Centrosome and retroviruses: The dangerous liaisons
© Afonso et al. 2007
Received: 13 February 2007
Accepted: 14 April 2007
Published: 14 April 2007
Skip to main content
© Afonso et al. 2007
Received: 13 February 2007
Accepted: 14 April 2007
Published: 14 April 2007
Centrosomes are the major microtubule organizing structures in vertebrate cells. They localize in close proximity to the nucleus for the duration of interphase and play major roles in numerous cell functions. Consequently, any deficiency in centrosome function or number may lead to genetic instability. Several viruses including retroviruses such as, Foamy Virus, HIV-1, JSRV, M-PMV and HTLV-1 have been shown to hamper centrosome functions for their own profit, but the outcomes are very different. Foamy viruses, HIV-1, JSRV, M-PMV and HTLV-1 use the cellular machinery to traffic towards the centrosome during early and/or late stages of the infection. In addition HIV-1 Vpr protein alters the cell-cycle regulation by hijacking centrosome functions. Enthrallingly, HTLV-1 Tax expression also targets the functions of the centrosome, and this event is correlated with centrosome amplification, aneuploidy and transformation.
Centrosomes were first described at the end of the 19th century by Theodor Boveri who had also the intuition of their central role in cell life .
Centrosomes are animal-specific non-membranous organelles that localize in close proximity to the cell nucleus for the duration of interphase. Their structure is highly conserved among higher eukaryotes. It usually consists of a pair of centrioles joined by fibers connecting their proximal ends which are embedded into a protein-dense matrix called the pericentriolar material (PCM) [2, 3]. The PCM is an ordered lattice that anchors a large number of microtubule (MT)-associated proteins, many of which bear putative coiled-coil domains, a tertiary structure known to facilitate protein-protein interactions . Centrioles are cylindrical corps formed by a radial array of nine MT-triplets, which are structurally similar to basal bodies of eukaryotic cilia and flagella [5, 6]. Centrioles play a role in the organization of the microtubular cytoskeleton, but they do not make direct contact with the MTs which nucleate from the γ-tubulin ring complexes (γ-TuRC) located within the PCM.
In animal cells, centrosomes represent the major microtubule-organizing structures (MTOC). The MTOC is responsible to direct the assembly and the orientation of MTs and to control MT-dependent processes such as trafficking of cytoplasmic vesicles and orientation of cellular organelles. At the onset of mitosis, centrosomes become the core structures of spindle poles and direct the formation of mitotic spindles. Upon cytokinesis, each daughter cell receives only one centriole, which duplicates once per cell cycle.
Centrosome duplication is tightly regulated and limited at once per cell cycle through a mechanism that prevents re-duplication. The complexes between Cyclin-dependent kinase 2 (Cdk2) and either Cyclin E or Cyclin A (Cdk2/CycE/CycA) and their substrate, nucleophosmin, have been proposed to act as licensing factors for centrosome duplication [8, 9]. However, this hypothesis has been recently challenged by Tsou and Stearns  who proposed that re-duplication is inhibited by a centrosome-intrinsic block . They suggested that the engaged/orthogonal conformation of the centrioles functions as a block for re-duplication. The cellular factor which promotes centriole-disengagement at the end of mitosis would therefore constitute the licensing factor .
Additionally several mitotic kinases of the Aurora, Polo and Nek families, which localize at least temporally to mitotic structures, have been shown to participate in the control of the centrosome cycle and mitotic spindle formation .
The tight control of centrosome number and duplication is essential for the stability of the genome. Consequently, any impairment in the regulation of centrosome number might lead to the assembly of multipolar spindles , which in turn might increase the frequency of aberrant mitosis and chromosome-segregation errors .
Recent studies demonstrated that MTOCs play a key role in cellular processes other than nucleation and organization of the MT network. Indeed, the centrosome provides a subcellular site where high local concentration of regulatory molecules in the proximity of their substrates is likely to increase the probability of specific interactions. Many regulatory molecules localize, at least temporally, at the centrosome. Thus, it has been suggested that the centrosome might act as a scaffold platform where integration of numerous cellular signaling pathways occurs, including control of cell cycle progression and completion of cytokenesis [13, 14].
