Transcription factor UBF depletion in mouse cells results in downregulation of both downstream and upstream elements of the rRNA transcription network

Abstract

Transcription/processing of the ribosomal RNA (rRNA) precursor, as part of ribosome biosynthesis, is intensively studied and characterized in eukaryotic cells. Here, we constructed shRNA-based mouse cell lines partially silenced for the Upstream Binding Factor UBF, the master regulator of rRNA transcription and organizer of open rDNA chromatin. Full Ubf silencing in vivo is not viable, and these new tools allow further characterization of rRNA transcription and its coordination with cellular signaling. shUBF cells display cell cycle G1 delay and reduced 47S rRNA precursor and 28S rRNA at baseline and serum-challenged conditions. Growth-related mTOR signaling is downregulated with the fractions of active phospho-S6 Kinase and pEIF4E translation initiation factor reduced, similar to phosphorylated cell cycle regulator retinoblastoma, pRB, positive regulator of UBF availability/rRNA transcription. Additionally, we find transcription-competent pUBF (Ser484) severely restricted and its interacting initiation factor RRN3 reduced and responsive to extracellular cues. Furthermore, fractional UBF occupancy on the rDNA unit is decreased in shUBF, and expression of major factors involved in different aspects of rRNA transcription is severely downregulated by UBF depletion. Finally, we observe reduced RNA Pol1 occupancy over rDNA promoter sequences and identified unexpected regulation of RNA Pol1 expression, relative to serum availability and under UBF silencing, suggesting that regulation of rRNA transcription may not be restricted to modulation of Pol1 promoter binding/elongation rate. Overall, this work reveals that UBF depletion has a critical downstream and upstream impact on the whole network orchestrating rRNA transcription in mammalian cells.

Keywords: C-MYC; RNA pol1; cellular growth; rRNA; transcriptional regulation.

Figure 1

Generation and validation of clones stably silenced for Ubf.A1, comparison by WB of UBF protein levels (UBF1 and UBF2) in WT, shUBF clone 16 and 72 (upper panel), and calnexin levels for normalization (lower panel) in samples from three independent experiments. A2 and A3, corresponding quantification of the three independent experiments shown in A1, depicting the average value of total UBF protein levels normalized to either calnexin (A2), or total protein (A3; as visualized by Ponceau staining of the immunoblot membrane shown in Fig. S1A), and expressed relative to the WT control value (set at 1) ± standard deviation (SD). The analysis with both normalization methods shows a marked reduction in UBF protein levels well below 50% of WT levels in both shUBF16 and 72. Statistical significance of differences was assessed by one-way ANOVA with Dunnett’s test. B1 and B2, quantification of relative Ubf mRNA levels by qRT-PCR, depicting respectively, a normalized average knockdown in shUBF16 (B1) or shUBF72 (B2), expressed as a fraction of the normalized average WT control values. Error bars correspond to the SD of three independent experiments. B2m+PumI housekeeping gene mRNA levels were used for sample normalization. Statistical significance of differences was assessed by Welch’s t test.

Figure 2

Immunofluorescence characterization of shUBF72. Representative images of UBF (red) and fibrillarin (green) immunodetection in the nucleolus, comparing WT (panels A1–A4) and shUBF72 (panels B1–B4). DNA is counterstained with Hoechst (blue) and an overlay of all fluorophores is displayed in the bottom row. Immunolabeling was carried out in parallel and image acquisition parameters were kept constant in all cases to allow direct comparison of immunofluorescence signals between cell lines. A pronounced reduction of the UBF signal is visible in shUBF72 (B1) while the overall architecture of the nucleolus is similar to WT. Scale bars 10 μm.

Figure 3

Quantification of rRNA in shUBF72 shows reduced expression.A, quantification of relative pre-rRNA precursor 47S rRNA expression by qRT-PCR comparing WT and shUBF72 cells grown in parallel under three different serum concentrations in culture medium (0.5%, 10%, and 20%), each from three independent experiments. Results show that both in WT and in shUBF72, production of 47S rRNA is serum dependent and its expression is reduced in the shUBF72 clone at all serum concentrations, compared with WT values. The statistical significance of differences was assessed by two-way ANOVA with Tukey’s correction. B, quantification of FUrd incorporation into newly transcribed RNA in the nucleolus, in the same experimental setup, mirrors the results in (A), displaying serum dependency of incorporation and decreased incorporation in shUBF72 at all serum concentrations, relative to WT. Statistical significance of differences was assessed by two-way ANOVA with Tukey’s correction. C1 and C2, Northern blot analysis of 47S rRNA levels (panel C1; 20 μg total RNA loaded per lane) and its quantification (panel C2). In C1, hybridization of a set of 4 independent experiments with a 47S rRNA-specific probe shows a visible reduction in signal (upper panel) with corresponding ethidium bromide staining of the gel, to display major rRNA species 28S and 18S (lower panel). Quantification confirms a significant reduction of 47S rRNA in shUBF72. D1 and D2, corresponding analysis for 28S rRNA in the same samples as in C1 (panel D1; 3 μg total RNA loaded per lane), reveals a significant reduction of 28S rRNA in shUBF72 (panel D2). The statistical significance of differences in C2 and D2 was assessed by Student’s t test.

