Mouse 3T3-PML^+/+ and 3T3-PML^-/- cells 26 , kindly provided by T. Hofmann (DKFZ Heidelberg) were cultured in Dulbecco' modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum in a 10% CO₂ atmosphere at 37°C. For live cell imaging experiments, cells were seeded on 42 mm glass dishes (Saur Laborbedarf, Reutlingen, Germany) and transfected with plasmid DNA one to two days before observation using FuGENE-HD Transfection reagent (Roche, Basel, Switzerland) according to the manufacturer' protocol.

The GFP-PML expression constructs have been described in detail previously 25 .

Whole cell extracts were produced from transiently or stably transfected cell lines, electrophoresed on SDS-PAGE and transferred to Protran nitrocellulose membrane (Schleicher & Schuell, Dassel, Germany). The membrane was incubated with primary antibodies (in PBS-T) and developed with a peroxidase conjugated secondary species-specific antibody (Jackson Immunoresearch, West Grove, PA, USA). Signal was detected using the ECL reagent (Amersham, Uppsala, Sweden) on imaging film (Biomax, Kodak, Stuttgart, Germany). Anti-GFP monoclonal antibody was from Santa-Cruz Biotechnology (Heidelberg, Germany). Anti-mouse PML monoclonal antibody (# 05-718), non-cross reactive with human PML protein, was purchased from Upstate.

Cells grown on 15 mm diameter coverslips were fixed with 4% formaldehyde for 10 minutes and permeabilized with 0.25% Triton-X100 for 3 minutes. Diluted anti-mouse PML mAB was incubated on cells for 45 minutes. After 3 washing steps with PBS, an anti-mouse secondary antibody coupled to Cy3 (Jackson Immunosearch, West Grove, USA) was incubated on cells for 45 minutes, followed by a DNA-staining step using ToPro3 or DAPI (Invitrogen, Carlsbad, USA) for 10 minutes and mounting with Prolong Gold antifade mounting medium (Invitrogen, Carlsbad, USA). For microscopy, a LSM 510Meta or LSM710 laser scanning confocal microscope (Carl Zeiss, Jena, Germany) was used.

Fluorescence correlation spectroscopy (FCS) measurements were performed at 37°C on a LSM 510Meta/ConfoCor2 combi system using a C-Apochromat infinity-corrected 1.2 NA 40× water objective (Carl Zeiss, Jena, Germany) as described in detail elsewhere 25 . Briefly, GFP-tagged proteins were spot-illuminated with the 488 nm line of a 20 mW Argon laser at 5.5 Ampere tube current attenuated by an acousto-optical tunable filter (AOTF) to 0.1%. The detection pinhole had a diameter of 70 μm and emission was recorded through a 505-530 nm band-path filter. For the measurments, 10 × 30 time series of 10 s each were recorded with a time resolution of 1 μs and then superimposed for fitting to an anomalous diffusion model in three dimensions with triplett function 27 using Origin Software (OriginLab, Northhampton, MA, USA). The diffusion coefficients and anomaly parameters were extracted from fit curves as previously described 25 .

Fluorescence Recovery after Photobleaching (FRAP) experiments were carried out on a Zeiss LSM 510Meta confocal microscope (Carl Zeiss, Jena, Germany). One or two image stacks were taken before the bleach pulse and 50-70 image stacks after bleaching of "regions of interest" (ROIs) containing one nuclear body each at 0.05% laser transmission to minimize scan bleaching. Image aquisition frequency was adapted to the recovery rate of the respective GFP fusion protein, usually a 20 second interval was applied. The pinhole was adjusted to 1 airy unit. The image stacks were maximum-projected into a single plane from which relative fluorescence intensities within the ROIs were quantitated according to 25 using Excel (Microsoft, Redmond, WA, USA) and Origin software (OriginLab, Northhampton, MA, USA).