As an example, following their injection into G2-arrested oocytes, centrosomes induce cell progression into mitosis . Likewise, in a Xenopus egg model, centrosomes can induce the activation of the mitosis promoting factor (MPF or Cyclin dependent kinase 1 (Cdk1) and Cyclin B complex), which is a major event in the initiation of mitosis . In mammalian cells, activation of MPF takes place at the centrosome during prophase and before any MPF-dependent H3-phosphorylation is detected in the nucleus . MPF activity is controlled by cyclin phosphorylation through the antagonistic actions of Cdc25 and Wee1, which are also regulated by phosphorylation [18, 19]. Finally, several positive and negative mitotic regulators which have distinct localization along the cell cycle, are also found at the centrosome during early mitosis (reviewed in ).
Centrosome amplification (more than three centrioles in a cell during the G1 phase) can result from different mechanisms: (a) duplication of centrosome more than once during a cell cycle, (b) failure of cell cytokenesis, (c) uncontrolled splitting of a centriole pair and (d) formation of acentriolar MTOCs . Centrosome amplification is often associated with genomic instability and therefore aneuploidy. Aneuploidy (i.e. the acquisition or loss of one or more chromosome from a diploid genome) is a very common feature of tumor cells [20–23]. After Boveri's hypothesis (the "aneuploidy hypothesis"), it has been admitted that cancer cells become aneuploid as a consequence of anomalous mitotic divisions. These defects were thought to result from centrosome amplification and transformation, and aneuploidy appeared likely to promote tumorigenesis, at least at low frequency. However, after the discovery of tumor suppressors and oncogenes, this assumption became debated [24–26].
A number of facts still support this hypothesis: first, aneuploidy frequently occurs before transformation. This is the case in a number of pre-cancerous lesions (cervix, colon, oesophagus etc.) [27–29]. Next, aneuploidy results in the de-regulation (up- or -down) of the expression of a number of genes. Third, transformation linked to aneuploidy requires several generations, which is coherent with the known incidence of cancers with age (for a review see ). Finally in silico modeling confirmed that the "aneuploidy theory" could explain how lymphocytes become transformed .
In the end, it was still technically challenging to test whether aneuploidy causes cancer or not until recently, since causing aneuploidy usually results in other cellular defects. Interestingly, in a very elegant study aimed at understanding whether aneuploidy drives tumorigenesis, contributes to tumor progression or is benign, Weaver and coll. demonstrated that aneuploidy acts both oncogenically and as a tumor suppressor : Low levels of chromosomal instability promote tumor initiation while higher levels are protective .
Foamy Virus (FV) and Human Immunodeficiency Virus type 1 (HIV-1) were reported to target the centrosome at an early phase of the viral life-cycle. Soon after entry in the host cell, both FV [37, 38] and HIV-1 [39–41] incoming virions en rout e to the nucleus accumulate in the peri-centrosomal area, in a MT and dynein-dynactin dependent manner (Figure 2).
An N-terminal coiled-coil motif within FV Gag (aa 150–180), is responsible for the centrosomal localization of the viral particles at this stage . In G0-arrested cells both FV Gag and the viral genome persist at this location for several weeks. Remarkably, FV life cycle resumes as soon as infected quiescent cells are stimulated to divide: integration and gene expression occur leading to the formation and the release of new progeny virions (Lehmann-Che submitted). Maintenance of viral capsids at the centrosome in quiescent cells could be a strategy that FVs have evolved to rapidly respond to stimuli received by the infected cell.
HTLV-1 requires cell-to-cell contact to spread efficiently. Interestingly, it has been observed that the MTOC of an HTLV-1-infected cell, which is involved in the formation of a two-cell conjugate, relocates in proximity of the adhesion site. In addition HTLV-1 genome and the viral proteins Gag and Env concentrate in the same area. Therefore, it has been suggested that polarization of the MTOC results in the orientation of many MT-plus ends towards the cell-cell junction, thus allowing the recruitment of viral Gag-containing complexes in proximity of the virological synapse where viral transmission ultimately takes place .
FV capsids form in the cytoplasm in the vicinity of the MTOC (Figure 2). Assembly depends on the integrity of the cytoplasmic targeting-retention signal (CTRS) sequence within Gag, which is responsible for the peri-centrosomal targeting of the polyprotein . The CTRS is an 18-amino acid long motif found within the MA domain of Gag, which functions as a dominant signal that directs intracytoplasmic capsid assembly of B/D-type retroviruses (such as FV and Mason-Pfizer Monkey Virus (M-PMV)) despite the presence of the bipartite membrane-targeting signal at the N-terminus of Gag .
