Nucleolar Localization of Nucleophosmin/B23 Requires GTP*

Incubation of HeLa cells with the IMP dehydrogenase inhibitors: ribavirin (100 pM, 4 h), tiazofurin (100 PM, 4 h), selenazofurin (100 pM, 4 h), or mycophenolic acid (10 p ~ , 4 h) resulted in -70% reduction in cellular GTP pools and shifting of nucleophosmin/B23 from nucleoli to nucleoplasm as detected by immunofluorescence (B23-translocation). Enzyme-linked immuno- sorbent assay and Western blot assay showed there is no loss or degradation of nucleophosmin/B23 protein during drug treatment. This translocation effect could be prevented by co-incubation with guanosine (100 pM) or reversed by addition of guanosine (100 p ~ ) to the culture medium after B23-translocation had been in- duced by these inhibitors. Under these conditions of guanosine supplementation, cellular GTP pool concen- trations were maintained at the control level. These results indicate that localization of nucleophosmin/B23 into the nucleolus is dependent on the cellular GTP level.

Incubation of HeLa cells with the IMP dehydrogenase inhibitors: ribavirin (100 pM, 4 h), tiazofurin (100 PM, 4 h), selenazofurin (100 pM, 4 h), or mycophenolic acid (10 p~, 4 h) resulted in -70% reduction in cellular GTP pools and shifting of nucleophosmin/B23 from nucleoli to nucleoplasm as detected by immunofluorescence (B23-translocation). Enzyme-linked immunosorbent assay and Western blot assay showed there is no loss or degradation of nucleophosmin/B23 protein during drug treatment. This translocation effect could be prevented by co-incubation with guanosine (100 pM) or reversed by addition of guanosine (100 p~) to the culture medium after B23-translocation had been induced by these inhibitors. Under these conditions of guanosine supplementation, cellular GTP pool concentrations were maintained at the control level. These results indicate that localization of nucleophosmin/B23 into the nucleolus is dependent on the cellular GTP level.
Nucleoli of rapidly proliferating cancer cells are large, pleomorphic, and hyperactive (1). These highly structured and specialized organelles are the sites of active transcription of rDNA and ribosome assembly. Numerous nucleolar proteins, RNA, and other elements are involved in these intricate processes to produce ribosomes (2,3).
Nucleophosmin/B23 (protein B23, N038, numatrin) is a major nucleolar phosphoprotein which is significantly more abundant in tumor and growing cells than in normal resting cells (4)(5)(6). Its localization in the granular region of the nucleolus (7) along with its association with preribosomal particles (8,9) suggest that it may be involved in ribosome assembly or processing. Induction of mitogenesis in B-lymphocytes is characterized by significant increases in nucleophosmin/B23 synthesis (lo), suggesting that this protein is associated with the transduction of mitogenic signals (6).
We have developed a B23-translocation assay which is a simple and rapid functional assay that directly indicates the effect of certain anticancer drugs, particularly the intercalators (9,(11)(12)(13)(14)(15)(16). The principle of this assay is based on the redistribution of nucleophosmin/B23 from nucleoli to nucleoplasm when cells are exposed to these anticancer drugs. B23translocation has been demonstrated in cell cultures to correlate with drug effects, and the assay is useful in detecting * This work was supported by National Cancer Institute Grant CA 42476. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore he hereby marked "aduertisement" in accordance with 18 U.S. C drug-resistant cancer cells (13)(14)(15)(16). B23-translocation has also been demonstrated to occur in an animal system, P388D1 leukemia implanted in DBA2 mice (17). The mechanism(s) of drug-induced redistribution of nucleophosmin/B23 is currently under investigation. The fact that only certain cytotoxic compounds have the capacity to induce B23-translocation (18) may provide some insight into the translocation mechanism, as well as into the function of nucleophosmin/ B23 in the cell. We now report that depletion of GTP pools by IMP dehydrogenase inhibitors (19)(20)(21)(22)(23)(24)(25)(26)(27) also causes B23translocation. This is the first report that adequate GTP levels are necessary for the localization of nucleophosmin/ B23 in the nucleolus.

