Nature Structural & Molecular Biology11, 1114 - 1121 (2004)
Published online: 3 October 2004; | doi:10.1038/nsmb837
PML bodies control the nuclear dynamics and function of the CHFR mitotic checkpoint protein
Matthew J Daniels, Alexander Marson
& Ashok R Venkitaraman
University of Cambridge, Cancer Research UK Department of Oncology and The Medical Research Council Cancer Cell Unit, Hills Road, Cambridge CB2 2XZ, UK.
Correspondence should be addressed to Ashok R Venkitaraman arv22@cam.ac.uk
Nuclear foci containing the promyelocytic leukemia protein (PML bodies), which occur in most cells, play a role in tumor suppression. Here, we demonstrate that CHFR, a mitotic checkpoint protein frequently inactivated in human cancers, is a dynamic component of PML bodies. Intermolecular fluorescence resonance energy transfer analysis identified a distinct fraction of CHFR that interacts with PML in living cells. This interaction modulates the nuclear distribution and mobility of CHFR. A trans-dominant mutant of CHFR that inhibits checkpoint function also prevents colocalization and interaction with PML. Conversely, the distribution and mobility of CHFR are perturbed in PML-/- cells, accompanied by aberrations in mitotic entry and the response to spindle depolymerization. Thus, PML bodies control the distribution, dynamics and function of CHFR. Our findings implicate the interaction between these tumor suppressors in a checkpoint response to microtubule poisons, an important class of anticancer drugs.
The PML protein is a target of the t(15;17) chromosomal translocation typical of promyelocytic variants of acute myeloid leukemia, which fuses PML reciprocally with retinoic acid receptor (RAR)1. Overexpression of PML in cancer cell lines induces growth arrest and apoptosis2. PML-RAR fusion proteins that interfere with the function of endogenous PML in a trans-dominant manner1,
3 cause leukemia in transgenic mice, whereas PML-/- mice develop a range of cancers including papillomas, carcinomas and lymphomas after exposure to carcinogens4; this is paralleled by the loss of PML in human cancers from diverse tissues5. Thus, PML is a potent tumor suppressor in vitro and in vivo, in many cell types.
PML accumulates in focal nuclear structures termed PML bodies2, with an average size of 1 m, which vary in number from 10−30 per nucleus. PML bodies are not formed in PML-/- cells4,
6, or in cells that express PML-RAR fusions1. Their disappearance is correlated with abnormalities in cell growth and death through effects on transcriptional regulation, as well as p53-dependent and independent apoptosis7. Because PML bodies contain several different molecules2, including the proapoptotic factor DAXX and the tumor suppressors p53, Rb, Chk2 and BRCA1, it has been proposed that they may serve as repositories for protein storage or modification8. However, the variable movement of proteins to and from PML bodies predicted by this hypothesis has yet to be directly demonstrated.
We investigated the intracellular trafficking of the checkpoint protein CHFR and report here its hitherto unrecognized interaction with PML bodies. CHFR, a RING- and FHA-domain-containing nuclear protein, is inactivated by epigenetic mechanisms in up to 30% of epithelial cancer cell lines9. CHFR inactivation abrogates the transient delay in passage through early mitotic prophase induced by microtubule-disrupting agents9. We find that the nuclear distribution and mobility of CHFR depend on its movement through PML bodies, revealing novel aspects of their putative role as depots for nuclear proteins. This regulation is perturbed by a trans-dominant inhibitory mutant of CHFR or in PML-/- cells, and is accompanied by abnormalities in mitotic progression and the response to microtubule depolymerization. Our findings identify a novel interaction between two tumor suppressor proteins that provides insight into the biology of PML bodies, and has implications for cancer therapy.
Results Alterations in GFP-CHFR distribution during the cell cycle Green fluorescent protein (GFP) was fused in-frame to the N terminus of human CHFR (Fig. 1a) to yield a GFP-CHFR fusion protein with a predicted molecular mass of 110 kDa (Fig. 1b). A similar GFP-CHFR fusion protein has been used previously, and can reconstitute function in CHFR-deficient cells10. We used it to examine the intracellular distribution of CHFR, because antibodies for fluorescent staining of the endogenous human protein have not yet been reported, nor have we been able to raise them. To provide evidence that GFP-CHFR behaves similarly to the endogenous molecule, we cloned the CHFR ortholog from Xenopus laevis (xCHFR), and raised an antibody against it that detects a single band of the appropriate size in cell extracts (Fig. 1c). Staining of endogenous xCHFR in the XR1 cell line during interphase (Fig. 1d) revealed a punctate pattern of nuclear distribution very similar to GFP-CHFR in HeLa cells (Fig. 1e). Moreover, endogenous x-CHFR and transfected GFP-CHFR colocalized extensively in XR1 cells (Fig. 1d). Collectively, these findings provide evidence that GFP-CHFR accurately reflects the behavior of endogenous CHFR.
