NSC348884 cytotoxicity is not mediated by inhibition of nucleophosmin oligomerization

Nucleophosmin (NPM) mutations causing its export from the nucleoli to the cytoplasm are frequent in acute myeloid leukemia (AML). Due to heterooligomerization of wild type NPM with the AML-related mutant, the wild-type becomes misplaced from the nucleoli and its functions are significantly altered. Dissociation of NPM heterooligomers may thus restore the proper localization and function of wild-type NPM. NSC348884 is supposed to act as a potent inhibitor of NPM oligomerization. The effect of NSC348884 on the NPM oligomerization was thoroughly examined by fluorescence lifetime imaging with utilization of FRET and by a set of immunoprecipitation and electrophoretic methods. Leukemia-derived cell lines and primary AML cells as well as cells transfected with fluorescently labeled NPM forms were investigated. Our results clearly demonstrate that NSC348884 does not inhibit formation of NPM oligomers neither in vivo nor in vitro. Instead, we document that NSC348884 cytotoxicity is rather associated with modified cell adhesion signaling. The cytotoxic mechanism of NSC348884 has therefore to be reconsidered.


Results
Interaction and stability of NPM with C21 point mutation. Evaluation of NSC348884 ability to affect NPM oligomerization in living cells and cell lysates requires sensitive methods for oligomerization monitoring as well as reliable positive and negative controls. To closely simulate the native conditions, we preferentially searched for point mutations that were reported to inhibit NPM oligomerization.
Point mutation in C21 was shown to be important for NPM oligomerization 54 . Recently, we have shown that conclusions about oligomerization of these mutants obtained under rather harsh conditions of the SDS-PAGE 54 might strongly differ from results obtained in living cells by fluorescence lifetime imaging (FLIM-FRET) and by immunoprecipitation in lysates 55 . We demonstrated that some C21 NPM-mutants, originally expected not to oligomerize 34,54 , are actually able to form complexes with the endogenous NPM in living cells. In addition, immunoprecipitation revealed interaction of the endogenous NPM with eGFP-labeled NPM bearing point mutation at C21 (G_C21), i.e. where C21 was substituted either to Ala (G_C21A), or to Phe (G_C21F). In the present work, we used fluorescence microscopy and native electrophoresis to characterize in detail the impact of Scientific Reports | (2021) 11:1084 | https://doi.org/10.1038/s41598-020-80224-1 www.nature.com/scientificreports/ these C21 mutations. Both G_C21 mutants exhibited nucleolar localization, identical to that of NPMwt (Fig. 1). Furthermore, the mRFP1-labeled NPM_C21 (R_C21) was found in the cytoplasm of HEK-293T (293T) cells co-transfected with eGFP-labeled NPMmut (G_NPMmut). The fraction of cells exhibiting mRFP1 signal in the cytoplasm was comparable for all R_NPM variants (Fig. 1). Both C21 mutants therefore seem to form heterooligomers with NPMmut, alike NPMwt in living cells.
To independently verify these findings, we examined native lysates from cells expressing G_C21A or G_C21F by semi-native and native electrophoresis 49 . Briefly, the lysates from cells harvested into non-reducing, nondenaturing buffer were directly subjected without boiling to acrylamide gel with and without SDS, respectively (see Material and Methods). These relatively gentle separation methods allowed for detection of both NPM monomers and oligomers and for estimation of their electrophoretic mobility. The monomer/oligomer band intensity ratio in semi-native conditions reflects the propensity of oligomers to dissociate into the monomers. Results are shown in Fig. 2a. Whereas C21A exhibited a high-MW band (presumably oligomers) identical to that of NPMwt under native conditions, the band from C21F was located at the position corresponding to the weak lower-MW fraction of NPMwt or NPMmut (presumably monomers). In these experiments, the band from endogenous NPM oligomers served as a loading and position control. The results from semi-native electrophoresis in Fig. 2b show markedly increased monomer/oligomer ratio of C21A compared to NPMwt and absence of C21F oligomers. These results suggest that although NPM oligomerization seems to be unaffected by the C21 point mutation in living cells, the stability of the oligomers is considerably attenuated. Similar results were obtained with mRFP1-labeled variants. Interestingly, mRFP1-labeled proteins displayed slightly lower mobility in the native conditions. Moreover, R_NPM oligomers were found to be somewhat more stable compared to G_NPM ones (Fig. 2b).
