Sub-cellular localization specific SUMOylation in the heart^☆

Abstract

Although the majority of SUMO substrates are localized in the nucleus, SUMOylation is not limited to nuclear proteins and can be also detected in extra-nuclear proteins. In this review, we will highlight and discuss how SUMOylation in different cellular compartments regulate biological processes. First, we will discuss the key role of SUMOylation of proteins in the extra-nuclear compartment in cardiomyocytes, which is overwhelmingly cardio-protective. On the other hand, SUMOylation of nuclear proteins is generally detrimental to the cardiac function mainly because of the trans-repressive nature of SUMOylation on many transcription factors. We will also discuss the potential role of SUMOylation in epigenetic regulation. In this review, we will propose a new concept that shuttling of SUMO proteases between the nuclear and extra-nuclear compartments without changing their enzymatic activity regulates the extent of SUMOylation in these compartments and determines the response and fate of cardiomyocytes after cardiac insults. Approaches focused specifically to inhibit this shuttling in cardiomyocytes will be necessary to understand the whole picture of SUMOylation and its pathophysiological consequences in the heart, especially after cardiac insults. This article is part of a Special Issue entitled: Genetic and epigenetic control of heart failure - edited by Jun Ren & Megan Yingmei Zhang.

1. Introduction

SUMOylation is an important post-translational modification in which one or more small-ubiquitin like modifier (SUMO) peptides are conjugated to a protein and contributes to the complexity of eukaryotic proteomes. SUMOylation is a dynamic and reversible process that requires conjugation and de-conjugation enzymes.

Reversibility of protein SUMOylation is achieved by de-SUMOylation enzymes called sentrin/SUMO-specific proteases (SENPs; SENP1-7). Certain SENPs, especially SENP1 and 2, contain both nuclear localization and export signal domains, and shuttling of SENPs from one compartment of the cell to another has an effect on altering SUMOylation levels in different sub-cellular regions. Here, we will review the role of SUMOylation in both extra-nuclear and nuclear compartments of cardiomyocytes and discuss the potential impact of de-SUMOylation enzyme’s shuttling on the control of the response and fate of cardiomyocyte upon cardiac insults.

1.1. SUMOyation (SUMO conjugation)

SUMOylation is a lysine-targeted post-translational modification (PTM). SUMO is a 10-kDa polypeptide [1,2] and is covalently conjugated to targeted molecules (substrates) at a lysine (K) residue within the specific consensus motif Ψ-K-x-D/E, in which Ψ is a hydrophobic residue, x is any amino acid, and D/E represent negatively charged aspartic acid or glutamic acid residue. There are four SUMO isoforms (SUMO 1–4) in human, and SUMO1 shares ~50% sequence identity with SUMO2, while SUMO2 and 3 differ by only three NH2-terminus residues (97% identity). Therefore, SUMO2 and 3 together are considered to form a subfamily, which is distinct from SUMO1. SUMO1–3 are ubiquitously expressed whereas SUMO4 is limited to kidney, lymph node, and spleen [2,3]. The functional role of SUMO4 remains controversial, because the C-terminus proline residue unique to SUMO4 inhibits the maturation of SUMO4 into the conjugatable form (or modifier) [4].

As shown in Fig. 1, SUMOylation utilizes an enzymatic machinery similar to that of ubiquitination, which has been reviewed extensively [3]. In brief, inactive precursor SUMO is maturated by SUMO proteases sentrin/SUMO-specific proteases (SENPs) to expose the di-glycine C-terminus motif and then activated by an E1 activating enzyme that is a heterodimer containing SAE1/SAE2 subunits. Next, activated SUMO binds to the E2 conjugating enzyme (Ubc9) via the transesterification reaction to facilitate SUMO protein attachment to lysine (K) residues on target proteins in conjunction with the E3 ligating enzyme. SUMO E3 ligase acts as an adaptor between Ubc9-SUMO and the substrate protein. The efficient and targeted SUMO modification of substrates can be accomplished by SUMO E3 ligase [2]. Several SUMO E3 ligases have been identified including the protein inhibitor of activated STAT (PIAS) family (PIAS1, PIAS2(x), PIAS3, PIAS4(y)), Polycomb-2 protein (Pc2), and RanBP2/Nup358 [5]. It is thought that SUMO E3 ligases determine the substrate specificity.

Fig. 1.

SUMOylation system and its regulatory mechanism. The SUMOylation system is a dynamic process of protein modification achieved by two enzyme systems; one which consisting of conjugates SUMO to substrates and one which de-conjugates. [176–179] SUMO proteins are covalently attached to certain residues of specific target substrates and change the function of these substrates. Prior to conjugation, the E1-activating enzyme, SAE1–SAE2 heterodimer, activates the mature form of SUMO [180]. SUMO is then transferred to Ubc9, an E2 conjugase, forming a thioester bond between Ubc9 and SUMO [181]. Lastly, SUMO E3 ligases, including a family of protein inhibitors such as activated STAT (PIAS1-4), transfer SUMO to the target substrate containing the free е-amino group of a lysine residue [182]. De-SUMOylation enzymes are also involved in the process of SUMOylation. Sentrin/SUMO-specific proteases (SENPs; SENP1–7) catalyze both de-conjugation of SUMOylated substrates and editing of the SUMO precursor into the matured form which now have a pair of glycine residues at the C terminus [183,184]. Adapted, reprinted, and modified from Heo et al. [22] with permission from ANTIOXIDANTS AND REDOX SIGNALING, April 2016, published by Mary Ann Liebert, Inc., New Rochelle, NY.

The majority of SUMO substrates are localized in the nucleus. Such nuclear substrates include proteins that regulate cell cycle progression, DNA repair and replication, and gene transcription. However, it is also becoming evident that many SUMOylated molecules are localized in extra-nuclear compartments, and they regulate protein functions including intracellular trafficking, apoptosis, protein stability, and enzyme activity [1,3,6]. SUMO-modification of substrates is known to play a major role in the following molecular events and processes: 1) SUMOylation provides a platform to recruit molecules that non-covalently bind to SUMO via SUMO-interacting motif (SIM), 2) SUMOylation can promote or block the association of molecules that interact with SUMOylated substrates, 3) SUMOylation can regulate substrate stability by competing for the lysine site with ubiquitination or degradation by recruiting the SUMO-targeted ubiquitin ligase (STUbL) family of proteins to the SUMOylated substrates, and 4) SUMOylation induces conformational changes in proteins so that their interaction with other molecules can be regulated [7,8] (Fig. 1).

1.2. De-SUMOylation (SUMO deconjugation)

SUMOylation is a dynamic and reversible modification of substrates and allows transient changes in signal transduction (Fig. 1). It is important to note that not only SUMOylation but also de-SUMOylation can play a key role in regulating cellular signaling pathways. SENPs are a family of enzymes with isopeptidase activity. In addition to cleaving the C-terminus of pro-SUMO to expose the d-glycine motif required for conjugation of mature SUMO as explained above, SENPs can catalyze de-conjugation of SUMO from target proteins. Six isoforms of SENP (1–3, 5–7) exist in humans and they exhibit preference toward distinct SUMO forms. SENP1 and 2 are involved in both maturation and de-conjugation of both SUMO1 and 2, although SENP2 has a lower processing (maturation) activity but a higher de-conjugating activity [9–12]. SENP3 and 5 favor SUMO2 and 3 over SUMO1 for both maturation and de-conjugation [13–15]. SENP6 and 7 cannot effectively remove monomeric SUMOs but are able to cleave polymeric chains of SUMO2/3 and edit lysine-linked polySUMO2/3 chains [11,16]. In addition to the SENP family, de-SUMOylating isopeptidase 1 and 2 (DeSI1 and 2) and ubiquitin-specific protease-like 1 (USPL1) can also function as SUMO-protease. So far, BTB-ZF is the only molecule identified as a target of DeSI1 [17], and USPL1 acts preferentially on SUMO2/3 and edits polySUMO2/3 chains with a limited capacity for the general pro-SUMO processing [18]. Since most SENPs have both pro-SUMO processing (maturation) and de-conjugation functions and since maturation can promote SUMO conjugation, the level of SUMOylation of each substrate may depend on the combined effects of SUMO ligases and SENPs. Further investigations will be necessary.

