Sirs

One of the major questions to emerge from the completion of genome sequence from mouse and human is the functional role of subnuclear three-dimensional organization in regulating signal transduction. In eukaryotic cells, coordinating the genome response to signal activation, in time and space, might equally depend on nucleus three-dimensional properties organized throughout distinct nuclear bodies (NB) or domains (Fig. 1). It could be hypothesized that the build-up of subnuclear space might contribute to the 'reactivation' of its components in such a way that it is a coding of units compartmentalization and an embodiment of the subnuclear space 'collective memory'. As the compartmentalization of the molecules transforms the subnuclear space into a domain of new experience, the subnuclear space, in the very process of its production, might become a means of control through its multiplicity of networks. Building up NBs at the molecular level therefore implies the presence of a code and the creation of a NB–memory, and so, a prototype NB should become a means of signal coordination by means of its multiplicity of networks. One of the most intriguing examples of subnuclear structures is the promyelocytic leukemia (PML) domain or PML body1,2,3. Whereas the biochemical properties of PML remain elusive, the PML–HAT-CBP association allowed us to postulate that the CBP-enriched PML NBs might act as intermediaries in signal-mediated transcription control4. Interest in PML domains and associated NBs became especially high once their molecular integrity and distinct spatiotemporal organization were correlated with the propagation efficiencies of both adenovirus-5 and HIV-1 in living cells5,6. These data suggested that CBP-enriched PML NBs might control function and higher-order chromatin organization of both DNA and RNA viruses and associated genome formations. We proposed that PML NBs might have similar roles in the normal cell cycle2.

Figure 1: Model depicting a mechanism for a coordinate regulation of signal transduction at the cellular level throughout distinct nuclear bodies (NBs) or domains.
figure 1

a,b | The hypothetical non-activated or 'stem NBs' switch to 'activated NBs' upon signal activation. c,d | Morphological distinct clusters of NBs or differentially activated NBs might target a unique cellular structure mediating the extracellular signal to the chromatin level. The schematic representation of this model is based on our observation that both adenovirus-5 and HIV-1 target PML domains, and that PML domains exert a regulatory role on both DNA and RNA viruses propagation properties. See text for a detailed explanation.

Indeed, we have been the first to demonstrate that adenovirus-5 immediate-early viral proteins from the E1 and E4 transcription units (the E1B-55kd, E4-ORF6 and E4-ORF3 proteins) coordinate a dramatic morphological change of PML NBs which results in a relocation of cellular proteins that interact with PML to the viral inclusions bodies, the sites of adenovirus replication and late RNA transcription5. Contrary to the statement in the Highlighted references in the Review by Nisole et al. recently published in this journal7, we were the first to demonstrate that PML expresses antiviral activities by showing that expression of wild-type PML considerably suppressed adenovirus-5 replication in living cells5. Our mutation analysis suggested that the PML-activated repression of adenovirus-5 replication in vivo reflects biological activities that are probably linked to compartmentalized macromolecular complexes, such as the PML domain, rather than to its free protein markers such as the PML protein. Similarly, we showed that treatment of the cells with interferon (IFN)-β, which enhanced the expression of PML/Sp100, modified the molecular properties of PML domains, blocked the action of adenovirus on PML-NB dissociation and suppressed adenovirus replication. Together, these results supported a regulatory role of PML NBs in the early-to-late switch of adenovirus-5 replication5. Therefore, these results gave rise to our first ideas of a new theory of how viruses infect cells and ultimately utilize cell-regulatory proteins for their own purposes. Because cells have many ways to defend themselves against infection, a necessary role of the virus is to disarm protective mechanisms that might otherwise normally block activation of the viral programme. In addition, viruses should organize the takeover of the cellular know-how that is essential for viral growth and propagation. If a cellular 'machine' exists that coordinates signal-activated cellular responses — both at the transcription, DNA-processing and translation levels — then the virus should be evolutionary well organized to detect and take over the know-how of such a 'machine'. So, one of the major questions that emerged from these studies on adenovirus-5 and PML-NB functional and structural association was whether the PML domain serves as a common target for multiple viruses or whether this process is specific for the adenovirus-5 replication. To extend our observation to other viruses, we examined PML NBs and HIV-1 propagation properties in living cells. Contrary to the viral cycle of adenovirus-5, which is divided into an early and late phase relative to initiation of viral replication8,9,10, the HIV-1 propagation depends on a reverse transcription and a nuclear-import process that ends with virus integration into the host-cell genome11.

If PML NBs crosstalk both with adenovirus-5 and HIV-1 propagation properties, then the virus–host interaction at the level of PML NBs should occur at the very early stage of HIV-1 infection and prior to the virus-genome integration12. As PML NBs were crucial for both late adenovirus transcription and replication, PML NBs should be crucial for HIV-1 genome organization prior to integration. On the basis of this hypothesis, we successfully showed, using different techniques, that the HIV-1 infection activates the nuclear exit of a PML-positive structure that co-localized with the pre-integration complex of the virus6. Even though mutation analysis of the HIV-1 genome suggested that the HIV-1 Tat transactivator would not be necessary for PML NBs and HIV-1 interaction6, Tat has been shown to target PML13 and vice versa (V. Doucas, unpublished observations). In agreement with other laboratories14, we also observed that other RNA viruses such as the human T-cell leukemia virus (HTLV) target and reorganize PML NBs12. Tax is known to stimulate nuclear factor (NF)-κB as well as transcription of the HTLV promoter. We have shown that the HTLV-1 Tax viral protein targets PML NBs, and that overexpression of the PML protein can block the regulatory effects of Tax15. These results further extended our previous observations and suggested that not only DNA viruses, but also RNA tumour viruses (including HIV), must take over cellular information controlled by the PML NBs.

We speculated that the PML domain might contain protective functions for the cell that must be overcome by the virus or, alternatively, that it carries an essential cellular factor which enables viral gene expression or template replication. There is also the possibility that the PML domain reflects the properties of a cellular structure with many signal-regulated images which mediates signal activation at the chromatin level (Fig. 1). If the PML–CBP association suggested an intermediate role of PML NBs in signal transduction, the growing list of proteins that shuttle and localize to these domains, such as PML, Sp100, CBP, Daxx, HDAC/mSin3a, p53, Rb and Notch/MAML1, established that the molecular integrity of the PML NBs depends on the extracellular signal2. Although answers to these profound questions are not yet completely known, it seems very probable that both DNA and RNA viruses target the same cellular structure, which might reflect a mechanism of fundamental importance to virology with potential implications for normal cell function.