We have not observed any changes in expression and in oligomerization state of the endogenous NPMwt in cells containing C21 mutants. Therefore, we further investigated the effect of C21F substitution on the stability of exogenous heterooligomers formed by a mixture of fluorescently labeled C21F and NPMwt. 293T cells were alternatively transfected with single fluorescent variants of C21F and NPMwt, or co-transfected with the combination of both. As shown in Fig. 3 and Supplementary Fig. S1, the exogenous NPMwt monomer/oligomer ratio is strongly affected by the presence of C21F. Band intensities attributed to C21F oligomers and NPMwt monomers are both considerably higher in traits corresponding to the co-transfected sample than the ones in traits from single transfected cells. Distribution of NPM molecules between oligomer (high-MW) and monomer (low-MW) bands likely depends on the NPMwt/C21F participation in the heterooligomers. Interestingly, endogenous NPM monomer/oligomer ratio seems to be unaffected by the presence of any exogenous NPM. Altogether, despite the C21F mutation does not completely abrogate NPM oligomerization, it clearly attenuates interaction affinities underlying the mixed oligomer formation in living cells.
Localization and oligomerization properties of NPM N-terminal deletion mutants. In view of the fact that the C21-point mutations do not cause any detectable changes of the NPM oligomerization in living Error bars represent ± SD of at least 3 independent experiments, p < 0.0001 (****) for the differences between the co-transfected and the single-transfected cells.     56 determined residues and regions accountable for NPM oligomerization, for its nucleolar localization, and for p14Arf binding. Their data disclosed that NPM lacking a part of the N-terminal domain was localized in the nucleoplasm and exhibited inability to interact with other NPM molecules. For negative controls in living cells, we therefore created fluorescently labeled N-terminal NPM mutants with deletions of the first 25, 100 and 117 amino acids (Δ25, Δ100, Δ117) and we analyzed their subcellular localization and oligomerization characteristics. Our confocal imaging experiments have revealed that all the truncated NPM forms reside both in the nucleoli and in the nucleoplasm (Fig. 4). As expected, the largest deletion resulted in an increased accumulation of the mutant in the nucleoplasm.
To monitor NPM complex formation, lysates from 293T cells transfected with fluorescently labeled deletion mutants were subjected to electrophoresis. As expected, inability of the truncated proteins to form oligomers was reflected by an absence of the high-MW bands under native and semi-native conditions (Fig. 5a). Further we searched for presence of NPM heterooligomers containing exogenous NPMwt and selected deletion mutants. 293T cells were co-transfected with plasmids ensuring expression of Δ117 and NPMwt and the lysates were analyzed ( Fig. 5b and Supplementary Fig. S2). It can be seen that bands from oligomeric NPMwt complexes are unaffected by the presence of Δ117 in the native as well as in the semi-native immunoblots. The result suggests that, in contrast to the C21 mutants, the presence of the deletion mutants does not affect the oligomerization of NPMwt. Again, expression of the endogenous NPMwt remained unchanged. To analyze the interaction potential of the N-terminal deletion mutants, 293T cells were transfected with plasmids encoding for GFP-labeled NPMwt, Δ25, Δ100 or Δ117. Then immunoprecipitation using GFP-Trap was performed and the precipitates were analyzed for the presence of endogenous NPM as well as for presence of other nucleolar proteins known to interact with NPM. Surprisingly, although the GFP-tagged deletion mutants displayed no ability to oligomerize, interaction of the endogenous NPM with any deletion mutant (except the plasmid encoding for free eGFP) was detected. The amount of co-precipitated endo-NPM was only slightly lower compared to the G_NPMwt precipitates (Fig. 5c). Simultaneously, level of co-precipitated nucleolin (NCL), which does not interact with the AML-related NPMmut 49 , was higher in precipitates of the truncated NPM forms. The level of another coprecipitated nucleolar protein, fibrillarin (FBL), also positively correlated with the extent of N-terminus deletion. On the other hand, the tumor suppressor p14Arf, which is known to interact with the N-terminal NPM domain 56 , clearly co-precipitated only with the G_NPMwt.