Different SENPs appear to have certain overlapping substrates, but they can also have a high degree of substrate specificity. SUMOylation of hypoxia-inducible factor 1α (HIF1α) decreases its stability leading to its degradation in response to hypoxia. SENP1 enhances HIF1α stability through de-SUMOylation. Indeed, HIF1α in SENP1^−/− mouse embryos shows increased levels of SUMOylation and reduced levels of stability. In contrast, HIF1α stability is not affected in SENP2^−/− embryos, indicating the specificity of SENP1 and 2 in regulating HIF1α SUMOylation [19,20].

It is interesting that both SENP1 and SENP2 are self-regulated through transcriptional feedback loops. For example, under conditions of genotoxic stress, there is an early increase in SENP2 mRNA followed by de-SUMOlyation of NEMO (NF-κB Essential Modulator)/IKK (inhibitor of κB kinase) γ by SENP2. NEMO then decreases SENP2 transcription thereby creating a negative feedback loop to prevent the survival of damaged cells [21].

1.3. SENPs sub-cellular localization

SENPs are mainly localized in the nucleus, but certain SENPs, especially SENP1 and 2, contain both nuclear localization and export signal (NLS and NES) domains and are able to shuttle between the nucleus and the cytoplasm. Such changes in the localization of SENPs within a cell have profound effects on SUMOylation levels in different cellular compartments (Fig. 2). Recently, we have reported the crucial role of SENP2 phosphorylation in regulating SENP2 nuclear export [22]. The NLS is located near the N-terminus of SENP2 and the leucine-rich, CRM1-dependent NES sequence in the central region. Because we found the involvement of both p90RSK kinase activity and SENP2 in disturbed flow (d-flow)–initiated p53 SUMOylation and subsequent apoptosis, we investigated whether p90RSK could directly phosphorylate SENP2. We subjected phosphorylated SENP2 to proteolytic digestion and liquid chromatography–tandem mass spectrometry and identified T368 as a SENP2 phosphorylation site by p90RSK. SENP2-T368A mutant inhibited p53 SUMOylation by d-flow, suggesting that SENP2-T368 phosphorylation is a key regulator of p53 SUMOylation and plays a critical role in EC apoptosis. Furthermore, we found that d-flow increased SENP2 nuclear export, which was significantly inhibited by the T368A mutation. In addition, although it was very low in the steady laminar flow area, a robust increase in anti-phospho-SENP2-T368 staining was detected in the cytoplasm of ECs in the d-flow areas of the mouse aorta. Taken together, these data suggest the crucial role of SENP2 T368 phosphorylation in regulating SENP2 nuclear export, which subsequently upregulates nuclear p53 SUMOylation (Fig. 2). The extra-nuclear localization of SENP1 [23], DeSI1, DeSI2 [17], and USPL1 [18] was also reported, but the regulatory mechanisms and functional consequences of different sub-cellular localization of these SUMO proteases are unclear. The role of SUMOylation in the nucleus and the cytoplasm can be very different. In the following two sections, we will review the role of SUMOylation in the extra-nuclear and the nuclear compartments and discuss how such compartmentalization can be achieved and what consequences it may have on the regulation of cellular functions.

Fig. 2.

SENP2 shuttling between the nuclear and extra-nuclear compartments: SENP2 contains a bipartite nuclear localization signal (NLS) sequence at the N terminus domain and a leucine-rich, CRM1-dependent nuclear export signal (NES) sequence in the central region. These NLS and NES sequences are involved in SENP2 shuttling between the nucleus and the cytoplasm, which regulates levels of SUMOylation of proteins in these compartments. Activated p90RSK associates with SENP2, and phosphorylates T368 site, leading to SENP2 nuclear export. This nuclear export diminishes the nuclear SENP2 de-SUMOylation function in the nucleus, and consequently up-regulates SUMOylation of p53 and ERK5in the nucleus. p53 SUMOylation increases p53 nuclear export and binds with Bcl-2 in the cytoplasm, which inhibits Bcl-2 anti-apoptotic effects, and induces apoptosis. ERK5 SUMOylation inhibits transcriptional activity of PPARs, which can be detrimental to regulating the cardiac function and remodeling after cardiac insult.

1.4. SUMOylation events in extra-nuclear compartments

In this section, we will discuss the functional consequences of various SUMOylated substrates and events localized in (a) the plasma membrane; K⁺ channel, (b) the sarcoplasmic reticulum (SR) membrane; SR Ca²⁺ ATPase 2a (SERCA2a), (c) the mitochondrial membrane; DRP1, and (d) the cytosol; NF-κB essential modulator (NEMO)/inhibitor of κB kinase (IKK)γ, protein kinase C α (PKCα), adenosine monophosphate-activated protein kinase (AMPK), and ubiquitin-proteasome system (UPS) (Fig. 3).

Fig. 3.

Diagram of signaling in a cardiomyocyte showing how SUMO modification differentially controls extra-nuclear and nuclear events, which can be coordinately regulated by the shuttling of SENPs between these two compartments. In the cytoplasm, up-regulation of SUMO modification on SERCA2a, NEMO, LKB1, and AMPKs up-regulates their functions, while SUMOylation of PKCa and DRP1 inhibits PKCa kinase activation and DRP1 function for inducing mitochondrial fragmentation. In the nucleus SUMOylation of proteins in the tissue factor complex including XBP-1s, ERK5, and PPARs inhibits their transcriptional activity. ERK5 SUMOylation inhibits ERK5-mediated ICER reduction by inhibiting CHIP E3 ligase activity, which can also inhibit TFs activity. SUMOylation of HDACs can augment their function, leading to accelerated inhibition of transcription. These data suggest that SUMOylation events in the extra-nuclear compartments in cardiomyocytes are cardio-protective, while the nuclear SUMOylation events are detrimental to these cells. We propose the shuttling of SENPs between the two compartments plays an important role in the response to various forms of cardiac insults and decides the fate of the cardiomyocytes.

1.4.1. Plasma membrane; potassium channels

1.4.1.1. Kv1.5 (potassium voltage-gated channel subfamily A member 5, KCNA5)

Kv1.5 is responsible for the I_Kur repolarizing current in atrial myocytes and also regulates vascular tone in peripheral vascular beds. It has been reported that de-SUMOylaion of Kv1.5 mediated by SENP2 leads to a substantial hyperpolarizing shift in the voltage dependence of steady-state inactivation [24]. Since significant V₅₀ shift of Kv1.5 in the depolarizing direction could not be detected by overexpressing SUMO3 and Ubc9, neither Ubc9 nor SUMO3 appears to regulate Kv1.5 function. However, the inhibition of Kv1.5 SUMOylation induced by the cytoplasm-targeted and constitutively active SENP2 with de-SUMOylation activity caused a significant hyperpolarization shift in the voltage dependence of inactivation without altering the total current density or voltage dependence of Kv1.5 activation, suggesting that changes in localization of SENP2 regulate Kv1.5 function. Kv1.5 is widely expressed in the cardiovascular system [25], and its role in the familial form of atrial fibrillation [26] has been reported [27]. Further investigation is necessary to determine the exact molecular events that lead to Kv1.5 SUMOylation and the role of SENP2 in this process.

