The PML protein was originally discovered in patients suffering from acute promyelocytic leukemia (APL), where a reciprocal chromosomal translocation resulting in a fusion of the PML protein to the retinoic acid receptor α turned out to be responsible for this hematopoietic malignancy [16]. PML, also known as TRIM19, belongs to the RBCC or tri-partite motif family (TRIM) of proteins that are comprised of a zinc-finger RING domain, one or two B-boxes (cysteine/histidine-rich motifs) and a predicted α-helical coiled-coil domain [17]. This motif, which allows PML to interact with other proteins as well as to homo-oligomerize, is essential for ND10 formation and for the function of PML as a growth- and transformation-suppressor [18]. Due to differential splicing of the PML gene transcript which consists of nine exons, at least seven different PML isoforms (I to VII) are expressed within cells, all sharing a common N-terminus but varying in their C-termini [17]. Since not all PML isoforms retain the nuclear localization signal in exon 6, beside nuclear ND10-localized PML also cytoplasmic species of PML exist [17]. So far, the functions of the PML splice variants are not fully understood. However, in this context it is of note that recent publications demonstrate an influence of herpesvirus infections on the splicing pattern of PML suggesting diverse roles for the individual PML isoforms with regard to viral replication [19,20]. Besides determining the subnuclear distribution of PML, alternative splicing could add new functional domains to the protein or may be an important mechanism for generating differential PML-binding interfaces for a variety of factors.

Moreover, all isoforms having molecular weights ranging from 48 to 97 kDa (kilo Dalton), are subject to posttranslational modifications like phosphorylation [21] or conjugation to the ubiquitin-homologous protein SUMO (SUMOylation) [22,23]. In addition to three SUMO modification sites at lysine residues 65 (in the RING finger domain), 160 (in the B1 box), and 490 (in the nuclear localization signal) [24], PML also contains a SUMO binding motif that enables it to interact noncovalently with SUMO [25]. Covalent as well as noncovalent SUMO modification is extremely important for the function of PML to orchestrate ND10 formation as clearly illustrated in case of the recruitment of hDaxx to this subnuclear structure [13,25,26].

HDaxx is a highly conserved nuclear protein that contains a serine/proline/threonine-rich domain, an acidic domain, a coiled-coil region, and two paired amphipathic helices [27-30]. It has been recognized as a regulator of both apoptosis and gene expression [31]. In gene regulation, hDaxx has been shown to function as a transcriptional corepressor. It negatively affects gene expression by suppressing the activity of several transcription factors, including Ets-1 (E twenty-six 1) [32], NF-κB (nuclear factor κB) [33], Pax3 [27], E2F-1 [34], Smad4 [35], p53 family members [36], and glucocorticoid, mineralocorticoid and androgen receptors [37]. Recent findings indicate that hDaxx associates via its newly identified SUMO-interacting motif (SIM) with SUMOylated DNA-binding transcription factors [26], thereby recruiting proteins involved in transcriptional repression, such as histone deacetylase 1 (HDAC1) [29], HDAC2 [38], DNA methyltransferase 1 (DNMT1) [39], or ATRX (α-thalassaemia/mental retardation syndrome X-linked) [40,41], a member of the SNF2 family of chromatin remodeling enzymes, to targeted promoters. The transrepressive effect of hDaxx, in turn, is modulated by its subnuclear compartmentalization. The SIM enables hDaxx to also noncovalently interact with SUMOylated PML, resulting in the sequestration of nucleoplasmic hDaxx to ND10 [25,26], which is accompanied by attenuation of its inhibitory function [29]. Beside ND10, hDaxx can also be targeted to the nucleolus, centromeres, or heterochromatin through its interactions with either the nucleolar protein MSP58 (microspherule protein of 58 kDa) [42], the centromeric protein CENP-C [30], or the heterochromatin-associated factor ATRX [40]. Taken together, these observations imply that, depending on its subnuclear localization, hDaxx seems to fulfill distinct functions.

