Viruses have evolved a variety of mechanisms to overcome host defenses and exploit cellular pathways to their own benefit. Sumoylation is one such reversible cellular pathway that intricately controls substrate protein function even upon mild fluctuations in their sumoylated status .
SUMO conjugation is mediated by a multistep enzymatic reaction, very similar to the ubiquitin pathway, catalyzed by a dimeric SUMO E1 activating enzyme (SAE1-SAE2 also known as Aos1-Uba2), one unique SUMO E2 conjugating enzyme (Ubc9), few SUMO E3 ligases, and several SUMO proteases that activate the SUMO precursor and remove SUMO from substrates, making the process reversible. Although high concentrations of Ubc9 are sufficient to transfer SUMO in vitro, specific E3 ligases are most often required in vivo for an efficient reaction.
Countless viral proteins have been shown to interact with SUMO components. Consequently, such target proteins of sumoylation can differently interact with partner proteins, extending therefore the impact of sumoylation on a multitude of downstream effector pathways and biological processes.
In particular, by selectively targeting the single E2 enzyme Ubc9, proteins encoded by viruses can modify their activities and/or potentially subvert cellular pathways to create a more conducive environment for their propagation. Virus exploitation of the SUMO pathway has been very recently the subject of some excellent reviews [2-4]. The focus of this review is to discuss how viral proteins can impact on the functions of Ubc9, and the possible consequences of Ubc9 deregulation in relation to its function as a SUMO-conjugating enzyme as well as functions that are independent of its catalytic activity.
Sumoylation–A Brief Overview
Sumoylation is a highly conserved reversible post-translational modification (PTM) first identified in 1996 as a novel ubiquitin-like modification, due to the significant functional relationship with ubiquitylation [5, 6]. Indeed, covalent conjugation of SUMO proteins to their substrates requires an enzymatic cascade, analogous to the ubiquitin pathway, comprising the sequential action of four enzymes: SUMO proteases (SENPs) to process the SUMO precursor, a modifier activating enzyme (E1), one conjugating enzyme Ubc9 (E2) and a number of the SUMO ligases (E3s).
SUMO Proteins and the SUMO E2 Conjugating Enzyme Ubc9
Yeast and invertebrates have only a single SUMO-encoding gene. However, vertebrate and plant genomes contain four SUMO genes, encoding for the SUMO paralogs 1–4. In humans, SUMO1, SUMO2, and SUMO3 are ubiquitously expressed in all tissues, whereas SUMO4 has been detected only in the kidneys, dendritic cells, and macrophages . SUMO2 and SUMO3 are nearly identical to each other, differing by only three N-terminal residues and therefore often referred to as SUMO2/3. SUMO1 however, shares only about 45% sequence identity with SUMO2/3 while SUMO4 is 86% homologous with SUMO2. Although the same enzymatic machinery can modify proteins equally with SUMO1 or SUMO2/3, the different SUMO paralogs have been shown to preferentially conjugate some substrates and therefore appear to play distinct functions in a particular cellular context . Another important difference between paralogs is that SUMO2/3 bears the canonical SUMO consensus motif and therefore can be sumoylated on itself, forming chains on substrate proteins . SUMO1 instead, usually acts as a terminator on the SUMO2/3 polymer .
Human Ubc9 is a 17kDa protein, which has a 100% sequence identity to mouse Ubc9  56% similarity to Saccharomyces cerevisiae Ubc9  and is 66% similar to Schizosaccharomyces pombe Hus5 . Ubc9 belongs to the Class I family of E2 enzymes based on sequence comparison. Four enzyme classes have been identified; Class I enzymes have a conserved catalytic core of 150–200 residues. Class II and III enzymes, contain a catalytic core with a C- or N-terminal stretch of amino acids, respectively. Class IV enzymes have both C- and N-terminal extensions . Since the catalytic core along with the amino acid extensions confer both localization and specificity, members of different enzyme classes or within the same enzyme class in turn form distinct functional sub-families .
At subcellular levels, in mammalian cells, Ubc9 is predominantly expressed in the nucleus and in the nuclear pore complex, but also around the nuclear envelope, in the cytoplasm, and in the cell membrane . Given its wide expression and its central role in governing a pleiotropic cellular pathway, it is not surprising how Ubc9 knockout or down-regulation has deleterious effects in a variety of organisms . In particular, Ubc9 depletion leads to defects in progression during mitosis [12, 18], in chromosome segregation , and developmental defects . Consistently, Ubc9 was found to be highly expressed in many types of human cancer cells , suggesting its crucial role in determining a selective growth advantage.