1.4.1.2. Potassium voltage-gated channel subfamily Q member (KCNQ)

Five KCNQ genes (KCNQ1 to KCNQ5) codify a family of 5 different voltage-gated potassium ion channels (KV7.1 to KV7.5), which are mainly expressed in the nervous and cardiac systems [28]. Yeh’s group has elucidated SENP2’s role in Kv7.2 SUMOylation using mice homozygous for the floxed SENP2 allele with a neomycin insert (SENP2^fxN/fxN) which induced a significant reduction in SENP2 transcription and protein expression. These SENP2^fxN/fxN mice appeared healthy at birth, but developed convulsive seizures followed by sudden death at 6–8 weeks of age. Reduced SENP2 levels in these mice created a hyper-SUMOylation environment and led to accumulation of SUMO-1 and SUMO-2/3 proteins in the brain and heart [29]. It is worth noting that the expression of SENP2 is abundant in the hippocampal region, which is a highly relevant area of the brain for seizure. Yeh’s group also found that SUMOylation of Kv7.2 was significantly enhanced in hippocampal neurons. Hyper-SUMOylation of this potassium channel protein diminished the M-current, which is conducted by Kv7, leading to a more positive resting membrane potential and increased excitability of hippocampal neurons. These data suggest the pathophysiological role of SENP2 in epilepsy via regulating the plasma membrane Kv7.2 function. The possible role of Kv7.2 SUMOylation in regulating cardiac arrhythmia is under investigation.

1.4.1.3. Potassium channel subfamily K member 1 (KCNK1)

KCNK1 is expressed in the heart and the central nervous system and regulates the sinus rhythm [30] and background leak currents stabilizing neuronal excitability [31]. KCNK1 can be SUMOylated at K274, and the point mutation of this lysine to glutamic acid (KCNK1 K274E) has been shown to increase KCNK1 current [32], suggesting that SUMOylation is inhibitory for this channel. Since overexpression of SENP1 inhibited KCNK1 SUMOylation and also increased KCNK1 current [33], SENP1 can be crucial for the regulation of KCNK1 SUMOylation and subsequent generation of K current. At present, however, the regulatory mechanism of KCNK1 SUMOylation in response to various stimuli remains unclear.

In summary, SENP1 and 2 are de-SUMOylation enzymes and play a major role in regulating the SUMOylation state of certain potassium channels and consequently their channel activities. However, it is not yet fully understood how the action of these de-SUMOylating enzymes is controlled in the cell. Obvious mechanisms would be to regulate expression levels of these enzymes, to regulate their enzymatic activity, or to control both of these aspects. Another mechanism, which is highlighted in this review, is to regulate intracellular localization of these SENPs without changing their levels of expression and enzymatic activity. SENPs are generally considered to function in the nucleus, but once they are exported from the nucleus, they can function in the extra-nuclear compartment including the plasma membrane. Such a mechanism may indeed play a role in the regulation of certain potassium channels by SENP1 and 2.

1.4.2. SR membrane; SERCA2a SUMOylation - SUMO1 vs. SUMO2/3

SERCA2a is a transporter found in the SR membrane of cardiomyocytes and transports Ca²⁺ into and out of the SR, and its dysfunction results in various types of heart failure. Hajjar’s group has reported the role of SERCA2a SUMOylation by SUMO1 in the process of heart failure [34] (Fig. 3). They found that SERCA2a was SUMOylated at two sites, K480 and K585, and that mutation of these lysine residues to arginine significantly decreased the ATP-binding affinity and ATPase activity of SERCA2a. In addition, they found that SUMOylation stabilized SERCA2a in the cell by inhibiting ubiquitination and subsequent proteosomal degradation. These data suggest that SERCA2a SUMOylation can improve cardiac function via up-regulating its activity and expression. To support these findings the authors also performed SUMO1 gene delivery experiments in both mice [34] and pigs[35] and used a small molecule activator of SERCA2a SUMOylation (N106) [34, 36] and found that these interventions which promoted SERCA2a SUMOylation improved cardiac function. Moreover, they showed that the improved cardiac function by elevated SUMO1 levels and that this improvement was diminished by the depletion of SERCA2a. These data appear to suggest a possible role of SERCA2a in the SUMO1-induced improvement of cardiac function. However, since it is well known that the depletion of SERCA2a inhibits the whole process of excitation-contraction coupling. Therefore, these experiments cannot exclude the involvement of other SUMOylated substrates, which can improve cardiac function independent on SERCA2a.

The role of SUMO1 in cardiac function was tested in cardiac muscle specific Sumo1-overexpressing transgenic mice, and the result showed that pressure overload-induced cardiac hypertrophy and dysfunction were inhibited by SUMO1 overexpression, suggesting the cardio-protective effect of SUMO1 [34]. In contrast, cardiac specific Sumo2 overexpressing mice showed dose-dependent cardiac hypertrophy and dysfunction [37]. In this study, the contribution of calpain 2 and calpastatin SUMOylation to cardiomyocyte apoptosis has been suggested. SUMO2-modified calpastatin was shown to be degraded more efficiently, which subsequently increased calpain 2 expression. SUMOylation of calpain 2 by SUMO2 increased its enzymatic activity. These results indicate that SUMO2 modification up-regulates the function of the calpastatin-calpain proteolytic system and consequently accelerates apoptosis. The SERCA2 SUMOylation in SUMO2 transgenic mice was not tested in this study. Since Kho et al. have only tested the level of co-immunoprecitation of SUMO2 with SERCA2a in human heart samples with no positive control [34], the contribution of SUMO2/3 in SERCA2a SUMOylation cannot be excluded. It is possible that SUMOylation of distinct sets of different substrates targeted by either SUMO1 or SUMO2/3 determines the beneficial and detrimental effects of SUMO1 and SUMO2, respectively, on cardiac function. However, as we will explain later, dynamin-related protein 1 (DRP1) SUMOylation by SUMO2/3 can also be protective of cardiac function. Therefore, we suggest that the existing contradictory data on the regulation of cardiac function by SUMOylation is not due so much to the difference between SUMO1 and SUMO2/3 but to the gene induction methods, which may force the expression of different SUMOs in different sub-cellular compartments. Since both SUMO1 and SUMO2 knock out mice are embryonic lethal [38,39], both are critical, at least, for embryonic development. Inducible cardiac specific SUMO1 and SUMO2 knock out mice may be necessary to clarify the role of SUMO isoforms in adult heart after cardiac insults.

1.4.3. Mitochondrial membrane; DRP1 SUMOylation - SUMO1 vs. SUMO2/3

Mitochondria are dynamic organelles whose overall morphology changes in response to cellular activity [40]. Such morphological changes include fission and fusion (i.e. elongation), which are under the tight regulation dictated by the physiological state of the cell [41,42]. Mitochondrial fusion is achieved by integrating outer and inner membranes of one mitochondrium with those of another mitochondrium and fission by pinching off of a mitochondrium in a manner similar to cytokinesis of animal cells. GTPase proteins from the dyamin family including mitofusin 1 (Mfn1), mitofusin 2 (Mfn2), and optic atrophy-1 (Opa1) are involved in the process of mitochondrial fusion [43,44], and ubiquitin ligases such as membrane-associated RING finger 5 (MARCH5) regulate the overall process [45]. Fission of mitochondria occurs in a sequential manner. First, constriction of mitochondrial tubules takes place. Next, the GTPase called DRP1 is mobilized from the cytosol to the outer membrane of mitochondria via several receptor proteins [46]. Upon reaching the outer membrane, DRP1 assembles into a scission complex by forming a spiral that surrounds the tubule. Then, in a GTP dependent manner, the DRP1 complex constricts the tubule to cause scission [46,47]. Lastly, the complex is disassembled.