(a) Schematic of the GFP-CHFR fusion protein highlighting the FHA, RING-finger and cysteine-rich (CR) domains. (b) An anti-GFP western blot of extracts from HeLa cells expressing the fusion proteins used here. The GFP CHFRFHA protein (relative molecular mass (Mr), 85 kDa) lacks the first 111 residues of CHFR. Lane 1, GFP-CHFR; lane 2, GFP-CHFRFHA; lane 3, PML-YFP. (c) Specificity of the antibody against X. laevis CHFR (X-CHFR). The antibody was used to detect X-CHFR by western blotting in control reticulocyte lysates without added cDNA (lane 1), in vitro−translated X-CHFR cDNA (lane 2), or oocyte extracts (lane 3). (d) Localization of endogenous X-CHFR in foci, and colocalization with transfected GFP-CHFR, in interphase XR1 cells. Results shown are typical of >50 cells microinjected with a plasmid encoding GFP-CHFR. Images are superimposed in the merge (DNA, anti-X-CHFR, GFP-CHFR) and two-pixel quantitative colocalization (anti-X-CHFR, GFP-CHFR) panels, with DNA in blue, anti-X-CHFR staining in red and GFP-CHFR in green. Areas of colocalization are yellow-green. Scale bar, 5 m. (e) Cell cycle−dependent localization of GFP-CHFR in HeLa cells. GFP-CHFR is green, DNA stained with DAPI is blue, and kinetochores visualized with CREST antibody are red. Scale bar, 5 m. (f) Localization of GFP-CHFR in interphase counterstained with DAPI (blue). Localization in large foci, small foci and diffuse staining in the nucleoplasm are visible. Scale bar, 5 m. (g) Numbers of large and small foci were enumerated per cell (n = 50). (h) Colocalization of GFP-CHFR with endogenous PML. Cells transfected with GFP-CHFR (green) were fixed and counterstained with anti-PML (red). Colocalization in large foci (PML bodies) appears yellow in the merge image. Scale bar, 5 m. (i) Anti-GFP or anti-PML immunoprecipitates from GFP-CHFR-transfected cells were blotted with monoclonal anti-PML. GFP-CHFR preferentially associates with the PML band corresponding to the predicted migration of PML-(SUMO)3. Anti-PML immunoprecipitates contain multiple PML-SUMO isoforms. As a negative control, the same immunoprecipitations were done using cells transfected with GFP alone (Supplementary Fig. 2 online).
In HeLa cells during interphase, GFP-CHFR accumulates, primarily in large and small focal structures that are excluded from the nucleolus, with a small amount distributed through the nucleoplasm. Coincident with mitotic prophase, this pattern changes (Fig. 1e). Many of the foci containing GFP-CHFR disperse, and the diffusely distributed protein surrounds but is excluded from the chromosome territories. A few small foci persist over chromatin. This mixed pattern continues after nuclear envelope breakdown and the progressive condensation of chromosomes during the remainder of mitosis. Similar results were obtained in SaOS2, HCT116 and 293T cells. So, the intracellular distribution of GFP-CHFR is regulated during progression through the cell cycle.
Large GFP-CHFR foci coincide with PML bodies GFP-CHFR foci in interphase cells range in size (Fig. 1f) from small (average diameter 0.31 0.065 m) to large (average diameter 0.77 0.1 m). Small foci occur (Fig. 1g) in about four-fold greater numbers (large foci, 6.9 4.4 (range 0−13) per nucleus; small foci, 34 9.4 (range 25−51) per nucleus). Similarities in the size and distribution of the large foci with the reported characteristics of PML bodies11 prompted us to examine the colocalization of the PML and CHFR proteins (Fig. 1h). Endogenous PML bodies revealed with an antibody against PML coincide with the large GFP-CHFR foci. Moreover, the SUMO-modified form of PML, which occurs in PML bodies and is required for their organization12, preferentially coimmunoprecipitates with GFP-CHFR (Fig. 1i).
Interaction of CHFR with PML by intermolecular FRET We established a method for intermolecular fluorescence resonance energy transfer (FRET) to investigate the physical interaction between CHFR and PML in living cells (Fig. 2a). FRET is the radiationless transfer of energy between a pair of suitably oriented, spectrally overlapping fluorophores in close physical proximity13. During FRET, the donor fluorophore's emission is quenched whereas the acceptor fluorophore's emission is stimulated. In our experiments, a CFP-CHFR fusion protein (whose intracellular distribution is indistinguishable from GFP-CHFR) acted as the FRET donor. The FRET acceptor was a PML-YFP fusion. Negative control experiments using cells transfected with a single fluorophore are shown in Supplementary Figure 1 online.
Figure 2. Intermolecular FRET between CFP-CHFR and PML-YFP.
(a) A schematic of intermolecular FRET. CFP excited with 405-nm light usually emits at 470 nm (no FRET, left). If correctly oriented CFP and YFP fluorophores are within nanometer proximity, FRET occurs, quenching the usual 470-nm emission from CFP, but exciting 535-nm emission from YFP (FRET, right). Note that 405-nm light cannot directly excite YFP (Supplementary Fig. 1 online). (b) Cells cotransfected with CFP-CHFR and PML-YFP were fixed and visualized using 405-nm laser light. Emission spectra for CFP (green) and YFP (red) are shown, as is the merged image. Scale bar, 5 m. (c) YFP was irreversibly photobleached with 514-nm light in the areas bound by the white line, before visualization of CFP and YFP emission excited by 405-nm light. Scale bar, 5 m. (d) A pseudo-colored FRET image (right) shows the ratio of donor (CFP) emissions before (left) and after (middle) YFP photobleaching, according to ref. 12. Scale bar, 5 m. The pseudo color scale marks the FRET ratio from 0 (blue) to 6.0 (red). The image depicts the magnitude of interaction between donor and acceptor fluorophores within PML bodies, and also reveals the geography of that interaction as heterogeneities in the pseudo-colored FRET ratios. Results are typical of at least three independent experiments.
Intermolecular FRET between CFP-CHFR and PML-YFP was demonstrated first in fixed interphase HeLa cells using 405-nm light, which was sufficient only to excite CFP. In the CFP-CHFR prebleach image in Figure 2b, large CHFR-containing foci are not visible, although small foci are clearly seen. By contrast, large foci are clearly visible in the PML-YFP image but small foci are not. In other words, donor (CFP) quenching and sensitized acceptor (YFP) emissions occur only in the large foci, demonstrating that FRET between CFP-CHFR and PML-YFP occurs exclusively in these structures—that is, in PML bodies.
