Viruses exploit the ubiquitin system by targeting cell-surface receptors recognized by immune cells for internalization, thereby evading recognition. We have characterized the KSHV (Kaposi's sarcoma-associated herpesvirus)-encoded E3 ubiquitin ligases, K3 and K5. We find their activities not only prevent recognition by cytotoxic T-lymphocytes, but also promote evasion of NK (natural killer) cells. NK cells recognize and lyse virus-infected cells expressing ligands for activatory receptors such as NKG2D (NK group 2D). K5 down-regulates cell-surface expression of the NKG2D ligands MICA/B (MHC class I-related chains A and B) by ubiquitination of MIC cytoplasmic tail lysine residues. Ubiquitination results in redistribution of MICA from the plasma membrane to an intracellular compartment, but does not result in an increased rate of degradation. Furthermore, K5 down-regulates cell-surface expression of another NK cell activatory receptor ligand, AICL (activation-induced C-type lectin). This activity requires the K5 RING (really interesting new gene)-CH domain and AICL cytoplasmic tail lysine residues. MICA and AICL down-regulation by K5 reduces NK cell-mediated cytotoxicity towards target cells, thus providing KSHV with an NK cell evasion mechanism.
The ubiquitin signalling system is a hallmark of eukaryotes and is not found in prokaryotic organisms . However, many pathogens have hijacked components of the ubiquitin pathway in order to modify the cellular environment and protect infected cells from the host immune system. An example of this piracy comes from the genome of KSHV (Kaposi's sarcoma-associated herpesvirus), which encodes the immune evasion genes K3 and K5. These genes are the founding members of a family of E3 ubiquitin ligases which possess a variant RING (really interesting new gene) (v-RING or RING-CH) domain. RING domain-containing proteins constitute the largest family of E3 ubiquitin ligases; 11 human RING-CH domain-containing proteins related to the viral immunoevasins have been identified [2,3].
K3 and K5 down-regulate MHC class I molecules through their E3 ubiquitin ligase activity [4,5]. However, this reduction may render KSHV-infected cells susceptible to NK (natural killer) cell activity. NK cells form part of the early innate immune response to infections, and are particularly important in defence against herpesviruses l [6,7]. NK cells can directly lyse infected cells and also induce and regulate the adaptive immune response; their cytotoxicity is controlled by a balance of activatory and inhibitory signals received through cell-surface receptors . This review will focus on new proteins targeted by K5 E3 ubiquitin ligase activity and how this provides KSHV with an NK cell evasion mechanism.
K5: a promiscuous E3 ubiquitin ligase
Investigation of K5 activity has shown that it acts on a number of cell-surface receptors. In addition to MHC class I, K5 has been shown to down-regulate B7.2, ICAM-1 (intracellular adhesion molecule-1), CD31, IFN-γ (interferon γ) receptor 1, CD166 and CD1d [9–13]. Like other RING E3 ligases, K3 and K5 are passive catalysts of ubiquitination. Interaction with their cognate E2 enzymes occurs through the RING-CH domain, allowing transfer of ubiquitin from E2 to substrate. Structural analysis has identified residues in the K3 RING-CH domain that are required for interaction with E2 enzymes , allowing the creation of K3 and K5 mutants lacking ligase activity [5,15]. Several studies have shown that the K5 RING-CH domain is absolutely required for ubiquitination of target protein cytoplasmic tail lysine residues [4,5,11,15]. However, it is not known how K5 substrates are selected. As yet, no direct interaction between a target protein and K5 has been detected, nor has a conserved K5-binding motif been identified. However, K5 has been found to associate with PACS-2 (phosphofurin acidic cluster sorting protein-2), and this interaction is required for down-regulation of CD31 and its degradation in the endoplasmic reticulum .
The E3 ubiquitin ligase activity of K5 affects proteins that are involved in diverse cellular pathways. There is evidence that B7.2 and ICAM-1 down-regulation by K5 can protect cells from NK cell cytotoxicity , but these data were challenged by others , leaving the role of K5 in NK cell evasion unresolved. This led us to investigate the ability of K5 to modulate expression of ligands for the NK activatory receptors, NKG2D (NK group 2D) and NKp80.
K5 down-regulates NK cell activatory ligands
NK cell activity is regulated by signalling through both activ-atory and inhibitory receptors. The majority of inhibitory receptors recognize MHC class I alleles, whereas activatory receptors recognize a range of ligands, including influenza-encoded haemagglutinin, but also ligands induced by stress and viral infection [17–19]. NKG2D is a widely expressed NK cell activatory receptor which has a central role in innate immunity. MICA/B (MHC class I-related chains A and B), which are well-characterized NKG2D ligands, are up-regulated upon viral infection and are important targets for viral evasion .