To evaluate the utility of the method for detection of the complexes containing both NPMwt and the deletion mutants, we performed GFP/RFP-immunoprecipitation from cells co-transfected with R_NPMwt and G_∆117. The co-transfection of R_NPMwt and G_C21F served as an interacting control. Whereas NPMwt was clearly detected in samples obtained by precipitation of the deletion mutant, the vice versa co-precipitation failed (Fig. 5d). Identical results were obtained for combination NPMwt+Δ100. No interaction was found between two color variants of the N-terminal mutant ( Supplementary Fig. S3).
The Förster resonance energy transfer (FRET) is a robust spectroscopic method for evaluation of a donor-acceptor proximity and works as a "ruler" on the nanometer scale. Due to the inverse 6th-power dependence of the transfer efficiency on the donor-acceptor distance 57 , the energy transfer occurs only between closely separated donor-acceptor pairs. This occurs e.g. within NPM oligomers comprising eGFP-and mRFP1-tagged subunits, where eGFP is a donor and mRFP1 acceptor. Complex formation therefore results in more efficient FRET, which is reflected in decreased fluorescence lifetime of eGFP. FRET-induced changes of the donor fluorescence lifetime can be mapped across the microscopic samples by fluorescence lifetime imaging (FLIM) 58 . FRET-FLIM is perfectly suited for live-cell imaging and we previously successfully used it to detect complexes formed by NPMwt and C21 mutants in living cells 55 . In the present work, the FRET between eGFP as a donor and mRFP1 as an acceptor attached to NPMwt and Δ117 (Δ100) was examined. Presence of FRET within the complex was detected by the mRFP1 photobleaching, since an increase in the eGFP lifetime upon the mRFP1 photodestruction is a strong positive indicator of the mixed-complex formation. Results are shown in Fig. 6. Panels A and B show the initial localization of the green and red signal within the cells, panels C and D the intensity ratio I red /I green before and after the photobleaching, respectively. Corresponding FLIM images and lifetime histograms from the analyzed nucleolar area are shown in the panels E, F and G, respectively. As expected, a significant lifetime increase resulting from the FRET cancellation was detected after photobleaching in cells cotransfected with G_NPMwt and R_NPMwt. On the contrary, virtually no effect was observed in cells transfected with the G_Δ117 + R_Δ117 mutants. In accord with the literature 56 and with our precipitation data, this result www.nature.com/scientificreports/ indicates inability of the Δ117 mutants to interact with each other and to form multimers. This protein pair can therefore serve as a negative control for the multimer formation in the NSC348884 experiments. As seen from Fig. 6, in samples containing combination of the Δ117 deletion mutants with NPMwt, the GFP-lifetime slightly, but still visibly, increases upon the acceptor photobleaching. The change corresponds with the results of our electrophoretic experiments and suggests some amount of the mixed multimer to be present in the cell. The statistical analysis from multiple experiments (n = 3-5) is presented in Fig. 6H, where cells transfected with G_Δ117 (the donor only transfection) served as a negative control. The figure clearly proves oligomerization of NPMwt. Oligomerization of Δ117 deletion mutants is clearly undetected and the presence of mixed NPMwt + Δ117 multimers is at the significance limit. Interestingly, the statistical evaluation also suggests the asymmetric character of interaction between NPMwt and Δ variants. These results are in a nice agreement with the immunoprecipitation.   41 . Furthermore, it was proved to inhibit proliferation, to upregulate p53 and to trigger apoptosis 41,50 . We thus investigated the influence of NSC348884 treatment in a panel of leukemia cell lines complemented with HeLa and 293T cells.
