DeSUMOylation: An Important Therapeutic Target and Protein Regulatory Event

Properties of the deSUMOylation proteases

The process of SUMOylation has dynamic reversible activity, and a rapid modification at even a small portion of target proteins is sufficient to produce tremendous functional changes (Hay, 2005). Therefore, a distinctive pattern of SUMO conjugates is always observed during the cell cycle (Kolli, et al., 2010). The reversible modification is due to the action of proteases, which remove attached molecules from the substrate as well as are responsible for SUMO maturation (Kim and Beak, 2009; Hickey et al., 2012). Some of the well-known protease families include Ulp1/2 in yeast and SENPs (SENP1–3 and SENP5–8) in mammals (Huang et al., 2009). Among these examples, SENP8 possesses nonspecificity to SUMO and has been identified as a specific protease for neddylation (Shin et al., 2011).

The C-terminus of Ulp/SENP contains different catalytic domains, whereas N-terminal sequence frequently dictates their subcellular localization as the N-terminus-truncated mutants of SENP3 exhibit similar proteolytic activity to wild-type enzymes, (Nishida et al., 2001). A list of various deSUMOylating enzymes and their properties are summarized in Table 1. Most SENPs tend to localize in the nucleus or colocalize in distinguishable subnuclear compartments (Kolli et al., 2010). For example, the closely related SENP1 and SENP2 reside in the nuclear envelope by linking the nuclear pore complex (NPC) (Zhang et al., 2002). Interestingly, their localization can undergo some shifting during cell cycle progression or cell stress (Bailey and O'Hare, 2004; Huang et al., 2009; Bawa-Khalfe et al., 2010), implicating that their activities are spatially regulated or may display divergent roles according to specific subcellular distribution. Indeed, the specific subcellular localization of mammalian SENPs partly contributes to their specificity (Han et al., 2004). Moreover, SENPs are capable of processing all SUMOs due to the existence of catalytic discrimination within the SUMO family (Hay, 2007). SENP1 and SENP2 can efficiently process and remove all three SUMO isoforms from their targets; however, SENP1 has a robust activity in deconjugating SUMO1-modified proteins during mouse embryonic development, while exhibiting limited efficiency in deSUMOylating SUMO-2/3 from its substrates (Sharma et al., 2013). SENP1 can also remove SUMO-1 from poly-SUMO-2/3 chains, thus it appears to have a specialized role in dismantling SUMO2/3 chain length (Sharma et al., 2013).

Besides the aforementioned SENPs, recently, new classes of deSUMOylase have been identified and characterized. DeSUMOylating isopeptidase (DeSI) is one of the family (DeSI1 and DeSI2) that recognizes different sets of substrates, for example, DeSI1 specifically deSUMOylates the transcriptional repressor, BZEL (GeneID: 72147) (Shin et al., 2012). USPL1 is another newly derived protease that is uniquely localized in glial bodies within the nucleus, implying a potential role in cell division apart from its catalytic functions (Schulz et al., 2012).

SENPs and gene expression regulation

Protein deSUMOylation is performed by SUMO proteases and any deactivation in them usually results in the accumulation of SUMOylated proteins in cells. Other than that SUMO proteases are also involved in numerous biological processes, including gene transcription, nucleocytoplasmic transport, cell proliferation, and early embryogenesis (Lu et al., 2009; Yeh, 2009). A list of substrates of individual SENPs and their intracellular implications are shown in Table 2. Despite the underlying molecular mechanisms of SENP functions remains enigmatic, SENP knockout or mutated mice are embryonically lethal, implicating that the role of SENPs is not redundant (Yamaguchi et al., 2005; Chiu et al., 2008; Kang et al., 2010).

Multiple regulators involved in transcriptional repression are targets for SUMOylation (Lindberg et al., 2010; Witty et al., 2010). SENPs positively regulate transcriptional activity of targets probably owing to counteraction of the canonical activity of SUMOylation or the maturation of SUMO precursors during transcriptional repression (Lyst and Stancheva, 2007). One classical role of SENP1 is its participation in maintaining the dynamic balance of Elk-1 SUMOylation. Depletion of SENP1 dampens transcriptional activation of Elk1 and enhances the activation of Wnt target genes by accumulating SUMOylated TBL1-TBLR1 (Kaikkonen et al., 2010; Choi et al., 2011). A recent study reveals that SENP1 can regulate mitochondrial biogenesis and functions through deSUMOylation of PGC-1α and subsequently promote its transcriptional activity (Yu et al., 2012).

Myocyte-specific enhancer factor-2 (MEF2) is a subset of transcriptional factors that play essential roles in embryonic development (Lu et al., 2013), and SENP2 is characterized as a major dominator for MEF2A transcriptional activity as it directly removes SUMO from the conjugated form and subsequently boosts its transcriptional activity (Qi et al., 2014). Similar mechanism is also shown by SENP2 in the regulation of ERK5 activity in endothelial cells (Heo et al., 2013).