It has been reported that SUMOylation of DRP1 regulates this ATPase and thus, also regulates mitochondrial fission (Fig. 3). Mitochondria-associated protein ligase (MAPL) is a 40 kDa protein located on the outer mitochondrial membrane[48]. Although MAPL can participate in the process of both ubiquitination and SUMOylation, under physiologic conditions MAPL preferentially functions as a SUMO E3 ligase for DRP1 SUMOylation (SUMO1) [49]. MAPL-mediated DRP1 SUMOylation (SUMO1) increases mitochondrial fission and hyper-fragmentation [50–52], leading to apoptosis in Cos7 and HeLa cells [48,49,53]. Cytochrome c functions as the terminal trigger for apoptotic cell death and is located in the intermembrane space [54]. Release of cytochrome c has been shown to depend on MAPL-mediated SUMOylation (SUMO1) of DRP1 [55]. The level of DRP1 SUMOylation can be reduced by de-SUMOylation enzymes including SENP5, SENP3, and SENP2 [56,57]. SENP5 is involved in mitochondrial fission via its interaction with DRP1 [52,58] and removes SUMO1, SUMO2, or SUMO3 [13,52]. DRP1 can be SUMOylated not only by SUMO1 but also by SUMO2 and 3. Interestingly, DRP1 modification by different SUMO isoforms has different functional consequences [56]. Depletion of SENP5 in COS7 and HeLa cells increases DRP1 SUMO1 modification by MAPL, leading to DRP1 association with mitochondria and increasing mitochondrial fragmentation and cellular apoptosis [50–52]. In contrast, cardiomyocyte-specific overexpression of SENP5 inhibits DRP1 SUMO2/3 modification and induces apoptosis via promoting the association of DRP1 with mitochondria [59]. These data suggest that DRP1 SUMO2/3 modification inhibits DRP1 recruitment to mitochondria whereas DRP1 SUMO1 modification promotes DRP1 binding to mitochondria and consequently up-regulates apoptosis [56]. Further studies are necessary to determine specific roles for the DRP1-SUMO1 and DRP1-SUMO2/3 conjugates in the process of mitochondrial fission.

Although SENP5 is localized primarily in the nucleus, a substantial amount of this enzyme is also found in the cytoplasm or the extra-nuclear compartment of the cell [51]. Zunino et al. have reported SENP5 translocation from the nucleus to the mitochondria during the G2/M transition [52]. Although the regulatory mechanism for this SENP5 translocation is unknown, this is another example that translocation of SUMO proteases between different intracellular compartments can occur, and such movements of de-SUMOyation enzyme may be able to alter SUMOylation states of various proteins located in different parts of the cell and regulate various cellular functions.

1.4.4. Cytosolic molecules; NEMO/IKKγ, protein kinase C α (PKCα), adenosine monophosphate-activated protein kinase (AMPK), and ubiquitin-proteasome system (UPS)

1.4.4.1. NEMO/IKKγ

NF-κB is a master regulator of cell survival and inflammation and plays an important role in various cardiac pathogenesis including heart failure [60], ischemic preconditioning [61], and apoptosis [62]. Under resting conditions, NF-κB is in an inactive state within the cytoplasm bound to IκB (Inhibitor of κB). When cells are stimulated by agonists such as TNF, IKK (IκB kinase) is activated and phosphorylates IκB, which then leads to degradation of IκB via ubiquitination. This not only frees NF-κB, but also unmasks its nuclear localization signal. NF-κB then translocates into the nucleus and transactivates pro-inflammatory genes [63]. SUMOylation has been reported to be involved at different levels of NF-κB regulation. IκBα can be modified by SUMO1 to protect it from ubiquitination and degradation, limiting NF-κB activation [64]. In contrast, modification of IκBα by SUMO2/3 confers the opposite effect and dissociates IκBα from NF-κB leading to NF-κB activation [65]. However, the exact pathophysiological role of IκBα SUMOylation remains unclear.

The IκB kinase (IKK) complex is an important regulatory kinase complex in regulating NF-κB signaling [66,67]. The complex consists of two kinases (IKKα and IKKβ) and a regulator subunit NEMO, also called IKKγ and is mainly expressed in the cytoplasm. NEMO/IKKγ does not possess catalytic activity but is required for IKK activation and its subsequent phosphorylation of IκB [68]. It has been reported that depletion of NEMO in cardiomyocytes promotes apoptosis and subsequent cardiac dysfunction via inhibiting the expression of anti-oxidant genes such as superoxide dismutase 2 and ferritin heavy chain. These data suggest that activation of NEMO/IKKγ-NF-κB signaling can be cardio-protective by inhibiting apoptosis [69]. Miyamoto’s group has reported that SUMOylation of NEMO during genotoxic stress leads to increased IKK activation and thus, NF-κB activation [70–76] (Fig. 3). Interestingly, a de-SUMOylation enzyme SENP2 is a downstream transcriptional target of NF-κB thereby creating a negative feedback mechanism for NF-κB activation through NEMO de-SUMOylation [21]. SENP6 attenuates Toll-like receptor-triggered inflammation in the endotoxin-induced murine sepsis model via de-SUMOylation of NEMO at K277 [77]. However these pathways are not well-characterized in the heart and await further investigation.

1.4.4.2. Protein kinase Cα

PKC contains multiple putative SUMOylation sites and modification by SUMO can affect its activity. Inactive PKCα is SUMOylated at the Lys465 site which can be de-SUMOylated by SENP1. Using rodent spinal cord neuronal cells, it was shown that PKCα activation by calcium was achieved only after de-SUMOylation by SENP1 [78]. This shows that SUMOylation can play an inhibitory role in PKCα kinase function. PKCα is known to negatively regulate cardiomyocyte contraction for which three specific extra-nuclear PKCα substrates play roles [79]. First, PKCα phosphorylates inhibitor 1 (I-1) at Ser67, which up-regulates protein phosphatase 1 activity, leading to greater phospholamban (PLN) de-phosphorylation and reduced activity of the SR Ca²⁺ ATPase (SERCA2) pump [80]. Second, PKCα activation increases G protein-coupled receptor kinase 2 (GRK2) phosphorylation and activity, and impairs β-agonist-stimulated ventricular function via abolishing cyclase activity [81]. Lastly, PKCα can phosphorylate cardiac troponin I (cTnI), cTnT, titin, and myosin binding protein C, the effect of which is to decrease the Ca²⁺ sensitivity and contractility of cardiomyocytes [79,82–85] (Fig. 3). These data suggest that inhibition of PKCα kinase activity by SUMOylation can be cardio-protective.

1.4.4.3. Adenosine monophosphate-activated protein kinase (AMPK) and ubiquitin-proteasome system (UPS)

AMPK is a stress-activated kinase, which can orchestrate the cellular response to a variety of stresses in the heart by regulating metabolism, protein synthesis, degradation, autophagy, and apoptosis [86]. AMPK is a complex of three subunits: a catalytic subunit (α) containing a serine-threonine kinase domain (KD) with a Thr172 phosphorylation site which is the target of liver kinase B1 (LKB1) and calcium-calmodulin-activated protein kinase kinase-β (CAMKKβ) and two regulatory subunits (β and γ) [86]. In unstressed cells, AMPK is mainly localized in the cytoplasm but can translocate to the nucleus after its activation [86]. Most studies have shown that endogenous AMPK activation is protective against cardiac insults including ischemia/reperfusion and pressure overload [86–90]. For example, depletion of AMPKα2 decreased glycolysis, lowered ATP levels, and impaired cardiac function after ischemia [91,92]. In addition, depletion of AMPKα2 resulted in greater cardiac hypertrophy and contractile dysfunction after pressure overload [93]. Rubio et al. have reported that AMPK SUMOylation (SUMO2) stimulates AMPK activation and inhibits its ubiquitin-dependent degradation [94] (Fig. 3). In addition, Yeh’s group has recently shown that LKB1 K178 SUMO1 modification promotes LKB1 association with AMPKα SIM, and accelerates AMPK activation [95] (Fig. 3). Taken together, these data suggest that the LKB1-AMPK SUMOylation can be cardio-protective.

