Sumoylation at the Host-Pathogen Interface

Wilson, Van G.

Finally, many viruses have developed mechanisms to disrupt sumoylation of PML and associated proteins such as Daxx. Given the important role of PML NBs in innate immunity it is to be expected that viruses would need to overcome this defense in order to establish a productive infection. The finding, that PML assembly into NBs and recruitment of associated proteins is highly dependent upon PML sumoylation [103, 104] , offers a rationale for why many viruses attack the sumoylation status of PML. There is currently a large list of viral proteins that are known to disrupt NBs, and this is a particularly prominent feature of the herpesvirus family, including the BZLF protein of Epstein-Barr virus [105] , the LANA2 protein of Kaposi's sarcoma-associated herpes virus [106] , human cytomegalovirus IE1 protein [107] , the herpes simplex ICP0 [108] , and the varicella zoster virus ORF61 protein [69] . Somewhat surprisingly the different herpes viruses utilize different mechanisms to disrupt PML sumoylation. For example, the EBV BZLF protein (also known as Zebra, Z, or Zta) is efficiently sumoylated at lysine 12 and appears to prevent PML sumoylation by out competing PML for a limited supply of free SUMO [105] . Consistent with this model, a lysine 12 mutant of BZLF is impaired in dispersing PML bodies when BZLF is limited compared to PML [109] . In contrast, as discussed in section 4.2, the herpes simplex ICP0 protein appears to be a SUMO-targeted ubiquitin ligase [67] . ICP0 contains SIM motifs that direct it to PML where the ICP0 ubiquitin ligase activity conjugates ubiquitin to PML and enhances its proteasomal degradation. Similar to ICP0, the varicella zoster virus ORF61 protein also possesses SIM motifs that are critical for interaction and dispersal of PML NBs [69] . However, the ORF61 protein alone is not sufficient to degrade PML, suggesting that its known ubiquitin ligase activity is not directly involved in modifying PML, so the precise mechanism for PML dispersal remains uncertain. Like ICP0 and ORF61, the LANA2 protein of KSHV has a SIM motif (residues 474-477) that is critical for interaction with and destruction of PML NBs in B cells [110] . Consistent with a SUMO-SIM interaction, the association of LANA2 and PML not only requires the LANA2 SIM motif, but also lysine 160 of PML which is a known sumoylation site. In addition to disruption of NBs, LANA2 causes PML levels to decrease via proteasomal degradation. These effects are associated with a LANA2 mediated increase in PML modification by SUMO2 with a concomitant increase in ubiquitination of PML. SUMO2/3 modification of PML is known to promote degradation of PML due to ubiquitination by the cellular RNF4 ubiquitin ligase [111] , suggesting that this may be the pathway triggered by LANA2 expression. A subsequent study found that LANA2 is sumoylated preferentially by SUMO2 and that a sumoylation deficient mutant of LANA2 was impaired for NB disruption [106] . This observation raises the possibility that SUMO moieties on LANA2 may be interacting with SIM motifs on PML or other proteins to further facilitate the formation of complexes which lead to PML degradation. Lastly, the human cytomegalovirus (HCMV) has two early proteins, IE1 and IE2, that are both localized to PML NBs [107, 108] . IE1 induces disruption of NBs without degradation of PML [112] , but with the loss of sumoylated forms of PML [113, 114] . However, IE1 has no SUMO protease activity in vitro, and purified IE1 does n