To confirm the specificity of intermolecular FRET, PML-YFP was irreversibly photobleached (Fig. 2c) before excitation of CFP-CHFR with 405-nm light. Here we observed the reverse of what is seen in Figure 2b. CFP-CHFR is clearly visible in the large foci (PML bodies) within the photobleached area, but PML-YFP is not. The loss of donor quenching and of sensitized acceptor emissions after photobleaching confirms the authenticity of FRET.
The endpoint of FRET imaging13, a ratio of donor emissions before and after acceptor photobleaching, was rendered as a pseudo-colored image (Fig. 2d). FRET ratios of 6.0 occur within the large foci (PML bodies) where CFP-CHFR and PML-YFP are in close physical proximity. Thin peripheral rims with a lower FRET ratio border the PML bodies, representing regions in which decreased FRET interaction might arise as a consequence of altered intermolecular orientation, displacement or stoichiometry at the periphery of the PML body.
Our observations reveal a novel feature of the organization of PML bodies. These structures are not homogenous in terms of the interaction between CFP-CHFR and PML-YFP. Instead, they consist of a central core and peripheral rim, which are demarcated as adjacent areas with different FRET ratios. This illustrates an application of FRET imaging for the interrogation of subcellular structures with high resolution.
Intermolecular FRET in living cells These results were validated and extended using intermolecular FRET in unfixed, living HeLa cells. Before acceptor photobleaching (Fig. 3a), quenching of the CFP-CHFR donor rendered large foci (PML bodies) less visible, whereas sensitized acceptor emissions through FRET made these structures visible in the adjacent PML-YFP image. This was reversed after acceptor photobleaching in a single large focus (PML body) (Fig. 3b), confirming that specific intermolecular FRET occurs in these structures. The heterogeneity in the interaction between CFP-CHFR and PML-YFP is not so apparent in the pseudo-colored image (Fig. 3c). Indeed, the apparent FRET ratio is considerably lower than in fixed cells, because of limitations imposed in living cells for imaging this FRET pair.
Figure 3. Intermolecular FRET between CFP-CHFR and PML-YFP in living cells.
(a) Cells cotransfected with CFP-CHFR and PML-YFP were visualized using 405-nm laser light. Emission spectra for CFP (green) and YFP (red) are shown, as is the merged image. Scale bar, 5 m. (b) 514-nm light was used to photobleach a single PML-YFP focus (white ring), before visualization of CFP and YFP emissions excited by 405-nm light, as described above. Scale bar, 5 m. (c) A pseudo-colored FRET image (right) shows the ratio of donor (CFP) emissions before (left) and after (central) YFP photobleaching. Scale bar, 5 m. The pseudo color scale marks the FRET ratio from 0 (blue) to 6.0 (red). Results are typical of at least three independent experiments. Comparison of the FRET ratios for the interaction between CFP-CHFR and PML-YFP in living and fixed cells (n = 10 for each) shows three-fold lower FRET in living cells (data not shown).
Traffic of GFP-CHFR to and from PML bodies Several lines of evidence (Figs. 1f,g, 2d and 3c) establish that a physical interaction between CHFR and PML occurs only in large foci (PML bodies). But CHFR also accumulates in small nuclear foci distinct from PML bodies. To define the relationship between these structures, we carried out three-dimensional reconstructive imaging of GFP-CHFR and endogenous PML (stained with anti-PML) in fixed HeLa cells. Approximately 60 images were take through a single cell (sufficient oversampling to ensure that each pixel was recorded in at least four focal z-planes), and the resultingimage set projected as a single three-dimensional composite (Fig. 4a).
Figure 4. Movement of small GFP-CHFR-containing foci to and from PML bodies.
(a) Cells expressing GFP-CHFR (green) were fixed and stained with anti-PML (red). Each image shows a single composite projection rendered from 60 individual z-planes, and highlights a single PML body. Regions of colocalization appear yellow in the merged images. GFP-CHFR-containing appendages, devoid of PML, occur on the surface of PML bodies. Scale bar, 1 m. (b) Time-lapse images show that small GFP-CHFR-containing foci apparently approach and coalesce with PML bodies (red circles), and also appear to be extruded from them (blue circles). Composite projections from serial z-plane images through the entire nucleus at each time point indicate that most CHFR-containing foci do not migrate out of the plane of focus (Supplementary Video 1 online). Results are typical of at least three independent experiments.
Notably, small CHFR foci occur not only in the nucleoplasm, but also in the form of appendages on the surface of PML bodies. Comparison of the FRET images in Figures 2 and 3 suggests that CHFR and PML do not interact in the paired appendages. On average, 4.2 1.3 pairs of GFP-CHFR foci decorate any given PML body, accounting for 20% of all small CHFR foci.
A dynamic interplay between GFP-CHFR appendages and PML bodies is apparent in time-lapse images of GFP-CHFR (Fig. 4b). Small GFP-CHFR-containing foci approach and fuse with PML bodies (Fig. 4b, 0−135 s). Conversely, paired GFP-CHFR-containing foci can apparently be extruded from PML bodies over time (Fig. 4b, 190−250 s). Similar fusion and extrusion events are observed when serial, 1 m z-plane images through entire cells at each time point are projected into a single plane (Supplementary Video 1 online), suggesting that the events truly occur at the surface of PML bodies, and are not simply the result of changes in the plane of focus of the foci. Collectively, these observations suggest that deposition and loading of CHFR from or into the smaller foci takes place on the surface of PML bodies, consistent with the altered interaction between CHFR and PML at the periphery of PML bodies revealed by FRET analysis (Fig. 2d).
Behavior of a trans-dominant inhibitory CHFR mutant A CHFR deletion mutant (CHFRFHA) lacking the N-terminal FHA domain (Fig. 1a) acts as a trans-dominant inhibitor of endogenous CHFR function9. When expressed in CHFR-proficient cells, this mutant abrogates the transient, CHFR-dependent delay in entry into mitosis triggered by microtubule poisons such as nocodazole. To explore the correlation between the localization of CHFR and its function, we investigated the effects of this trans-dominant mutant on GFP-CHFR distribution.