It has been well established that accumulation of misfolded proteins contributes to the pathogenesis of heart failure. The ubiquitin-proteosome system (UPS) and selective autophagy are the two major mechanisms responsible for the removal of misfolded proteins, and both processes mainly occur in the cytoplasm. Robbins and his colleagues showed that depletion of Ubc9, the SUMO E2 conjugating enzyme, in cardiomyocytes caused accumulation of protein aggregates inside these cells and impaired cardiac function [96]. In addition, they also showed that Ubc9-mediated SUMOylation increased autophagy, which led to reduction of protein aggregate formation, fibrosis, and hypertrophy while at the same time, improving cardiac function and survival [97]. These studies were done using mice expressing a mutant α-B-crystallin in a cardiomyocyte specific manner (Fig. 3). It is unclear how Ubc9 regulates UPS and autophagy, but it is tempting to speculate the involvement of AMPK in this process, because AMPK SUMOylation increases AMPK activation which can upregulate UPS and autophagy.

In summary, most of the cardiac extra-nuclear SUMOylation events, except for a potassium channel and the calpastatin-calpain proteolytic system, are cardio-protective against cardiac insults (Table 1). It is important to state here that Wykoff et al. have reported that by utilizing a collection of epitope-tagged yeast strains, they found 82 proteins associated with SUMO [98]. In addition, by using immobilized metal affinity chromatography, Vertegaal et al. have identified 53 SUMO-conjugated proteins including 44 novel SUMO targets in HeLa cells [99]. These data suggest the possible existence of unstudied SUMOylated proteins in the heart. Of note, they also showed that SUMOylation was strongly related to transcription because nearly one-third of the identified target proteins are putative transcriptional regulators [99]. Therefore, to understand the whole picture of how SUMOylation regulates cardiac function, it is crucial to know the functional role of SUMOylation in regulating nuclear molecules including transcriptional factors. We will discuss the pathophysiological role of SUMOylation events in the nucleus in the next section.

Table 1.

Summary of transgenic models, phenotypes, and effects of SUMOylation (extra-nuclear events).

Molecule	Transgenic model	Model characteristics	Phenotypes and the effect of SUMOylation
KCNA5 (Kv1.5)	Kcna5 Replace mKv1.5 with 4-aminopyridine (4-AP)-insensitive channel rat Kv1.1 (SWAP mice) [162]	Non-lethal	4-AP-sensitive component of I (K, slow) was absent No α-dendrotoxin-sensitive current and no 4-AP-induced prolongation of QTc. No increase in arrhythmia during ambulatory telemetry monitoring [162]. De-SUMOylation of KCNA5 induced hyperpolarizing shift in the voltage dependence of steady-state inactivation [24]
KCNK1	Kcnk1−/− [163]	Non-lethal	Impaired regulation of Pi transport in proximal tubule and water transport in medullary collecting duct [163]. KCNK1 regulates sinus rhythm after bran ischemia [31]. KCNK1 SUMOylation inhibited KCNK1 current [32].
SERCA2a	Cardiac specific-Serca2a−/− [164]	Approximately 20% increase in embryonic and neonatal mortality. Major cardiac malformations were observed in embryonic lethal mice [164].	Adult mice showed a reduced spontaneous nocturnal activity and developed a mild compensatory concentric cardiac hypertrophy with impaired cardiac contractility and relaxation. Ca2+ uptake levels were reduced. Relaxation and Ca2+ removal by the SR were significantly reduced [164]. SERCA2a SUMOylation stabilized SERCA2a and improve cardiac function via upregulating its activity and expression [34].
DRP1	Cardiac specific-inducible Drp1−/−	Inducible knock out.	Mitochondrial enlargement, lethal dilated cardiomyopathy, and cardiomyocyte necrosis. DRP1 SUMO1 modification increase mitochondrial fission and hyper-fragmentation, leading to apoptosis in Cos7 and HeLa cell. DRP1 SUMO2 modification inhibited apoptosis via reducing the association of DRP1 with mitochondria.
NEMO	Nemo−/− Cardiac specific Nemo−/− [69]	Males homozygous for targeted null mutations exhibit embryonic lethality by embryonic day 13.5 from apoptotic liver damage [165].	Heterozygous females show dermatopathy (patchy skin lesions with granulocyte infiltration), which is similar to the human X-linked disorder incontinentia pigmenti [166]. Sever liver degeneration and lack of NF-κB activation [165]. Adult-onset dilated cardiomyopathy. Increasing oxidative stress, inflammation and apoptosis [69]. NENO SUMOylation increases IKK and NF-κB activation [76].
PKCα	PKCα−/− Cardiac specific PKCα−/− [80]	Non-lethal.	Enhance insulin signaling through PI3-K [167]. Increases cardiac contractility and protects against heart failure [80]. PKCα SUMOylation inhibits PKCα kinase activation [78].
AMPK	Ampkα1−/− [168,169] Ampkα2−/− [168,170] Ampkβ1β2−/− Skeletal and heart specific (under the control of muscle creatine kinase promoter) Ampkβ1β2−/− [171] Kinase dead Ampkα2 transgenic mice with muscle creatine kinase promoter (skeletal and cardiac expression) [87].	Non-lethal Non-lethal Homozygous embryonically lethal Non-lethal Non-lethal	Anemia, reticulocytosis, splenomegaly, increased erythrocyte turnover, and elevated plasma erythroprotein levels [169]. Hyperglycemic, hypoinsulemic, and insulin resistance [168,170]. Physically inactive with reductions in skeletal muscle mitochondrial content. Impaired contraction-stimulated glucose uptake [171]. Maintain glucose uptake and glycolysis during ischemia and reperfusion in vitro, and show no increase of fatty acid oxidation during reperfusion in vivo. Impaired recovery of left ventricle contractile function after ischemia/reperfusion via increasing apoptosis and lowering ATP content in vivo [87]. AMPKβ2 SUMOylation (SUMO2) increases AMPK activation and inhibits its ubiquitin-dependent degradation [94].
LKB1	Lkb1−/− [172]	Homozygous embryonically lethal	Neural tube defects, mesenchymal cell death, and vascular abnormalities. Tissue specific deregulation of vascular endothelial growth factor (VEGF) expression [172]. LKB1 SUMOylation (SUMO1) promotes LKB1 association with AMPKα and accelerates AMPK activation [95].
UBC9	Cardiac specific Ubc9 overexpression [97]	Non-lethal	Increased autophagy. Reduced aggregate formation, decreased fibrosis, reduced hypertrophy, and improved cardiac function and survival in a mutant α-B crystallin mice (CryAB(R120G)) [97]. UBC9 is a SUMO E2 conjugating enzyme.

1.5. SUMOylation in the nuclear compartment

In this section, we will discuss functional consequences of various SUMOylated substrates in the nucleus including (a) endoplasmic reticulum (ER) stress-mediated transcription factors, XBP-1s (spliced X-Box Binding Protein 1), (b) peroxisome proliferator-activated receptors (PPARs), (c) ERK5 (extracellular-signal regulated kinase 5)-CHIP (carboxyl terminus of HSP70-interacting protein)-ICER (inducible cAMP early repressor) complex, and (d) HDACs (histone deacetylases). Notable effects of SUMOylation in the nucleus are due to its trans-repression activity against a variety of transcriptional factors and SUMO modification of transcription factors and cofactors, which results in transcription repression [3,100] (Fig. 3). One of the possible inhibitory mechanisms is that covalent attachment of SUMO provides a new interaction interface that mediates recruitment of transcriptional corepressors [101]. For example, SUMOylation of transcriptional factor Elk-1 increases its affinity to HDAC2, which then induces histone de-acetylation and transcriptional repression of the c-fos promoter [102]. Another possible mechanism may involve REST corepressor 1 (RCOR1). RCOR1 plays a role as a corepressor by recruiting the RCOR1/KDM1 (histone lysine-specific demethylase)/HDAC1 and 2 complex to transcription factors and by increasing de-acetylation and de-methylation of histone tails to generate a repressive chromatin structure [103]. RCOR1 binds directly to SUMO2 with its non-consensus SUMO-interaction motif (SIM) [103]. This association is crucial for recruiting the RCOR1-KDM1-HDAC1/2 complex to various covalently SUMO-modified transcription factors, altering the acetylation and methylation status of histone and leading to transcriptional repression. As we will explain in this section, SUMOylation also has significant effects on the function of HDACs, which adds more complexity to the mechanism of SUMOylation-mediated trans-repression and its functional consequence.

