Monitoring of nucleophosmin oligomerization in live cells

Adherent cell line HEK-293T (a gift from Š Němečková's lab, Institute of Hematology and Blood Transfusion) was cultured under standard cultivation conditions in DMEM (SIGMA) supplemented with 10% FBS (SIGMA), 37 °C and 5% CO₂ atmosphere.

Co-transfection efficiency was tested by flow cytometry (LSR Fortessa, BD Biosciences). Cells were detached from cultivation dish with Trypsin (SIGMA), dissolved in DMEM medium to block Trypsin, span down and then resuspended in PBS. Fluorescence originating from eGFP and mRFP1 tags was registered in FITC and PE channels, respectively. Green and red -positive events were assessed for at least 20 000 cells for each sample.

Transfected cells expressing fluorescent proteins were processed after 40h-incubation. GFP-Trap_A system (Chromotek) was used following the manufacturer's instructions as described in [27]. Briefly, eGFP expressing adherent cells were scrapped from dish in ice-cold PBS and extensively washed with PBS. The cell pellet was lysed in the lysis buffer (10 mM Tris/Cl pH 7.5, 150 mM NaCl, 0.5 mM EDTA, 0.5% NP-40, protease and phosphatase inhibitors) on ice for 30 min and centrifuged at 20.000 g/10 min/4 °C. The lysate was applied on the GFP-Trap_A beads and rotated for 1 h at 4 °C. Then the samples were centrifuged and extensively washed in the diluting buffer (10 mM Tris/Cl pH 7.5, 150 mM NaCl, 0.5 mM EDTA) to be resuspended in 2xSDS-sample buffer (100 mM Tris pH 6.8, 4% SDS, 200 mM DTT, 20% glycerol), boiled for 10 min and centrifuged at 2.500 g/2 min/4 °C. Supernatant was stored at −20 °C until used for SDS-PAGE. Western blotting was carried out according to [27]. Mouse monoclonal antibodies against GFP and NPM (clone 3F291) were from Santa Cruz Biotechnology and they were used at a dilution 1:500. Anti-mouse HRP-conjugated secondary antibody was purchased from Thermo Scientific and used at concentrations 1:50 000. Blots were visualized with ECL Plus Western Blotting Detection System (GE Healthcare) and evaluated by G-box iChemi XT4 digital imaging device (Syngene Europe).

Plasmids for expression of fluorescently tagged proteins were constructed by standard molecular cloning techniques (cutting with restriction enzymes (Thermo Scientific) and ligation (NEB)). Plasmids for NPM and NCL expression were prepared as described in [46]. Fragments, PCR-amplified from cDNA library (Jurkat cells, Origene), were subcloned to vectors peGFP-C2 or pmRFP1-C2 (originally Clontech) using XhoI and BamHI unique restriction sites. Mutations of Cys21 were introduced to the sequence of WT NPM using merged primers containing appropriate mutation and XhoI restriction site (table 1). Following PCR amplification, the resulting fragments were subcloned into the peGFP-C2 and the pmRFP1-C2 as in the case of the WT NPM form.

The constructed plasmids were amplified in E. coli and purified with PureYield Plasmid Miniprep System (Promega). For transfection, cells were seeded to 1 × 10⁵/ml cell density 24 h prior transfection. Expression plasmids were transfected into HEK-293T cells using jetPrime transfection reagent (Polyplus transfection) according to manufacturer's protocol. Growth medium was replaced 4 h after transfection and cells were then grown 40 h prior analysis.

The NPM constructs and their combinations which were prepared or used in this study are summarized in table 2. In addition, combination of NCL constructs is mentioned.

The cells were grown on glass bottom Petri dish (Cellvis). Fluorescence experiments were carried out at a room temperature after sealing the Petri dish with parafilm to prevent CO₂ leakage. One Petri dish sample was measured typically within 1 h. The subcellular distribution and colocalization of eGFP- or mRFP1-fused NPM variants were observed under a confocal laser scanning microscope FluoView FV1000 (Olympus Corporation) as in [46]. Fluorescence images were processed by FluoView FV10-ASW 3.1 software and by ImageJ-FiJi.