Like PML, the Sp100 protein represents a permanent ND10 constituent which is expressed via alternative splicing from a single gene transcript giving rise to four different isoforms designated as Sp100A [11], Sp100B [43], Sp100C [44], and Sp100-HMG [45,46]. All Sp100 splice variants share a common N-terminus harboring an HSR (homogeneously staining region) motif responsible for homo-oligomerization of Sp100 and its targeting to ND10 as well as a binding site for the nonhistone chromosomal DNA-binding protein HP1 (heterochromatin protein 1). The larger splice variants Sp100B, -C, and -HMG encode additional functional motifs at the C-terminus such as a SAND domain (common to all three longer isoforms), a PHD finger-bromodomain (Sp100-C), or an HMG box (Sp100-HMG). All three motifs represent potential DNA-binding domains which can^{frequently be found in proteins affecting chromatin structure [47]. Together with the observation that Sp100 interacts with heterochromatin protein HP1 [45], which plays a central role in establishing a stable heterochromatic network, this suggests a role of Sp100 in transcriptional regulation. Indeed, the isoform Sp100B has been shown to function as a transcriptional repressor of both cellular and viral promoters in transient expression experiments [48]. In addition, as recently reported, Sp100B preferentially associates via its SAND domain with DNA sequences containing unmethylated CpG dinucleotides, since methylation of cytosines in the CpG context abrogates DNA binding of Sp100B [49]. This finding prompted the authors to speculate that the preference of Sp100B for nonmethylated CpGs could provide a mechanism to specifically target this isoform to foreign DNA (including viral genomes), which is predominantly hypomethylated [49].}

Like many ND10 constituents, including PML and hDaxx [23,50,51], Sp100 is also subject to posttranslational modification by SUMO [51]. While SUMOylation of Sp100 is not required for its ND10 localization [52], it has been shown to enhance the interaction with HP1 and thus to stabilize Sp100-HP1 complexes suggesting that the functional interplay of Sp100 between ND10 and heterochromatin could be regulated in this way [44]. Recent data indicate, that the SUMO modification status as well as the relative expression level of Sp100 are to some extent regulated, either directly or indirectly, by PML [14]. SiRNA-mediated knockdown of PML resulted in an apparent reduction in the abundance of Sp100-HMG and the SUMO-modified form of Sp100A, with a concomitant increase in unmodified Sp100A [14]. Thus, Sp100 metabolism seems to be closely linked to that of PML.

HSV-1, a member of the neurotropic alpha sub-family of herpesviruses, is a common human pathogen that causes recurrent infections through its ability to establish^{a latent state in sensory ganglia after a primary infection of epithelial cells. It was the first virus to be shown to affect ND10 morphology during infection. The initial observation that HSV-1 disrupts ND10 but an ICP0-deficient mutant does not [85], rapidly led to the identification of ICP0 being necessary and sufficient for this effect [88,89]. ICP0 is a RING finger protein that is very important in certain cell types for initiating viral lytic infection and contributes to the reactivation of quiescent viral genomes in cultured cells as well as of latent virus from neurons in mouse models [90]. ICP0 initially precisely colocalizes with ND10 at early times upon infection and subsequently mediates the disaggregation of PML-NBs by inducing the degradation of the SUMO-1 modified forms of PML and Sp100 [91-94], leading to the release and dispersal of other ND10 proteins. This effect of ICP0 on ND10 is dependent upon ICP0 having an intact RING finger motif [88] and reflects its ability as an E3 ubiquitin ligase to recruit the E2 ubiquitin-conjugating enzymes UbcH5a and UbcH6 in order to target specific proteins for degradation by the proteasome [95,96]. However, it is likely that theapparent loss of the low molecular weight isoforms of Sp100 (except Sp100-A) duringwt HSV-1 infection is a consequence of ICP0-induced degradationof PML, rather than being a direct effect of ICP0 on Sp100 itself (since the electrophoretic pattern of Sp100 mimics what is observed in PML-depleted cells) [14]. Thus, it is tempting to speculate that by downregulating PML, ICP0 indirectly targets both major ND10 components.}