Prior to characterization of the NSC348884 effect on NPM oligomerization, we performed the analysis of cell viability and apoptotic markers. The cell viability in the presence of NSC348884 was monitored by propidium iodide (PI) exclusion test (Fig. 7a). Caspase-3 fragmentation as well as changes of p53 expression were investigated by immunoblotting to assess the extent of apoptosis in NSC348884-treated cells (Fig. 7b). For the majority of the cell lines, the EC50 value was within the interval from 2 to 10 µM. The viability drop correlated with increased caspase-3 fragmentation indicating the onset of apoptosis (Fig. 7b). Simultaneously, NSC348884induced increase in the p53 level was detected in some of the cell lines possessing wild-type p53. Contrarily to previously reported results 41 , the majority of cell lines with NPMwt was more sensitive than the cell line with NPMmut (OCI-AML3). Comparable sensitivity (from caspase-3 fragmentation) to NSC348884 treatment was found also for the primary cells of AML patients regardless of their NPM mutational status (Fig. 7c). Unexpectedly, native PAGE experiments revealed no influence of NSC348884 on NPM oligomerization (Fig. 8). Endogenous NPM oligomers were found to be stable in KG-1, HL-60, MV4-11, and HeLa cell lines, which exhibited extensive apoptosis after NSC348884 treatment, as well as in OCI-AML2, OCI-AML3 and 293T, which were substantially more resistant to the treatment.
To further investigate effect of NSC348884 in vivo, we co-transfected 293T and HeLa cells with R_NPMwt and G_NPMmut. Then the cytoplasmic localization of R_NPMwt was monitored for 2 h after addition of 10 µM NSC348884 (Fig. 9). In agreement with our previous results 36,59 , detectable fraction of R_NPMwt was found in the cytoplasm of both cell lines at the starting time point. Lower cytoplasmic fraction of R_NPMwt in HeLa cells (compared to 293T) likely results from a higher endogenous NPM level 36,60 . Importantly, the cytoplasmic localization of R_NPMwt remained unchanged for at least 2 h after the treatment suggesting independence of NPM oligomerization on the presence of NSC348884 in vivo (Fig. 9). Simultaneously, there was an obvious effect of NSC348884 on cell-surface adhesivity. The effect is clearly visible in transmitted light images (DIC). Whereas the 293T cells progressively rounded and finally lost their contact with the glass surface of the culture dish, the  www.nature.com/scientificreports/ HeLa cells detached from the surface individually. In any case, mitotic cells that detached from the surface for cell division have never re-adhered. The oligomerization of fluorescently labeled NPM was finally tested by the native PAGE and by immunoprecipitation in cell lysates and by FRET in living cells. First, we tested whether the low-MW band attributed to NPM monomers appears in native lysates of 293T cells expressing a combination of G_NPMmut and R_NPMwt after the NSC348884 treatment. Lysate from cells co-expressing G_NPMmut and weakly oligomerizing R_C21F was used to mark the position of the low-MW band ( Fig. 10a and Supplementary Fig. S4). No difference between the control and the NSC348884-treated sample was found either under native or semi-native conditions. Similar results were obtained from cells co-transfected with alternative combinations, i.e. with G_NPMwt + R_NPMwt or with G_NPMmut + R_NPMmut (Fig. 10b). Identical samples were afterwards subjected to immunoprecipitation (GFP-and RFP-Trap). All the exogenous NPM forms as well as the endogenous NPM were detected in all GFP-and RFP-precipitates regardless the NSC348884 addition (Fig. 11). In agreement with our previous work 49 , control experiment revealed that NCL co-precipitated with NPMwt and it did not co-precipitate with any form of NPMmut. Again, the NPM-NCL interaction was not affected by the presence of NSC348884 in any experiment.
Oligomerization in living cells was independently tested by the resonance energy transfer. As seen from Fig. 12, FLIM-FRET experiments reveal unchanged eGFP fluorescence lifetime upon NSC348884 treatment of cells co-transfected with donor-and acceptor-labeled NPMwt and NPMmut. Prolonged eGFP-fluorescence lifetime after mRFP1-photobleaching confirmed the complex formation in control cells without NSC348884 (column 1 and 2). The NSC348884-treatment did not affect the lifetime pattern (column 3) and the second round of the acceptor bleaching confirmed persistence of heterooligomers despite the presence of NSC348884 (column 4). Lower FRET extent in the NPMmut co-transfected cells (the second row) is likely a result of lower cytoplasmic NPMmut concentrations and consequent lower complex formation. Nevertheless, presence of FRET is still detected. NSC348884 activity resulting in cell rounding and loss of their contact with the glass surface is visibly documented by the morphology screening during the FLIM experiments (columns 5, 6), similarly to Fig. 9. No lifetime change following mRFP1-photobleaching or NSC348884 treatment was detected in the control sample ( Supplementary Fig. S5), i.e. in cells expressing two color variants of NCL, where FRET should not be detected 55 . Altogether, FLIM-FRET experiments confirmed that NSC348884 does not affect NPM oligomerization, although it influences apoptosis and cell adhesion.