PIE-ccN&B cross-correlation experiments utilized dual-color laser excitation set-up with a pulsed 'blue' source (Picoquant, LDH-DC-470, 470 nm) and a 'green' continuous wave source (Uniphase, He-Ne laser, 543 nm). The excitation beam of both wavelengths was guided into the sample via 488/543/633 dichroic mirror (Olympus) and water immersion objective (Olympus 60x, NA 1.2). The emitted fluorescence was then guided through a pinhole (120 μm), coupled into the multimode optical fiber and at its exit it was split into two detection channels separated by a 535 DCXC dichroic mirror (Chroma filters). The 'blue' and 'green' emission channels were equipped with the bandpass filters transmitting wavelengths of 515 ± 15 nm and 605 ± 20 nm (Chroma filters), respectively. Signal was detected with two single photon counting modules (tau-SPAD, Picoquant) and the arrival times were recorded with multichannel event timer with TCSPC option (Hydraharp 400, Germany) synchronized with the Galvo scan unit (Fluoview 1000, Olympus). This arrangement enabled us to operate the experiments in the PIE mode adjusted for the combination of pulsed and continuous wave laser. Specifically, the 'blue' pulsed laser was operated at 10 MHz repetition rate (i.e. approximately 100 ns temporal spacing between the adjacent pulses). As the lifetime of the eGFP fluorophore is about 2.5 ns, the mRFP1 emission excited exclusively by the continuous wave 'green' source can be extracted by means of the time-gating (i.e. the signal coming in last 50 ns within the TCSPC histogram was taken), which suppressed completely the undesired cross-talk artifact. For the PIE-ccN&B cross-correlation experiments, we adapted the protocol originally developed for the Raster image correlation spectroscopy (RICS) [54]. For each experiment, 100 frames consisting of 84 × 82 pixels were collected. The scanner was operated at the speed of 100 μs/pixel with a step of 50 nm. The image stacks were taken only from the nucleoplasm which displayed relatively uniform fluorescence intensity.

The values of eGFP and mRFP1-brightness (B₁ and B₂, respectively) and cross-brightness (B_cc) can be calculated for each pixel within the recorded set of images as described in [55, 56]. In brief, prior to the brightness calculation the images were corrected for bleaching by a detrend procedure accounting for the Poissonian distribution of the recorded signal. The cross-variance σ_cc of the fluorescence signal, and subsequently the cross-brightness B_cc, were then evaluated for the eGFP and mRFP1 channels as follows:

$$\begin{eqnarray}&&{\sigma }_{cc}^{2}=\,\displaystyle \frac{{\sum }^{}({I}_{1}-\langle {I}_{1}\rangle )(({I}_{2}-\langle {I}_{2}\rangle )}{K}\,\,\,\,\,\,\end{eqnarray} \tag{ 1 }$$

$$\begin{eqnarray}&&{B}_{cc}=\displaystyle \frac{{\sigma }_{cc}^{2}}{\sqrt{\langle {I}_{1}\rangle \langle {I}_{2}\rangle }},\end{eqnarray} \tag{ 2 }$$

where ${I}_{1}$ and ${I}_{2}$ stand for the fluorescence intensity in the eGFP and mRFP1 channel, respectively, $\langle {I}_{1}\rangle$ and $\langle {I}_{2}\rangle$ stand for their means and K is the number of evaluated frames.

FLIM experiments were carried out on an inverted IX83 microscope with FV1200 confocal scanner (Olympus). The microscope was equipped with a FLIM add-on comprising picosecond semiconductor lasers, fibre-coupled GaAsP hybrid detectors, and TimeHarp 260PICO TCSPC detection electronics (all PicoQuant, Berlin, Germany). The instrument was complemented with frequency-doubled OPO (Chameleon compact), laser (Coherent, Santa Clara, USA) for tunable VIS excitation used in photobleaching experiments. Specifically, cellular FLIM experiments were performed with a UPLSAPO 60x NA 1.2 water immersion objective (Olympus). eGFP fluorescence was excited at 485 nm by a LDH-DC-485 laser head (PicoQuant), emission decays were collected on the pixel-by-pixel basis in the epi-fluorescence mode using combination of 560 nm short-pass dichroic and Semrock 520/35 bandpass filters in the descanned detection path. To avoid pile-up, the data collection rate was kept below 5% of the laser repetition rate. FLIM data were analyzed by the SymPhoTime64 software (PicoQuant). The lifetime images were generated by a robust 'fast-FLIM' approach when mean pixel lifetimes were calculated by a method of moments [57]. Specifically, the mean lifetime τ_av for each pixel was determined as the difference between the barycentre of the fluorescence decay recorded at one position of the scanner and the offset t_offset at the steepest uprise of the decay curve:

$$\begin{eqnarray}&&{\tau }_{av}=\displaystyle \frac{\displaystyle \sum I{}_{i}t_{i}}{\displaystyle \sum {I}_{i}}-{t}_{offset},\end{eqnarray} \tag{ 3 }$$

where I_i stands for the intensity recorded in time t_i. Least-squares reconvolution was applied for accurate analysis of cumulative decays from larger ROIs (cell, nucleus, nucleolus). Fluorescence of eGFP was assumed to decay bi-exponentially [53, 58]:

$$\begin{eqnarray}&&I(t)={a}_{1}\cdot {e}^{-t/{\tau }_{1}}+{a}_{2}\cdot {e}^{-t/{\tau }_{2}},\end{eqnarray} \tag{ 4 }$$

where τ_i are lifetime components and a_i stand for corresponding amplitudes. The intensity-weighted mean fluorescence lifetime was calculated as:

$$\begin{eqnarray}&&{\tau }_{mean}=\displaystyle \sum {f}_{i}\cdot {\tau }_{i},\,{f}_{i}={a}_{i}{\tau }_{i}/\displaystyle \sum {a}_{i}{\tau }_{i},\end{eqnarray} \tag{ 5 }$$

where f_i are intensity fractions of the ith lifetime component. Correlation plots were done in Matlab.

2.7.1. Statistical analysis

Methods for the NPM oligomerization monitoring based on the cross-correlation FFS (ccFFS) require dual-color labeling. We transfected HEK-293T cells with 1:1 mixtures of plasmids (table 2), as in [46], to ensure simultaneous expression of eGFP- and mRFP1-tagged NPM inside the cell. Typical confocal images of cells expressing NPM variants labeled with both colors are shown in figure 1.

Zoom In Zoom Out Reset image size

Good controls are an essential prerequisite for reliable identification of interacting proteins by means of the ccFFS methods. In cross-correlation experiments based in the PIE-ccN&B, we used cells expressing R_NPM_G (table 2) carrying both eGFP and mRFP1 on the same protein as a positive control and cells expressing combination of G_NPM (table 2) with free mRFP1 as a negative control. Confocal images of these controls are presented in figure 2. The eGFP and mRFP1 proteins bound within the R_NPM_G should drift together in the cell nucleus. R_NPM_G was found to be localized both in the nucleus and the nucleolus (figure 2(A)), with weaker preference for the nucleolus compared to the WT NPM (figure 1(A)). The common movement should result in positive value of cross-brightness B_cc obtained by the ccN&B (see equation (2)). For the negative control we used the G_NPM complemented with free mRFP1 expressed from the empty cloning vector. To compensate for higher expression of mRFP1 compared to G_NPM, either cells with lower mRFP1 signal were selected for the experiment or cells were transfected with lower amount of the mRFP1 plasmid.

Zoom In Zoom Out Reset image size

The dual labeling of cells was verified by flow cytometry (figures 2(C), (D)). Cells expressing the R_NPM_G served as a reference. Since each fused protein carries pair of the flurescent proteins each cell expressing the bi-colored NPM should contain the same amount of eGFP and mRFP1. Indeed, events detected for cells expressing the R_NPM_G lay approximately on a diagonal (figure 2(C)). Additionally, events registered for cells expressing G_NPM/R_NPM mixture (1:1) exhibit broad intensity distribution along a curve fitted to the R_NPM_G events (figure 2(D)). This confirms simultaneous expression of both eGFP- and mRFP1-tagged NPM which is a good prerequisite for successful application of the ccN&B in detection of bi-colored aggregates.