Figure 4

Growth signaling and cell cycle factors are reduced in UBF silencing conditions.A1 and A2, representative WB analysis (panel A1) and quantification of three independent experiments (panel A2) for pS6 kinase, expressed as a fraction of total S6K (pS6K/S6K), comparing WT and shUBF72 and using mTOR inhibitor rapamycin (rap) and MAPK inhibitor U0126 (U0). In the WB, the detection of pS6K, total S6K, and calnexin (used as loading control for normalization of all samples) is displayed. The quantification shown in A2 depicts the average normalized values of pS6K/S6K protein and reveals a significant decrease in the fraction of pS6K protein compared with WT cells (at 0.5 and 10% serum). Upon rapamycin treatment, pS6K/S6K is reduced in both shUBF72 and WT cells, compared with 10% serum conditions. Upon U0126 treatment, pS6K/S6K is reduced in WT cells while slightly increased in shUBF72. Shown are the average values ± SD of three independent experiments. B1 and B2, equivalent experiment as above, analyzing the pEIF4E/EIF4E fraction, shows a significant decline in the ratio of shUBF72 at both 0.5 and 10% serum. C1 and C2, equivalent analysis for cell cycle regulator RB, expressed as a pRB/RB fraction. Although pairwise differences (reduction in shUBF72) are not always statistically significant, a reduction of pRB/RB in shUBF72 at 0.5% and 10% serum and also in the presence of U0126, relative to WT, is clear. The statistical significance of differences in A2-C2 was assessed by two-way ANOVA with Tukey’s correction.

Figure 5

Quantification of Ubf gene expression and protein levels, levels of the pUBF fraction and RRN3 in shUBF72, and their serum dependency.A, quantification in three independent experiments of relative expression of Ubf by qRT-PCR comparing WT and shUBF72 cells grown in parallel under three different serum concentrations in culture medium (0.5, 10, 20%), shows insignificant fluctuations within each cell type, albeit with great and statistically significant reduction in shUBF72, relative to WT, at all serum concentrations. B1, corresponding WB analysis from three independent experiments of protein levels of total UBF confirms this finding (upper panel). The same samples were re-run on a new blot and pUBF signal was detected with a phospho-S484-specific antibody (bottom panel). B2–B4, quantification of results from these 3 independent experiments with normalization of the total UBF signal (UBF1+2) to same-lane calnexin (B2), normalization of the total pUBF signal (pUBF1+2) to same-lane calnexin (B3), and calculation total pUBF signal as a fraction of total UBF (pUBF/UBF) (B4). C1 and C2, representative WB analysis (C1) and quantification (C2) of protein levels of RRN3 factor in three independent experiments, under the three different serum concentrations and normalization to same-lane calnexin, shows serum-dependency and a significant reduction in shUBF72, relative to WT samples. The statistical significance of differences was assessed by two-way ANOVA with Tukey’s correction in all panels.

Figure 6

Analysis and quantification of UBF binding to rDNA chromatin.A, WB analysis of fractions used in the UBF ChIP experiments shown in B2. Representative WB: “Input” corresponds to 20% of the total input samples derived from cross-linked cell lysates; “UBF ChIP” samples correspond to an equal amount of the amount used for ChIP in B2. The IgG sample corresponds to negative control and the (/) symbol denotes no sample loading in that lane. Strips were probed with anti-pUBF (upper panel) or anti-UBF antibodies (lower panel). B1, schematic representation of the functional 45kb mouse rDNA repeat unit (upper panel) to illustrate the constituent parts, and detail of the enhancer/promoter area (lower panel), to indicate the DNA sequence elements to which oligonucleotide primers were targeted for the qPCR in the UBF ChIP quantification shown in B2 (IGS3, intergenic spacer 3; SpPr, spacer promoter; Tsp, terminator spacer; To/Pr, To promoter proximal terminator/promoter; 28S, 28S rRNA coding sequences; T1-T10, terminator sites 1–10). B2, UBF ChIP quantification of 3 independent experiments to compare UBF occupancy in shUBF72 and WT at known UBF binding sites across the rDNA unit shows reduced occupancy in shUBF72 at all sites. For example, for the SpPr hotspot UBF binding is 3.09 ± 0.91 for shUBF72 vs. 6.48 ± 0.33 SD (or 47% of WT). The statistical significance of differences was assessed by multiple t tests using the Holm-Sidak correction with α = 0.05. C1 and C2, representative example of Southern blot analysis of “open`’ (or active/potentially active, transcriptionally competent) rDNA chromatin segment (upper band) and “closed” (or inactive/silent) rDNA chromatin segment, following psoralen incorporation and crosslinking experiments, comparing WT and shUBF samples (C1; Ctrl, control reference of non-crosslinked DNA). Quantification of four independent experiments shows a statistically significant decrease in the fraction of open chromatin, by 23.41% relative to WT, in shUBF72 (C2). The statistical significance of differences was assessed by Welch’s t test.

