Differential Phosphorylation and Localization of the Transcription Factor UBF in Vivo in Response to Serum Deprivation IN VITRO DEPHOSPHORYLATION OF UBF REDUCES ITS TRANSACTIVATION PROPERTIES*

We have analyzed the expression, phosphorylation, and localization of the ribosomal DNA transcription factors UBFl and UBF2 in Chinese hamster ovary cells in response to serum deprivation. In vivo labeling experiments demonstrate that UBFl and UBFB are phos- phoproteins. Phosphoamino acid analysis of the in vivo labeled proteins demonstrate that UBF is phosphoryl- ated on serine residues. Following serum deprivation there is no alteration in the cellular levels of UBFl and UBF2 as determined by Western blotting, but there is an 80% reduction in the level of phosphorylation of UBF compared with logarithmically growing cells. Following serum deprivation there is a redistribution of UBF between the nucleolus, the nucleus, and the cytoplasm. Phosphatase-treated UBF demonstrated a reduced ability to rescue transcription by RNA polym- erase I from the rDNA spacer promoter in vitro. These findings suggest that phosphorylation of UBF is a pre- requisite for transactivation of RNA polymerase I. were purified through the CM-Sephadex chromatography step (1). In uitro transcription of the rat rDNA spacer promoter (6) was carried out as previously described (1, 6) including the addition of a radiolabeled internal standard to monitor the recovery of nucleic acids (1, 6). Treatment of the purified UBFlIUBFB with bacterial alkaline phosphatase in buffer C20 (1) was performed essentially as described (12).

We have analyzed the expression, phosphorylation, and localization of the ribosomal DNA transcription factors UBFl and UBF2 in Chinese hamster ovary cells in response to serum deprivation. In vivo labeling experiments demonstrate that UBFl and UBFB are phosphoproteins. Phosphoamino acid analysis of the in vivo labeled proteins demonstrate that UBF is phosphorylated on serine residues. Following serum deprivation there is no alteration in the cellular levels of UBFl and UBF2 as determined by Western blotting, but there is an 80% reduction in the level of phosphorylation of UBF compared with logarithmically growing cells. Following serum deprivation there is a redistribution of UBF between the nucleolus, the nucleus, and the cytoplasm. Phosphatase-treated UBF demonstrated a reduced ability to rescue transcription by RNA polymerase I from the rDNA spacer promoter in vitro. These findings suggest that phosphorylation of UBF is a prerequisite for transactivation of RNA polymerase I.
The promoters of vertebrate ribosomal RNA genes consist of elements with similar functions: a core promoter element (CPE),' an upstream promoter element (UPE), and a promoter-proximal terminator site (To) (Ref. 1,and references therein). Transcription of ribosomal RNA genes requires at least two defined factors referred to as SL-I and UBF (1-3). Although the CPE is sufficient for transcription i n uitro, the UPE is required for transcription in uiuo (4). I n uitro experiments indicate that the transcription factor UBF footprints over the UPE of the 45 S and spacer promoters (1-6). Whereas it has been established that SL-I interacts with both the CPE and the 5' boundary of the UPE (1,3) the factor has not been characterized in detail. UBF has been isolated from a number of species, including human, rat, mouse, and Xenopus laeuis. In the cases of human, rat, and mouse the purified factors * This work was supported by the Geisinger Foundation. 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. Section 1734 solely to indicate this fact.
Previous studies on the regulation of ribosomal RNA synthesis in response to serum deprivation in eukaryotes have led to the model that transcription by RNA polymerase I is regulated by the post-translational modification of either RNA polymerase 1 (12) and/or a factor which is tightly associated with the polymerase, such as T F l C (13). In this respect we have analyzed the expression, localization, and phosphorylation of UBF in cultured Chinese hamster ovary (CHO) cells in response to serum deprivation. Cells deprived of serum, and the growth factors contained therein, cease to progress through the cell cycle and shut off ribosomal RNA synthesis (Ref. 12, and references therein). Our evidence demonstrates that both UBFl and UBFB are phosphoproteins and that the degree of phosphorylation of UBF is significantly reduced when CHO cells are serum-deprived. Immunolocalization studies show loss of UBF from the nucleolus following serum deprivation. In uitro transcription experiments demonstrated that treatment of UBF with alkaline phosphatase reduces the transcriptional activation properties of UBF.