The main and best-characterized Ubc9 function is to specifically transfer, with equal efficiency , SUMO paralogs to substrates. In contrast to the large number of the ubiquitin E2, Ubc9 is the sole SUMO conjugating enzyme that directly binds all sumoylated proteins known to date.
Ubc9 in SUMOylation
SUMO is cleaved by sentrin-specific proteases (SENPs)  to expose the C-terminal diglycine motif, before adenylation and ATP-dependent activation by the SUMO E1 activating enzyme SAE1/SAE2 (Sumo Activating Enzyme, also known as AOS1-Uba2). Once activated, the SUMO moiety from the E1-SUMO thioester is transferred by a lateral trans-esterification reaction onto Cys93 of Ubc9  where the C-terminal Glycine of SUMO is conjugated by thioester bond before its transfer to target proteins. The same Ubc9 domain is also responsible for substrate recognition, enabling proper orientation of the acceptor Lysine , and for E1 binding . Although Ubc9 can transfer SUMO to targets by itself , specific SUMO E3 ligases facilitate the reaction by direct transfer of SUMO from Ubc9 to substrates or by correctly orientating the SUMO-E2 complex to promote substrate specificity . In addition, Ubc9 not only forms a thioester bond with SUMO but also interacts with SUMO noncovalently in its N-terminal region, enabling in this way the formation of SUMO polymeric chains .
Substrate sumoylation often occurs on the Lysine residues within the consensus sequence ψKx(D/E) (where ψ is a large hydrophobic residue) which is directly recognized by Ubc9 . Sumoylation on consensus Lysine residues has also been shown to depend on adjacent phosphorylation sites as in the case of the phosphorylation-dependent SUMO motif (PDSM) or on downstream acidic residues as in the case of negatively charged amino acid-dependent sumoylation motif (NDSM; (29, 30). Alternatively, sumoylation can also be mediated by SUMO interacting motifs (SIMs), short sequences of <10 amino acids . These motifs can interact with SUMO in the Ubc9-SUMO thioester enabling Ubc9 to conjugate SUMO to a target Lysine, not necessarily within a SUMO consensus sequence. Two other SUMO interaction modules have been recently described: the ZZ zinc finger of HERC2  and the SUMO1-specific SUMO-binding arm of dipeptidyl peptidase 9 . Finally, the SENP family of DeSumoylating-Isopeptidases  and the recently identified Ubiquitin specific protease-like 1 (USPL1) selectively remove SUMO from substrates allowing free SUMO to enter another conjugation cycle .
Catalysis-Independent Functions of Ubc9
Although the SUMO conjugation function of Ubc9 has been studied on several target proteins of sumoylation, the catalysis-independent function of Ubc9 was demonstrated when the RanGAP1-SUMO1 heterodimer, unable to bind Ubc9, failed to localize to the nuclear pore complex . Subsequently biochemical reconstitution of Ubc9 with sumoylated RanGAP1 and RanBP2 revealed the role of Ubc9 as a structural constituent of a functional multisubunit E3 ligase complex . In S. cerevisiae, Ubc9 sumoylated at K153, which has reduced catalytic activity, has been shown to be essential to position SUMO proteins, by back-side binding to SUMO-charged Ubc9 proteins, enabling SUMO chain formation. The functional consequence of Ubc9 that cannot be modified by SUMO, results in reduced SUMO conjugates during meiosis and impaired synaptonemal complex formation . In addition, Ubc9 may also function as a cellular chaperone [39, 40], transcriptional coregulator , protein stabilizer [40, 42], or can determine miRNAs expression .
Regulation of Ubc9
Ubc9 undergoes autosumoylation at K14 in mammals and K153 in S. cerevisiae, as shown by mass spectrometric and mutation analyses. In vitro SUMO modification of Ubc9 at K14 has been shown to enhance the interaction between Sp100 SIM and SUMO . More recently, work carried out in budding yeast explored the possibility of sumoylated Ubc9 as a cofactor to the SUMO-Ubc9 thioester for the formation of sumoylated intermediates during meiosis. In an elegant study carried out very recently , immunoprecipitating endogenous sumoylated proteins using monoclonal SUMO1 and SUMO2 antibodies followed by peptide elution identified the preferential covalent conjugation of Ubc9 with SUMO2. This approach has been the closest to physiological states of identifying sumoylated conjugates since overexpression systems although enable the detection of SUMO substrate proteins, are likely to interfere with SUMO paralog preference. To this effect, we have recently demonstrated the significance of this paralog specificity on histone deacetylase 1 (HDAC1) stability in transformed and untransformed cells . In addition, it was recently shown that Ubc9 could also be acetylated at K65 in the presence of SIRT1 (sirtuins) inhibitors. Ubc9 acetylation preferentially reduces sumoylation of substrate proteins containing the NDSM, while other substrates remain unaffected . Regulation of Ubc9 activity is also influenced by another PTM. Indeed, CDK1/cyclin B kinase in vitro is able to specifically phosphorylate Ubc9 at Serine 71, leading to enhanced Ubc9 ability to conjugate SUMO to different substrates . Moreover, it has been demonstrated that the same phosphorylation also affects Ubc9 stability . Finally, regulation of Ubc9 activity has also been observed upon oxidant damage, resulting in Ubc9-SAE2 crosslinks, reduced Ubc9 thioester formation, and global desumoylation .