Early work by Hunter and coll. showed that M-PMV, the prototypic D-type retrovirus, assembles at a perinuclear location , which has recently been identified as the MTOC . Specific targeting of M-PMV Gag polyproteins to the pericentriolar region is mediated by the CTRS, which interacts in a co-translational manner with the dynein-dynactin motor complexes . Nascent Gag polyproteins still associate to the polysomes, accumulate near the centrosome where both partially and fully assembled spherical capsids are visualized by electron microscopy. Gag was suggested to interact with endocytosed M-PMV Env glycoproteins trafficking into pericentriolar recycling endosomes at this subcellular location, thus allowing efficient migration of the immature capsids towards the budding sites at the plasma membrane . Of note, centrosomal targeting does not appear to be absolutely required for M-PMV assembly: Indeed, a point mutation (R55W) within the CTRS abolishes centrosomal accumulation of M-PMV Gag. Yet, Gag mutant still assembles at the plasma membrane following a type C morphogenesis pattern .
Similarly, the Gag polyprotein of Jaagsiekte Sheep Retrovirus (JSRV), another β-retrovirus localizes near the MTOC of infected cells and its centrosomal targeting is a prerequisite for the subsequent transport of JSRV virions to the plasma membrane but not for assembly . Indeed co-expression of Gag from JRSV and enJS56A1, a sheep endogenous retrovirus closely related to JSRV, results in the delocalization of JSRV Gag from the MTOC. Chimeric viral particles composed of both JSRV and enJS56A1 Gag polyproteins still assemble in the cytoplasm but they cannot reach the plasma membrane .
The centrosome has been proposed to be the subcellular site where HIV-1 Gag polyproteins are synthesized and bind to the viral genomic RNA, thus initiating encapsidation and viral particle assembly. Indeed siRNA-mediated depletion of heterogeneous nuclear ribonucleoprotein A2 (hnRNP A2), a cellular protein which expression levels regulate the nucleocytoplasmic trafficking of HIV-1 genomic RNA, almost results in the complete pericentriolar accumulation of the viral genomic RNA without affecting Gag-expression levels . HIV-1 genomic RNA and newly synthesized Gag molecules colocalize near the centrosome in a manner that depends on the presence of the packaging signal region (ψ), a high-affinity binding site for Gag within the viral RNA genome. These observations suggest that ψ acts as a targeting signal which specifically directs initiation of HIV-1 encapsidation to this subcellular domain .
The centrosome has also been suggested to represent an advantageous site for virus assembly because of the high local concentration of chaperons [52, 53] which have been shown to participate in M-PMV Gag folding and thus in viral capsid formation .
In addition, because of its role as a cellular MT-organizing center and its perinuclear localization, the centrosome represents an optimal site (1) through which incoming viruses can easily gain access to the nucleus and (2) through which newly synthesized viral components traffic to reach the viral assembly and budding sites (Figure 2). The precise mechanisms that viruses and viral components use to transit from the MTOC to the nucleus and then back are still under investigation, but it has been suggested that importins or the dynein-mediated trafficking might be involved in this process .
Early studies reported that HIV-1 14kDa Viral protein R (Vpr) was an oncogenic protein and suggested that it could be responsible for some AIDS-associated cancers . At that time, several groups reported the presence of multiple centrosomes, and consequently aneuploidy as well as micronuclei in Vpr-expressing cells . Even if it has now been clearly demonstrated that Vpr is not oncogenic per se, it still would be interesting to understand the molecular bases of such centrosomal dysfunctions.
Indeed, HIV-1 infection impairs cell-cycle progression and infected cells accumulate in G2 in vitro . Although it has been shown that several HIV-1 proteins independently block the cell cycle , Vpr is considered the major viral determinant responsible for such arrest. The G2 arrest might then allow the survival of abnormal cells, such as polyploid ones [60, 61]. Noteworthy, the amount of Vpr present in a viral particle is sufficient to induce cell cycle arrest . By mutational analysis, it has been demonstrated that the C-terminal domain of Vpr is responsible for the cell cycle arrest [63, 64]. Moreover it has been established that Vpr phosphorylation is needed for this function . Following Vpr-expression, the accumulation of G2-arrested cells correlates with the inactivation of the MPF, which normally occurs at the centrosome [57, 66, 67]. Vpr-mediated cell cycle arrest is accompanied by an hyperphosphorylation of Cdk1 [60, 66, 67]. The published results do not show any linear correlation between the amount of Vpr and the inhibition of MPF, which is consistent with the fact that Vpr does not bind directly to MPF components .