1.5.1. Endoplasmic reticulum (ER) stress-mediated transcription factors; XBP-1s

ER stress is one of the endogenous sources of cellular stress, which occurs following the accumulation of misfolded proteins in the ER. To counterbalance the ER stress, cells utilize unfolded protein response (UPR), a three-pronged signaling pathway to restore the proper ER function. This entails inhibiting general protein synthesis but promoting ER processing in order to reduce misfolded protein aggregation in ER. Depending on the degree of stress, UPR can be either pro-survival or pro-apoptotic [8]. The three-pronged UPR pathway is mediated by three distinct ER-localized trans-membrane proteins: inositol-requiring kinase 1 (IRE1), PRKR-like ER kinase (PERK), and activating transcription factor 6 (ATF6) [104]. It has been reported that IRE1 and PERK pathways are regulated by SUMOylation [105]. IRE1 possesses both kinase and endonuclease activities. Unfolded proteins in ER activate IRE1 and increase its auto-phosphorylation, which up-regulates IRE1 endoribonuclease activity and excises a 26 base pair fragment from unspliced XBP1 mRNA. Following religation by a putative tRNA ligase, active forms of spliced XBP1 mRNA are generated and XBP1 protein is synthesized. It has been reported that SUMOylation of XBP1s (by both SUMO1 and SUMO2/3) by PIAS2, a SUMO E3 ligase, inhibits transcriptional activity of XBP1s and subsequent UPR target gene expression [106] (Fig. 3). In addition, SENP1 has been suggested to play a role in maintaining ER-stress-mediated XBP1 activity by de-SUMOylating it [107]. XBP1 knock out mice are embryonic lethal [108]. In addition, it has been reported that overexpression of dominant negative XBP1 increases apoptosis in isolated cardiomyocytes in response to ischemia/reperfusion [109], suggesting that activation of ER stress may be cardio-protective. It appears, therefore, that inhibition of transcriptional activity of XBP1s by SUMOylation in the nucleus is detrimental after cardiac insult. Future study is necessary to verify this prediction.

1.5.2. Peroxisome proliferator-activated receptor isoforms (PPARs)

Cardiac energy substrate utilization is critically controlled by the PPAR family of transcription factors including PPARα, PPARβ/δ, and PPARγ, and all three isoforms are exclusively localized in the nucleus. Systemic PPARα null mice display decreased cardiac fatty acid oxidation rates while glucose oxidation rates and reliance on glucose metabolism for ATP production are increased [110,111], the condition of which induces cardiac dysfunction [112]. Cardiomyocyte specific deletion of PPAR-δ in mice showed reduced expression of key fatty acid oxidation genes and also reduced basal myocardial fatty acid oxidation, leading to progressive myocardial lipid accumulation, cardiac hypertrophy, and congestive heart failure with reduced survival [113]. Cardiomyocyte specific PPARγ null mice exhibited progressive cardiac hypertrophy with mitochondrial oxidative damage, and most mice died from dilated cardiomyopathy [114]. All three PPAR isoforms can be SUMOylated, and SUMO modification of the ligand binding domain of PPARα and γ inhibits their transcriptional activity [115,116] (Fig. 3). Although the specific SUMOylation sites of PPARβ/δ were not determined, SUMOylation of PPARβ/δ promoted their degradation and inhibited their transcriptional activity [117]. Although the precise role of SUMOylated PPARs in the heart remains unclear, these data suggest that SUMOylation of PPAR isoforms may be detrimental to maintaining cardiac function after various cardiac insults.

1.5.3. ERK5-PPARs

Although the localization of overexpressed ERK5 was cytosolic in resting cells and shifted to the nucleus after activation [118], Raviv et al. reported that endogenous inactive ERK5 was also exclusively expressed in the nucleus [119], which we have independently confirmed (data not shown). ERK5 plays a critical role in regulating the apoptotic pathway of the cardiomyocyte. ERK5 is a unique kinase that also possesses transcriptional activity and functions as a transcriptional co-activator for PPARs and myocyte enhancer factor-2 (MEF2), down-regulating the expression of proteins involved in apoptosis in endothelial cells and cardiomtocytes (Fig. 4). Mice with cardiomyocyte-specific ERK5 knockouts exhibit accelerated dysfunction and apoptosis of cardiac muscle cells after thoracic aorta constriction [120]. We have reported that SUMOylation at the N-terminus region (K6 and K22) of ERK5 significantly inhibits the C-terminus ERK5 transcriptional activity, which is independent on its kinase activation [121]. We found that deletion of PIAS1 significantly inhibited cardiomyocyte apoptosis induced by reactive oxygen species (ROS) and that overexpression of ERK5 K6R/K22R SUMOylation mutants also reduced ROS-induced apoptosis. These data suggest that ERK5 SUMOylation plays a major role in regulating cardiomyocyte apoptosis [122] (Fig. 4). Diabetic mediators (H₂O₂ and AGEs) and ischemia under the diabetic condition also increase ERK5 SUMOylation and promote apoptosis in cardiomyocytes in both in vitro and in vivo [121–123]. The contribution of PPARs to the ERK5 transactivation-mediated anti-inflammatory effect has been established [121,124], but their role in cardiomyocyte apoptosis mediated by SUMOylated ERK5 needs further investigation.

Fig. 4.

The primary structure of ERK5 and its regulation. ERK5, also called a big MAP kinase, is twice the size of other MAPKs and hence the largest kinase within its group. It possesses a catalytic N-terminus domain including the MAPK-conserved threonine/glutamic acid/tyrosine (TEY) motif in the activation loop with 50% homology with ERK1/2, and a unique C-terminus tail with transactivation domains. The activation of ERK5 occurs via interaction with and dual phosphorylation in its TEY motif by MEK. On the other hand, inflammatory stimuli or athero-prone flow (d-flow) leads to ERK5 deactivation via phosphorylation of Ser486 or Ser496, respectively. SUMO modification of the N-terminus K6 and K22 sites inhibit its own transactivation. B. Insulin growth factor-1 (IGF-1) or pre-conditioning activates ERK5 kinase activity in cardiomyocytes, leading to phosphorylation of the TEY motif and de-SUMOylation of the two sites which then fully activates ERK5 transcriptional activity. In contrast, ischemia under the diabetic condition (DM + MI), reactive oxygen species (ROS), and advanced glycation endo-products (AGE) increase ERK5 SUMOylation and ERK5 Ser496 phosphorylation, inhibiting ERK5 transcriptional activity and promoting ERK5 degradation via ERK5 Ser486 phosphorylation. CHIP; carboxyl terminus of HSP70-interacting protein, p90RSK: p90 ribosomal S6 kinase; PKCζ, protein kinase C-ζ; and PPARs, peroxisome proliferator-activated receptors. Adapted, reprinted, and modified from Heo et al. [22] with permission from Antioxidants and Redox Signaling, April 2016, published by Mary Ann Liebert, Inc., New Rochelle, NY.