Disruption of ND10 by ICP0 correlates with its role in stimulating HSV-1 lytic infection and reactivation from quiescence or latency [90]. Furthermore, it has been shown that an ICP0-negative HSV-1, which fails to disrupt ND10 is hypersensitive to the effects of IFN in certain cultured cell lines [97,98] and exhibits low pathogenicity in normal mice in vivo, while infection of mice that are unable to mount an interferon response restores its replication capacity [99]. Although neither infection with wt nor ICP0-deficient HSV-1 is enhanced in the absence of PML in PML^-/- murine embryonic fibroblasts (MEFs) compared to control PML^+/+ MEFs, the replication of the ICP0-deletion virus is substantially compromised by IFN treatment of these cells in the presence but not absence of PML [100]. These results suggest that PML contributes to the cellular IFN-mediated restriction of HSV-1 infection and that one function of ICP0 is to efficiently counteract the repressive effect of PML. However, neither exogenous expression of PML isoforms III, IV, nor VI turned out to affect virus yield, even though PML overexpression blocked the dispersal of ND10 in response to HSV-1 infection [91,101,102]. In light of the increasing evidence that different PML isoforms possess distinct properties, one possible explanation could be that the PML variants used in those studies may play a minor role in the host IFN-mediated antiviral response. In addition, in contrast to the prior fixed cell analyses, live-cell imaging experiments clearly demonstrated that large ND10 aggregates as a consequence of high-level expression of PML are nonetheless subject to extensive modification in terms of both content and morphology during^{HSV-1 infection [101].}

In principle, it was not until the generation of physiologically more relevant knockdown cells being derived from the natural human host, in which individual ND10 components were depleted, that substantial advances in understanding the biological relevance of ND10 for HSV-1 replication were made. Although infection of cells with an shRNA-mediated down-regulation of either PML or Sp100 had no effect on wt HSV-1 replication, it resulted in a significant increase in the efficacy of gene expression and plaque formation of an ICP0-null mutant virus [14,103]. On the same theme, Negorev et al. previously reported that the restrictive activity of Sp100 is based on isoforms B, C, and HMG having an intact SAND domain with which they are capable of repressing viral IE transcription at the promoter level [104,105]. In accordance with this, only overexpression of Sp100B but not Sp100A suppressed HSV-1 gene expression [48,105]. Moreover, by specifically depleting the repressive Sp100 variants B, C, and HMG using RNA interference, these proteins could be confirmed as an essential part of the IFN-mediated^{restriction of HSV-1 replication [105]. In continuative studies Negorev et al. [104]., in addition, demonstrated that IFN treatment of cells leads to a change in the splicing pattern of Sp100 transcripts in favor of the repressive isoform Sp100C}

Interestingly, simultaneous knockdown of both ND10 proteins, PML as well as Sp100, further stimulated gene expression and plaque forming ability of the ICP0-knockout virus but did not completely eliminate its replication defect indicating the involvement of additional repressive factors [103]. Indeed, unpublished data presented by the Everett group at the 34^th International Herpesvirus Workshop suggest repressive roles for further ND10 constituents such as the transcriptional repressor hDaxx or the chromatin-remodeling enzyme ATRX which cooperatively contribute to the ND10-based inhibition of HSV-1 infection by forming a chromatin-remodeling complex (Everett, R., personal communication). Taken together, these findings add more weight to the concept of ND10 as an intrinsic antiviral defense mechanism of the cell as individual ND10 components participate independently in the silencing of viral gene expression. Furthermore, a recent publication suggests that not only the nuclear ND10-associated PML variants but also the cytoplasmic isoforms could potentially contribute to an antiviral response against HSV-1 as they are capable of sequestering the viral transactivator protein ICP0 to the cytoplasm thereby limiting viral protein accumulation and replication [20]. Interestingly, HSV-1 infection has been shown to induce changes in the splicing of PML pre-mRNA resulting in a selective enrichment of the suppressive cytoplasmic PML variants [20]. In this regard, it is noteworthy that splicing of PML pre-mRNA is also affected by the closely related herpes simplex virus type 2 (HSV-2) [19]. Nojima and coworkers recently identified the viral effector protein ICP27 as an alternative splicing regulator of the PML transcripts which has the ability to actively alter the PML expression profile during HSV-2 infection [19].