In addition to the modulation of its sumoylation activity, the transcriptional regulation of Ubc9 in ER positive breast cancer cells treated with 17β-estradiol was also investigated. This study identified putative binding sites for the estrogen receptor on the Ubc9 promoter and engaging the ER signaling pathway was shown to increase mRNA and protein levels of Ubc9 .
Interestingly, Ubc9 expression can also be finely tuned by microRNA, small noncoding RNA sequences that act as translational repressors of gene transcripts. In particular, miR-30 [52, 53], mir-200, mir-182 families , and miR-214  directly target Ubc9, suggesting that physiopathological Ubc9 expression could be at least in part mediated by miRNAs.
Viral Interplay with Sumoylation–A Brief Overview
Since targeting the sole E2 conjugating enzyme that controls the SUMO pathway has a wide impact in cell functions, viruses extensively manipulate Ubc9 to their own advantage. In the following section, we will briefly discuss some of the latest paradigm of viral protein exploiting Ubc9. Viral proteins have been shown to regulate Ubc9 in several ways; either by way of direct interaction, Ubc9 protein degradation, or Ubc9 mis-localization. The consequences of this regulation can impact the global cellular sumoylation/desumoylation states or can be independent of it–either mechanism proving beneficial to infectivity and viral pathogenesis (discussed in sections “Virus Benefits From Impaired Sumoylation,” “Virus Benefits From Sumoylation,” and “Virus Benefits From Ubc9 but Not Catalysis”). Viral proteins have also shown to be targets of the cellular sumoylation machinery influencing its protein localization and transcriptional functions (discussed in section “SUMO Modification of Viral Proteins”).
Virus Benefits From Impaired Sumoylation
A well-studied example of a viral protein affecting the SUMO pathway comes from studies carried out in our lab on Gam1. Gam1, an essential protein of the CELO virus  decreases levels of Ubc9 in vitro and in vivo in a proteasome-dependent manner . Although the significance of Ubc9 decrease has not been studied directly, the Gam1 deleted CELO virus has been shown to be replication incompetent . Additionally, Gam1 also targets SAE1/SAE2 for degradation  and desumoylates PML-NDs . Another viral protein that was shown to decrease global levels of sumoylation is the E6 oncoprotein of HPV16/18 (Human Papillomavirus Type 16/18) . Since E6 has not been shown to target other members of the SUMO pathway, it is likely that Ubc9 decrease contributes to low levels of SUMO conjugates in cells. Expression of E6 proteins in H1299 cells, which have high levels of the E3 ligase E6AP, versus K3 cells that contain low levels of E6AP showed that Ubc9 decrease was most prominent in H1299 cells, demonstrating the possible involvement of E6AP in degrading Ubc9. Interestingly, low risk E6 had no effect on Ubc9 or sumoylation, suggesting a possible relationship between Ubc9 degradation and cervical cancer . Finally, also the coxsachie B5 viral protein exploits Ubc9 to decrease the sumoylation status of selected proteins. Although the viral protein does not alter Ubc9 levels, infection coincides with redistribution of Ubc9 and SUMO from being a localized puncta in uninfected HeLa cells to a dispersed morphology. This redistribution correlated with HDM2 degradation promoted by its loss of sumoylation . Thus, the tethering of Ubc9 to different cellular compartments might serve as crucial indicators that can be investigated to reveal altered Ubc9 functions.
Virus Benefits From Sumoylation
While Gam1, E6, and B5 viral proteins decrease sumoylation, Epstein-Barr virus (EBV) latent membrane protein 1 (LMP1) exploits Ubc9 to increase SUMO conjugation in infected cells. EBV is associated with a variety of lymphomas and epithelial derived malignancies  and in almost all EBV positive tumors, LMP1 is considered the main viral oncogene. The C-terminal domain of LMP1 contains three C-terminal activating regions (CTAR 1–3) and CTAR3 has been shown to interact with Ubc9. Deletion of this region in LMP1 (Δ33) abrogates Ubc9 binding and is incapable of increasing cellular levels of sumoylation upon SUMO1 and SUMO2/3 overexpression. The deletion mutant (Δ33), also showed reduced cell migration by scratch assays . Moreover, the Interferon Regulatory Factor 7 (IRF7) has been recently identified as a specific target of LMP1-induced sumoylation and promotes IRF7 nuclear accumulation, stability, and limits IRF7-transcriptional activity in latently infected EBV cells . Since IRF7 is a master regulator of the innate immune response, LMP1-mediated Ubc9 manipulation might favorably mask virus detection in the host cells.
