COH000 does not compete with ATP or SUMO1 binding (Figure 2A). Cys30 is fully buried in previously determined crystal structures of SUMO E1s (Lois and Lima, 2005, Olsen et al., 2010), and the analogous sites are buried in previously published Ubl E1 crystal structures (Lee and Schindelin, 2008, Noda et al., 2011, Walden et al., 2003). Crystal structure of COH000 in complex with SUMO E1, indeed, revealed a more extensive conformational flexibility than seen previously for this family of enzymes (Lv et al., 2018). Therefore, binding of COH000 to the SUMO E1 requires a conformational state of the enzyme not previously observed. Sequence comparison of E1 enzymes revealed that the covalent modification site and the amino acid sequence surrounding it is conserved among the human Uba1 (ubiquitin E1), Uba2 (SUMO E1), Uba3 (Nedd8 E1), Uba4 (Urm1 E1), Uba7 (ISG15 E1), and Atg7 (E1 for Atg8 and Atg12) (Figure S1C). Thus, the covalent allosteric inhibition mechanism exemplified by COH000 could be applicable to other E1 enzymes. Findings described here also raise a question of whether E1 enzymes can be allosterically regulated by small-molecule metabolites that bind to the same site as COH000.