With the exception to directly dictate gene expression, some SENPs exert indirect roles by affecting DNA epigenetic modification. One case is SENP2, which regulates Gata4 and Gata6 transcription through altering the interaction of Pc2/CBX4 on its promoters. In SENP2-deficient embryo, a higher SUMOylation level of Pc2/CBX4 markedly elevates the level of Pc2/CBX4 on PcG target gene promoters and boosts methylation of H3K27me3, resulting in Gata4 and Gata6 transcription repression and consequently embryonic lethality (Zhang et al., 2004; Kang et al., 2010). Similarly, SENP3 can deSUMOylate RbBP5 (Nayak et al., 2014), a component of the SET1/MLL regulatory module, which additionally comprises WDR5, Ash2L, and DPY-30 (Zhang et al., 2013). SENP3 depletion attenuates deposition of Ash2L on the DLX3 gene and compromises subsequent H3K4 methylation by SET1/MLL, leading to DLX3 transcription suppression (Nayak et al., 2014). Based on these evidences, further studies on other SENPs presumably reveal their nonredundant roles in epigenetic modification.

P53 is a well-known tumor suppressor, which plays an essential role in preventing aneuploidy by triggering checkpoints that handle the damages or induce apoptosis to eliminate the affected cells (Sherr et al., 2005; Andreou and Tavernarakis, 2010). Cumulative works discovered pivotal implications of SENPs on the regulation of p53 activities. The paramount elucidated case is the SENP2-Mdm2-p53 pathway (Fig. 1), which modulates p53/Mdm2 circuit in mice trophoblast layer development (Chiu et al., 2008). The full-length isoform of SENP2 is indispensable and sufficient to negatively regulate p53 activity; however, when ablated, it apparently perturbs SUMO modification and subcellular distribution of Mdm2 by diminishing the target p53 degradation. Interestingly, reintroduction of SENP2 alleviates this deficiency by deconjugating SUMOylated Mdm2, thereby decreasing the p53 level (Jiang et al., 2011; Heo et al., 2013). Notably, the nucleolar SUMO-specific protease, SMT3IP1/SENP3, also participates in the Mdm2-p53 pathway. Overexpression of SENP3 initiates accumulation of Mdm2 in the nucleolus and stabilizes p53 protein by competing with p53 for Mdm2 binding, thereby suppressing Mdm2-mediated ubiquitin–proteasome degradation of p53. It is noteworthy that the role of SENP3 in the facilitation of p53 stabilization is independent of its deSUMOylation activity (Nishida and Yamada, 2011).

OPEN IN VIEWER

FIG. 1.

The functions of sentrin-specific protease (SENP)2 and SENP3 in P53 transcription regulation during the Mdm2-P53 ubiquitin–proteasomal degradation pathway. (A) Under physiological conditions, Mdm-2 itself functions as a ubiquitin ligase E3 toward p53 and, when SUMOylated, it can interact with P53 following the nuclear export of p53, which entitles P53 to be degraded by 26S proteasome-mediated proteolysis. (B) SENP2 overexpression leads to the deconjugation of small ubiquitin-like modifier (SUMO) isoform(s) from Mdm-2, which attenuates the association of Mdm-2 and P53 and abrogates the Mdm-2-mediated P53 degradation. Both of them contribute to the stabilization of P53 and subsequential transcription of P53 target genes. (C) When compared with SENP2, SENP3 stabilizes P53 through a competitive inhibition and deSUMOylation activity independent mechanism. SENP3 can restrict Mdm-2 into the nucleolus and has the ability of competing with P53 for Mdm-2 binding, which prevents Mdm-2-mediated ubiquitin–proteasomal degradation of P53. Color images available online at www.liebertpub.com/dna

SENPs and ribosome biogenesis

Another charming role of SENPs is evidenced by its participation in ribosomal biogenesis. Eukaryotic ribosome maturation is a tightly coordinated multistep process (Zemp and Kutay, 2007; Thomson et al., 2013). Intensive evidence from in vitro investigation of SENP3 in mitosis cells acknowledges deSUMOylation as a pivotal regulatory process in coordinating ribosome formation in time with the physiological state of cells.