Myc epitope-tagged CHFRFHA was coexpressed in HeLa cells with the GFP-CHFR fusion protein. Coexpression prevents the typical focal accumulation of GFP-CHFR in interphase cells (Fig. 5a). Neither large nor small foci occurred. Instead, both proteins assumed a diffuse nuclear distribution. Thus, inhibition of CHFR function by the CHFRFHA mutant is accompanied by the loss of localization in nuclear foci.
Figure 5. CHFRFHA prevents recruitment of CHFR, but not of PML, into PML bodies.
(a,b) Cells cotransfected with Myc-epitope-tagged CHFRFHA, and either GFP-CHFR (a) or PML-GFP (b) were fixed and stained with anti-Myc. Myc-CHFRFHA expression prevents GFP-CHFR from localizing to PML bodies (a), but has no effect on the accumulation of PML-GFP in these structures (b). Scale bar, 5 m.
However, the CHFRFHA mutant had no effect on the distribution of PML (Fig. 5b). In cells coexpressing PML-GFP with the myc-tagged CHFRFHA mutant, PML-GFP continued to accumulate in characteristic PML bodies. This implies that the trans-dominant CHFRFHA mutant works to prevent the localization of CHFR to PML bodies and not vice versa. Corroborating this conclusion, CHFRFHA did not coimmunoprecipitate with PML or SUMO-modified forms of PML (data not shown). Collectively, our findings suggest that CHFR recruitment to PML bodies requires its FHA domain, a known phosphopeptide-binding motif, and—because the CHFRFHA mutant suppresses CHFR function as a dominant-negative9—that this recruitment is relevant to biological function.
Control of CHFR dynamics by PML To investigate how the PML-CHFR interaction might regulate function, we used fluorescence recovery after photobleaching (FRAP) to determine the dynamics of movement of GFP-CHFR and the GFP-CHFRFHA mutant between PML bodies and the nucleoplasm (Fig. 6). Cumulative results of 30 experiments are tabulated (Table 1). Fluorescence recovery of GFP-CHFR in the nucleoplasm exceeds 90% of the prebleach intensity, indicating that the molecule is highly mobile. This behavior is shared with the GFP-CHFRFHA mutant, which localizes exclusively to the nucleoplasm. In contrast, GFP-CHFR localized to PML bodies has restricted mobility. Over the 20-s time frame of our measurements, fluorescence recovery is no more than 50% of the prebleach intensity, indicating that at least half of the GFP-CHFR in PML bodies is in an immobile pool. The retardation of GFP-CHFR mobility in PML bodies is further substantiated by the reduced coefficient of diffusion. Thus, our results clearly demonstrate that the PML-CHFR interaction reduces the mobility of—or, in other words, sequesters—GFP-CHFR in PML bodies.
FRAP analysis of living HeLa cells transfected with the indicated constructs. Fluorescence intensity was measured in the regions ringed in white before (prebleach) or at different times after photobleaching to calculate flourescence recovery. Scale bar, 5m.
The dynamic behavior of GFP-CHFR sequestered in PML bodies is not identical to that of PML or SUMO-modified PML in these structures. The fluorescence recovery of YFP-SUMO1 or of PML-GFP fusion proteins in PML bodies (14% or 4%, respectively) is far lower than that of GFP-CHFR (50%) there (Table 1). This is true whether just half of a PML body is photobleached, or whether an entire one is (data not shown), indicating that PML-GFP in PML bodies does not undergo exchange either with the nucleoplasm, or within the structure itself. We infer from these findings that PML or SUMO conjugates within PML bodies are immobile, whereas 50% of CHFR in PML bodies is mobile and can undergo exchange with other compartments. By distinguishing the dynamics of structural components of PML bodies from that of a protein that traffics through them, our findings provide new insight into the role ascribed to PML bodies as repositories for protein deposition and mobilization during biological processes.
Distribution and dynamics of CHFR in PML-/- cells The trans-dominant inhibitory mutant, CHFRFHA, prevents the interaction of CHFR with PML but does not perturb PML body integrity (Fig. 5), suggesting that the traffic of CHFR through PML bodies is necessary for function. This prompted us to examine the distribution and dynamics of CHFR in primary cultures of PML-/- murine embryo fibroblasts (MEFs), in which PML bodies are absent4. MEFs from litter mate, PML+/+ embryos were used as controls.
In PML-/- cells, GFP-CHFR did not accumulate in foci in the nucleus of interphase cells (Fig. 7a), but instead was diffusely distributed in the nucleoplasm. Analysis by FRAP reveals that nucleoplasmic GFP-CHFR in PML-/- cells recovers to >90% of its prebleach intensity (data not shown), suggesting that it resides in a freely mobile pool. The distribution and dynamics of GFP-CHFR in PML-/- cells closely resemble those of the GFP-CHFRFHA mutant, which does not interact with PML, in both PML-/- and control cells. These findings confirm that the PML-CHFR interaction in PML bodies is required for the control of CHFR distribution and dynamics.
Figure 7. Perturbed CHFR localization and mitotic checkpoint abnormalities after PML disruption.