Inhibition of the NPM oligomerization by mutation of Cys21 to Ala or Phe was already reported [16, 25]. To have a possibility to test NPM oligomerization monitoring by the PIE-ccN&B, NPM gene in plasmids for FP-tagged expression was mutated by molecular cloning to provide the Cys21-mutated NPM variants. HEK-293T cells were transfected with 1:1 mixtures of the plasmids to express the WT NPM and the NPM mutants fused either with eGFP or mRFP1. Localization of the two color NPM mutants in cells was verified by the confocal fluorescence microscope compared to WT NPM (figure 1). The WT NPM localizes preferentially to the nucleolus as expected (figure 1(A)). Compromising the oligomerization by the substitutions in the Cys21 should result in delocalization of NPM to the nucleoplasm [31]. However, we observed that all the NPM variants labeled with eGFP or mRFP1 are localized mainly in nucleoli (figures 1(B), (C)). Hence, contrary to the expectations, the Cys21 mutants do not show any signs of delocalization from the nucleolus to the nucleoplasm. The tagged mutants are matching the tagged WT NPM in their localization and nucleoplasma levels. Therefore, the inhibition effect of the Cys21 substitutions on the NPM oligomerization is challenged by these microscopy observations in live cells.

Provided that joint aggregates consisting of both eGFP- and mRFP1-labeled NPM are formed in transfected cells, the aggregates emit green and red fluorescence simultaneously. Therefore, green and red fluorescence signals that originate from double illuminated confocal volume should fluctuate simultaneously in separate detection channels as the two-color aggregates drift in and out the volume. The cross-correlation between the signals can be obtained e.g. from equation (2). The FFS-based methods can see exclusively the oligomerization of the mobile fraction of the NPM, which is a source of the fluctuation in the signal. To eliminate any crosstalk between channels, we collected the data in the PIE mode that enables entire separation of eGFP and mRFP1 signals. The measuring frame of 84 × 82 pixels was positioned in the nucleoplasm to obtain as homogenous distribution of the fluorescence signal as possible. The cross-correlation brightness B_cc was calculated for each pixel in a series of raster images using equation (2). The mean B_cc value was then determined across the whole measuring frame. The controls were used for validation of the PIE-ccN&B protocol in live cells. Typical pixel-based distribution of B_cc in a cell expressing the R_NPM_G construct is demonstrated in figure 3(A). Typical Gaussian fits obtained for the controls and the G_NPM/R_NPM mix are shown in figure 3(B). Results clearly indicate that the distribution of B_cc corresponding to the G_NPM/R_NPM can be found between the distributions corresponding to the positive and the negative controls, being close to the positive one. Mean B_cc values calculated for individual cells of the controls and the G_NPM/R_NPM mix (n = 9 for the R_NPM_G, n = 7 for the G_NPM/R_NPM, n = 4 for G_NPM/R) are shown in figure 3(C). The positive control R_NPM_G and the G_NPM/R_NPM mix are significantly separated from the negative control G_NPM/R. The elevated mean B_cc values calculated for the G_NPM/R_NPM mix point to formation of drifting two-color NPM aggregates.

Zoom In Zoom Out Reset image size

Data acquired for the dual labeled Cys21 NPM mutants in the PIE-ccN&B experiments are presented together with data for the WT NPM in figure 3(C). There is no statistically significant difference between B_cc values acquired for the two-color mixtures of NPM mutants and B_cc values of the WT protein mixture (p = 0.83 for C21A and 0.87 for C21F). The points obtained for all mixtures are clearly separated from the negative control G_NPM/R. Also this result undermines the inhibitory effect of the Cys21 substitutions on the NPM oligomerization in live cells.

We applied FLIM-FRET for tracking donor-acceptor proximity in mixed NPM oligomers to confirm independently the oligomerization of the FP-tagged NPM observed in live cells. The proximity should be manifested by shortening of eGFP fluorescence lifetime since excitation is then transferred from the donor eGFP to the acceptor mRFP1. A typical result obtained measuring cells expressing the G_NPM/R_NPM mix is presented in figure 4. NCL like NPM localizes preferentially into the nucleolus [46]. It exhibits also comparable fluorescence intensities in the nuclei of transfected cells. So NCL was used for comparison with NPM in the FLIM-FRET experiments (figures 4(A), (B)). In FLIM images (figures 4(C), (D)), we can clearly see significant shortening of the eGFP lifetime in cells expressing the G_NPM/R_NPM compared to the G_NCL/R_NCL. Pixel-by-pixel correlation analysis of the FLIM images is shown in figure 4(E). Pixel-based average eGFP lifetime τ_av is plotted against the corresponding fluorescence intensity. G_NPM/R_NPM τ_av values are well below G_NCL/R_NCL values. Therefore the G_NCL/R_NCL mix is used as a negative dual-color reference which helps to demonstrate presence of FRET in the case of the G_NPM/R_NPM.