NPM1/B23 is a plethora of shuttling phosphoproteins, which associates with 60S preribosome and plays an important role in the 28S rRNA maturation process (Haindl et al., 2008; Maggi et al., 2008). Given the crucial role of NPM1, there exist two conflicting reports. Both studies confirmed a concise effect of SENP3 and SENP5 on 60S periribosome maturation and export. One study concludes that SENP3 and SENP5 are necessary for efficient rRNA processing through their proteolytic purpose of disengaging the SUMO isoform from its molecular chaperone, NPM1. Depletion of SENP3 leads to constitutive SUMOylation of NPM1, which severely impinges nucleolar ribosomal RNA processing coupled with accumulation of 32S pre-rRNA particles, suggesting a defect in splicing the internal transcribed spacer-2 (ITS-2), which resides between 5.8S and 28S. More importantly, SENP5 affects rRNA processing as well and, when depleted, decreases the production of primary 47S rRNA transcripts (Haindl et al., 2008; Burger, 2013). The second study illustrates that B23/nucleophosmin forms a complex with SENP3 and SENP5, and depletion of B23/nucleophosmin conspicuously diminishes the level of both SENPs. This indicates that the function of SENP3 and SENP5 in ribosome biogenesis is conferred by physical interaction with B23/nucleophosmin, which entitled B23/NPM1 to regulate their abundance and spectrum of SUMOylated conjugates within nucleoli (Yun et al., 2008). These results implicate that although both SENP3 and SENP5 colocalize with granular components of the nucleolus, they are presumably responsible for deSUMOylation of a specific group of species, which are crucial for any particular process in the ribosome biogenesis pathway.

The mechanism of SENP3 involved in ribosome maturation has been explored owing to the identification of a protein complex comprising PELP1-TEX10-WDR18 in mammals. It interacts with NPM1, indispensable for 28S rRNA formation and nucleoplasmic export of 60S ribosomal intermediates (Finkbeiner et al., 2011). PELP1 is preferentially modified by SUMO2/3 and this modification is reversed by SENP3. Similarly, SENP3 knockdown relocalizes the expression of PELP1 from the nucleolus to the nucleoplasm (Nair et al., 2010). In SENP3 deficiency cells, the 60S preribosome subunit retains in the nucleolus, suggesting that SENP3-mediated removal of SUMO2/3 from the PELP1 complex determines its subnuclear distribution. However, a SUMOylation-deficient mutant of PELP1 still exports from the nucleolus upon SENP3 depletion, indicating that PELP1 is not the only target of SENP3 during this process (Finkbeiner et al., 2011). We hypothesize that the coordinated SUMOylation of several components of the PELP1 complex emerges as a checkpoint that hinders premature assembly of the PELP1-WDR18-TEX10 complex to 60S rRNA or serves as a signal for its release from these structures.

Another prominent example is LAS1L, a nucleolar protein that is essential for cell proliferation, 60S ribosomal subunit synthesis, and 28S rRNA maturation (Castle et al., 2010). The loss of LAS1L-associated complex leads to increased abundance of the 32S pre-rRNA subunit. LAS1L also couples with SENP3, PELP1, TEX10, NOL9, and WDR18 and cosediments with the pre-60S ribosomal particle. Further studies reveal that SUMOylation of LAS1L is dictated by SENP3 and either SENP3 or NPM1 depletion boosts LAS1L SUMOylation and its relocalization from the nucleolus to the nucleoplasm (Castle et al., 2011). This indicates that SENP3 is partially involved in ribosome biogenesis by governing sublocalization of LAS1L, PELP1, and SENP3 and probably serves as a checkpoint to restrict the accessibility of SUMO conjugates to 60S.

One landmark step during ribosome biogenesis is preribosome export through nuclear pores. Previous studies reveal that ribosomal export is an energy-dependent process, which utilizes the export factor Crm1 and RanGTPase system, and depletion of any of these factors results in rRNA accumulation within the nucleus (Johnson et al., 2002; Fromont-Racine et al., 2003). Despite the unequivocal significance of Crm1p to ribosome nuclear export, we can still envisage that a single export receptor is insufficient to translocate large assembled cargo through NPC. Nucleophosmin (NPM1) contains a functional Crm1-dependent nuclear export signal that enables its nucleocytoplasmic shuttling; furthermore, NPM1 is spatiotemporally controlled by the Ran-Crm1 complex, suggesting that it might contribute to presubunit export (Wang et al., 2005). Indeed, pre-60S export is definitely disturbed after overexpression of either mutants of nonshuttling NPM1 (Yu et al., 2006). Both SENP1 and SENP2 display their role in NPC localization, proper configuration, and function in the maintenance of nucleoporin homeostasis (Chow et al., 2014), thereby we could not preclude the possibility that SENP1 and SENP2 are potent factors in ensuring ribosome maturation.