EXPERIMENTAL PROCEDURES
Materials-All DNA and RNA modifying enzymes and isopropyl-@-D-thio-galactopyranoside (IPTG) were purchased from Promega; Taq DNA polymerase was from Perkin-Elmer/Cetus. "'I-Labeled goat anti-rabbit IgG, F(ab'), fragment, [3'P]orthophosphate, [a-32P] UTP, and [y-3'P]ATP were purchased from Du Pont-New England Nuclear. Nitrocellulose and Immobilon-P were purchased from MSI Inc. and Millipore, respectively. RIBI adjuvant was from Immuno-Chem Research, Inc. RPMI-1640 media was from Mediatech; protein G-agarose, newborn calf serum, and dialyzed newborn calf serum were from GIBCO/BRL. Prestained protein size standards were purchased from Bio-Rad. Bacterial alkaline phosphatase was obtained from Pharmacia LKB Biotechnology Inc. Rhodamine-conjugated, affinity-purified, goat anti-rabbit IgG was obtained from Organon Teknika Corp. Antiserum to fibrillarin was purchased from Sigma. TLC cellulose plates were obtained from Kodak.
Expression of Recombinant UBF2 in Escherichia coli and Antibody Production-A cDNA fragment coding for amino acids 53-390 of rUBF2 was amplified by PCR from the plasmid template p405rUBF (8) and cloned into the vector pET3a (14), generating pETrUBF53-390. IPTG-induced expression of rUBF53-390 in the host E. coli BL21(DE3) was performed essentially as described (14,15). The recombinant protein rUBF53-390 was purified from induced cultures by preparative SDS-PAGE (16). Rabbits were immunized with 100 pg of rUBF53-390 protein in RIBI adjuvant following an immunization scheme recommended by the manufacturer. The antiserum crossreacted with both recombinant rUBFl and rUBF2 synthesized in E. coli (data not shown).
Tissue Culture of CHO Cells and in Vivo Labeling with P'P] Orthophosphate-CHO cells were cultured in RPMI-1640 plus 10% newborn calf serum (NBS). For serum deprivation studies, cells were initially plated a t a density of 3.5 X lo5 cells/60-mm dish. After 24 h in complete media, the cells were washed with RPMI-1640 and serumstarved for either 24,48, or 68 h by incubation in RPMI, 0.05% NBS. Cell viability, determined by the trypan blue dye exclusion assay, was greater than 90% after 24 or 48 h of serum starvation.
For in vivo phosphorylation studies, logarithmically growing and serum-starved cells were plated a t a density of 3.5 X lo5 and 5.5 X lo5 cells, respectively, per 60-mm dish in RPMI, 10% NBS (3 ml) and grown for 24 h. Cells were then washed in phosphate-free RPMI media. Log growing cells were labeled for 16 h in phosphate-free RPMI, 20% dialyzed NBS containing 1 mCi/ml ["P]orthopho~phate. Where specified, separate cultures were either serum-starved (RPMI, UBF Phosphorylation and Ribosomal RNA Synthesis 0.05% NBS) for either 24, 48, or 72 h or were serum-starved (RPMI, 0.05% NBS) for 24 or 48 h and refed (RPMI, 10% NBS) for 8 h prior to labeling. Immunoprecipitation and Phosphoamino Acid Analysis of "P-Labeled UBFI and UBF2 from CHO Cells-Cell lysates were prepared as described (17). Anti-UBF serum or preimmune serum was added to the cell lysates at a final dilution of 1:50. Immune complexes, precipitated with protein G-agarose as described by the supplier, were fractionated by SDS-PAGE, blotted onto Immobilon-P membrane, and subjected to autoradiography (17). For quantitation, the immunoprecipitated bands were cut from the membranes and subjected to liquid scintillation spectrophotometry. Phosphoamino acid analysis was carried out as described (18).