Zoom In Zoom Out Reset image size

An alternative analysis of the same data is shown in figure 5. Fluorescence decays were averaged across individual nuclei and rigorously analyzed by least-squares reconvolution analysis yielding nucleus-based mean fluorescence lifetime τ_mean. This analysis confirmed significant shortening of the eGFP lifetime in cells expressing the G_NPM/R_NPM mix compared to control cells expressing G_NPM only, i.e. without presence of the acceptor. eGFP lifetimes observed for the G_NCL/R_NCL reference are even slightly longer compared to the G_NPM control. The absence of lifetime shortening comparable to the G_NPM/R_NPM in the G_NCL/R_NCL case indicates that FP crowding in a dense nucleolus environment does not necessarily mean marked eGFP lifetime decrease due to nonspecific FRET.

Zoom In Zoom Out Reset image size

For further confirmation of the FRET, we performed acceptor photobleaching, which is expected to restore the donor lifetime (figure 6). Two similar dual-color G_NPM/R_NPM cells exhibiting shortened eGFP lifetime were imaged. The mRFP1 acceptor in one of the cells was photobleached by a high-intensity laser light (550 nm) (figure 6(A)). The eGFP lifetime was restored from ∼2.2 ns to ∼2.4 ns as seen in the FLIM images (figure 6(B)), which corresponds to the lifetimes presented in figure 5. The pixel-by-pixel analysis of eGFP lifetimes in dependence on acceptor/donor intensity ratio (figures 6(C), (D)) and in dependence on eGFP intensity (figures 6(E), (F)) clearly shows that donor lifetimes and intensities shifted after the acceptor bleaching, which indicates FRET confirming the donor/acceptor proximity. Therefore in accordance with the PIE-ccN&B data, the observed FRET points to proximity of the labeled NPM in hetero-aggregates formed in live cells. We attribute the observed aggregates to the NPM oligomerization expected in live cells.

Zoom In Zoom Out Reset image size

Having validated the in vivo FLIM-FRET method with the WT NPM, we used it to investigate the influence of the Cys21 substitutions on the NPM oligomerization in live cells. Pixel-by-pixel correlation analysis of FLIM images showed that eGFP lifetimes have similar distribution in cells expressing the WT and the Cys21-mutated NPM (figures 7(A)–(C), respectively). The significant lifetime decrease with increasing donor/acceptor ratio points to FRET in mixed oligomers regardless from what FP-tagged NPM variant the oligomers are formed.

Zoom In Zoom Out Reset image size

Both the FLIM-FRET and the PIE-ccN&B experiments reveal that the Cys21-mutants exhibit oligomerization behavior in vivo similar to the WT NPM. To further support these findings, we applied in vitro co-immunoprecipitation utilizing a pull-down of eGFP-labeled proteins by the GFP-Trap beads [27]. The co-immunoprecipitation can be used for identification of interaction partners of the eGFP-labeled NPM. It should therefore detect endogenous NPM co-residing in the hetero-oligomers.

We transfected cells with plasmids for expression of free eGFP, G_NPM, G_NPM-C21A and G_NPM-C21F. The eGFP-labeled proteins are co-immunoprecipitated with their interaction partners (figure 8). The endogenous NPM is detected in each precipitate which points to tight interaction of the native NPM with the G_NPM and with both of the eGFP-tagged Cys21-NPM mutants. No interaction is found in the control sample containing free eGFP. The GFP-Trap proves that the eGFP-labeled NPM is able to form complexes with the endogenous NPM irrespective of the Cys21 mutations, which were expected originally to inhibit the NPM oligomerization.

Zoom In Zoom Out Reset image size