1.5.4. ERK5-CHIP-ICER

Previously, our group has reported reduced expression of cAMP hydrolyzing enzymes including phosphodiesterase 3A (PDE3A) and increased expression of inducible cAMP early repressor (ICER) in failing hearts [125,126]. ICER is a pro-apoptotic transcriptional repressor, which inhibits transactivation of cAMP response element binding protein (CREB), and thus downregulates Bcl2. Furthermore, ICER represses PDE3A gene transcription, leading to increased cAMP availability and upregulation of PKA signaling, forming an autoregulatory positive feedback loop. Angiotensin II and isoproterenol (β-adrenergic receptor agonist) activate this mechanism by downregulating PDE3A and upregulating ICER, providing a mechanism for the observation that activation of neurohormonal systems affects myocyte apoptosis [126,127]. Interestingly, ERK5 activation induced by insulin growth factor 1 decreased ICER protein stability through ubiquitin-mediated degradation [128], and we found an important contribution of CHIP (carboxyl terminus of HSP70-interacting protein), an E3 ubiquitin (Ub) ligase, to this process [129] (Figs. 3, 4). It has been reported that CHIP has an important cardioprotective function as it reduces myocardial injury from ischemia/reperfusion after MI by inhibiting cardiac muscle cell apoptosis [130]. Indeed, depletion of CHIP increases infarct sizes and decreases survival [131]. Our group has reported that ICER is a CHIP in mice substrate and that in order for CHIP to ligate Ub to ICER, CHIP must form a complex with de-SUMOylated, hence activated, ERK5 as disruption of ERK5-CHIP binding by a small peptide fragment completely inhibits CHIP Ub ligase activity and consequently up-regulates ICER expression [132] (Fig. 3, 4). The precise mechanism of the de-SUMOylation dependent ERK5-CHIP association needs further investigation, but our data may suggest that ERK5 SUMOylation can be detrimental to the cell via inhibiting the association of ERK5 and CHIP.

1.5.5. Histone deacetylase (HDAC) 1 and 2

Histone acetylation is controlled by two types of enzymes, histone acetyltransferases (HATs) and HDACs [133–136]. HATs catalyze the addition of an acetyl group to specific lysine residues of histone, whereas HDACs do the opposite, catalyzing the removal of acetyl groups. In most cases, histone acetylation is associated with transcription of genes [133,137,138]. It has been reported that inhibitors of Class I HDACs (especially 1 and 2) can inhibit cardiac hypertrophy and preserve cardiac function [139]. For example, depletion of HDAC2 inhibits cardiac hypertrophy after pressure overload [140], while cardiomyocyte specific overexpression of HDAC2 promotes cardiac hypertrophy by reducing the expression of INPP5F (Inositol Polyphosphate-5-Phosphatase F), the gene encoding phosphatidylinositide phosphatase SAC2, which is a negative regulator of the Akt/GSK-3β pathway [140]. In addition, an inhibitor for HDAC1 and 2, 3-(4-substituted phenyl)-N-hydroxy-2-propenamide (SK-7041), [141] completely inhibits the hypertrophic response of the heart after pressure overload [142]. Not only against cardiac hypertrophy, it has also been reported that trichostatin A, pan-HDAC inhibitor, reduces infarct size and improves cardiac function after ischemia/reperfusion [143]. These data suggest that HDAC1 and 2 may have an exacerbating effect on cardiac hypertrophy and dysfunction after cardiac insults.

It has been reported that both HDAC1 and 2 can be SUMOylated [144,145]. Especially, HDAC1 SUMOylation with SUMO1, but not SUMO2, at K444 and K476 promotes HDAC1 stability and up-regulates its expression and activity [145]. Given that HDAC1 has a negative pathological effect, these data suggest that HDAC1 SUMOylation can be detrimental (Fig. 3). HDAC2 SUMOylation can increase p53 de-acetylation and promote NF-κB activation [144,146], which is anti-apoptotic. Of note, the involvement of p90RSK in SUMOylated HDAC2-mediated NF-κB activation has been suggested [146]. p90RSK increases NF-κB signaling via modulating cytosolic events such as IκB or NR-κB p65 phosphorylation [147]. It is possible that HDAC2 SUMOylation discussed here may occur in the extra-nuclear compartment (i.e. in the cytoplasm). Further investigation is necessary to clarify these issues.

As we have described in this section, SUMOylation can inhibit most of the transcriptional factor activity, and subsequently be detrimental after cardiac insult (Table 2). Therefore, although SERCA2 SUMOylation and overexpression of Ubc9 are cardio-protective, these events may not reflect the nature of SUMOylation in the whole heart. It is important to emphasize here that the SUMOylation process can be regulated by the localization changes of de-SUMOylation enzymes including SENP2. As we have explained above, since both SENP1 and 2 contain multiple NLS and NES domains, changes of sub-cellular localization of de-SUMOytion enzymes can be one of the important ways to regulate the “function” of these enzymes to coordinately control the cardiomyocytes responses to various cardiac insults in nuclear and extra-nuclear events.

Table 2.

Summary of transgenic models, phenotypes, and effects of SUMOylation (nuclear events)

Molecule	Transgenic model	Model characteristics	Phenotypes and the effect of SUMOylation
XBP-1	Xbp-1−/− [108,173]	Homozygous embryonically lethal	Cellular necrosis of cardiac myocytes [108]. Hypoplasmic fetal livers, resulting in death from anemia via reducing hematopoiesis [173]. Spliced XBP-1 SUMOylation inhibits transcriptional activity of XBP-1s and subsequent UPR target gene expression [109].
PPARs	Pparα−/− [110–112] Cardiac specific Pparγ−/− [114] Cardiac specific Pparδ−/− [113]	Non-lethal Non-lethal Non-lethal	Decreased cardiac fatty acid oxidation rates, while glucose oxidation rates and reliance on glucose metabolism for ATP production are increased [110–112]. Progressive cardiac hypertrophy with mitochondrial oxidative damage, and most mice died from dilated cardiomyopathy [114]. Reduced expression of key fatty acid oxidation genes and also reduced basal myocardial fatty acid oxidation. Progressive myocardial accumulation, cardiac hypertrophy, and congestive heart failure [113]. PPARα and γ SUMOylation inhibits PPARα and γ transcriptional activity [115,116]. PPARδ SUMOylation promotes its degradation and inhibits PPARδ transcriptional activity [117].
ERK5	Erk5−/− Cardiac-specific Erk5−/−	Embryonically lethal Cardiac-specific ERK5 KO flox mice under α-MHC promoter No cardiac abnormalities in morphology or function up to 6 months of age	Loss of vascular integrity with EC misalignment and leakage due to defects in angiogenesis [174] Attenuates cardiac hypertrophy remodeling with accelerated cardiac apoptosis and dysfunction after TAC [120] Reduces CHIP Ub ligase activity and increased ICER post-MI[132] ERK5 SUMOylation inhibits ERK5 transcriptional activity [121].
CHIP	Chip−/−	Non-lethal	Increases infarct size via up-regulating apoptosis after ischemia and reperfusion [131].
HDAC1 and 2	Hdac1−/− [175] Hdac2−/− [175]	Homozygous embryonically lethal Survival until the perinatal period	Obliteration of the lumen of the right ventricle, excessive hyperplasia, apoptosis of cardiomyocytes, and bradycardia [175]. HDAC1 SUMOylation promotes HDAC1 stability and upregulates its expression and activity [145]. HDAC2 SUMOylation increases p53 de-acetylation and promotes NF-κB activation [146].