For the mathematical model, the structural complexity of a PML body has been approximated assuming that molecules undergoing binding and unbinding to and from the body do so at the surface, and molecules situated more towards the inside of the body cannot unbind before moving to the surface. There is thus a reservoir of tightly bound "inner" molecules and one of loosely bound "outer" ones. Exchange between these reservoirs is modeled by linear kinetics, i. e. the more molecules there are, the more will move inside or out with rate constants k _in and k _out, respectively. Binding and unbinding to the PML body is treated similarly, with rate constants k _on and k _off, respectively. The experimental set-up provided that bleach ROI and FRAP ROI is similar, and so are assumed to be the same for modelling purposes. Diffusion inside and out of the ROI was modelled as a linear two-way process with a rate constant proportional 2D/r², where D is the diffusion coefficient measured by FCS, and r is the ROI's radius. This constant yields the effective exchange rate through the boundary of a circular area. The model considers only fluorescent molecules of one type at a time, so no interactions between different molecular species are considered. It distinguishes between molecules outside the ROI, inside the ROI but diffusing freely, loosely bound at the surface of the PML body, and tightly bound inside the body. The fluorescence outside the ROI is all but unchanged by the bleaching pulse and subsequent diffusion, so this value was used to normalize all concentrations in the equations. Describing normalized concentrations of the fluorescent protein in free diffusion (x), loosely bound (y) and tightly bound (z), the reaction system results in the model equations

The differential equations were numerically solved using an explicit Runge-Kutta formula (method ode45 in MATLAB). To fit the parameters of the model, an Evolution Strategy with Covariance Matrix Adaptation (CMA-ES) was employed 28 .

We determined the ratio p = 20 of steady state fluorescence in the body vs. the background by confocal microscopy of GFP-PML isoforms and pixel intensity evaluation using MetaMorph software (Molecular Devices, Sunnyvale, USA).

This enabled us to express the observable fluorescence in the ROI as

Since the amount of bleached molecules is small compared to the overall amount of molecules in the nucleus, it is plausible to assume that after a sufficiently long time, fluorescence returns to the value measured before the photobleach. This constraint removes one degree of freedom from the model, yielding

These values were used to normalize the concentrations used in the model. The mathematical model treats normalized concentrations of fluorescent molecules in free diffusion (x), loosely bound (y) and tightly bound (z) to the PML body.

It is important to note that the concentration values x, y and z have been normalized by the background fluorescence outside the region of interest. Since we have determined the ratio p between equilibrium fluorescence inside the body and outside the ROI, and because the data used to fit the model was given in relative fluorescence intensity with an equilibrium value of 1, the background flourescence is given by 1/p. Therefore, the observed RFI value w is related to the model concentrations by

Over a very long time, all molecular species in the PML-body will eventually turn over, so that it is reasonable to assume w(t) tends towards 1 for large t. This means that for equilibrium conditions, in which , w(t) = 1. In this case, we have

and thus, from the first three lines, we get

Finally, taking this together with the conditition x + y + z = p, we find

This enabled us to remove one degree of freedom, k _on, from the model.

We had previously analyzed the dynamics of component exchange at PML nuclear bodies in human cells expressing endogenous PML proteins 25 . The objective of the current study was to study the formation of PML NBs in the absence of endogenous PML expression. All six human PML isoforms (Fig. 1A) were therefore expressed as GFP fusion proteins in mouse 3T3 control cells (3T3-PML^+/+) or mouse 3T3 cells derived from PML knock-out mice (3T3-PML^-/-). All GFP-PML constructs are functional in human cells 25 and were expressed as full-length proteins in 3T3 cells, as judged by western-blotting (data not shown).

Immunofluorescence analyses showed that all six human GFP-PML isoforms localized to enogenous PML nuclear bodies in 3T3-PML^+/+ mouse cells (Fig. 1B). Importantly, in 3T3-PML^-/- mouse cells, each individual human GFP-PML isoform was able to form nuclear bodies (Fig. 1B). This confirmed that the nuclear body formation ability of PML resides within sequences encoded by exons 1 to 6, represented by GFP-PML VI, and suggests that the C-terminal extensions of PML isoforms (exons 7 to 9) do not alter this function (Fig. 1B). Because GFP-PML VI which does not contain the SUMO-interacting motiv (SIM) (Fig. 1A) is still able to form nuclear bodies in the absence of endogenous PML bodies we conclude that a SIM is not essential for nuclear body formation by PML as previously suggested 22 .