Like HSV-1, varicella zoster virus (VZV), the causative agent of chicken pox (varicella) in children or shingles (zoster) upon reactivation in elderly or immunocompromised individuals, is characterized as a human alpha-herpesvirus [106]. Comparable to the situation with HSV-1, new findings indicate that VZV infection is also negatively regulated by components of ND10 [107]. The analysis of VZV propagation in cells lacking the major ND10 factors revealed that replication of wt VZV is accelerated in case PML or hDaxx expression is silenced, while depletion of Sp100 had no effect on viral growth [107]. Like all members of the alpha-herpesvirus subfamily, VZV codes for an ortholog of HSV-1 ICP0 termed ORF61p which, similar to ICP0, transcriptionally activates viral promoters and enhances the infectivity of viral DNA [108,109]. Since accumulating evidence suggests that the transactivator function of ICP0, at least in part, is based on the ability of ICP0 to inactivate the cellular intrinsic defense instituted by ND10, a similar role was assumed for ORF61p. However, in contrast to ICP0 which is known to downregulate both restriction factors, PML as well as Sp100, ORF61p was found to specifically reduce Sp100 protein levels only in order to allow efficient virus replication [107,110]. Consequently, a chimeric HSV-1 mutant, which expressed ORF61p in place of ICP0 only partially rescued the phenotype of an ICP0-null virus as ORF61p fails to target PML, which remains available to restrict virus replication [110]. These findings suggest that ICP0 and ORF61p have evolved separately to provide different functions during virus infection. Thus, although alpha-herpesviruses like HSV and VZV are considered to be very similar, the studies presented by the Silverstein group demonstrate that they have evolved unique and specialized ways to interfere with the ND10-mediated host cell response to ensure optimal viral replication.

In common with all herpesviruses, EBV from the gamma-herpesvirus subfamily, possesses a biphasic life cycle consisting of a lytic and a latent phase. However, in contrast to HSV-1, its latent state is characterized by the expression of a set of proteins that are necessary for establishment and maintenance of latency [111]. Latent EBV infection, in addition, is strongly associated with several types of cancer, including nasopharyngeal carcinoma (NPC) [141]. While during latency, EBV genomes can be found in close association with interphase chromosomes, activation of the lytic replication cycle results in a spatial redistribution of viral DNA to ND10 [142]. Using in situ hybridization the authors also showed the development of replication centers starting at ND10 as replicating EBV genomes were frequently found beside this nuclear domain [142]. Subsequent studies identified the ori_lyt sequence as being required for the appearance of replicating episomes in association with PML-NBs [143]. However, the integrity of ND10 is no longer maintained as soon as productive EBV infection is initiated. The expression of lytic proteins is accompanied by a sequential redistribution of ND10 proteins starting with the dispersal of Sp100, hDaxx, and NDP55 which is then finally followed by PML relocalization [142-144]. Induction of productive EBV replication goes along with the expression of BZLF-1, the principal inducer of the lytic gene expression program [111]. Indeed, experiments carried out by Adamson and Kenney [144] illustrated, that BZLF-1 expression is sufficient to induce ND10 disruption by reducing the level of SUMOylated PML. BZLF-1, which is itself SUMO-1 modified, has been shown to compete with PML for limiting amounts of SUMO-1 [144]. In accordance with this, mutation of the SUMO-modification site of BZLF-1 abrogates desumoylation of PML [144]. However, this does not annihilate the capability of BZLF-1 to disperse ND10 after transient expression suggesting the existence of additional mechanisms for BZLF-1-mediated ND10 disruption. Further mutational analysis of BZLF-1 revealed a correlation between the capability of this protein to disperse Sp100 from ND10 and its efficiency to activate transcription via interaction with the ND10-located factor CBP [145].

Although initial studies suggested that EBV infection has no influence on ND10 during latency [142], recent evidence indicates that the latently expressed protein EBNA-LP (also designated as EBNA5) is likewise modifying the composition of ND10 accumulations. Already over ten years ago, EBNA-LP, which is important for EBV-mediated B-cell immortalization by functioning as a potent gene-specific coactivator of the viral transcriptional activator EBNA2 [146], was reported to colocalize with ND10 in EBV-positive lymphoblastoid cell lines [147]. Novel findings imply that EBNA-LP coactivates EBNA2 through binding of the ND10 factor Sp100, thereby selectively displacing it together with HP1α from PML-NBs [148]. Since both HP1α and Sp100 have been shown to possess transcriptional repressor activity [44,45,149], and expression of a mutant form of Sp100, that was unable to associate with ND10, resulted in coactivation of EBNA2 even in the absence of EBNA-LP [148], it has been speculated that the EBNA-LP-induced ND10 rearrangements may mitigate transcriptional barriers that prevent efficient expression of viral latent genes important for establishing latency. Furthermore, as IFN-αbtreatment of cells did not compromise EBNA-LP´s coactivator function it was suggested that EBNA-LP might play a role in EBV-evasion of IFN-mediated antiviral responses [150]. Interestingly, novel findings from Sivachandran et al. demonstrate that EBV latent infection in NPC cells is also associated with a modification of ND10 integrity and that EBNA1 is responsible for this effect by decreasing PML protein levels in a proteasome-dependent manner as well as through interaction with the cellular ubiquitin-specific protease USP7/HAUSP [151]. In addition, the data support a model in which EBNA1-mediated ND10 disruption promotes the survival of cells with DNA damage [151]. Thus, EBNA1 is considered as a viral factor that contributes to the establishment of NPC. In summary, these results are regarded as the first indication that modulation of ND10 might also be required for the establishment of nonproductive or latent herpesvirus infections as well as for the development of virus-induced cancers.