Rather, the viral-induced cell cycle dysfunctions seem to correlate mostly with an alteration of the functions of MPF regulators, such as Wee1 and Cdc25. This is supported by the finding that Vpr interacts with these proteins, at least in a yeast two-hybrid assay . Cdc25 is a component of a family of phosphatases that activate MPF by antagonizing the effects of Wee1. It has been reported that Vpr alters Cdc25 activity either directly  or through the interaction with Cdc25-inhibitors such as 14-3-3σ and PPA2 [70, 71] (Figure 3). 14-3-3 localizes at the centrosome during mitosis and inhibits the activity of Cdc25 by binding to the phosphorylated form of the protein . Vpr-binding allows 14-3-3 to interact with the unphosphorylated form of Cdc25C. This trimeric complex delays the entry into mitosis [73, 74] (Figure 3).
The prolonged cell cycle arrest induces apoptosis, which could be eventually responsible for the CD4+ depletion that is observed in vivo during AIDS progression .
In contrast to HIV-1, which is responsible for the depletion of the CD4+ cell population, HTLV-1, the etiological agent of Adult T cell leukemia/lymphoma (ATLL), causes lymphocyte transformation. Generally, lymphoma occurrence is linked to cell-cycle alterations. During the course of HTLV-1 infection, the alteration of the cell-cycle regulation is tightly linked to the expression of the viral protein Tax, which largely carries out the transforming capacity of HTLV-1. Tax can activate the NF-κB, the CREB/ATF and the SRF pathways. Both CREB/ATF and NF-κB pathways have been involved in the Tax-mediated immortalization/transformation (for a review see ).
The G1/S checkpoint regulation is tightly associated with the activation of the transcription factor E2F. In resting cells, E2F is inactive due to the formation of a complex with the Retinoblastoma protein (Rb). Once Rb is phosphorylated, either by the cyclinD-cdk4 complex at early G1, or by the cyclinD-cdk6 complex at late G1, it is degraded and E2F turns active. In Tax expressing cells, the G1/S transition is altered. A decade ago, Schmitt et al. demonstrated that the increased proliferation of Tax-expressing cells was correlated with an increased activity of both cdk4 and cdk6 . Such activation has been since demonstrated to be a consequence of direct interaction of cdk4 and cdk6 with the amino-terminal domain of Tax [78, 79]. Tax can also act directly on Rb. It binds directly to hypophosphorylated Rb and sends it to the proteasome where it is degraded .
Because it alters the MPF regulation, Tax can also disrupt the G2/M checkpoint. MPF regulation is dependent on Cdc25. The function of Cdc25 is modulated by Chk-proteins and the activity of both Chk1 and Chk2 is impaired in presence of Tax, in vitro and in vivo [81, 82]. The action of Tax is a consequence of its ability to bind directly to these latter proteins. Consequently, in HTLV-1 infected cells, Cdc25 activity is not repressed and the progression into mitosis occurs earlier than scheduled .
Altogether, these cell-cycle alterations may facilitate the accumulation of errors during mitosis therefore inducing centrosomal alterations and aneuploidy.
This hypothesis is consistent with the reports of HTLV-1-leukemic cells carrying multi-lobulated nuclei (also known as "flower cells"). These cells possess an abnormal number of centrosomes and chromosomes [84–88]. The frequency of aneuploid cells is significantly increased in acute or classical ATLL patients, even if compared to chronic ATLL [86, 89]. This is also accompanied by structural chomosomes abnormalities such as translocations and deletions. The cytogenetic abnormalities that are found in ATLL cells are not specific of these disease. They are however more frequent in the acute and lymphoma types than in chronic or smoldering types. They include various karyotypic abnormalities including translocations on chromosome 14 (14q32, 14q11), deletions of 6q but also numerical abnormalities such as trisomies 3,7 and 21 as well as monosomy of the × chromosome or loss of an Y chromosome .
As Tax is the HTLV-1 protein that alters the cell-cycle, its expression was suspected to be necessary and sufficient to induce aneuploidy and centrosome multiplication. Nitta et al. using JPX-9 cells that expresses Tax upon stimulation with cadmium  showed that Tax expression indeed allowed the accumulation of cells with an abnormal number of centrosomes : 10% of Tax-expressing cells, displayed an abnormal number of centrosome, whereas only 5% of JPX-9 untreated cells showed centrosome amplification. Interestingly the authors also showed that the micronuclei formation was consecutive to the centrosome amplification and that up to 2% of Tax-expressing JPX9 cells exhibited micronuclei at 72 h of treatment.