To further investigate the mechanism of NSC348884 action, we analyzed changes in the cell-surface contact after NSC348884 addition with help of Electric Cell-Substrate Impedance Sensing (ECIS) technique. ECIS allows www.nature.com/scientificreports/ for non-invasive real-time monitoring of cell interaction with the surface of the sample well. Small microelectrodes embedded in the bottom of the ECIS plate serve to measure the impedance in a range of frequencies of the sensing electric current, and the signal is then automatically decomposed into resistance and capacitance.
In our experiments, the capacitance component at a high frequency (64 kHz) mirrored that of the resistance at 2 kHz and reflected mainly the area of the cell-surface contact. Before the drug addition, the progressive increase of the signal reflects cell attachment and proliferation. With this device, we were able to follow the time course of the adhesivity decrease after NSC348884 addition (Fig. 13). We have reported previously, that inhibition of SRC family kinases by dasatinib resulted in a rapid drop of ECIS signal, which corresponded to cell shrinkage, and we thus used dasatinib as a reference compound 61 . In both adherent cell lines, 293T and HeLa, NSC348884 induced large, dose-dependent changes in the resistance signal, which were similar to those produced by IPA-3 62 , an inhibitor of p21-activated kinases (PAK). PAK are key regulators of adhesion signaling, which have been proposed as therapeutic targets in different kinds of cancer including leukemias 63,64 . We thus analyzed possible effect of NSC348884 on expression and activity of PAK1, as well as of Cofilin, which governs actin remodeling during changes of cell shape. Indeed, Ser144 phosphorylation of PAK1 reporting on its kinase activity was reduced after 2 h of NSC348884 treatment whereas total PAK1 expression remained unchanged (Fig. 14). Simultaneously, inactivating phosphorylation at Ser3 of Cofilin was detected.

Discussion
The N-terminal region of NPM is essential for its oligomerization as well as for its chaperone function as numerous proteins interact with NPM through this domain. Since AML-related NPM mutation does not substantially affect its ability to form oligomers, NPM-interacting proteins become frequently mislocalized together with aberrantly localized NPMmut. Targeting the NPM oligomerization offers a possibility to manipulate localization of the interacting partners. Simultaneously, profiting from slightly different oligomerization properties of NPMmut and NPMwt, a fine control of NPM oligomerization by appropriate concentrations of oligomerization-inhibiting drugs might have a therapeutic effect in the AML with NPM1 mutation. Although several alterations of NPM N-terminal domain were reported to disrupt NPM oligomerization in vitro, results thoroughly describing NPMwt and NPMmut oligomerization in vivo are missing. We have previously documented that C21 point mutations   www.nature.com/scientificreports/ do not disrupt NPM oligomerization in living cells 55 . Here we demonstrate that the cytoplasmic mislocalization of R_C21 and R_NPMwt in cells co-expressing G_NPMmut is very similar (Fig. 1). This strongly indicates an existence of interaction between natural NPM forms and the C21 mutants. Our results from native and semi-native electrophoreses allowed us to evaluate the aggregation potential of C21 mutants depending on the substituting aminoacid (Fig. 2) in vitro, in agreement with the results of Prinos et al. 54 obtained mainly in experiments with recombinant proteins. Whereas C21F substitution resulted in disruption of NPM oligomers under native conditions ( Fig. 3 and Supplementary Fig. S1), the effect of the C21A substitution was only detectable under reducing conditions. NPM oligomers were reported to consist of five NPM molecules 30 , and formation of heterooligomers containing NPMwt and NPM mutants were found to be highly frequent 35,36 . We therefore suggest that the in vivo stability of NPM heterooligomers is permitted by a sufficient number of NPMwt molecules in the heterooligomeric complex. Thus, in living cells, the C21 point mutations do not have potential to fully disrupt these complexes, although their ability to retain in oligomers is compromised under in vitro conditions. In agreement with results of Enomoto et al. 56 , partial or complete deletion of NPM oligomerization domain (aa1-117) led to delocalization of the truncated protein from the nucleoli to the nucleoplasm (Fig. 4). However, even the NPM with completely deleted N-domain (Δ117) exhibited higher concentration in the nucleoli compared to the nucleoplasm. This is likely due to the fact, that nucleolar localization signal as well as nucleic acid binding domains remain unaffected by the deletion. We have found that oligomerization of the deletion mutants was