We found that K5 down-regulates cell-surface expression of the NK cell activatory ligands MICA and MICB (Figure 1A) . However, a truncated allele, MICA *008, which is expressed in HeLa cells, was resistant to down-regulation (Figure 1B) . Interestingly, MICA *008 is a common allele found at high frequency in many populations [20,21]. The resistance to down-regulation by K5 was due to the loss of cytoplasmic tail lysine residues . Full-length MICA was ubiquitinated on its cytoplasmic tail lysine residues in the presence of K5, but not the K5 RING-CH mutant, K5 W/I (Figure 1C) . To determine the fate of ubiquitinated MICA, K5-expressing HeLa cells transfected with GFP (green fluorescent protein)–myc-tagged MICA were analysed by [35S]methionine radiolabelling pulse–chase experiments. HeLa cells express the endogenous truncated MICA *008 allele, which is unaffected by K5 expression, and thus provided an internal control, allowing comparison of the fate of the ‘K5-sensitive’ and ‘K5-insensitive’ MICA alleles under similar conditions (Figure 1D) . The degradation of GFP–myc–MICA was not significantly increased compared with the truncated endogenous MICA allele *008 (Figure 1D) . Immunofluorescence of HeLa cells showed GFP–myc–MICA predominantly at the plasma membrane (Figure 1E) . In the presence of K5, GFP–myc–MICA was redistributed to punctate intracellular vesicles (Figure 1E) . Therefore K5 expression caused MICA ubiquitination and redistribution to an unidentified intracellular compartment, but did not enhance degradation.
K5 down-regulates MICA, MICB and AICL through its E3 ubiquitin ligase activity
Like NKG2D, NKp80 is a C-type lectin-like homodimeric receptor that stimulates NK cell cytoxicity, but it is expressed on NK cells alone and is absent from rodents. It was recently described to recognize AICL (activation-induced C-type lectin), found on the cell surface of myeloid cells . KSHV productively infects monocytes within Kaposi's sarcoma lesions, and can undergo lytic replication in these cells . We found that K5 could down-regulate the NKp80 activatory ligand AICL from the cell surface (Figure 1F) , which led to degradation of AICL in an endolysosomal compartment . This activity required the K5 RING-CH domain, in addition to lysine residues in the AICL cytoplasmic tail .
K5 promotes evasion of NK cell cytotoxicity
To investigate whether the down-regulation of MICA/B and AICL by K5 allowed NK cell evasion, NK cell-mediated cytotoxicity towards K5-expressing cells was examined. Cells stably expressing MICA and a K5-resistant lysine-less MICA (MICA Kless) were tested for killing by primary human NK cell lines in the presence and absence of K5 (Figure 2A) . The basal level of NK-mediated killing of targets was increased following expression of MICA (Figure 2A) . K5 expression reduced the killing to background levels, an effect not seen on MICA Kless-K5-expressing target cells. These results suggested that K5 protects cells from NK killing and this effect is absolutely dependent on the down-regulation of cell-surface MICA and not on other cell-surface receptors, such as ICAM-1 and B7.2, which are also affected by K5.
K5 protects cells from NK cytotoxicity by down-regulation of MICA and AICL
Owing to the lack of cytoplasmic tail lysine residues, K5 cannot down-regulate the truncated MICA allele *008, which contains a modified transmembrane domain and is mislocalized in polarized epithelia . In addition, MICA alleles are reported to have differential NKG2D binding, depending on the presence of a methionine or valine residue in the MICA extracellular domain . Alleles with a methionine residue at position 129 are reported to have higher binding to NKG2D than alleles with a valine residue at the same position . Given these reports in the literature, the ability of various MICA alleles to trigger NK cytotoxicity in a 51Cr-release NK cytotoxicity assay was compared. MICA *004 (129-Val), MICA *001 (129-Met), MICA *018 (129-Met), MICA *008 (129-Val and truncated) were expressed in a target cell line. No significant difference was seen between the specific lysis of target cells expressing MICA-129-Met alleles compared with MICA-129-Val alleles, or between full-length alleles compared with the truncated allele (Figure 2B). As MICA *008, the naturally occurring truncated allele, can efficiently trigger NK cell cytotoxicity towards target cells, but is not down-regulated by K5, this suggests that possession of this allele may have some protective effect during KSHV infection. Given the high frequency of MICA *008 in many populations, this raises the possibility that viral infection has driven selection for this allele. In addition, MICA alleles with either a methionine or valine residue at position 129 in the MICA extracellular domain can trigger NK cell cytotoxicity efficiently, under these experimental conditions.
To determine the effect of AICL down-regulation by K5, we expressed K5 in U937 cells. U937 is a myeloid cell line which expresses AICL, but not MICA, and is sensitive to NK cell lysis in an NKp80-dependent manner . K5 expression reduced the specific lysis of U937 cells by primary NK cells (Figure 2C) . This protection from NK-mediated lysis was similar to that seen with NKp80 blocking antibody, and no additional protection was seen in the presence of K5 when NK cells were NKp80-blocked (Figure 2C) . This suggests that the protection of U937-K5 cells compared with U937 cells is NKp80-dependent and probably results from the down-regulation of AICL by K5 (Figure 2C) .
The hijacking of an E3 ubiquitin ligase from cellular genes has led to KSHV acquiring the ability to regulate a very broad spectrum of proteins within its host cells. K5 is one of the most versatile immunoevasins so far described: it is capable of targeting molecules involved in CD8+, CD4+, NK-T and NK cell activation, the IFN-γ pathway and endothelial cell migration. In the present article, we have reviewed a role for K5 in the evasion of NK cell recognition. The removal of cell-surface MICA/MICB and AICL expression by the K5 gene provides KSHV with protection from NK cell cytotoxicity, which may be an essential function for K5 in vivo.
Ubiquitin and Ubiquitin-Like Modification in Health and Disease: Biochemical Society Irish Area Section Annual Meeting held at Royal College of Surgeons in Ireland, Dublin, Ireland, 22–23 November 2007. Organized and Edited by Caroline Jefferies (Royal College of Surgeons in Ireland).