(a) PML+/+ or PML-/- MEFs transfected with either GFP-CHFRFHA or GFP-CHFR were fixed and visualized. CHFR focus formation is impaired in PML-/- MEFs. Scale bar, 5 m. (b) Enumeration of mitotic stages (G2−Pro, cells in late G2 phase or prophase, with intact nuclear membranes; Prometa, prometaphase; Meta, metaphase; Ana, anaphase; Telo, telophase) in asynchronously dividing PML+/+ or PML-/- MEFs. The frequency of cells in late G2−prophase is elevated in PML-/- cells. Over 4,000 cells were enumerated in each of two independent experiments. (c) Cells in different mitotic stages were enumerated before (0), and at 90 and 270 min after, treatment with 10 M nocodazole as described14. Cells in metaphase, anaphase or telophase were grouped together as 'late mitosis' for convenience. In contrast to PML+/+ cells, which show an arrest in late G2−prophase, PML-/- cells breach the CHFR-dependent checkpoint to accumulate in prometaphase. Their accumulation in prometaphase indicates that the spindle assembly checkpoint triggered by nocodazole, not dependent on CHFR, is unaffected by PML disruption. Over 4,000 cells were enumerated in each of two independent experiments.
Mitotic abnormalities in PML-/- cells CHFR-deficient cells, or cells transfected with the CHFRFHA mutant that does not interact with PML, do not exhibit the transient delay in mitotic entry that occurs in wild-type cells after challenge with agents that disrupt the mitotic spindle9. So, we tested whether PML bodies are also required for this response. The mitotic distribution of asynchronously growing PML-/- and control cells was first determined (Fig. 7b). Cells with condensed chromosomes but an intact nuclear envelope characteristic of the late G2−prophase transition were over-represented in PML-/- cells as compared with wild-type controls, suggesting that prophase entry, duration or exit might be affected by PML disruption. Indeed, the CHFR-dependent checkpoint response is known to regulate this phase, delaying entry into mitosis after exposure of cells to microtubule poisons9. To test the effect of PML disruption on prophase entry more directly, we used previously described methods14. In contrast to control cells, which progressively accumulate in late G2−prophase with an intact nuclear membrane after exposure to depolymerizing concentrations of nocodazole, PML-/- MEFs continue to enter mitosis and accumulate in prometaphase, consistent with spindle assembly checkpoint arrest (Fig. 7c). Together, these observations indicate that PML is required to enforce the early, CHFR-dependent checkpoint that delays entry into mitosis after spindle disruption, but not for the later prometaphase arrest that is independent of CHFR.
Discussion We report here that the intranuclear distribution and dynamics of CHFR are controlled by its interaction with PML, identifying a new mechanism in the mitotic checkpoint response to microtubule poisons, and providing insight into the nature and function of PML bodies. We used a GFP-CHFR fusion to track the intracellular distribution of CHFR. Antibodies for fluorescent staining of the endogenous human protein have not been reported, and our own attempts to raise rabbit anti-human CHFR antisera have not so far yielded a reagent that works for fluorescent staining. However, several lines of evidence indicate that GFP-CHFR reliably reports the behavior of the endogenous protein. Antibody staining of the X. laevis ortholog of CHFR in X. laevis cells reveals accumulation in large and small foci closely similar to GFP-CHFR distribution in human cells, as well as colocalization of endogenous CHFR with transfected GFP-CHFR. In transfected human cells, this focal distribution is not observed with other GFP fusion proteins, or with GFP alone. Deletion of the FHA domain of CHFR is sufficient to prevent focal accumulation. Moreover, overexpression of the CHFRFHA mutant blocks the transport of coexpressed GFP-CHFR to foci in a trans-dominant fashion, consistent with the known behavior of this mutant as a functional 'dominant negative.' Collectively, these findings suggest that the focal distribution of GFP-CHFR is relevant to the biological properties and function of the endogenous protein.
Large CHFR-containing foci are coincident with PML bodies, and the interaction between these proteins occurs only in these structures, involving a distinct pool of CHFR. Intermolecular FRET between CFP-CHFR and PML-YFP occurs exclusively in PML bodies, and not in small nuclear foci that contain CHFR, or in the nucleoplasmic pool. FRAP experiments indicate that 50% of CHFR in PML bodies undergoes exchange with other cellular pools. In fact, small CHFR-containing foci move to and from PML bodies. These observations suggest that intracellular CHFR cycles through PML bodies, with a substantial fraction resident in these structures at any given time.
Traffic through PML bodies is essential for CHFR's biological function. Perturbation of the CHFR-PML interaction by the trans-dominant CHFRFHA mutant, or in PML-/- cells, not only alters the distribution and dynamics of CHFR, but is also accompanied by an inability to enforce the CHFR-dependent checkpoint response to microtubule disruption. Protein activation in PML bodies putatively occurs via post-translational modifications such as acetylation, phosphorylation or SUMO conjugation8. Indeed, CHFR undergoes phosphorylation in mitotic cells15. We have not yet been able to detect such modifications on Myc-epitope-tagged CHFR or a GFP-CHFR fusion protein during interphase (data not shown), either because of the limitations of our experimental system, or because CHFR is not modified during its transit through PML bodies.
Recruitment of CHFR to PML bodies requires its N-terminal FHA domain, a phosphopeptide-binding module that recognizes motifs surrounding modified serine or threonine residues. Our findings suggest that phosphorylated PML may itself be the anchor for CHFR recruitment, because intermolecular FRET shows the molecules to be in close physical proximity in PML bodies. Indeed, PML can be phosphorylated on tyrosine and serine residues16. Notably, the trans-dominant CHFRFHA mutant not only fails to interact with PML but also prevents the recruitment of GFP-CHFR to PML bodies, indicating the formation of a CHFR−CHFRFHA heterodimer. Dimerization may reflect the fact that CHFR is active as an E3 ubiquitin ligase through its RING domain10,
15,
17, because these enzymes usually comprise hetero- or homodimers of RING-containing proteins18. PML itself contains a RING domain1, prompting speculation that the PML-CHFR interaction might have enzymatic activity.