To study the binding properties of individual PML isoforms at nuclear bodies we employed FRAP. A spherical region containing one PML nuclear body was bleached to background levels and fluorescence recovery was monitored for 20 min (Fig. 2). Similar to human cells 25 , all six GFP-PML isoforms exhibited protein-specific FRAP curves in 3T3 cells expressing endogenous PML nuclear bodies (Fig. 3, black curves). In particular, we could confirm that exchange of GFP-PML V is extremely slow: fluorescence recovered to only ~40% after 20 min indicating a very slow turn over of PML V molecules at nuclear bodies (Fig. 3E). This observation corroborates our previous conclusion that PML V may act as a hyper-stable scaffold component within PML nuclear bodies 25 .

Human cells contain six nuclear PML isoforms, whereas for mouse cells only two PML transcripts have been described so far, and the corrsponding mouse PML sequences are more than 80% similar to human PML isoform I 29 . It should therefore be pointed out that PML isoforms II to VI have no direct counterparts in the murine system with respect to alternative expression of exons 6 to 9. Nevertheless, the analysis of these human isoforms in mouse cells may still deliver valuable information on PML nuclear body formation, particularly in a PML-negative background. Based on the high sequence similarity the human GFP-PML I construct has the potential to functionally (and thus dynamically) act in a similar fashion as mouse PML in murine cells. Indeed, the exchange dynamics of human GFP-PML I protein at nuclear bodies of PML-positive murine cells was similar to human cells (Fig. 3A) 25 . In contrast, GFP-PML I was alomst immobile at nuclear bodies in 3T3 cells lacking endogenous PML (Fig. 3A). The same phenomenon was observed for GFP-PML isoform III and in a subpopulation of cells expressing GFP-PML isoforms II or IV (Fig. 3, B-D). Thus, in the absence of endogenous PML proteins, GFP-PML I to IV form stable aggregates. These observations suggest that these isoforms require the presence of other PML isoforms or endogenous PML-binding proteins for efficient exchange at nuclear bodies. They also suggest that the capacity to contribute to nuclear body stability is inherent to all of these PML isoforms. We also observed a minor population (< 20%) of cells in which GFP-PML isoforms II and IV showed dynamic exchange at nuclear bodies (Fig. 3B, and 3D, red curves). These observations suggest a cell cycle dependent behavior of PML II and PML IV at nuclear bodies which likely originate from interaction of these isoforms with as yet unknown binding sites outside nuclear bodies. However, since human PML II and PML IV are not conserved in mouse cells, such interactions might be biologically non-relevant. In contrast to GFP-PML I to IV, the dynamics of GFP-PML V and VI were almost unaltered in PML negative cells (Fig. 3E, and 3F).

The presence of slow exchanging populations of GFP-PML I to IV is consistent with the idea that these isoforms are also able to provide nuclear body stability, as concluded for PML V. In human cells, GFP-PML I and III exhibit much higher exchange rates at nuclear bodies 25 than observed here in mouse cells (Fig. 3).

That GFP-PML I and III do not exchange with soluble nucleoplasmic populations in mouse cells may indicate that the human isoforms can not be dissociated from nuclear bodies through interaction with soluble mouse PML-binding proteins outside nuclear bodies. Since the SUMOylation status of PML also regulates the exchange rate at nuclear bodies 25 , changing SUMO patterns on GFP-PML I to IV may also explain their changing exchange rate at NBs.