completely abrogated and no interaction between two truncated NPM forms was detected ( Fig. 5 and Supplementary Figs. S2, S3). Nevertheless, immunoprecipitation revealed presence of both exo-and endogenous NPMwt in the G_Δ117 precipitates indicating that the truncated NPM yet participates in complexes that are possibly too large to enter the native gel. The existence of mixed Δ117/NPMwt complexes is also supported by the FRET results monitoring the Δ117-NPMwt interaction in living cells (Fig. 6). Enhanced level of nucleolar proteins NCL and FBL co-precipitated with the deletion mutants suggests better accessibility of the NPM region responsible for binding of these proteins. Both NCL and FBL were previously found to reside in the nucleoli of cells with NPM mutation 46,49,59 and they may therefore mediate the nucleolar localization of the deletion mutants. Complexes containing NCL and/or FBL together with NPMwt and Δ117mutants thus represent a potential pool of proteins that can co-precipitate with the deletion mutants.
NSC348884 is declared to inhibit NPM oligomerization 50 . Its structure was obtained by in silico screening using a small molecular library. As the crystal structure of the human NPM was not available, Xenopus NO38core chaperone structure (residues 1-107) was used for the screening. The sequence identity of this structure with equivalent part of human NPM oligomerization domain is 77%. NSC348884 was identified as the best candidate for inhibition of the dimerization interface of this polypeptide. Authors of the original paper 50 used native electrophoresis to demonstrate NSC348884-induced oligomer disruption in LNCaP and HCT116 cell lines. They detected diminished bands near 121 kDa, which they attributed to NPM oligomers. Surprisingly, the intensity of monomer bands remained unchanged. Other complex phenomena like induction of apoptosis and p53 upregulation were also found in NSC348884-treated cells 41,50 . From the available data it seems that function www.nature.com/scientificreports/ of NSC348884 as an inhibitor of the full-length human NPM oligomerization was not unequivocally proven. We therefore analyzed effect of NSC348884 on various leukemia cell lines and on cells expressing fluorescently labeled NPM constructs. First, we tested cell viability and apoptotic signatures in order to determine the range of proper NSC348884 concentrations for the live-cell experiments. Concentrations required for a substantial viability decrease and caspase-3 fragmentation fell into the interval of 2-10 µM for the majority of the cell lines (Fig. 7). Neither the OCI-AML3 cell line nor the primary cells of AML patients with NPM mutation displayed enhanced sensitivity to NSC348884 treatment. Consistently with previous results 41,50 , NSC348884 induced p53 upregulation in some cell lines. We also noticed NSC34884-induced downregulation of p14Arf in 293T and HeLa cells (data not shown). In contrast to results of Balusu et al 41 , none of the tested cell lines nor primary cells from AML patients displayed any change in the NPM oligomerization upon treatment with efficient NSC348884 concentrations when investigated by the native and semi-native electrophoreses (Fig. 8). Similarly, oligomers containing fluorescently labeled NPMwt and NPMmut were not affected by the NSC348884 treatment ( Fig. 10 and Supplementary Fig. S4). These results were further verified by immunoprecipitation where both exogenous and endogenous NPM co-precipitated with both GFP-and RFP-labeled NPMwt and NPMmut, despite the presence of NSC348884 (Fig. 11). Also the fluorescence microscopy revealed sustained fraction of NPMwt in the cytoplasm of NSC348884 treated cells co-expressing NPMwt and NPMmut, which witnesses for their interacton. Accordingly, the FLIM-FRET proved persisting interaction between fluorescently labeled NPM molecules upon the NSC348884 treatment (Fig. 12). Cells expressing two fluorescent variants of likely noninteracting NCL were used as a control (Supplementary Fig. S5). As expected, NCL molecules labeled with the eGFP donor and the mRFP1 acceptor on their N-termini did not exhibit any FRET, which was proved by a zero lifetime change upon the acceptor photobleaching. The result was independent of the NSC348884 treatment. Compared to NCL, the presence of FRET in the cells with fluorescently labeled NPM is clearly detectable both before and after the NSC348884 treatment. We conclude that contrary to the published data 50 , NSC348884 does not act as an oligomerization inhibitor and does not affect formation of NPM oligomers under physiological conditions. This finding is extremely important in view of the fact that this drug has been recently reported to cause numerous cellular effects, which were ascribed, in accordance with its declared function, to disruption of NPM oligomerization [51][52][53] .