The altered functions of several centrosomal proteins seems also to be involved in the Tax-driven aneuploidy. As an example, HsMAD1 (also known as TXBP181) functions are impaired in Tax expressing cells  (Figure 5). HsMAD1 acts at the G2-M-checkpoint. Since HsMAD1 localizes to the centrosome during metaphase, it is tempting to speculate that the loss of HsMAD1 functions could be linked to the loss or to the modification of the centrosomal activity.
Lately, another partner of Tax, the centrosomal TAX1BP2 protein (also known as TXBP121) , was also implicated in the Tax-dependent initiation of aneuploidy  (Figure 5). Remarkably, the authors observed first at least a 5-fold difference in the number of Tax expressing JPX-9 cells that display centrosome-amplification vs. untreated JPX-9. This is significantly higher than the 2 fold increase observed by Nitta et al in the same experimental settings. Then, they demonstrated that Tax binds to and colocalizes with endogenous TAX1BP2, forming perinuclear dots. In the absence of Tax, the overexpression of TAX1BP2 leads to a reduction in the number of cells that contain supernumerary centrosomes. On the contrary, depletion of endogenous TAX1BP2 induces centrosome amplification. Therefore, Tax and TAX1BP2 have opposite effects. Besides, a Tax mutant that does not interact with TAX1BP2 can no longer induce centrosome duplication. The authors concluded that Tax targets TAX1BP2 to cause aneuploidy.
In the end, a series of questions remain to be solved: Do Tax, RanBP1 and TAX1BP2 form a complex? Alternatively, do Tax and TAX1BP2 compete for RanBP1-binding? How does Tax target TAX1BP2? What is the level of TAX1BP2 in HTLV-1 infected cells?
One important fact does not fit perfectly well with all these experimental results: ATLL development is not a rapid process but takes decades [97, 98]. We therefore believe there is an alternative and more provocative interpretation for these results. It is possible that aneuploidy occurs normally in a stochastic manner in cells which will then die. When Tax is expressed, and because of its pleiotropic effects, these aneuploid cells would then undergo a series of processes, such as p53 transcriptional inhibition for example, that would circumvent apoptosis induction. In other words, Tax would not induce aneuploidy but rather would allow aneuploid cells to survive.
Viruses have evolved different strategies to traffic within an infected cell. Active transport along the cytoskeleton networks, in particular the MTs, has been demonstrated for a series of nuclear replicating viruses such as retroviruses. For these latter, the centrosome seems to play a central role both during early and late stages of the replication cycle. It will be important to understand the functional meaning of the centrosomal localization of incoming FV and HIV-1 following infection. Is it just a mandatory/compulsory route to reach the nucleus following trafficking along the MT network, or is it (also) a transforming platform which selectively modifies the incoming viral material thus allowing a successful integration into the host genome ?
In addition, the centrosome is not a mere spectator of the cell cycle but exerts a significant control over it. By providing a scaffold for many cell cycle regulators and their activity, it influences cell-cycle progression, especially during the G1 to S-phase transition [99, 100]. To this end, this organelle receives and integrates signals from outside the cell and facilitates their conversion into cellular functions.
The role of this central organelle in retrovirus replication and pathogenicity is still mysterious and will certainly require more consideration.
Human T cell leukemia virus type 1
Human Immunodeficiency virus type 1
C-terminal Src kinase-homologous kinase
Microtubule organizing center
Adult T cell leukemia/lymphoma
Serum responsive factor
cAMP response element-binding protein
Mitosis promoting factor
Viral protein R
Mitotic arrest-defective 1
Cell division cycle
mouse embryonic fibroblasts
A2: heterogeneous nuclear ribonucleoprotein A2
cytoplasmic targeting-retention signal
Mason-Pfizer Monkey Virus
Jaagsiekte Sheep Retrovirus
PA and AZ contributed equally to the work. PA, AZ, RM and AS wrote this manuscript. AS and RM share senior authorship on this work. RM and PA thank the "Programme Transversal de Recherche 214" from the Pasteur Institute for its financial support. RM is supported by INSERM. AZ is supported by a grant from ANRS. PVA is supported by le Ministère de la Recherche & des Technologies. The authors thank Dr Mohamed Nejmeddine and Pr. Charles Bangham for providing the micrographs presented in Figure 4.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.