Figure 7

Protein levels of RNA Pol1 are reduced upon UBF silencing.A1 and A2, analysis of protein levels of RNA Pol1 subunit POLR1E by WB shows a profound decrease in both shUBF16 and shUBF72, relative to WT cells. WB of three independent experiments is displayed in panel A1 and the corresponding quantification in panel A2 with protein levels normalized to calnexin. See Fig. S3panels A1+A2 for alternative normalization to total protein, as visualized by Ponceau staining, and Fig. S3panels B1–B3 for WB and quantification of actin, as a negative control (no change) from the same experimental set. B1 and B2, WB (panel B1) and corresponding quantification (panel B2) of RNA Pol1 subunit POLR1A in 3 independent experiments. Relative POLR1A protein levels are normalized to calnexin, and the decrease is large in both shUBF cell lines. See Fig. S3panels C1+C2 for alternative POLR1A quantification to Ponceau staining from the same experimental set. The statistical significance of differences in A2 and B2 was assessed by one-way ANOVA with Dunnett’s method. C1–D2, immunofluorescence detection of POLR1A (green) confirms a clear reduction of signal in shUBF72, compared with WT cells (compare panels C1 and D1). Cells were counterstained for DNA (blue; panels C2 and D2), and image acquisition parameters were identical per fluorochrome, across samples. Scale bars 10 μm.

Figure 8

RNA pol1 is modulated at the transcriptional level by extracellular growth signals in different WT cells and its expression is markedly affected by UBF silencing.A, quantification in three independent experiments of relative expression of Polr1B subunit of RNA Pol1 by qRT-PCR comparing WT and shUBF72 cells grown in parallel under three different serum concentrations in culture medium (0.5%, 10%, and 20%), reveals a serum-dependent increase in WT and a similar tendency and a clear corresponding reduction of Polr1E mRNA levels in shUBF72 at all serum concentrations. Statistical significance of differences was assessed by two-way ANOVA with Tukey’s correction. B1 and B2, representative WB (panel B1) and quantification of POLR1E protein levels (panel B2) in three independent experiments of an equivalent setup confirms the serum-dependent modulation of RNA Pol1 in both WT and shUBF72. Sample normalization was to same-sample total protein as revealed by Ponceau whole-membrane staining (see Ponceau stained gel in Fig. S4A). Statistical significance of differences was assessed by two-way ANOVA with Tukey’s correction. C, RNA Pol1 ChIP quantification of 3 independent experiments to compare Pol1 occupancy in shUBF72 and WT at known binding sites across the rDNA unit (with oligonucleotide primers shown in Fig. 6C1) shows reduced occupancy in shUBF72 at all sites to a range of 40% of WT or lower. Statistical significance of differences was assessed by multiple t-tests using the Holm-Sidak correction with α = 0.05. D, representative WB of POLR1E levels in WT mouse IMCD cells, grown at different serum concentrations, and quantification of 5 independent experiments with sample normalization to its Ponceau staining (see Ponceau-stained gel in Fig. S4B). E, equivalent experiment using mouse Neuro2A cells, with corresponding Ponceau gel in Fig. S4C. F, equivalent experiment using human HeLa cells, with corresponding Ponceau gel in Fig. S4D. Statistical significance of differences in (D–F) was assessed by one-way ANOVA with Dunnett’s method. G, quantification of relative gene expression of the C-myc oncogene in WT and shUBF72 cells in 3 independent experiments, shows a significant decrease of C-myc mRNA levels upon UBF silencing. Statistical significance of differences was assessed by Welch’s t test.

Figure 9

Assessment byqRT-PCR of how UBF silencing affects gene expression of factors that are critical for rRNA transcription. A, quantification in three independent experiments of mRNA levels of Cyclin D in WT control and shUBF72 cells grown in parallel at three different serum concentrations in the culture medium (0.5%, 10%, and 20%). B–E, quantification in the same samples of mRNA levels of components of the PIC: Taf1a (panel B), Taf1b (panel C), Taf1c (panel D) and Tbp (panel E). F, quantification in the same samples of mRNA levels of PIC-interacting factor Rrn3. G and H, quantification in the same samples of mRNA levels of termination factor Ttf1 (panel G) and upstream element-binding and chromatin remodeler factor Ctcf (panel H). To be directly comparable, values were calculated using the common base method (71) and expressed as a ratio to the normalized mean WT value at 10%, with error bars corresponding to SD. The statistical significance of differences in all panels was assessed by two-way ANOVA with Tukey’s correction.