Western Blotting and Immunofluorescence Studies-Whole cell lysates (100 fig of protein) from CHO cells cultured as described were fractionated by SDS-PAGE, blotted onto nitrocellulose, and probed with the anti-UBF serum (1:500 dilution) (17). For immunofluorescence studies, CHO cells were grown on coverslips and were fixed and stained with either anti-UBF serum (1:80 dilution) or antiserum to fibrillarin essentially as described (19). The second antisera were used as recommended by the suppliers.
Analysis of the mRNAs Coding for UBFl and UBF2 in CHO-First strand cDNA was synthesized using total RNA (20), random hexaoligonucleotide primers, and reverse transcriptase followed by amplification of the UBF coding sequences by PCR and analysis by Southern blotting using a mixture of 5' end-labeled, internal oligonucleotide probes as described (8).
In Vitro Transcription Studies-UBF1 and UBFZ were purified through the CM-Sephadex chromatography step (1). In uitro transcription of the rat rDNA spacer promoter (6) was carried out as previously described (1, 6) including the addition of a radiolabeled internal standard to monitor the recovery of nucleic acids (1, 6). Treatment of the purified UBFlIUBFB with bacterial alkaline phosphatase in buffer C20 (1) was performed essentially as described (12).

Analysis of the Expression of UBFl and UBF2 in Different
Species and in Response to Serum Deprivation-Western blot analysis of cell lysates prepared from Novikoff (rat), HeLa (human), COS (monkey), CHO (hamster), and BFE (bovine) cells showed that in each case the anti-UBF serum recognized two bands at 97 and 94 kDa (Fig. LA), confirming that both UBFl and UBFZ are highly conserved in the different eukaryotic species tested. Western blot analysis (Fig. 1B) of cell lysates from CHO cells which were either grown in the presence of serum, serum-starved for 24, 48, or 68 h, or were serum-starved and refed for the specified time periods showed that, with the exception of cells which were serum-starved for 68 h (Fig. lB, lane 7), there was no significant change in the levels of UBFl and UBF2 in the CHO cells grown under these different conditions. Analysis of the UBF coding sequences in RNA isolated from logarithmically growing CHO cells, serum-starved CHO cells, and starved and refed CHO cells showed that there was no substantial change in the steadystate levels of the UBF coding mRNAs (Fig. IC). Taken together, these results indicate that the level of expression of UBFl and UBFS does not change in response to serum deprivation.
Amlysis of the Immunolocalization of UBF in Response to Serum Deprivation-Immunolocalization studies using the anti-UBF antibody confirmed that UBF in CHO cells is a nucleolar protein (Fig. 2, panels A and B) as indicated by the specific staining of the nucleolus in logarithmically growing cells. However, following serum starvation for 24 h, the distribution of UBF was altered. The nucleolar fluorescence decreased, and a diffuse nuclear and cytoplasmic staining was evident (Fig. 2, panel C). When cells serum-starved for 43 h were refed for 5 h, UBF was again located in the nucleolus (Fig. 2, panel D). In post-confluent cells UBF was again distributed between the nucleolus, the nucleus, and the cytoplasm (Fig. 2, panel E). In contrast, the nucleolar localization of fibrillarin was not altered under these growth conditions although the level of nucleolar fluorescence decreased upon serum starvation and when the cells were confluent (Fig. 2, panels A-E, respectively). The Phosphorylation of UBF Is Altered in Response to Serum Deprivation-In order to test whether UBFl and UBFB are phosphorylated, CHO cells were labeled with ["PI orthophosphate, and UBFl and UBFS were immunoprecipated, fractionated by SDS-PAGE, and blotted onto Immobilon. A typical precipitate formed in the presence of the anti-UBF antiserum but not preimmune serum contained two radioactive bands at 94 and 97 kDa (Fig. 3A, lane 2).