Notably, FRET analysis reveals heterogeneity in the interaction between CHFR and PML within PML bodies. A central core region with a high FRET ratio is bounded by a thin peripheral rim in which the FRET interaction is diminished. Indeed, the movement of small CHFR-containing foci to and from PML bodies involves the peripheral rim, suggesting that the PML-CHFR interaction might be structurally or quantitatively altered in this region. Whether these differences are reflected in visible ultrastructural features of the organization of PML bodies into a dense core with a central aperture19 is not known, but our findings suggest that this warrants further study.
We find using FRAP analysis that two structural components in PML bodies (PML itself, and PML body−bound SUMO conjugates) are comparatively immobile, in that only 5−15% of these molecules can be exchanged with surrounding compartments. In contrast, at least one dynamic component, CHFR, is relatively more mobile. Our observations provide insights into the role of PML bodies as sites for protein storage or modification. They suggest that dynamic and structural components of PML bodies are neither stoichiometrically nor permanently associated with one another, but instead are in a dynamic equilibrium. This feature could facilitate the movement of proteins like CHFR to and from PML bodies. In this light, the relative immobility of PML and SUMO conjugates in the PML body is notable, given that they are believed to work directly in protein storage or modification. Our findings suggest that the regulated mobilization of PML and/or SUMO by cellular signals might be essential to induce protein release or modification.
The intranuclear distribution and dynamic properties of GFP-CHFR are altered in PML-/- cells, providing an independent line of evidence to support our observations on the CHFRFHA mutant. Notably, cell cycle progression in PML-/- cells is subtly perturbed, in that late G2−prophase cells are more abundant, suggesting alterations in mitotic entry or prophase progression. This is not a reported feature of CHFR deficiency9, raising the possibility that it could reflect a function of PML that is independent of CHFR.
CHFR is the only known component of the checkpoint mechanism that delays entry into mitosis in response to microtubule poisons, an important and widely used class of anti-cancer drugs. The checkpoint is frequently inactivated in human epithelial cancers, by mutations or epigenetic silencing of CHFR20,
21,
22,
23. We report here that PML-/- cells fail to exhibit the delay in mitotic entry after microtubule disruption that characterizes this CHFR-dependent checkpoint, indicating that PML also participates in the mechanism. Moreover, the CHFRFHA mutant — which suppresses the checkpoint in a trans-dominant manner — not only fails to accumulate in PML bodies, but also prevents CHFR from doing so. Thus, multiple lines of evidence indicate that the PML-CHFR interaction is essential for this checkpoint. It will therefore be important to ascertain whether alterations in the expression of PML, or mutations which affect its ability to control CHFR, also compromise the checkpoint in human cancer cells, influencing their sensitivity to treatment with microtubule-disrupting drugs. This is particularly pertinent in light of the observation that PML is frequently inactivated in human epithelial cancers from diverse tissues5.
Methods cDNA cloning. cDNAs encoding full-length CHFR and the CHFRFHA mutant were isolated by S. Anand (Hutchison/MRC Center) through RT-PCR reactions on RNA prepared from 293T cells using RNAzol. The oligonucleotide primers used for RT-PCR incorporated EcoRI (5') and XhoI (3') restriction sites for cloning into the vector pCS2MT, thereby adding 6x in-frame Myc epitope tags to the 5' end. The cDNAs (without Myc tags) were subcloned as EcoRI (5') to XbaI (3') fragments into pEGFP-C1 (Clontech). The primers used were (Forward) 5'-ACGCGTGAATTCCATGGAGCGGCCCGAGGAAG-3' and (Reverse), 5'-TGCAAGATCTCTCGAGTTAGTTTTTGAACCTTGTCTGTTCACAGATA-3', for full-length CHFR. The same Reverse primer was paired with (Forward) 5'-ACGCGTGAATTCCATGGAACCGGAACACAACGTGGCATACCTC-3' for CHFRFHA. CFP and YFP were subcloned in place of EGFP between NheI and EcoRI sites. PML-YFP-N1 was a gift from T. Rich. YFP was replaced with GFP from pEGFP-N1 (Clontech) via BamHI and NotI restriction sites. Constructs were verified by nucleotide sequencing.
Xenopus laevis CHFR isolation and analysis. To clone a cDNA encoding x-CHFR, the National Center for Biotechnology Information database of X. laevis ESTs was probed with the protein sequence of human CHFR using the translated BLAST algorithm. RT-PCR was used to assemble a full-length clone from X. laevis RNA prepared from embryos, supplemented by the RACE method to complete the 5' end. The sequence and properties of the protein will be reported separately. A rabbit antiserum was raised against residues 277−420 expressed as a glutathione S-transferase fusion protein, and affinity-purified against antigen. The affinity-purified antibody was validated in western blotting against oocyte extract, using as a positive control in vitro translated protein prepared using the TnT reticulocyte lysate system (ProMega) from full-length x-CHFR cDNA. XR1 cells24, a gift from W.A. Harris (University of Cambridge), were transfected with the GFP-CHFR construct by DNA microinjection for colocalization experiments. Antibody staining and microscopic imaging were done according to the methods below.
Transfection. Cell lines or primary MEFs were transfected with Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions and analyzed by western blot, immunofluorescence or live cell microscopy at 36 h after transfection.
Antibodies. Anti-PML mouse monoclonal (PG-M3) antibody was purchased from Santa Cruz Biotechnology and used at 1:200 for western blot, immunoprecipitation and immunofluorescence. Purified anti-GFP polyclonal (ab290-50, Abcam) antiserum was used for immunoprecipitation, and anti-GFP monoclonal (8362-1, Clontech) for western blot. Monoclonal anti-Myc (9E10) was purchased from Santa Cruz Biotechnology and used at 1:1,000 for western blot and immunofluorescence. Fab fragment secondary antibodies for immunofluorescence were purchased from Molecular Probes and used at 1:200.