Besides preribosomal export, the involvement of SENPs in other ribosome biogenesis processes during ribosome biogenesis is categorically necessary for functional ribosome maturation. It is reported that overexpression of SENP8 destabilizes a subset of ribosomal proteins and reduces the level of ribosomal protein L11, possibly by coregulating ubiquitin–proteasomal degradation, implying that SENP8-mediated neddylation is related to ribosome biogenesis (Xirodimas et al., 2008). Attractive avenues of future research on the possible link between SUMOylation and ribosome biogenesis will be to identify unknown SUMOylation substrates involved in ribosomal maturation, which will elucidate the potential cross talk of SUMO and other ubiquitin-related modifiers. This will also improve our current understanding of whether and how the molecular pathways that affect ribosome maturation are interrelated.

SENPs and mitochondrial dynamics

The mitochondrion, a multifunctional organelle that can dynamically undergo fusion and fragmentation to coordinate cell cycle progression, exerts crucial roles in numerous cellular processes, including energy metabolism, apoptosis, and senescence (Lee et al., 2014). It is pertinent to state that mitochondrial dysfunction has been determined as a vital contributor to various human diseases (Knott et al., 2008; Guo et al., 2013). Thus, the regulation of mitochondrial dynamics during mitosis might be a determinant of cellular survival, particularly in oocytes, of which developmental competence is substantially dependent on energy metabolism (Zeng et al., 2014). Despite the functional relevance of mitochondrial dynamics that has been recognized under the condition of mitochondria-dependent apoptotic cell death (Karbowski and Youle, 2003), the regulatory machinery of mitochondrial remodeling is far less fully understood. Recently, researchers are focused on the possible role of SUMOylation in diagnosis of different diseases and their therapeutics. Several elegant discoveries are made in the connection between SUMOylation and mitochondrial morphology regulation.

Drp1, a dynamin-related fission GTPase that is important for mitochondrial fission, can be SUMOylated, which is required for Drp1 recruitment to the mitochondrial membrane (Zunino et al., 2007; Guo et al., 2013); SENP5, a primarily nucleoli localized SUMO protease with substantial residues within the cytosol, can translocate from nucleoli to the mitochondrial surface to facilitate deSUMOylation of Drp1 and alter mitochondrial morphology at G2/M transition (Zunino et al., 2009). Upon silencing of SENP5, the cell cycle is significantly compromised. Furthermore, SENP5 overexpression can alleviate SUMOylation-mediated mitochondrial fragmentation through deconjugation of SUMO-1 from Drp1, while SENP5 depletion turns mitochondrial morphology fragmented. Additionally, SENP5 downregulation leads to a dramatical increase in reactive oxygen species (ROS) production, which can be attenuated by silencing of endogenous Drp1. All these data reveal SUMOylation as well as SENP5-mediated deSUMOylation both as master regulators of mitochondrial metabolism and subsequent cell cycle progression.

Besides SENP5, SENP3 is recently shown to be a regulatory switch for Drp1 through modulating SUMO-2/3 removal from Drp1, which promotes mitochondrial fragmentation, releases cytochrome c, and apoptosis that represents a classical example of therapeutic targets (Guo et al., 2013). Similarly, the model for ischemic resistance is classically reviewed by Anderson and Blackstone (2013). Of note, Drp1 can be SUMOylated by both SUMO-1 and SUMO-2/3, in which future work is concentrated on characterizing additional mitochondrial SUMO targets and other SUMO proteases involved in mitochondrial morphology that might provide more insights into the functional link between mitochondrial dynamics and cell cycle progression. All of these efforts will definitely promote the identification of a new therapeutic target related to mitochondrial dysfunction.

SENPs and cell cycle regulation

With regard to cell cycle progression, SENP3 is reported to balance SUMOylation of Borealin couple with RanBP2 (Klein et al., 2009). Borealin is essential for a number of biological processes, including chromosome segregation, spindle assembly checkpoint, and cytokinesis (Welburn et al., 2010; Meyer et al., 2013). Our laboratory is currently concentrating on different roles of SENP3. We recently explored nucleolus periphery localization of SENP3 in mouse oocytes and found that SENP3 depletion exacerbated the aberrant spindle formation and disrupted meiosis progression, implying a potent role of SENP3 in different reproductive defects (unpublished data). In case of HIF1, an important transcriptional factor in hypoxia, SENP3 has been reported to have biphasic redox sensing, resulting in variable SUMOylation of its substrate, p300, which ultimately deactivates HIF1 (Wang et al., 2012). This study indicated its important roles in ROS generation, carcinogenesis, and its cure. Significantly, apart from SENP3, almost all other SUMO proteases could interfere with the cell cycle to some extent (Wang et al., 2010; Wang et al., 2013), although the molecular mechanisms of some of them are not fully known (Table 2). Based on these studies, deSUMOylation agents could be potential therapeutic targets against important ischemic and cancerous diseases. The attempts to make inhibitors to control deSUMOylation are in progress, but no pharmacological agents are available to date.