MATERIALS AND METHODS
Drugs and Antibodies-Ribavirin, tiazofurin, and selenazofurin were synthesized (28)(29)(30). Toyocamycin was kindly provided by the Drug Synthesis Branch of the National Cancer Institute. Mycophenolic acid, actinomycin D, guanosine, guanine, hypoxanthine, xanthine, adenosine, inosine, and xanthosine were purchased from Sigma. Fluorescein isothiocyanate-conjugated goat anti-mouse IgG was also obtained from Sigma.
Cell Culture Studies-HeLa cells were grown as a monolayer in minimum essential medium supplemented with 10% newborn bovine serum in a 5% CO, humidified incubator at 37 "C. For immunofluorescence studies, cells were subcultured over glass slides in a Petri dish 2 days before use. Fresh media were added 2 h before the beginning of each experiment. Generally, cells were either incubated with single drugs or co-incubated with drugs and normal nucleosides for 4 h prior to assessment of B23-translocation. In reversal studies, cells were first induced to B23-translocation, and then guanosine or other purines were added to the culture media for 2 to 4 additional hours.
Immunofluorescence Staining-Immunofluorescence staining was performed as previously described (13). Briefly, cells were fixed in 2% paraformaldehyde and then permeabilized in cold acetone prior to overnight incubation with monoclonal antibody (diluted 1:16 in PBS)' that recognizes only nucleophosmin/B23. Fluorescein-conjugated goat anti-mouse IgG served as the second antibody. Slides were examined for the localization of nucleophosmin/B23 using fluorescence microscopy.

5823
of Nucleophosmin/B23 Requires GTP solution). The sheet was then washed (3 X 10 min) in TBST buffer 100 p~ ribavirin for 4 h compared to that observed in unbefore incubation with Promega substrates for color development.

treated control cells. Control cells have large and pleomorphic
then scraped and suspended in RSB buffer (10 mM Tris, pH 7.4, 10 ELZSA Studies-Monolayer HeLa cells were rinsed with PBS, nucleoli with bright fluorescence (Fig. 1, A and c), and the mM NaCl, 1.5 mM MgC12, 0.1 mM phenylmethylsulfonyl fluoride, 10 nucleoplasmic fluorescence is relatively weak-In contrast, p~ leupeptin, 0.1 mM p-chloromercuriphenylsulfonic acid, and soni-ribavirin-treated cells show a uniform and higher nucleocated with an Ultrasonic cell disrupter for -30 s. After centrifugation plasmic fluorescence with decreased nucleolar size and fluo-(30 min, 27,000 X g) to remove cell debris, the supernatant (whole rescence (Fig. 1, B and D) which indicates that nucleophoscell extract) was serially diluted in PBS containing 0.001% bovine m i n /~2 3 has shifted from nucleoli to nucleoplasm. L~~~~ serum albumin and loaded into wells of a microtiter plate (Immulon 11). The ELISA was performed as previously described by Yung et al. dosages (25 and 50 pM for h, and shorter times (9). The monoclonal antibody to nucleophosmin/B23 was diluted (100 p M for 2 h) resulted in noticeable but lesser effects (data 1:20,000 in PBS before use. not shown). To determine whether B23-translocation resulted Measurement of Cellular GTP Pools-High performance liquid from a reduction of the protein, the content of nucleophoschromatography (HPLC) was used to quantitate GI"I' in HeLa cell min/B23 in whole cell extracts before and after drug treatextracts after drug treatments. Preparation of cell extracts and chroment was determined by ELISA assay. Fig. shows that there et al. (31). Briefly, following incubation of cells in culture medium is no significant decrease Of nuc1eophosmin/B23 in drug-matography were performed essentially as described by Matsumoto containing the compounds, cells were scraped from the flask, counted, treated cells. Nucleophosmin/B23 was also analyzed by Westand centrifuged to pellet. The macromolecules were precipitated by ern blot assay. A single immunoband at M, = 37,000 was g for 5 min, and the acid extract was neutralized with trioctylamine lo6 cells). The precipitates were removed by centrifugation at 3000 X There is no detectable degradation of n u c~e o p~o s m~n /~~3 in freon (32). The nucleotides were separated by anion exchange after the drug treatment. Taken together, these results indichrolnatography on an Altex Ultrasil-Ax column (4.6 x 250 mm) at cate that ribavirin treatment causes redistribution of nucleoa flow rate of 1.5 ml/min with a gradient of solution A (10 mM phosmin/B23 from nucleoli to nuc~eop~asm. and the eluent was nos~lselenazole-4-carboxamide), were also tested for the abilmonitored a t 254 nm. Retention times for ATP and GTP were ity to induce B23-translocation. Both analogs were found to obtained by using authentic standards. be effective (Table I). Ribavirin and its analogs were previously reported to inhibit IMP dehydrogenase (19)(20)(21)(22)(23)(24)(25)(26). To
These results indicate that inhibition of IMP dehydrogenase causes B23-translocation. Depletion of GTP by IMP Dehydrogenase Inhibitors and B23-translocation-IMP dehydrogenase catalyzes the con version of inosine monophosphate (IMP) to xanthosine monophosphate (XMP), which is a precursor of guanosine monophosphate (GMP). Inhibition of this enzyme causes a reduction in de novo GMP synthesis resulting in depletion of cellular guanine nucleotide pools (33). To investigate the relationship between BPS-translocation and GTP pools, the GTP content of drug-treated cells was determined. Results are shown in Table 11. We found that the GTP pool concenwhile cells treated with ribavirin (100 PM, 4 h) is 0.24 f 0.01 nmol/106 cells, a reduction of -70%. The reduction in GTP concentration persists after 6 and 8 h of drug treatment (Table  11) as does the B23-translocation.
If depletion of GTP pools induced by ribavirin is the cause of B23-translocation, then replenishment of GTP should reverse the translocation. As shown in Fig. 40, addition of guanosine together with ribavirin effectively prevented B23translocation. At the same time, we observed that the GTP content of these cells was slightly higher than in untreated controls (Table 11). B23-translocation induced by tiazofurin, selenazofurin, or mycophenolic acid could be similarly blocked by guanosine (Table I). Of the other purines tested (adenosine, inosine, hypoxanthine, xanthine, xanthosine, and guanine), 20 only co-incubation with guanine effectively prevented ribavipm. A and C, control cells; I3 and D, ribavirin-treated cells (100 PM, rin-induced B23-translocation. 4 h); A and R, immunofluorescence; C and D, phase contrast.