In order to understand PML nuclear body assembly in a more quantitative way we applied a kinetic modeling approach established previously 25 (Fig. 4A). Diffusion inside and out of bleached regions was modelled as a linear two-way process as described in the materials and methods section. The diffusion coefficients of the GFP-PML isoforms, as determined by fluorescence correlation spectroscopy, ranged between D = 1 - 3 μm²s^-1 (data not shown). Table 1 contains for each of the PML protein isoforms the binding and unbinding rate k _on and k _off, and the rate of movement to the inner core and to the outer surface of the body, k _in and k _out. From these, we computed the residence time (R.t.), i.e. the mean time a molecule spends bound to the nuclear body, and the fraction of molecules bound in the inner and outer region of the body, bnd_in and bnd_out. This model provided good fits to the measured FRAP curves of all PML isoforms (Fig. 4, B-O). This quantitative evaluation confirmed that GFP-PML V had the longest residence time (947 s) of all isoforms in PML-positive cells (Table 1). The residence times of GFP-PML isoforms I to IV ranged between 221 and 336 seconds which is very similar to their residence times at nuclear bodies in human cells 25 . Thus, each PML isoform exhibits individual exchange characteristcs at nuclear bodies. Since the molecular structure of PML isoforms are identical over two-thirds of the sequence at the N-termiuns, the observed differences in the dynamic behaviour must originate from their C-terminal varying parts.

Compared to PML-positive cells, the residence time at nuclear bodies of GFP-tagged PML isoforms I to IV was increased several-fold in PML-negative cells (Table 1). This result suggests that the exchange rate of overexpressed human PML isoforms is influenced by the dynamics of the endogenous mouse PML proteins. Thus, in the absence of endogenous mouse PML protein, the unrelated human isoforms tend to form more insoluble aggregates. Interestingly, this is not true for the shortest PML isoforms V and VI, the residence time of which is almost identical independent of the presence or absence of endogenous PML bodies (Table 1). This individual property of GFP-PML V and VI argues in favor of a more structural role for these isoforms at PML nuclear bodies.

PML contains three Lysine residues (K65, K160, and K490; Fig. 1A) at which all isoforms can be SUMOylated in vivo 30 . To determine the impact of SUMO modifications we analyzed the localization and dynamics of a GFP-PML IV construct in which K160 and K490 were mutated to Arginine. This mutant protein localized diffusely in the nucleoplasm of both PML-positive and PML-negative cells (Fig. 5, and data not shown) indicating that SUMOylation at K160 and K490 are required for PML nuclear body binding. FRAP analysis of the SUMO mutant within a nucleoplasmic region revealed fast and complete recovery within seconds suggesting a predominant diffusion type of mobility (Fig. 5). We had shown previously that the K160/490R mutant is still able to bind to nuclear bodies in human cells but its residence time was only 5.8 s compared to the residence time of wild-type GFP-PML IV, which is 7.6 min 25 . Thus, the human K160/490R mutant very transiently interacts at 'human' PML bodies 25 but does not bind to 'mouse' PML bodies and is not able to form nuclear body structures in the absence of any PML protein (Fig. 5). Bacterially expressed PML protein is not SUMO-modified but still able to form nuclear body-like structures in vitro through self-assembly of the N-terminal RBCC region of PML 31 . Because GFP-PML-K160/490R is not able to form nuclear body structures in the absence of other mouse or human PML proteins (Fig. 5) we conclude that SUMOylation is a critical determinant for PML body formation. In future studies it will be intersting to analyze the impact of single, double and the triple SUMO mutants of PML in nuclear body formation.

FRAP analyses of subnuclear domains such as speckles, Cajal bodies and nucleoli had revealed that their component parts rapidly exchange with nucleoplasmic pools 32 33 34 35 36 37 38 39 . Typical residence times of proteins within these compartments are in the seconds range (Fig. 6). These observations have led to the conclusion that nuclear body proteins undergo repeated and rapid cycles of association and dissociation between the nuclear body and the nucleoplasm 40 . As a consequence, a nuclear body is in perpetual flux and its structure is determined by the ratio of on-rate versus off-rate of its components 41 . While this assembly mechanism is certainly true for speckles, Cajal bodies and nucleoli, the work presented here and previously 25 demonstrate that the stabilty of PML nuclear bodies relies on very long residence times of specific PML isoforms, in particular PML V 25 (Fig. 6), and probably also PML VI (Table 1). These observations strongly support the idea of a scaffold function of PML nuclear bodies 42 .