Poised transcription factories prime silent uPA gene prior to activation

Abstract

The position of genes in the interphase nucleus and their association with functional landmarks correlate with active and/or silent states of expression. Gene activation can induce chromatin looping from chromosome territories (CTs) and is thought to require de novo association with transcription factories. We identify two types of factory: "poised transcription factories," containing RNA polymerase II phosphorylated on Ser5, but not Ser2, residues, which differ from "active factories" associated with phosphorylation on both residues. Using the urokinase-type plasminogen activator (uPA) gene as a model system, we find that this inducible gene is predominantly associated with poised (S5p(+)S2p(-)) factories prior to activation and localized at the CT interior. Shortly after induction, the uPA locus is found associated with active (S5p(+)S2p(+)) factories and loops out from its CT. However, the levels of gene association with poised or active transcription factories, before and after activation, are independent of locus positioning relative to its CT. RNA-FISH analyses show that, after activation, the uPA gene is transcribed with the same frequency at each CT position. Unexpectedly, prior to activation, the uPA loci internal to the CT are seldom transcriptionally active, while the smaller number of uPA loci found outside their CT are transcribed as frequently as after induction. The association of inducible genes with poised transcription factories prior to activation is likely to contribute to the rapid and robust induction of gene expression in response to external stimuli, whereas gene positioning at the CT interior may be important to reinforce silencing mechanisms prior to induction.

Figure 1. Activation of the uPA gene by TPA treatment induces large-scale chromatin repositioning in the nucleus of HepG2 cells.

(A) Diagram illustrating the genomic context of the uPA locus and the genomic region detected by the BAC probe (RP11-417O11; ∼228 kb) used for FISH experiments. CAMK2G, calcium-calmodulin-dependent protein kinase II gamma; VCL, vinculin. Arrows indicate the 5′-3′ transcription direction. (B) Kinetics of the transcriptional induction of the uPA gene with TPA. uPA RNA expression was assessed by quantitative RT-PCR after treatment with TPA for different times. Values were normalized to 18S rRNA and expressed relative to uninduced cells. (C) Induction of uPA protein expression with TPA. Indirect immunofluorescence analyses of uPA protein expression (red) before and after TPA treatment (3 h). Nuclei were counterstained with DAPI (blue). The proportion of uPA-expressing cells is indicated (n = 188 and 144 for untreated and treated cells, respectively). Bar: 20 µm. (D) TPA induces large-scale repositioning of the uPA locus (green) to the exterior of its own chromosome 10 territory (CT10; red). The position of the uPA locus relative to CT10 was determined in HepG2 cells, before and after TPA activation for 3 h, by cryoFISH using a whole chromosome 10 paint (red) and the digoxigenin-labelled BAC probe (green) represented in (A). Nucleic acids were counterstained with TOTO-3 (blue). Arrowheads indicate uPA loci. The positions of uPA loci relative to the CT were scored as “inside,” “inner-edge,” “edge,” “outer-edge,” and “outside”; error bars represent standard deviations. Bars: 2 µm.

Figure 2. The regulatory regions of the uPA locus display an open chromatin conformation before and after TPA activation.

(A) Micrococcal nuclease (MN) digestion progressively cleaves cross-linked, sonicated bulk chromatin to mononucleosomes. Material from the 50 min digestion time-point was used for MN-ChIP experiments. (B) Diagram illustrating the regulatory region of the uPA gene, including the enhancer (E), promoter (P), and the position of amplified genomic fragments at and around the enhancer and promoter regions. Roman numerals indicate exons in the coding region (white and black boxes represent untranslated and translated regions, respectively). (C) Chromatin-associated features of the uPA enhancer and promoter regions as revealed by PCR amplification patterns of MN-digested chromatin DNA before and after transcriptional activation. HepG2 cells were grown ±TPA for 3 h, before chromatin preparation. Enhancer fragment 1 (E1), but not E2 and E3, is resistant to MNase cleavage after 50 min of digestion. Two promoter fragments larger than a single nucleosome (P and Px, 320 bp and 464 bp, respectively) are detected at the same digestion time point in the promoter region, consistent with protection due to additional bound proteins such as transcription factors and RNAP . (D) MN-ChIP experiments identify histone modifications associated with open chromatin at the enhancer and promoter regions before and after TPA activation. Chromatin was cross-linked, sonicated, and treated with MN for 50 min, prior to immunoprecipitation with antibodies against specific histone modifications associated with open (H3K4me2, H3K9ac, H3K14ac) or closed (H3K9me2) chromatin. Control immunoprecipitations were performed with polyclonal anti-uPA receptor (uPAR) antibodies (unrelated). To improve the resolution of MN-ChIP in the promoter region, two smaller, overlapping genomic fragments spanning the entire P fragment (uP and dP, 140 and 199 bp, respectively) were also amplified.

Figure 3. Inactive uPA loci associate with poised transcription factories rich in the initiating form of RNAP (S5p), prior to activation.

