PU.1 expression is modulated by the balance of functional sense and antisense RNAs regulated by a shared cis-regulatory element

Abstract

The transcription factor PU.1 is an important regulator of hematopoiesis; precise expression levels are critical for normal hematopoietic development and suppression of leukemia. We show here that noncoding antisense RNAs are important modulators of proper dosages of PU.1. Antisense and sense RNAs are regulated by shared evolutionarily conserved cis-regulatory elements, and we can show that antisense RNAs inhibit PU.1 expression by modulating mRNA translation. We propose that such antisense RNAs will likely be important in the regulation of many genes and may be the reason for the large number of overlapping complementary transcripts with so far unknown function.

Figure 1.

Both strands of the PU.1 gene are transcribed, and homology region H3 confers promoter activity in the antisense orientation. (A, top panel) Linear diagram showing the positions of the PU.1 gene locus homology regions URE (H1 + H2), proximal promoter (PrPr), and H3, as well as exons 1–5 (E1–E5). (Bottom panel) Diagram showing the position of the sense- and antisense-specific primers in exon 3 (block arrows), and the ATSS (star), corresponding to antisense RNA (AS; long black arrow). The long white arrow indicates the orientation of the sense RNA (S; PU.1 mRNA). Two short black arrows indicate orientation of the homology region H3 fragment–luciferase constructs (H3-S and H3-AS, sense and antisense, respectively). (B) Quantification of the PU.1 gene antisense transcripts by real-time strand-specific RT–PCR in HL-60 and RAW 264.7 cells. Strand-specific RT was performed for sense RNA (PU.1 mRNA) and for antisense RNA (AS) using sense- and antisense-specific primers (as shown in A), followed by PCR with the respective primer pair (see the Supplemental Material for oligonucleotide sequences and the Materials and Methods and legend to Supplemental Fig. S1 for a detailed protocol for strand-specific RT–PCR). Expression values are shown as the percent of sense transcript expression (100%). Mean values and standard deviation (error bars) based on three real-time PCR runs are displayed. (C) Antisense RNAs (AS) are more stable than sense RNAs (S). Shown is a time course of sense and antisense RNA levels isolated from bone marrow mononuclear cells derived from wild-type mice. Cells were treated with Actinomycin D (Alexis Biochemicals) at 10 μg/mL. An equal amount of viable cells (0.5 million), counted after Ficoll gradient purification, were collected at 0, 1, 6.5, and 8 h of treatment. Total RNA was isolated as described in the Materials and Methods and subjected to sense- and antisense-specific quantitative RT–PCR. The primers used for quantitative RT–PCR are described in the Supplemental Material, “Primers Used for Real-Time RT–PCR,” and shown in Supplemental Figure S1A. Correlation coefficients (R²) and equations of the exponential regression curves are indicated. (D) Homology region H3 confers promoter activity in the antisense orientation. The 215-bp fragment of homology region H3 was inserted into the pXP2 luciferase reporter vector in both sense (H3-S) and antisense (H3-AS) orientation. The PU.1 proximal promoter (PrPr) construct was described previously (Li et al. 2001). The pXP2 luciferase vector alone, as well as vectors including the PU.1 proximal promoter or H3 constructs were transiently transfected into HL-60 (left panel) and Jurkat (right panel) cells. Shown is the fold change in luciferase activity compared with pXP2 alone. Data are presented as the mean ± standard deviation based on three to six data points for each construct.

Figure 2.

The URE is instrumental in PU.1 antisense RNA synthesis. (A) The linear diagram shows the position of the PU.1 gene locus homology regions: the URE (H1 + H2), removed by homologous recombination (Rosenbauer et al. 2004), proximal promoter (PrPr), and H3; initiation sites and orientation of the sense (long white arrow) and antisense (long black arrow) transcripts; and the position of the antisense-specific probe (pentagon). (B) Northern blot analysis illustrates down-regulation of the antisense RNAs in the URE^Δ/Δ mouse model (Rosenbauer et al. 2004). RNAs were isolated from bone marrow of wild-type (+/+) and hypomorphic (Δ/Δ) animals. Two micrograms of polyA(+) were run side by side and hybridized with the probe specific to antisense RNAs (pentagon, A). The top panel demonstrates down-regulation of antisense RNAs. The middle panel demonstrates down-regulation of the PU.1 sense mRNA in the URE^Δ/Δ mouse model, using a double-stranded murine PU.1 cDNA probe. The bottom panel depicts GAPDH mRNA levels as loading controls (see the Supplemental Material for the probe information). (C,D) The URE regulates expression of both sense and antisense RNAs. Relative sense and antisense RNA levels in bone marrow of wild type and animals with an URE deletion were quantitated by Northern blot phosphorescence imaging (normalized to GAPDH mRNA) (C) and, as corroboration, by strand-specific RT–PCR (graph values calculated as mean ± standard deviation of three real-time RT–PCR runs, normalized to GAPDH mRNA, measured by TaqMan real-time analysis) (D). See the Supplemental Material for the TaqMan probes and primers used for strand-specific RT–PCR.

Figure 3.