Human herpesvirus 8 (HHV8), a gamma-herpesvirus also known as Kaposi´s sarcoma-associated herpesvirus (KSHV), is a lymphotropic virus involved in the pathogenesis of Kaposi´s sarcoma (KS), primary effusion lymphoma (PEL), and a subset of multicentric Castleman´s disease (MCDs). KSHV also encodes an early protein, K8, which is a distant evolutionary homologue of the EBV regulatory factor BZLF-1, that has been found to localize to ND10 and to establish replication compartments in association with this nuclear substructure [152,153]. However, unlike BZLF-1, K8 does not induce any changes to ND10 composition [153]. Consequently, since the genome replication of KSHV also occurs in close contact to PML-NBs, it seems to be a general feature of herpesviruses to initiate their replication in the near vicinity of ND10. Nevertheless, for a long time, KSHV was considered to be unique among herpesviruses in that it might not target ND10 for destruction. However, very recently, Marcos-Villa et al. identified the LANA2 protein of KSHV as a viral regulatory factor that is capable of inducing the disruption of ND10 [154]. As a mechanism for this, it was proposed that LANA2 mediates the proteasome-dependent degradation of PML by promoting SUMO2-based ubiquitylation of PML [154]. Moreover, the authors presented evidence that depletion of PML caused by LANA2 relieves PML-mediated transcriptional repression of cellular survivin, a host protein that directly contributes to the malignant progression of KSHV-infected PEL cells [154]. Taken together, these data strongly support the notion that, similar to EBV (see 8.1.), KSHV has developed strategies to neutralize ND10 function in order to promote cellular transformation and carcinogenesis.

Like KSHV, MHV-68 (also known as murid herpesvirus 4) belongs to the herpesvirus genus rhadinovirus. It is emerging as a suitable model to study basic biological questions of gamma-herpesvirus host interactions due to the fact that it naturally infects mice and establishes chronic infections in them.

Intriguingly, recent data suggested that the MHV-68 infection cycle is also negatively influenced by ND10 as viral replication results in the disassembly of this subnuclear structure [155,156]. The virion tegument protein ORF75c of MHV-68, a polypeptide with homology to the cellular formylglycinamide ribotide amidotransferase (FGARAT) enzyme, was identified as being responsible for this effect by inducing the proteasomal degradation of the main structural ND10-organizer PML [155]. Consequently, an ORF75c deletion mutant failed to destroy ND10 accumulations [156]. In accordance with the assumption that the ORF75c-induced depletion of PML is a means to annihilate the antiviral activity of ND10, MHV-68 replication turned out to be more robust in murine fibroblasts being deficient of PML [155]. However, although Ling and colleagues claimed that the degradation of PML by ORF75c is limited to murine fibroblasts and cannot be observed with human PML [155], we could also detect a complete elimination of the human PML variants upon infection of primary human fibroblast with MHV-68 (Tavalai, Full, Ensser and Stamminger, unpublished). Furthermore, the entire loss of PML as a consequence of MHV-68 infection presumably also accounted for a simultaneously observed disappearance of the low mobility isoforms of Sp100, comparable to the situation in HSV-1 infected cells (see 6.1.) or after depletion of the defining ND10 factor in PML-kd HFFs [14]. At the same time, MHV-68 also induced a significant reduction in hDaxx protein levels when compared to mock-infected cells. Although this effect on hDaxx is much less pronounced in comparison to MHV-68-mediated PML downregulation, it nevertheless mimics the findings for HCMV, where we were likewise never able to detect a total depletion of the transcriptional repressor hDaxx during the HCMV replicative cycle [129]. Thus, MHV-68 negatively affects the abundance of all three major ND10 constituents directly upon infection in order to efficiently circumvent the intrinsic antiviral response instituted by ND10. Given the fact, that MHV-68 naturally infects mice, it could represent a useful model system for investigation of ND10-herpesvirus interactions in vivo.