(A, D) MN-ChIP analyses detect the initiating (S5p) form of RNAP bound at the uPA promoter and enhancer before TPA activation (A). Following activation the enhancer is associated with the elongating form of RNAP (S2p), while the promoter is associated with both forms (S5p and S2p) of the enzyme (D). HepG2 cells were grown ±TPA for 3 h, before chromatin preparation and MN-ChIP with antibodies specific for S5p and S2p forms of RNAP. Control (unrelated) antibodies were polyclonal anti-uPA receptor antibodies. (B, E) The position of the uPA locus (red) relative to S5p and S2p sites (green) was determined by immuno-cryoFISH before (B) and after (E) TPA activation for 3 h, using a rhodamine-labelled BAC probe containing the uPA locus and antibodies specific for RNAP phosphorylated on S5 or S2 residues. The association of uPA genes with RNAP (S5p or S2p) was scored as “associated” (signals overlap by at least 1 pixel) or “separated” (signals do not overlap or are adjacent; see Figure S8 for additional examples). Arrowheads indicate the position of uPA loci. Nucleic acids were counterstained with TOTO-3 (blue). Bars: 2 µm. (C, F) Frequency of association of uPA loci with RNAP-S5p or -S2p sites before (C) and after (F) TPA activation. The decrease in association with S5p factories and the increase in association with S2p factories observed after activation were both statistically significant (χ² tests, p = 0.0006 and p<0.0001, respectively).

Figure 4. Factories containing S2p are also marked by the S5p modification and are less abundant than sites marked by S5p.

(A–C) RNAP-S5p sites are more abundant than RNAP-S2p sites. Cryosections (∼140 nm thick) from HepG2 cells grown ±TPA for 3 h were indirectly immunolabelled with antibodies specific for RNAP-S5p or -S2p (green), as indicated (A, B). Nuclei were counterstained with TOTO-3 (red). Representative images from TPA-treated cells are shown. RNAP-S5p and -S2p detection was optimised by using the highest concentrations of antibodies that give little detectable background in sections treated with alkaline phosphatase (see Figure S2E, S2F). Bars: 2 µm. Measurement of the number of S5p and S2p sites per unit area in the nucleoplasm (C) reveals a larger population of S5p than S2p sites, both before and after activation (Student t-test, p<0.0001 for both cases; number of nuclear sections analysed were for S5p, n = 45 and 41, and for S2p, n = 38 and 46, respectively, for − and +TPA). The decrease in S5p sites after TPA activation is statistically significant (p = 0.012), whereas no statistically significant difference was observed in the number of S2p sites (p = 0.44). (D–F) Most S2p sites also contain the S5p modification. Cryosections were first indirectly immunolabelled in the absence (D) or presence (E) of an antibody against RNAP-S5p (4H8; red) or in the presence of an unrelated anti-biotin antibody (F; red) that detects mitochondria in the cytoplasm (F; arrowheads). After formaldehyde cross-linking to preserve the first immunocomplex, sections were indirectly immunolabelled with an antibody against RNAP-S2p (H5; green). Nucleic acids were counterstained with TOTO-3 (insets), and confocal images collected using the same settings without signal saturation. Bars: 2 µm. Pre-incubation with 4H8 reduces the intensity of S2p signal throughout the nucleoplasm, except at interchromatin regions (E, arrows), in comparison to control samples incubated in the absence of 4H8 (D). Incubation with anti-biotin control antibody before labelling with H5 antibody (F) has no effect on S2p distribution throughout the nucleoplasm. (G) Measurements of average S2p intensity across the nucleoplasm show a ∼3-fold decrease in S2p detection after blocking with 4H8. Omission of 4H8 or pre-incubation with unrelated antibodies does not affect the level of the S2p signal in the nucleoplasm. Dotted line indicates the background intensity in the S2p channel measured from sections incubated with all antibodies except H5 (−H5; images unpublished). Number of nuclear profiles was >20 for each sample.

Figure 5. CAMK2G and VCL genes are transcriptionally active and associated with active transcription factories independently of TPA activation.