During the live-cell experiments, we noticed apparent changes in cell adhesivity. The cell-surface contact area during the NSC348884 treatment was therefore monitored by Electrical Cell-Substrate Impedance Sensing (ECIS) (Fig. 13). The rapid onset of changes in the ECIS signal indicated that the cell shrinkage and detachment was not a secondary effect accompanying apoptosis. As the course of the ECIS signal was similar to that induced by the inhibitor of p21-activated kinases, IPA-3 62 , we investigated also activity and expression of PAK1 and Cofilin, a known actin regulator 63 (Fig. 14). The observed changes of both PAK1 and cofilin phosphorylation indicate that NSC348884 interferes with adhesion signaling. Further research is required to elucidate mechanistic role of NSC348884 in this process and its potential for anticancer therapy.

Conclusion
We have shown that a proposed inhibitor of NPM oligomerization, NSC348884, does not affect NPM oligomer formation in any of the examined leukemia cells. Moreover, the cell sensitivity to NSC348884 treatment is not potentiated by AML-associated NPM mutation. On the other hand, we have uncovered so far unknown effect of NSC48884 on the cell-surface adhesion, which could play a key role in the complex cellular response to the NSC48884 treatment.
In addition, our findings prove that point mutations in Cysteine 21 slightly potentiate oligomer dissociation but the overall NPM interaction potential with other NPM molecules remains conserved in living cells. Deletion mutants lacking part of the NPM N-terminal domain completely lose their oligomerization ability, but they partially retain the interaction with NPMwt, possibly through enhanced interaction with other nucleolar proteins in complexes with NPMwt.

Material and methods
All methods were carried according to Declaration of Helsinki. www.nature.com/scientificreports/ Plasmid construction and transfection. As we described in detail previously 36,59 , the gene for NPM was amplified from cDNA library (Jurkat cells, Origene) by PCR and inserted to vectors peGFP-C2 and pmRFP1-C2 (originally Clontech), designed for expression of protein chimeras with a fluorescent protein connected to the N-terminus of the target protein, by standard methods of molecular cloning. NPM mutants were constructed by PCR using extended primers targeting NPM1 sequence neighboring regions cut from the N-terminus or containing the mutated part of the exon 12 of the NPM1 gene complemented with appropriate restriction sites 36 . After amplification in E. coli, the plasmids with subcloned genes were purified with PureYield Plasmid Miniprep System (Promega) and transfected into adherent cell lines using jetPRIME transfection reagent (Polyplus Transfection).
Cell lysis and western blotting. Cell lysis As described previously 49  Native and semi-native PAGE Lysates were mixed with 2xnative buffer (NB, 50 mM Tris pH6.8, 10 mM DTT, 10% glycerol) and subjected without boiling to 7.5% AA Tris-glycin gel without SDS for native electrophoresis, or to the gel with SDS (2%) for semi-native electrophoresis.
Immunoprecipitation. Immunoprecipitation using GFP-or RFP-Trap (Chromotek) was performed according to manufacturer´s instruction as described previously 36 . Briefly, cells were harvested and washed with PBS, lysed in LB for 30 min/4 °C and centrifuged at 20.000 g/4 °C for 10 min. The lysate was mixed with GFP/RFP-nanobeads and rotated for 1 h/4 °C. The beads were extensively washed with diluting buffer (10 mM Tris/Cl pH7.5, 150 mM NaCl, 0.5 mM EDTA), resuspended in SB, boiled at 95 °C for 10 min and centrifuged 20.000 g/4 °C for 10 min. Supernatant was stored at − 20 °C until used for SDS-PAGE.