When cells were serum-starved for 24 or 48 h there was an 80% reduction in the degree of UBFlIUBF2 phosphorylation (Fig. 3A, lanes 3 and 4 ) in comparison with cells grown in the presence of serum (Fig. 3A, lane 2). When cells were serumstarved for 24 or 48 h and refed with complete medium for 8 h and labeled with ["P]orthophosphate for 16 h, the level of phosphorylation of UBFl and UBF2 recovered to that observed in logarithmically growing cells (Fig. 3A, lanes 6 and  7). In contrast, cells which were serum-starved for 72 h contained no detectable levels of phosphorylated UBF (Fig.  3A, lane 5 ) even upon refeeding (Fig. 3A, lane 8) Analysis of the phosphorylation of UBF in response to altered growth conditions. A, UBFl/UBF2 were immunoprecipated from ": P metabolically labeled CHO cells using the anti-UBF antiserum, fractionated by SDS-PAGE, blotted onto nitrocellulose, and subjected to autoradiography. Lanes 1 and 2, immunoprecipitates from logarithmically growing cells using the preimmune and anti-UBF antiserum, respectively. Lunes 3-5 are immunoprecipates from cells serum-starved for 24, 48, and 72 h, respectively. Lanes 6-8 are immunoprecipates from 24-, 48-, and 72-h serum-starved cells which were refed for 8 h in complete medium and labeled for 16 h, respectively. Equivalent numbers of cells were used for each analysis. The position and size (kDa) of the molecular weight size standards are indicated. B, phosphoamino acid analysis of '"P metabolically labeled, immunoprecipated UBFl/UBF2 was performed as described under "Experimental Procedures." Lunes 1 and 4 are from 24-h serumstarved CHO cells; lanes 2 and 5 are from 24-h serum-starved/%h refed CHO cells; and lanes 3 and 6 are from logarithmically growing CHO cells. Approximately equal counts/min of metabolically labeled samples of UBF were subject to phosphoamino acid analysis. The position of the standards phosphoserine (Ser), phosphothreonine (Thr), and phosphotyrosine ( T y r ) are indicated. * denotes radiolabeled inorganic phosphate. The arrowhead indicates the direction of chromatography.
indicated that there was not a wholesale alteration in either the gross amount or the pattern of phosphorylation (data not shown). These experiments demonstrated that ( a ) both UBFl and UBFP are phosphoproteins; (6) the degree of phosphorylation of UBFl and UBFZ is reduced following serum deprivation; and (c) following serum starvation and refeeding the 1 2 3 4 5

Trans.
Int. Std. level of phosphorylation of UBFl and UBFZ increases dramatically.
In Vitro Transcription Studies Using Bacterial Alkaline Phosphatase-treated UBFI/UBFZ-To establish whether the phosphorylation of UBF plays a direct role in the ability of UBF to activate transcription, the spacer promoter (which is transcriptionally inactive in the absence of UBF (6, repeated in Fig. 4, lune I ) was used as a template in in uitro transcription assays. Transcription from this promoter is dependent upon the presence of UBF in this assay (6,Zl). UBF rescued transcription from this promoter (Fig. 4, lanes 2 and 3 ) . Bacterial alkaline phosphatase-treated UBF (Fig. 4, lane 5) demonstrated a reduced ability (67% reduced) to activate transcription from the spacer promoter. Addition of equivalent amounts of bacterial alkaline phosphatase directly to the transcription reaction did not significantly affect UBF-dependant transcription (Fig. 4, lune 4 ) . DISCUSSION The results presented here demonstrate that: 1) the RNA polymerase I transcription factor UBF is a phosphoprotein, 2) the modified amino acid is serine, 3) the phosphorylation of UBF is reduced approximately 80% in serum-starved CHO cells relative to the level of phosphorylation of UBF in logarithmically growing cells, 4) the subcellular distribution of UBF shifts from being predominantly nucleolar to a dispersed nucleolar/nuclear/cytoplasmic distribution following serum starvation, and 5 ) treatment of purified UBF with bacterial alkaline phosphatase reduced the transactivation properties of UBF in an in vitro RNA polymerase I transcription system.