B23 in control and ribavirin-treated HeLa cells. Bar in A,
In other experiments, guanosine was capable of reversing adding ice-cold 0.4 N perchloric acid to the cell pellet (100 p1 per 2 X observed in both control and ribavirin-treated cells (Fig. 3).  translocation previously induced by ribavirin. Fig. 4A shows HeLa cells incubated with ribavirin (100 p~) for 4 h then "rescued" by addition of 100 MM guanosine (Fig. 4, G and H).

TABLE I B23-translocation induced in HeLa cells bv selected cvtotoric wents
Nucleophosmin/B23 had almost completely relocalized into nucleoli within 2 h after addition of guanosine (Fig. 4G), and, within 4 h (Fig. 4H), cells exhibited control-like immunofluorescence. Similarly, we observed that the GTP content of these cells recovered to a higher level after the guanosine rescue (Table 11). There is a direct relationship between  Table I1 also summarizes the ATP levels in HeLa cells following drug treatment. While the GTP content decreases during ribavirin treatment, the ATP content is not affected (2.88 f 0.37 nmol/106 treated cells uersus 2.81 k 0.58 nmol/ lo6 control cells). The ratio of ATP/GTP in ribavirin-treated cells is higher than in control cells (-11 uersus 3.5). Although there are variations in the cellular ATP and GTP content in the guanosine supplementation experiments, the ATP/GTP ratios are low (mean = 1.9). The significance of the ATP/ G T P ratio for B23-translocation remains to be investigated.