Immunofluorescence. Cells grown overnight on 0-thickness glass coverslips (VWR) were fixed for 5 min at room temperature in 4% (w/v) formaldehyde, washed with TBS containing 0.1% (v/v) Triton-X100 and blocked in AbDil (2% (w/v) BSA, 0.1% (w/v) sodium azide in TBS with 0.1% (v/v) Triton X-100). AbDil was used to dilute primary and secondary antibodies, before incubation at room temperature for 3 h or 1 h, respectively. Slides were mounted in Vectashield with or without DAPI.
Immunoprecipitation. Transfected cells (3 106) were lysed in 150 mM NaCl, 0.1% (v/v) NP40, 100 M sodium orthovanadate, 100 M PMSF, 1 mM DTT, on ice. Cell debris was pelleted by centrifugation before the cleared supernatant was aliquoted for immunoprecipitation with 2 g of anti-PML, anti-GFP polyclonal or control rabbit IgG. Antibody capture was done with protein G−Sepharose beads (Sigma), which were washed extensively in lysis buffer before elution of immunoprecipitates into SDS-PAGE loading buffer.
Microscopy. All images were acquired on a Zeiss Axiovert 200M inverted microscope equipped with a Zeiss 510 META confocal head, using Zeiss software. Three-color confocal images were obtained by laser excitation using appropriate band-pass filters. For live cell imaging, a heated stage unit (Heating Insert P, Zeiss) was used to incubate 1-thickness borosilicate chambered coverglass slides (Nunc) layered with cells growing in Liebowitz L-15 medium without phenol red (Invitrogen) supplemented with 10% (v/v) fetal calf serum. Two-pixel images showing quantitative colocalization of X-CHFR with GFP-CHFR were generated using a standard function in the Zeiss LSM510 software. Briefly, a bivariate scatter plot was produced by the software, displaying fluorescence intensities for each channel for each image pixel, before threshold determination and quantitative colocalization analysis were done as described25 to generate the colocalization image.
FRET. FRET analysis by acceptor photobleaching was carried out as described26, with slight modifications. Briefly, HeLa cells transfected with either CFP-CHFR or YFP-PML alone were plated onto coverslips and emission spectra obtained (data not shown). Cells doubly transfected with both fusion proteins were used for acceptor photobleaching hetero-FRET experiments. Lambda-mode excitation using 405-nm light from a blue diode laser at 9.9% laser power was used to extract spectral emissions for CFP and YFP. FRET efficiency was estimated (reviewed in ref. 13) for each image according to EFRET = ICFPpost / ICFPpre.
FRAP. Cells were visualized with a 40 1.3-NA oil objective lens with 10 digital zoom, imaged at 256 256 8-bit resolution with a scan speed of 245 ms using 488-nm argon laser light at 1% power. A 2-m bleach spot was generated after 5 scans, using 30 iterations of 100% power 488-nm light, before image acquisition to monitor fluorescence recovery. Data were corrected for the loss of fluorescence owing to imaging as described27. Each data point represents the average of 30 observations acquired from 3 independent experiments. The effective diffusion coefficient was estimated from the solution of D = 2 / t, where accounts for the relative mobility of each component as described28.
Mitotic progression and arrest. MEFs from PML-/- embryos or PML+/+ litter mate controls4 in passage 3 of culture were plated at 500,000 cells per well in a six-well plate containing a 0-thickness glass coverslip in DMEM supplemented with 10% (v/v) FCS. After 16 h, the coverslips were fixed in 0.5% (v/v) glutaraldehyde for 7 min at room temperature, and quenched in 0.2% (w/v) sodium borohydride. The coverslips were washed extensively with TBS 0.1% (v/v) Triton-X-100 and mounted in Vectashield plus DAPI (Vecta Labs). The classic mitotic phases29 were scored on the basis of chromosome morphology, with >4,000 cells counted per sample. To assess the mitotic response to spindle disruption, cells were treated with a final concentration of 10 M nocodazole (Sigma) added from a 1.7 mM stock for the indicated times before processing as above, as described14.
Received 15 July 2004; Accepted 3 September 2004; Published online: 3 October 2004.