The phosphorylation state of certain transcription factors is important in modulating their transcriptional properties (Refs. 22 and 23, and references therein). The results presented here strongly suggest that the phosphorylation state of UBF may affect its transactivation properties. The reduced transactivation properties of phosphatase-treated UBF in our in vitro transcription system suggest that the phosphorylation of UBF may be a prerequisite for transcriptional activation. The finding that authentic UBF can be phosphorylated i n vitro by casein kinase I1 (data not shown) and the observation that the level or extent of phosphorylation of UBF actually increases when serum-starved cells are refed suggests that the degree of phosphorylation of UBF is regulated.
There is also the possibility that the phosphorylation of UBF is necessary for its nucleolar localization. The events that control protein translocation in response to various conditions, including serum starvation, have recently been addressed (Ref. 24, and references therein). It appears that the nuclear localization of several proteins, e.g. c-Fos (24), NF-KB (25), and SV-40 T-antigen (26), is regulated by phosphorylation. We have seen that concomitant with a reduction in the level of UBF phosphorylation, UBF redistributes between the nucleolus, the nucleus, and the cytoplasm. Such an event would effectively reduce nucleolar, UBF-dependant RNA polymerase I transcription in serum-starved cells. Relevant to this hypothesis is the observation that serum starvation is associated with the translocation of another nucleolar phosphoprotein, B23 or N038, from the nucleolus to the nucleoplasm (27).
Several laboratories have reported that when cell growth is shut off in response to stimuli such as starvation (12, 28), glucocorticoids (13), and serum deprivation (29) that the ability of RNA polymerase I extracted from those cells to correctly initate transcription in vitro is reduced. This has been ascribed to a modification of RNA polymerase I itself (12,28,29) or to the modification of a factor such as TFlC (13) that is tightly associated with the polymerase. Our results suggest that cells respond to serum starvation by altering the phosphorylation state of UBF as well. Although the above mentioned studies on the regulation of eukaryotic rDNA transcription reported no changes in the activities of the other transcription components in the whole cell extracts of the down-regulated cells one must remain cognizant of the facts that 1) standard in vitro transcription reactions of the mouse and rat 45 S rDNA promoter do not appear to require UBF (1, 3) and 2) those studies used whole cell extracts which would contain a small amount of phosphorylated UBF. Our results indicate that the amount of UBF present in downregulated cells is unchanged and suggest that the phosphorylation of UBF affects its role in transcription and in its subcellular localization, thus regulating its activity in transcription in two ways. Therefore, our observations may complement earlier results, as it is formally possible that the same enzyme or signal transduction cascade responsible for regulating the modification of RNA polymerase I, TIF-lA, TIF-lB, or TIF-1C may also be involved in regulating the phosphorylation of UBF. We are currently attempting to identify the kinase(s) and/or phosphatase(s) which modify UBF in vivo in response to serum and to ascertain the possible roles of such systems in coordinately regulating the activity of the individual components required for efficient rDNA transcription. UBF, which has potential phosphorylation sites for casein kinase I1 in its carboxyl-terminal acidic tail, is phosphorylated by casein kinase I1 i n vitro (data not shown). The identification and systematic mutagenesis of the phosphorylated amino acid(s) in UBFl and UBF2 should allow us to define the mechanism(s) by which the phosphorylation of UBF modulates its role in transactivation of RNA polymerase I.