DISCUSSION
IMP dehydrogenase catalyzes the conversion of IMP to XMP. Inhibition of this enzyme results in depletion of cellular guanine nucleotide pools. The current studies show that depletion of cellular GTP and/or alteration in the ATP/GTP ratio by IMP dehydrogenase inhibitors causes the redistribution of nucleophosmin/B23 from nucleoli to nucleoplasm (B23-translocation). This translocation process is reversible because it can be prevented or reversed by addition of guanine or guanosine to the culture medium, which allows the cell to utilize salvage pathways to replenish the depleted GTP pools. The mechanism relating G T P depletion to B23-translocation is not presently clear.
Since G T P is one of the precursors of the nucleic acids, and its depletion can cause reduction of DNA and RNA synthesis, an obvious question is whether nucleic acid synthesis inhibition per se causes the B23-translocation. We found that certain cytotoxic agents (including tubercidin, cytosine P-Darabinofuranoside, and 5-fluorouracil) which effectively inhibit RNA and/or DNA synthesis (34) do not induce B23translocation (18): These results suggest that nucleic acid synthesis inhibition and B23-translocation do not constitute a cause/effect relationship.  GTP and GDP are modulators of G-proteins which are involved in the signal transduction mechanisms of hormones, growth factors, and cytokines. The quantity of nucleophosmin/B23 in nucleoli increases with increased cell growth as observed in hypertrophic rat liver (35) and tumor cells (36). Also, a 5-to 20-fold increase in nucleophosmin/B23 is observed when B-lymphocytes are activated by mitogens (6). The increased nucleophosmin/B23 concentration in proliferating cells is associated with higher nucleolar activity. In general, GTP binding activates G-proteins. It is possible that depletion of GTP indirectly inhibits or interrupts the continuity of a G-protein-coupled "nucleolar activation pathway." One of the end results of this interference is the failure of nucleophosmin/B23 to localize in the nucleolus. There are at least two signal transduction pathways affected by G-proteins: 1) activation of adenylate cyclase which raises intracellular cAMP levels and 2) stimulation of phospholipid breakdown and the subsequent activation of protein kinase C and calcium channels. Depletion of GTP could ncgnt,ivaly affect tl~exe pathways. Preliminary results indicate that the cAMP level is unaffected by ribavirin treatment.
The reported Kd values for proteins that bind GTP range from 3 x to M (37). For example, Ras and EF-Tu have Kd values in the to lo" M range, while phosphoenolpyruvate carboxykinase has a Kd of 30 pM (38). The intracellular concentration of GTP in HeLa cells as determined in this study (0.81 nmol/1O6 cells, Table 11) is about 128 p~ (based on 160 X lo6 cells/ml). While a 3-4-fold reduction of this concentration would not affect the GTP binding of Ras or EF-Tu, it might affect the GTP binding of those proteins with high Kd values. The fact that a 3-4-fold reduction of GTP causes B23-translocation suggests that if a G-protein is indeed involved in the pathway for nucleolar activation, it would be one with a relatively high Kd and therefore sensitive to the prevailing GTP concentration. Regardless of which signal transduction pathway is involved in activation of the nucleolus, the end result is increased production of ribosomes. Intricate mechanisms that mobilize hundreds of proteins, RNAs, and other elements in the nucleolus are involved in ribosome production (3). The making of ribosomes is dependent on a balanced supply of rRNA and ribosomal proteins (39) and a proper conformation of rRNA (40). Multiple steps are involved, such as transport of ribosomal proteins to the nucleolus and binding to nascent pre-rRNA, splicing of the 45 S RNA, packaging of ribosomal proteins into ribosomal subunits, and delivery of the preribosomal ribonucleoprotein to the cytoplasm. Interference with any one of these processes could result in abortion of ribosome synthesis and B23-translocation. Actinomycin D and toyocamycin induce BZ3-translocation (Refs. 9 and 11, Table I, and Fig. 4, B and F ) . However, these drugs do not affect cellular GTP or ATP levels in a similar way as ribavirin (Table 11). In addition, translocation induced by these drugs is not reversed or prevented by guanosine (Fig. 4, E and F , and Table I). Since B23-translocation induced by toyocamycin and actinomycin D is not associated with reduced GTP pools, we are led to conclude that there is more than one mechanism of B23-translocation. Recent studies show that the nucleolus may also play an important role in virus replication (41) and transcription. Two HIV proteins, Tat and Rev, which are essential for the early regulation of viral gene expression (42), are found localized in nucleoli (43)(44)(45). Fankhauser et al. (46) reported that nucleophosmin/B23 is specifically bound by the HIV Rev protein which is responsible for the transport of unspliced mRNA to the cytoplasm (47,48). The virus may utilize the nucleolar machinery during the early stages of replication, and nucleophosmin/B23 could serve as a shuttle protein (49) for the transport of viral as well as ribosomal proteins. This process may require GTP. Ribavirin is a clinically useful, broad spectrum antiviral agent (19,(50)(51)(52). The fact that ribavirin induces B23-translocation suggests that one of the drug's actions is in the nucleolus, possibly to limit the virus's usage of this GTP-dependent transport mechanism.