(A) Diagram depicting the uPA locus and the location of the fosmid probes used for the detection of CAMK2G (covering ∼46 kb of genomic sequence), uPA (∼44 kb), and VCL (∼42 kb) genes in cryoFISH experiments. Arrows indicate the 5′-3′ transcription direction. (B) Detection of primary transcripts for the CAMK2G, uPA, and VCL genes in HepG2 cells ±TPA, incubated in the presence or absence of the RNAPII inhibitor α-amanitin. Total RNA was extracted and the levels of primary transcripts determined by qRT-PCR using primers that amplify the exon1-intron1 junctions. Error bars represent the standard deviation of three independent replicates. (C, D) Frequency of association of CAMK2G, uPA, and VCL with RNAP-S5p (C) or -S2p (D) sites before and after TPA treatment for 3 h. The position of each fosmid signal relative to S5p and S2p sites was determined by immuno-cryoFISH, using rhodamine-labelled fosmid probes and antibodies specific for RNAP phosphorylated on S5 or S2 residues (images unpublished). The association of uPA and its flanking (CAMK2G and VCL) genes with RNAP (S5p or S2p) was scored as “associated” (signals overlap by at least 1 pixel) or “separated” (signals do not overlap or are adjacent as in Figure 3B, 3E). The association with S5p sites is similar for all genes across the locus before and after activation (C; n _uPA = 91 and 93; n _CAMK2G = 95 and 95; n _VCL = 90 and 92, for − and +TPA, respectively). CAMK2G and VCL are also associated with S2p sites independently of TPA treatment (D; n _CAMK2G = 104 and 121; n _VCL = 79 and 86, for − and +TPA, respectively), whereas the association of the uPA gene with S2p specifically increases upon TPA induction (D; n _uPA = 254 and 225). The increase in association with S2p factories observed after activation was statistically significant (χ² test, p<0.0001).

Figure 6. The induction of uPA gene expression is associated with an increase in the number of active alleles.

(A) TPA induction of uPA transcripts at the transcription site. uPA gene transcription was detected in HepG2 cells ±TPA by combining RNA and DNA-FISH in the same cryosection. Sections were first hybridized with mixture of five Cy3-labelled fiftymer oligonucleotide probes mapping at introns 4, 5, and 10, and exons 8 and 9 and the signal amplified with fluorescent antibodies. After cross-linking the immunocomplexes detecting uPA RNA, sections were hybridized with the fosmid uPA probe. uPA DNA and RNA signals were scored as “active” (signals overlap or adjacent to each other) or “inactive” (signals do not overlap) to determine the state of activity of the uPA alleles in ±TPA-treated cells. Arrowheads indicate the position of uPA-DNA (green) associated with or separated from uPA-RNA (red) signals. The use of exon probes results in the detection specific cytoplasmic RNA signals. Nucleic acids were counterstained with TOTO-3 (blue). Bars: 2 µm. TPA treatment increases the frequency of uPA allele association with uPA RNA signals from 13% to 27% (n = 200 and 216, respectively; p = 0.002). (B) The frequency of co-association of the uPA gene (blue) with RNAP-S2p (green) and uPA-RNA (red) was determined by triple labelling, using the fosmid uPA probe, antibodies specific for RNAP phosphorylated at S2 residues (H5), and Cy3-labelled oligonucleotide probes. The frequency of association of active uPA alleles with S2p sites was scored as “associated” (signals overlap by at least 1 pixel) or “separated” (signals do not overlap and adjacent signals). Arrowhead indicates the position of uPA-DNA (blue) and uPA-RNA (red) relative to S2p (green). Nucleic acids were counterstained with TOTO-3 (inset). Bar: 2 µm. Most uPA genes that are actively transcribed are also associated with RNAP-S2p, confirming that this RNAP modification marks active factories.

Figure 7. Increased association of uPA locus with active transcription factories upon TPA activation is independent of the large-scale repositioning of the uPA locus relative to its chromosome territory.

(A) The position of the uPA locus (uPA-DNA, red) relative to chromosome 10 (CT10, blue) and to S5p and S2p sites (green) was determined in HepG2 cells ±TPA activation for 3 h, by immuno-cryoFISH using a digoxigenin-labelled BAC probe containing the uPA locus, whole chromosome 10 paint, and antibodies specific for RNAP phosphorylated on S5 or S2 residues. Arrowheads indicate the position of uPA loci. Nuclei acids were counterstained with DAPI (images unpublished). Bars: 2 µm. (B) The frequencies of uPA loci which are associated or separated from RNAP-S5p (left hand column) or RNAP-S2p (right hand column) were measured at each position relative to the CT (“inside,” “inner-edge,” “edge,” “outer-edge,” and “outside”) before and after TPA induction. uPA locus association with S5p and S2p sites was detected in all position analyzed. (C) The proportion of uPA loci that associate with RNAP-S5p (left hand column) or RNAP-S2p (right hand column) before and after TPA activation was determined at each CT position. Association of the locus with S5p or S2p sites before and after activation are independent of its position relative to the CT. (D) Combined detection of the chromosome 10 (CT10, blue), the uPA gene (uPA-DNA, green), and uPA RNA (uPA-RNA, red) was performed by RNA- and DNA-FISH using a whole chromosome paint, the uPA fosmid probe, and Cy3-conjugated oligonucleotide probes. Nuclei acids were counterstained with TOTO-3 (insets). Arrowheads indicate the position of uPA-RNA and uPA-DNA signals. Bar: 2 µm. (E) The proportion of uPA genes that associate with uPA RNA before and after TPA activation was calculated at each CT position. Prior to activation, uPA genes at the CT interior are less frequently transcribed than loci outside the territory. After TPA treatment, the frequency of transcriptional events is the same at all CT positions.