Quantitative 3C assays demonstrate the close physical proximity of the URE to the proximal promoter and H3 antisense promoter. (A) Diagram showing the genomic position of the homology regions (H1–H3 and proximal promoter [PrPr]), position and orientation of the primers used in the final 3C PCRs (common primer from the URE region: XB1; variable downstream primers: short horizontal arrows marked as 1–8), and locations of XbaI restriction enzyme sites (vertical arrows); block arrows marked as 1–8 represent calculated interaction frequencies. (B,C) Southern blot analysis of the linear amplification of the control and test libraries. Interaction frequencies were calculated by dividing the test 3C library (in the murine macrophage RAW 264.7 line) signals by control library (BAC) signals. The graphs represent results of two independent PCR/Southern blot experiments. (D) Schematic model of spatial organization of the regulatory elements of the PU.1 gene. The model depicts (1) the close physical proximity of the PU.1 gene locus homology regions (the URE [H1 and H2], proximal promoter [PrPr], and H3) and (2) the initiation and orientation of sense and antisense transcripts (long arrows). Short block arrows indicate genomic orientation of the PU.1 gene locus.

Figure 4.

RNAi targeting antisense transcripts leads to an increase in PU.1 protein and PU.1 mRNA levels associated with decreased levels of polysomal antisense transcripts. (A) Diagrams showing position of sites in the proximal promoter and homology region H3 used to design complementary siRNA (black and white rhomboids, respectively). Long arrows indicate orientation of the sense (PU.1 mRNA) and antisense (AS) transcripts. (B,C) Knockdown of antisense transcripts using siRNAs complementary to RNAs covering the proximal promoter and homology region H3 in HL-60 cells leads to ∼105% and ∼210% increases in PU.1 protein, respectively. Cytoplasmic and nuclear extracts (marked C and N, respectively) (15 μg each) isolated from cells transfected with antisense-specific and unrelated siRNAs were used for Western blot analysis. (B, top panel) Result of probing the membrane with PU.1 antibody. (Bottom panels) Controls used were β-tubulin and β-actin. (C) Relative PU.1 protein levels in nuclear fractions of cells transfected with antisense-specific siRNAs and unrelated siRNAs. The vertical bars indicate the relative amount of PU.1 protein in nuclear fractions normalized to β-actin and shown as the percent of PU.1 protein levels in cells transfected with unrelated siRNAs (100%) (NIH ImageJ version 1.37k). (D,E) Knockdown of antisense transcripts using siRNAs complementary to the proximal promoter or homology region H3 in HL-60 cells leads to 125% ± 11% (P < 0.001) and 145% ± 12% (P < 0.001) increase in PU.1 mRNA, respectively. (D) Northern blot analysis of the PU.1 mRNA levels in total RNA samples using a double-stranded human PU.1 cDNA probe. GAPDH mRNA levels and ethidium bromide staining were used as loading controls (see the Supplemental Material for probe information). (E) Histograms showing relative PU.1 gene mRNA levels in cells either untreated or transfected with antisense-specific siRNAs or unrelated siRNAs (Invitrogen, Cat. No. 12935-400) as a control, normalized to 18S. Expression values are shown as the percent of PU.1 mRNA levels in untreated cells (100%). Mean values and standard deviations (error bars) based on three real-time PCR runs are displayed. (F) Knockdown of antisense transcripts using siRNAs complementary to the proximal promoter (PrPr) and homology region H3 in HL-60 cells leads to ∼30% and ∼55% decreases in PU.1 antisense RNAs, respectively (left panel) in the cytoplasmic subfractions, and to ∼175% and ∼160% increases in PU.1 sense mRNA, respectively (right panel). Expression values of PU.1 mRNA and antisense RNAs in the soluble cytoplasmic subfractions isolated from the cells transfected with antisense-specific siRNAs are shown as the percent of those transfected with unrelated siRNAs (100%). Mean values and standard deviation (error bars) based on four experiments are displayed. (G) Knockdown of antisense transcripts using siRNAs complementary to the proximal promoter and homology region H3 in HL-60 cells leads to ∼58% and ∼78% decreases in PU.1 antisense RNAs, respectively (left panel), and to no changes in PU.1 gene mRNA levels (right panel) in the polysomal fractions. Expression values of PU.1 mRNA and antisense RNAs in the polysomal fractions isolated from the cells transfected with antisense-specific siRNAs are shown as the percent of those transfected with unrelated siRNAs (100%). Mean values and standard deviations (error bars) based on four experiments are displayed. (H) Immunoprecipitation of sense/antisense PU.1 RNAs-translation factor complexes from total polysomal fraction RNA–translation factor complexes (pRIP) using antibodies against eIF4A (translation initiation factor), eEF1A (translation elongation factor), and 4E-BP1 (inhibitor of translation initiation). HL-60 cells were HCHO-cross-linked, and total polysomal fraction was isolated as shown in Supplemental Figure 9A. pRIP was performed as described in the Supplemental Material. Histograms show relative PU.1 gene mRNA and antisense RNA levels in the immunoprecipitates, normalized to GAPDH (left and right panels, respectively). Mean values and standard deviations (error bars) based on three real-time PCR runs are displayed. (I) Antisense PU.1 gene RNAs function by stalling the translation between the initiation and elongation steps. HL-60 cells were treated with antisense-specific and unrelated siRNAs, cross-linked (1% HCHO, 10 min), and the polysomal fractions was analyzed by pRIP with antibodies against eIF4A and eEF1A followed by real-time RT–PCR.