Live-cell imaging. The cells were seeded in the 2.5 mm culture dish with glass bottom (Cellvis) for 24 h and then transfected with plasmids containing fluorescent variants of the desired genes. After another 24 h, the transfected cells were observed under confocal laser scanning microscope FluoView FV1000 (Olympus Corporation) using 543 nm excitation for RFP fluorescence and 488 nm excitation for GFP and for differential interference contrast (DIC) observation. UPlanSAPO 60 × NA1.35 oil-immersion objective was used for imaging. For long-term monitoring, the culture dish was sealed by parafilm to prevent CO 2 leakage and it was placed into microscopy chamber tempered to 37 °C. NSC348884 was added just before the start of the measurement. Fluorescence images were processed by the FluoView software FV10-ASW 3.1.
Lifetime imaging and acceptor bleaching. The apparatus used for lifetime imaging is described in detail elsewhere 65 . Briefly, we used inverted IX83 microscope equipped with a FV1200 confocal scanner (Olympus, Germany), cell cultivation chamber (Okolab) and FLIM add-ons from PicoQuant. Fluorescence was excited by a pulsed diode laser (LDH-DC-485, 485 nm, PicoQuant) running at 20 MHz repetition rate. Light was coupled to the microscope by a single-mode optical fiber and reflected to the sample by 488 nm long-pass dichroic mirror (Olympus). Typically, UPLSAPO 60XW NA 1.2 water-immersion objective (Olympus) was used for imaging. Fluorescence was directed via multimode optical fiber to a cooled GaAsP hybrid PMT (PicoQuant) through the 520/35 bandpass filter (Semrock). Signal was processed by the TimeHarp 260-PICO TCSPC card and the SymPhoTime64 software (both PicoQuant). To avoid pile-up artifacts, the data collection rate at brightest pixels was kept below 5% of the excitation frequency. FLIM images were collected in a few minutes with the excitation power around 0.1 μW. Acceptor photobleaching was done by a 561 nm semiconductor CW laser with a multi-mW power at the focal point. All experiments were done at 37 °C.
Lifetime data processing. Lifetime images were generated in the SymPhoTime64 by the "fast-FLIM" method when pixel lifetimes were calculated by a method of moments 66 . Specifically, pixel lifetimes τ avg were determined as the difference between the barycenter of the fluorescence decay and the time-offset t offset of the steepest growth of the decay at each pixel: www.nature.com/scientificreports/ where I i is a decay intensity at time t i . Exported FLIM images were further processed and visualized by the Fiji software 67 . An accurate analysis of the cumulative decays from larger area of interest was done by the least-squares reconvolution also in the SymPhoTime64. Fluorescence was assumed to decay multiexponentially according to the formula: where τ i and α i are the fluorescence lifetimes and the corresponding amplitudes, respectively. Typically, 2 decay components were sufficient for acceptable fit. The intensity-weighted mean fluorescence lifetime was calculated as: where f i are fractional intensities of the ith lifetime component: Electrical cell-substrate impedance sensing (ECIS). Impedance measurements were performed using the ECIS Zθ device (Applied Biophysics). The wells of 8W10E+ plates were filled with 200 µl culture medium and the baseline was monitored for several hours before cell addition. HeLa or 293T cells were seeded at 120.000 cells/well and monitored overnight, the inhibitors were added after 20-24 h. One well from each plate was left empty (medium only), and the signal from this well was used as the baseline for the other wells of the same plate. The instrument automatically decomposes the impedance signal into resistance and capacitance. As the course of capacitance at 64 kHz mirrored that of resistance at 2 kHz, the observed evolution of the resistance signal reflects changes in the cell-surface contact area. The ECIS records were exported to Microsoft Excel and processed using the GraphPad Prism software: the background was set to zero at a time point shortly before cell seeding, and the baseline (empty well) was subtracted. The curves shown in the graphs represent the averages from replicate wells, which were run in parallel.

Statistical analyses.
As described in our previous work 36 , the majority of experiments were performed using cell lines and repeated until the observed differences between groups reached statistical significance. A p value of 0.05 or lower was pre-set to be indicative of a statistically significant difference between groups compared. In diagrams, arithmetic means of replicates of all experiments were plotted with SD error bars. Significance levels (p values of ANOVA or Student's t-test) were determined using InStat Software (GraphPad Software).