REFERENCES
Melnick, A. & Licht, J.D. Deconstructing a disease: RAR, its fusion partners, and their roles in the pathogenesis of acute promyelocytic leukemia. Blood93, 3167–3215 (1999) | PubMed | ISI | ChemPort |
Takahashi, Y., Lallemand-Breitenbach, V., Zhu. J. & de Thé, H. PML nuclear bodies and apoptosis. Oncogene23, 2819–2824 (2004). | Article | PubMed | ISI | ChemPort |
Jensen, K., Shiels, C. & Freemont, P.S. PML protein isoforms and the RBCC/TRIM motif. Oncogene20, 7223–7233 (2001). | Article | PubMed | ISI | ChemPort |
Wang, Z.G. et al. Role of PML in cell growth and the retinoic acid pathway. Science279, 1547–1551 (1998). | Article | PubMed | ISI | ChemPort |
Gurrieri, C. et al. Loss of the tumor suppressor PML in human cancers of multiple histologic origins. J. Natl. Cancer Inst.96, 269–279 (2004). | Article | PubMed | ChemPort |
Zhong, S., Muller, S., Freemont, P.S., Dejean, A. & Pandolfi, P.P. Role of SUMO-1-modified PML in nuclear body formation. Blood95, 2748–2753 (2000). | PubMed | ISI | ChemPort |
Salomoni, P. & Pandolfi, P.P. The role of PML in tumor suppression. Cell108, 165–170 (2002). | Article | PubMed | ISI | ChemPort |
Zhong, S., Salomoni, P. & Pandolfi P.P. The transcriptional role of PML and the nuclear body. Nat. Cell Biol.2, E85–E90 (2000). | Article | PubMed | ISI | ChemPort |
Scolnick, D.M. & Halazonetis, T.D. Chfr defines a mitotic stress checkpoint that delays entry into metaphase. Nature406, 430–435 (2000). | Article | PubMed | ISI | ChemPort |
Chaturvedi, P. et al. Chfr regulates a mitotic stress pathway through its RING-finger domain with ubiquitin ligase activity. Cancer Res.62, 1797–1801 (2002). | PubMed | ISI | ChemPort |
Muller, S., Matunis, M.J. & Dejean, A. Conjugation with the ubiquitin-related modifier SUMO-1 regulates the partitioning of PML within the nucleus. EMBO J.17, 61–70 (1998). | Article | PubMed | ISI | ChemPort |
Rieder, C.L. & Cole, R. Microtubule disassembly delays the G2-M transition in vertebrates. Curr. Biol.10, 1067–1070 (2000). | Article | PubMed | ISI | ChemPort |
Bothos, J., Summers, M.K., Venere, M., Scolnick, D.M. & Halazonetis, T.D. The Chfr mitotic checkpoint protein functions with Ubc13-Mms2 to form Lys63-linked polyubiquitin chains. Oncogene22, 7101–7107 (2003). | Article | PubMed | ISI | ChemPort |
Chang, K.S., Fan, Y.H., Andreeff, M., Liu, J. & Mu, Z.M. The PML gene encodes a phosphoprotein associated with the nuclear matrix. Blood85, 3646–3653 (1995). | PubMed | ISI | ChemPort |
Kang, D., Chen, J., Wong, J. & Fang, G. The checkpoint protein Chfr is a ligase that ubiquitinates Plk1 and inhibits Cdc2 at the G2 to M transition. J. Cell Biol.156, 249–259 (2002). | Article | PubMed | ISI | ChemPort |
Aravind, L., Iyer, L.M. & Koonin, E.V. Scores of RINGS but no PHDs in ubiquitin signaling. Cell Cycle2, 123–126 (2003). | PubMed | ChemPort |
Boisvert, F-M., Hendzel, M. & Bazlett-Jones, D.P. Promyelocytic leukemia (PML) nuclear bodies are protein structures that do not accumulate RNA. J. Cell Biol.148, 283–292 (2000). | Article | PubMed | ISI | ChemPort |
Corn, P.G. et al. Frequent hypermethylation of the 5' CpG island of the mitotic stress checkpoint gene Chfr in colorectal and non-small cell lung cancer. Carcinogenesis, 24, 47–51 (2003). | Article | PubMed | ISI | ChemPort |
Toyota, M. et al. Epigenetic inactivation of CHFR in human tumors. Proc. Natl. Acad. Sci. USA100, 7818–7823 (2003). | Article | PubMed | ChemPort |
Mariatos, G. et al. Inactivating mutations targeting the chfr mitotic checkpoint gene in human lung cancer. Cancer Res.63, 7185–7189 (2003). | PubMed | ISI | ChemPort |
Satoh, A. et al. Epigenetic inactivation of CHFR and sensitivity to microtubule inhibitors in gastric cancer. Cancer Res.63, 8606–8613 (2003). | PubMed | ISI | ChemPort |
Sakaguchi, D.S., Coffman, C.R., Gallenson, N. & Harris, W.A. A glial cell line promotes the outgrowth of neuritis from embryonic Xenopus retina. Acta Biol. Hung.39, 201–209 (1988). | PubMed | ISI | ChemPort |
Manders, E.M., Stap, J., Brakenhoff, G., van Driel, R. & Aten, J.A. Dynamics of three-dimensional replication patterns during the S-phase, analysed by double labelling of DNA and confocal microscopy. J. Cell Sci.103, 857–862 (1992). | PubMed | ISI | ChemPort |
Bastiaens, P.I., Majoul, I.V., Verveer, P.J., Soling, H.D. & Jovin, T.M. Imaging the intracellular trafficking and state of the AB5 quaternary structure of cholera toxin. EMBO J.15, 4246–4253 (1996). | PubMed | ISI | ChemPort |
Ellenberg, J. et al. Nuclear membrane dynamics and reassembly in living cells: targeting of an inner nuclear membrane protein in interphase and mitosis. J. Cell Biol.138, 1193–1206 (1997). | Article | PubMed | ISI | ChemPort |
Yu, D.S. et al. Dynamic control of Rad51 recombinase by self-association and interaction with BRCA2. Mol. Cell12, 1029–1041 (2003). | Article | PubMed | ISI | ChemPort |
Acknowledgments M. Daniels generously assisted us with XR1 cell culture and preparation of X. laevis extracts, and T. Mills with microinjection. T. Rich (Department of Pathology, University of Cambridge) and P.-P. Pandolfi (Memorial Sloan-Kettering Institute, New York) supplied the PML-YFP construct, and PML-/- and control MEFs, respectively, and provided constructive comments on this paper, as did L. Ko-Ferrigno. M.J.D. was supported by an Astra-Zeneca studentship, through the MB/PhD program at the University of Cambridge, and by the UK Medical Research Council. A.M. was the receipient of a Herchel Smith Harvard scholarship. Work in A.R.V.'s laboratory is supported by the Medical Research Council and Cancer Research UK.
Competing interests statement:
The authors declare that they have no competing financial interests.
(RAR
1 
m, which vary in number from 10−30 per nucleus. PML bodies are not formed in PML-/- cells
FHA protein (relative molecular mass (Mr), 85 kDa) lacks the first 111 residues of CHFR. Lane 1, GFP-CHFR; lane 2, GFP-CHFR
0.065 
FHA prevents recruitment of CHFR, but not of PML, into PML bodies.
106) were lysed in 150 mM NaCl, 0.1% (v/v) NP40, 100 