Virus Benefits From Ubc9 but Not Catalysis
One of the earliest reports on viral protein interaction with Ubc9 was shown in a yeast two-hybrid screen, which identified the Adenoviral protein E1A as a binding partner of mouse Ubc9 . The interaction was shown to be direct, dependent on the amino acid sequence within the adenoviral protein E1A and the N-terminal of Ubc9. The interaction was preserved in a C93S mutant and was thus independent of the catalytic competence of Ubc9. More importantly, the article also showed that when the E1A-Ubc9 interaction was lost by mutating specific residues within E1A, the function of the viral protein to make bigger and fewer PML-NDs (promyelocytic leukemia nuclear domains) was lost as well . Since differences in PML-ND number and size have been correlated to different states of viral infection by HSV-1, EBV, and CMV , the requirement for Ubc9 interaction to E1A is particularly significant. Thus, it appears that the interaction of Ubc9 per se potentiates the effects of E1A upon infection rather than the sumoylation function of the protein. Similarly, the Moloney Murine Leukemia virus capsid protein was shown to interact with Ubc9. Mutations that target this interaction interface have been shown to block viral replication .
Another example of a viral protein that benefits from Ubc9 but independent of its catalytic function is the HIV-1 envelope protein gp120. In the absence of Ubc9 the gp120 levels could be rescued by lysosomal inhibition, pointing to the role of Ubc9 in stabilizing gp120 after trafficking out of the trans-golgi network. However, virions released from Ubc9 depleted cells show an 8–10 fold decrease in infectivity. The Ubc9-Gag interaction, which has been shown to be important for gp120 stability, also occurs with the catalytically dead Ubc9. Furthermore, compared to Ubc9 depleted cells, the catalytically dead Ubc9 mutant was unable to affect levels of viral infectivity [68, 69].
SUMO Modification of Viral Proteins
A number of viral proteins are themselves the targets of sumoylation, enhancing their pathological functions in infected cells. Here are a few examples where viral protein sumoylation by the host SUMO machinery influences either viral protein subcellular localization or transcriptional activity.
UL44 is a human cytomegalovirus (HCMV) DNA polymerase subunit involved in HCMV replication. UL44 strongly binds Ubc9 and is extensively sumoylated by SUMO1 and 2/3 both in vitro and at a later time during HCMV replication. Remarkably, Ubc9-mediated UL44 sumoylation alters its subcellular localization, resulting in decreased UL44 expression in the viral DNA replication compartments, and increased viral replication and viral production, suggesting that UL44 sumoylation could support later functions important for viral propagation . Similar to the mechanism adopted by UL44 is the nucleocapsid protein (NP) of Hantan virus, which interacts with Ubc9 and is sumoylated in infected cells. NP functions to mediate viral assembly of structural proteins with viral genomic RNAs  and NP sumoylation is fundamental in determining its localization at the perinuclear region, crucially affecting viral replication and assembly occurring in this site .
An example of a viral protein where Ubc9 interaction influences its transcriptional activity is the influenza A virus nonstructural protein NS1. NS1 sumoylation affects NS1 protein levels and its Ubc9-mediated sumoylation, specific for the SUMO paralog 1, increases NS1 stability causing therefore a slight acceleration in viral replication rate. Notably, SUMO1 conjugation seems to be a common feature of most NS1 proteins, further underlining the importance of this modification in influenza virus infection . HPV E2, the Human Papillomavirus early protein E2 is also a target for Ubc9 mediated sumoylation by all the three main SUMO paralogs. An E2 mutant lacking the sumoylation site is rendered less capable of promoting the transcription of its target genes, without affecting E2 subcellular localization or its DNA-binding ability, revealing the importance of E2 sumoylation in the transcription of target genes . EBV Rta is another example of a viral protein whose transcriptional activity is strictly regulated by Ubc9-mediated sumoylation. EBV is usually maintained under latent conditions after infection and undergoes a lytic cycle phase to proliferate -mainly regulated by Rta . Nonconsensus Rta sumoylation mediated by Ubc9, increases its transactivation activity and promotes viral replication of the latent EBV . Moreover, it has been reported that PIAS1, PIASxβ, and RanBPM can stimulate SUMO conjugation to Rta, acting therefore as E3 ligases [76-78].