2. SUMOylation vs. ubiquitination

Although SUMOylation and ubiquitination uses a similar set of E1-E3 enzyme cascade, they use different set of E1–E3 enzymes. In addition, SUMO only shares ~18% homology with ubiquitin. SUMO is approximately 11 kDa in size, compared to the 8 kDa ubiquitin (Ub)molecule [148]. Previously, it has been suggested that SUMOylation acts as an antagonist of Ub, but it also promotes degradation by recruiting the SUMO-targeted Ub ligase as discussed above. It is important to state here that Ub can regulate protein functions by both degradation-dependent and independent mechanisms. When proteins bind to the lysine-48 (K48)-linked polyubiquitin chain, the protein is targeted to the proteasome for degradation. In addition to K48, the K63-linked poly-ubiquitin chain (K63-Ub chain) regulates proteins through a degradation-independent mechanism. For example, the IKK complex is activated when linked to the K63-Ub chain [149–152] (Fig. 5). We would like to briefly discuss here the functional role of Ub chains in regulating NF-kB activation, and its relationship to SUMOylation.

Fig. 5.

Ubiquitin-mediated TAK1 and IKK complex activation. Hypoxia/reperfusion activates the ubiquitin E3 ligase TRAF6, and TNF/TNFR activates TRAF2 and 5. TRAF6 increases K63-linked polyubiquitin chains, which associate with the TAB2 subunit of TAK1 kinase and activate TAK1 kinase. The K63-Ub chains also bind NEMO to recruit the IKK complex, thereby accelerating the phosphorylation of IKK Adapted, reprinted, and modified from Abe et al. [185] with permission from Circ Res, Jun 7, 2013, published by Wolters Kluwer Health, Inc.

Under Toll-like receptor (TLR) activation, Interleukin 1 receptor associated kinase (IRAK) 1 is phosphorylated by IRAK4 and then associates with and activates the ubiquitin E3 ligase TNF receptor associated factor 6 (TRAF6). TRAF6 complexes with the ubiquitin E2 complex composed of Ub conjugating enzyme (Ubc) 13 and Ub-conjugating enzyme variant (Uev) 1A [152,153] catalyzes the formation of K63-linked poly-ubiquitin chains on itself and other protein (Fig. 5). TRAF6 forms K63 Ub-chains, which promote the binding of the TAB2 (TGF-β-activated kinase 1 and MAP3K7-binding protein 2) subunit of the TAK1 (TGF-β-activated kinase 1) kinase complex and NEMO, leading to TAK1 activation [152,154]. Next, the K63-linked poly-Ub chain binds NEMO to recruit the IKK complex, activate IKK, phosphorylate IκB, and cause its degradation, and consequently activate NF-κB [155]. In addition to the K63-Ub chain formation, SUMO-1 modification of NEMO is required for NF-κB activation in response to genotoxic stress inducers [76]. First, DNA damage accelerates PIAS4 interaction with NEMO and preferentially stimulates site-selective modification of NEMO by SUMO1. DNA damage also activates ATM kinase and phosphorylates NEMO. Both SUMOylation and phosphorylation of NEMO promote its translocation to the cytoplasm, which is subsequently incorporated into the IKK complex. At the same time, K63-linked ubiquitination of BIRC2 (Baculoviral IAP repeat-containing protein 2) promotes NEMO mono-ubiquitination at K285, which is crucial for IKK complex activation [156]. These data suggest the importance of a cross-talk between ubiquitination and SUMOylation in NF-κB signaling [157].

The possible role of the SUMO chain has also been suggested [158]. SUMO can interact with substrates by covalently linking to specific lysine residues of substrate proteins or non-covalently associating with substrates via SUMO-interacting motifs (SIMs) [159]. These interactions are analogous to the ubiquitin system, in which a ubiquitin chain forms on the covalently attached ubiquitin and recruits ubiquitin-binding molecules, and such a molecular complex is able to efficiently activate signaling [158]. The molecules recruited to the ubiquitin chain are non-covalently bound to ubiquitins. Similarly, SUMO can form SUMO chains, to which molecules with SIMs are able to attach non-covalently, forming a SUMO chain–mediated molecular complex [159–161]. A growing number of proteins have been identified for which covalent SUMO association with substrate is regulated by non-covalent interaction of SUMO via SIM [158]. These data suggest that multiple SUMO-interacting molecules can be coordinately regulated by SUMO chain formation, but the exact regulatory mechanism and role of SUMO chain formation remain less clear than those of Ub chain formation. Further investigation will be necessary to determine the relationship between Ub chain and SUMO chain formation.

3. Conclusion

In this review, we have highlighted the role of SUMOylation in different sub-cellular locations particularly in heart. Available data show that SUMOylation events in the extra-nuclear compartment are overwhelmingly cardio-protective, while many SUMOylation events in the nucleus are detrimental to cardiac function, mainly because of the transrepressive nature of SUMOylation on transcription factors. As we also discussed, SUMOylation is an important and dynamic posttranslational protein modification occurring at different sub-cellular compartments, which is tightly regulated by the localization of SUMOylation and de-SUMOylation enzymes. In particular, SUMO proteases including SENP1 and 2 containing both NLS and NES and can shuttle between the cytoplasm and the nucleus. It is likely that shuttling of de-SUMOylation enzymes between the two compartments regulates SUMOylation events after cardiac insults (Fig. 3). Although conventional cardiac knockout or overexpression and gene transfer methods are powerful tools to determine the role of each SUMO-related molecule in cardiac dysfunction and remodeling, these methods are less useful for determining a dynamic regulation of SUMOylation mediated by shuttling of de-SUMOylation enzymes between the nuclear and extra-nuclear compartments. To determine the exact role of SUMOylation in the heart, it is necessary to clarify the precise molecular mechanism of de-SUMOylation enzyme shuttling. Experimental methods that can only inhibit the shuttling of de-SUMOylation enzyme but not SUMO ligase must be developed. Using such methods, we will be able to obtain more precise data on how SUMOylation within different compartments of cells is regulated by differential localization of SUMOylation and de-SUMOylation enzymes, which will aid us to determine the response and fate of cardiomyocytes by SUMOylation after cardiac insults.

Abbreviations

AMPK

adenosine monophosphate-activated protein kinase

CAMKK

calcium-calmodulin-activated protein kinase kinase

CHIP

carboxyl terminus of HSP70-interacting protein

CVD

cardiovascular disease

cTnI

cardiac troponin I

d-flow

disturbed flow

DeSI

de-SUMOylating isopeptidase

DRP1

dynamin-related protein 1

ER

endoplasmic reticulum

ERK5

extracellular-signal regulated kinase 5

GRK2

G protein-coupled receptor kinase

HAT

histone acetyltransferases

HDAC

histone deacetylase

HIF

hypoxia-inducible factor

ICER

inducible cAMP early repressor

IKK

inhibitor of κB kinase

KCNK1

potassium channel subfamily K member 1

KDM1

histone lysine-specific demethylase

LKB1

liver kinase B1

MAPL

mitochondria-associated protein ligase

NEMO

NF-κB Essential MOdulator

NF-κB

nuclear factor kappa B

NES

nuclear export signal

NLS

nuclear localization signal

p90RSK

p90 ribosomal S6 kinase

PDE3A

phosphodiesterase 3A

PKC

protein kinase C

PIAS

protein inhibitor of activated STAT

PLN

phospholamban

PPAR

peroxisome proliferator-activated receptor

PTM

post-translational modification

RCOR1

REST corepressor 1

ROS

reactive oxygen species

SENP

sentrin/SUMO-specific protease

SERCA2a

sarcoplasmic reticulum Ca²⁺ ATPase 2a

SIM

SUMO-interacting motif

SR

sarcoplasmic reticulum

STUbL

SUMO-targeted ubiquitin ligase

SUMO

Small Ubiquitin-related Modifier

Ub

ubiquitin

UPS

ubiquitin-proteasome system

USPL

ubiquitin-specific protease-like

XBP-1s

spliced X-Box Binding Protein 1